61275 v2 @2011 The International Bank for Reconstruction and Development/The World Bank 1818 H Street, NW, Washington, D.C. 20433 USA Disclaimer This report has been discussed with the government of India but does not bear their approval for all its contents, especially where the Bank has stated its judgment/opinion/policy recommendations. The findings, interpretations and conclusions expressed in this paper are based on staff analysis and recommendations and do not necessarily reflect the views of the Executive Directors of The World Bank. Rights and Permissions The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/The World Bank encourages dissemination of its work and will normally grant permission promptly. All queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street, NW, Washington, DC20433, USA, fax: 202-522-2422, email: pubrights@worldbank.org. Photo Credits Source: World Bank, New Delhi Sunai Consultancy (P) Ltd., Patna, Bihar Designed and Printed by Macro Graphics Pvt. Ltd. www.macrographics.com Acknowledgements This report, in two volumes, was prepared at the Kennedy School of Government), and Mamta Murthi request of Government of India. Volume II contains the (ECSHD). full detailed report with all analyses and findings, while Volume I is a more condensed version highlighting The report draws on background papers prepared by main conclusions. The report team was led by Philip N.C. Saxena (Anti-Poverty Programs), Mahendra Dev, O'Keefe (EASSP, then SASSP), and the core team K. Subbarao, C. Ravi, and Prof. Galab (Safety net program included Puja Vasudeva Dutta, Mohammed Ihsan performance, drawing on the three state SP survey), Ajwad, Kalanidhi Subbarao, Robert Palacios, Rinku Puja Vasudeva Dutta (Poverty and vulnerability; Social Murgai, and Dina Umali-Deininger. Mansoora Rashid, pensions), Martin Ravallion and Rinku Murgai (ex-ante Sector Manager Social Protection for South Asia, simulations for MGNREG), Soumya Kapoor and Vidhya guided its preparation, and Julian Schweitzer was Soundararajan (MGNREG program performance), the then Human development Sector Director. The Sumita Chopra (Smart cards) Mohammed Ihsan Ajwad report was thoroughly updated by Philip O'Keefe, Puja (national SP program performance, drawing on the Vasudeva Dutta, Robert Palacios and John Blomquist IHDS data), Rinku Murgai and Jyotsna Jalan (2002 BPL in 2010 to reflect recent developments in Social System; Social pensions); Dina Umali-Deininger and Protection (SP) policy and programs, more recent data for some programs, and government feedback. Vidhya Klaus Deininger (PDS), Robert Palacios and Sangeeta Soundararajan provided extensive research support to Goyal (Social security for unorganized sector); Navolina the team during revisions to the report. Renu Gupta, Patnaik (Welfare funds); Philip O'Keefe (Workfare and Savita Dhingra and Tanusree Talukdar provided team social security); Jayashree Balachander, Philip O'Keefe support throughout preparation. Peer Reviewers are and Puja Vasudeva Dutta (SP review for Jharkhand); Shubham Chauduri (EASPR); Lant Pritchett (Harvard R. Badiani, S. Dercon and P. Krishnan (chronic and transient Acknowledgements poverty, capture in development programs, drawing development, GoI), Dr. Pani (then Secretary Rural on ICRISAT data) and S. I Rajan (Social pensions). It has Development, GoI), Arjun Sengupta, K.P. Kannan, also benefited greatly from two surveys: (i) a national and Ravi Srivastava (NCEUS), Amita Sharma (Joint survey conducted by NCAER in 2005 in collaboration Secretary, MoRD, GoI), Neelam Sawhney (Joint with the University of Maryland. This was a survey of Secretary MoRD, GoI), Abhijit Sen (Member, Planning many human development issues and included a Bank- Commission), Santosh Mehrotra (then Advisor, financed set of safety net questions. The team is most Planning Commission), Anil Swarup (DG, MoLE, grateful to Amaresh Dubey (then NCAER) and Professor GoI), Mr. Prashant (then Director, DEA, GoI), K. Raju Sonalde Desai (University of Maryland) for their efforts (then Principal Secretary Rural Development, GoAP), in producing the data and insights into the survey Pronab Sen (Chief Statistician and Secretary, Ministry process, and to Suman Bery (NCAER) for support; and of Statistics and Programme Implementation, GoI), (ii) a three-state SP survey commissioned for this report. A. Mishra (then Special Secretary Planning, GoUP), This was carried out by AC Nielsen ORG-MARG under R.V. Singh (Secretary Planning, GoO), A. Singh (Secretary the leadership of Sumit Kumar, and with inputs and Social Welfare, GoR), Dr. Sharma (Secretary Food, GoO), oversight by a team from Centre for Economic and Social K. Saha (then Secretary Food, GoB), A. Mukerkji (then Studies, Hyderabad, led by Professor Mahendra Dev, Principal Secretary, Rural Development, GoB), and and including Professors C. Ravi and Galab. Kalanidhi many other officials at state and district levels. Others Subbarao was also a key participant in design and who provided their time and insights included Alakh analysis. The study was financed from a DFID Trust Fund, Sharma (IHD Delhi) and Geeta Unikrishnan (DFID, support from which is gratefully acknowledged. New Delhi), Marc Socquet (ILO, New Delhi), Robert Jenkins, Ramya Subramanian, Annemieke Burkmeister The scope of the report was defined following a launch and Rajib Ghosal (UNICEF, New Delhi and Lucknow), workshop in New Delhi in late 2004, which was jointly and World Bank colleagues Shonali Sen, Gaurav organized with Government of India, World Bank, Datt, Christine Allison, Maitreyi Das, Tara Vishwanath, and World Food Program, in particular with Nisha Sumita Chopra and colleagues in the South Asia Social Srivastava. The report has also benefited enormously Protection group. The report has also greatly benefitted from interactions with and feedback from a range of from formal comments received from the Ministries Government of India and state level officials, and with of Rural Development (September 2008), Consumer researchers and civil society representatives. They Affairs, Food & Public Distribution, Panchayati Raj and include Renuka Vishwanathan (then Secretary Rural Labour and Employment (April 2010). Regional Vice President Isabel Guerrero, SACVP Country Director N. Roberto Zhaga, SACIN Sector Director Michal Rutkowski, SASHD Sector Manager Mansoora Rashid, SASSP Task Team Leader Philip O'Keefe, EASSP (then SASSP) Social Protection for a Changing India: Volume II Acronyms AABY Aam Aadmi Bima Yojana AAY Antyodaya Anna Yojana ACA Additional Central Assistance APL Above Poverty Line BPL Below Poverty Line CAG Comptroller and Auditor General CBO Community-based Organization CCT Conditional Cash Transfer CSS Centrally Sponsored Scheme CSO Civil Society Organization DEA Department of Economic Affairs DPC District Planning Committee DRDA District Rural Development Agency EPFO Employees Provident Fund Organization FCI Food Corporation of India FM Financial Management FPS Fair Price Shop GDP Gross Domestic Product GoB Government of Bihar GoD Government of Delhi GoI Government of India GP Gram Panchayat GoR Government of Rajasthan GoUP Government of Uttar Pradesh HD Human Development HH Households IAY Indira Awaas Yojana ICDS Integrated Child Development Scheme ICRISAT International Crops Research Institute for the Semi-Arid Tropics ICT Information and Computer Technology IEC Information and Education Campaign IRDA Insurance Regulatory and Development Authority Acronyms JBY Janshree Bima Yojana LIC Life Insurance Corporation of India LPG Liquified Petroleum Gas M&E Monitoring and Evaluation MDM Midday Meal MEGS Maharashtra Employment Guarantee Scheme MFI Micro-finance Institution MIS Management Information System MKSS Mazdoor Kisan Shakti Sangathan MOLE Ministry of Labour and Employment MoRD Ministry of Rural Development NCEUS National Commission on Enterprises in the Unorganized Sector NFFWP National Food for Work Program NGO Non-governmental Organization IGNOAPS/NOAPS Indira Gandhi National Old Age Pension Scheme MGNREGA Mahatma Gandhi National Rural Employment Guarantee Act NRLM National Rural Livelihood Mission NSAP National Social Assistance Program NSS National Sample Survey JNNURM Jawaharlal Nehru National Urban Renewal Mission OBC Other Backward Classes PDS Public Distribution System PEM Public Expenditure Management PEO Performance Evaluation Office PFMA Public Financial Management and Accountability PMGY Pradhan Mantri Gramodaya Yojana PMT Proxy Means Test POS Point-of-service PRFDA Pension Fund Regulatory and Development Authority PRI Panchayati Raj Institution RBI Reserve Bank of India RSBY Rashtriya Swasthya Bima Yojana RD Rural Development SC Scheduled Caste SFC State Food Corporation SGRY Sampoorna Grameen Rozgar Yojana SGSY Swarnajayanti Gram Swarozgar Yojana SHG Self-Help Group SJSRY Swarna Jayanti Shahari Rozgar Yojana SP Social Protection SRM Social Risk Management ST Scheduled Tribe ULB Urban Local Bodies UTI Unit Trust of India UWEP Urban Wage Employment Program VAMBAY Valmiki Ambedkar Awaas Yojana v Social Protection for a Changing India: Volume II Table of Contents Executive Summary xv Chapter 1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 1 A. Poverty and Inequality 4 B. Vulnerability 10 C. Implications for Social Protection Policy 21 Chapter 2: The Social Protection Policy and Program Mix 25 Chapter 3: Protective Programs 31 A. Public Distribution System (PDS) and Antyodaya Anna Yojana (AAY) 33 B. Social Pensions 44 C. Targeted Housing Programs 48 D. Determinants of Program Participation 52 E. Conclusions and Recommendations 55 Chapter 4: Public Works and Promotional Programs 67 A. Public Works Programs 70 B. Promoting Movement out of Poverty in the Short and Long Term 88 C. Conclusions and Recommendations 100 Chapter 5: Social Security: Closing the Coverage Gap 105 A. Social Protection Coverage and the Nature of the Unorganized Labor Force 109 B. Social Security Schemes Operating in India 111 C. Recent Initiatives to Expand Social Security Coverage 116 D. Expanding Social Security Coverage in the Coming Decade 124 E. Conclusions and Recommendations 128 Chapter 6: Financing and Institutions for Social Protection 131 A. Financing of Social Protection Programs 134 B. Institutional Roles and Responsibilities in SP Programs 145 Table of Contents v C. Political economy of Institutional Reform in Social Protection 152 D. Conclusions and Recommendations 154 Chapter 7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 159 A. Program Awareness and Outreach 161 B. The Application Process and Eligibility Determination 165 C. Operational Aspects of Program Administration 171 D. Conclusions and Recommendations 180 Chapter 8: Targeting mechanisms ­ BPL and Beyond 185 A. What Does "Poor" Mean in Terms of Public Programs? 188 B. The 2002 BPL Methodology 190 C. Alternative Targeting Methods 202 D. How Do Targeting Methods Stack Up? 208 E. Options for Improving Targeting in SP Programs 210 Chapter 9: Getting More from India's Social Protection System ­ Directions for the Future 219 A. Policy Reform in Social Protection 222 B. Improving Implementation of Social Protection Programs 225 C. Political Economy of Social Protection Reform 253 Annexes 255 Annex 1: India Human Development Survey (IHDS) ­ II (2005) 257 Annex 2: Social Protection Survey in Three States 261 Annex 3: Cumulative Coverage Across Major Programs 265 Annex 4: Unbundling SP Service Delivery Activities 266 Annex 5: Methodology for Comparing BPL "Poor" and NSS "Poor" at the Household Level 269 References 271 List of Tables Table 1: Major central social protection schemes, 2009 xviii Table 2: Summary of SP program performance (all-India) xxii Table 1.1: Inequality trends 7 Table 1.2: Migration rates (%) 15 Table 1.3: High incidence of shocks among rural households 18 Table 1.4: The poor cope differently with shocks than the rich 20 Table 2.1: Major central social protection schemes, 2009 28 Table 2.2: Major SP programs by objectives and spending rank 29 Table 3.1: Possession of ration cards by type and state, 2004/05 (% of households) 36 Table 3.2: Household-level offtake of PDS grains - All households and rural/urban, 1993/94, 1999/00 and 2004/05 (% HH) 37 Table 3.3: Household-level offtake of PDS grain by quintile by state, 2004/05 39 v Social Protection for a Changing India: Volume II Table 3.4: Share of PDS grains captured by consumption quintiles, 2004/05 40 Table 3.5: Planning commission estimates of BPL grain leakage in PDS, early 2000s 41 Table 3.6: Household PDS offtake as share of official offtake by state, 2004-05 (%) 42 Table 3.7: Coverage rates of target group for NOAPS, annapurna, widow and disability pensions by state, 2004-05 45 Table 3.8: Household coverage rates and benefit incidence of social pensions by wealth, location and social category, 2004/05 (%) 46 Table 3.9: Schemes to provide support for housing, latrine or chulha construction by state (rural only), 2004-05 49 Table 3.10: Coverage and median benefits of targeted housing programs by wealth, location and social category (rural areas only), 2004-05 50 Table 3.11: Coverage rates and median benefits of urban housing by quintile and social category, 2004-05 51 Table 3.12: Program participation determinants by wealth quintiles, 2004/05 53 Table 3.13: Program participation determinants by social category, 2004/05 53 Table 3.14: Program participation determinants by selected socio-economic characteristics, 2004/05 54 Table 4.1: Evolution of public works programs in India, 1980 to 2006 70 Table 4.2: Pre-MGNREG public works coverage rates from state-specific surveys 74 Table 4.3: Coverage of MGNREG using administrative data, 2006/07 to 2008/09 74 Table 4.4: Coverage of SGRY/FFW by expenditure/wealth and social group, 2004/05 77 Table 4.5: Average and marginal odds of participation in Indian public works programs, 1993/94 78 Table 4.6: Rural daily minimum and average agricultural wages and MGNREG average wage cost 83 Table 4.7: Coverage of SGSY by wealth, social category and location, 2004/05 89 Table 4.8: School stipend coverage and receipts (rural and urban areas), 2004/05 93 Table 4.9: Coverage and receipts of school stipends by wealth, social category and location, 2004/05 94 Table 4.10: Coverage of midday meals, 2004/05 98 Table 4.11: Coverage of MDM by per capita expenditure and social category, 2004/05 98 Table 4.12: Determinants of participation ­ Karnataka, Orissa and Madhya Pradesh (2006) 99 Table 4.13: Infrastructure for MDM 100 Table 5.1: Distribution of unorganized sector workers aged 20-50, by earnings decile, 2004/05 109 Table 5.2: Unorganized sector workers that are members of groups by earnings decile 110 Table 5.3: Key indicators of mandated social security programs 111 Table 5.4: Selected insurance and pension programs of the unorganized sector 116 Table 5.5: Key parameters of RSBY 117 Table 6.1: Spending on major SP programs, 2002/03 - 2009-10 134 Table 6.2: Allocation rules for selected SP CSS 138 Table 6.3: Core funding shares by state for rural anti-poverty programs 138 Table 6.4: Formula share for SP CSS and actual allocations (% of total by state), 2004/05 139 Table 6.5: Expenditure releases per poor rural household for major SP CSS, 2006/07 142 Table 6.6: Spending on the five main CSS for 2006/07 and adjusted allocation norms 143 Table of Contents v Table 6.7: Average household spending on SGRY/NFFW, SGSY and IAY by region, 2002-05: BPL and below the poverty line households 144 Table 7.1: Awareness about Programs, Orissa, Madhya Pradesh and Karnataka (%) 162 Table 7.2: Households Aware of the Safety Net Programs by Social Groups, Orissa, Karnataka and MP combined (%) 163 Table 7.3: Village level determinants of program awareness, Orissa, Karnataka and MP, 2006 164 Table 7.4: Overall awareness of MGNREG, various MGNREG states, 2006 164 Table 7.5: Awareness of specific elements of MGNREG entitlements by state, 2006 165 Table 7.6: Main reasons for not having ration card by state, 2004/05 (% households) 166 Table 7.7: Main reasons for no ration card by wealth, location and social category, 2004/05 167 Table 7.8: Main problems in getting social pension sanctioned, Rajasthan, 2006 167 Table 7.9: Determinants of use of middleman to access benefits, 2006 169 Table 7.10: Reliance on intermediaries to access benefits by program and intermediary types, Orissa, Karnataka and MP, 2006 169 Table 7.11: Share of GPs maintaining registers of social pensioners, Rajasthan, 2006 175 Table 8.1: Discrepancies between different estimates of "BPL" households (rural and urban combined) 189 Table 8.2: Poverty rate and targeting errors in the 2002 BPL classification, by state 192 Table 8.3: Poverty rate and targeting errors in the BPL classification, by expenditure class (1999/00) 193 Table 8.4: Under-coverage rates of BPL method for the poor, selected states 196 Table 8.5: Poverty and under-coverage rates across different targeting methods 197 Table 8.6: Improvement over BPL method in coverage rates for lowest two deciles by state across different targeting methods (%) 198 Table 8.7: De facto targeting even worse than de jure 200 Table 8.8: Possession of ration cards, by type and socio-economic status (% households) 201 Table 8.9: Determinants of ration card holding 202 Table 8.10: Type of targeting by SP programs 203 Table 8.11: Comparing community-based and BPL targeting 206 Table 8.12: Comparing community-based and expenditure-based identification of poor for Orissa 207 Table 8.13: Cumulative share of beneficiaries of major programs by wealth quintile, 2004/05 (%) 208 Table 8.14: Cumulative share of total benefits captured by quintile by program, 2004/05 (%) 209 List of Figures Figure 1: Correlation between major CSS SP program share in total allocations (LHS) and releases (RHS) by state and state share of total poor/poverty rate, 2004/05 xx Figure 2: Diversion and leakage of BPL PDS grains (as % of total) by state, early 2000s xxi Figure 3: Share of poor household by state misclassified as non-poor by 2002 BPL method xxv Figure 1.1: Evolution of poverty since the early 1970s 5 Figure 1.2: Poverty rates - rural, small urban centers and large cities 8 Figure 1.3: Inter-state disparities in poverty 9 Figure 1.4: A large proportion of the population is clustered around the poverty line 10 v Social Protection for a Changing India: Volume II Figure 1.5: Health problems and natural calamities are the most common shocks affecting households 18 Figure 3.1: Household-level offtake of PDS grains by quintile, various years 38 Figure 3.2: Household-level offtake of PDS grains by quintile and rural/urban, 2004/05 38 Figure 3.3: Share of TPDS grains in total household grain consumption among households accessing TPDS, various years 39 Figure 3.4: Proportion of explained program determinants accounted for by state location, various programs, 2004/05 54 Figure 4.1: SGRY workdays per agricultural worker and per BPL household by state, 2003/04 73 Figure 4.2: Household coverage rates of public works by state and all-India, 1987/88 to 2004/05 73 Figure 4.3: Seasonality in MGNREGs work provision, 2006/07 to 2007/08 76 Figure 4.4: Coverage of MGNREG by state, 2008/09 76 Figure 4.5: Participation of women, SC and ST workers in MGNREG, 2006/07 to 2008/09 78 Figure 4.6: Main types of works under MGNREG, 2008/09 85 Figure 4.7: Coverage of midday meals according to administrative data (among children enrolled in primary school), 2007-08 97 Figure 5.1: Life insurance and pension coverage by income decile 112 Figure 5.2: Household enrolment in RSBY by month, February 2008 ­ August 2009 122 Figure 6.1: Share of main CSS in total central SP spending, 2002/03, 2006/07, and 2008/09 135 Figure 6.2: Spending on social assistance and social insurance by region, early 2000s (% GDP) 137 Figure 6.3: Correlation between major CSS SP program share in total allocations by state and state share of total poor in India, 2004/05 140 Figure 6.4: TPDS BPL/AAY offtake by state, 2002-05 141 Figure 6.5: SGRY offtake on cash and food, 2004-05 141 Figure 6.6: Average spending per poor rural HH on main SP programs and poverty headcount by state, 2006-07 143 Figure 6.7: Intra-state SP spending 144 Figure 8.1: BPL misclassification of the Poor and Rural Poverty across states 193 Figure 8.2: Predicted targeting errors in the BPL classification (1999/00) 193 Figure 8.3: Distribution of individual BPL indicator scores across expenditure classes 195 List of Boxes Box 1.1: Lost in translation ­ Invisible populations under - served by public programs 6 Box 1.2: Approaches to assess vulnerability empirically 11 Box 3.1: Policy evolution of the PDS 34 Box 3.2: Some PDS terminology 41 Box 3.3: PDS food coupon reform in Bihar ­ Better results in a poor state 57 Box 3.4: Smart cards in food programs 61 Box 4.1: Key features of MGNREG 71 Box 4.2: Some state-level innovations in implementing MGNREG 80 Box 4.3: MGNREG implementation agencies 81 Table of Contents x Box 4.4: Potential channels of impact of MGNREG 87 Box 4.5: Employment Generation and Marketing Mission (EGMM) 92 Box 4.6: International experience with Conditional Cash Transfers (CCTs) 94 Box 4.7: Overview of conditional cash transfers (CCTs) in India 96 Box 5.1: Main models of health insurance for unorganized workers in India 115 Box 5.2: Recent failed attempts to expand insurance coverage to unorganized workers in India 115 Box 5.3: Illustration of a contribution based old age pension scheme 124 Box 6.1: Mission convergence: Institutional reform of targeted programs in Delhi 150 Box 6.2: Guiding principles for assignment of institutional responsibilities 157 Box 7.1: Accessing IAY ­ The view from the field in three states 170 Box 7.2: Innovations in payment systems 174 Box 7.3: Smart cards in public programs: Experience and issues 177 Box 7.4: Case study of cross-checking for duplicate beneficiaries in Karnataka 178 Box 7.5: Community monitoring of social programs: Experience from Rajasthan and AP 179 Box 8.1: Main types of targeting methods 203 Box 8.2: Participatory Identification of the Poor (PIP) in Andhra Pradesh 205 Box 8.3: Combining indicator-based targeting with community validation and finalization of the poor in Kerala 205 Box 8.4: Community targeting in Bangladesh ­ How well does it identify the poor? 206 Box 8.5: International findings on targeting methods 211 Box 8.6: International experiences with proxy means testing 214 Box 8.7: Targeting under mission convergence for urban Delhi 216 Box 8.8: Operational considerations in scaling up community-based targeting methods 217 x Social Protection for a Changing India: Volume II Executive Summary Executive Summary A. The SoCiAl PRoTeCTion of the social protection system. India's SP system is one which has the resource base and institutional capacity ConTexT in inDiA of a lower-middle income country, but expectations from its population and the political establishment that india's surge in growth and rapid expansion in increasingly approach those of a middle income power. public spending in the past decade has created new Matching fiscal and institutional capacity with societal possibilities for its social protection system. The expectations will be a major challenge for policy makers growing importance of social protection (SP) is reflected in the coming decade. While sustained spending in the Government of India (GoI) Common Minimum increases on SP have been seen in recent years and Program and 11th Five Year Plan which commit to should continue to be possible with robust growth, a institutionalization of programs as legal rights (as in challenge will be to avoid the growth dividend being the case of public works, through the National Rural diluted on programs which do not effectively address Employment Guarantee Act), continued upscaling of the needs of India's poor. interventions (e.g., social pensions and midday meals), and proposals to expand new types of SP interventions while the indian economy has undergone a sea-change to the large unorganized sector (e.g., social security). since the 1970s, the country's social protection Increased resources and political priority for social policies and implementation practices have only in very recent years begun an overdue transition. protection are at the core of India's 11th Plan commitment to a more inclusive growth model. Social protection systems need to evolve to meet the needs of their populations and developments in in addition to presenting opportunities, india's the economy. In this respect, the policy mix of the SP growth has raised expectations from the population system has until very recently been to a significant Executive Summary x extent founded on the needs of the India of the 1970s.1 one of considerable dynamism in the SP arena in India This was a period when around half the population was when compared to the preceding decades. Notably, chronically poor, the country was in aggregate food MGNREG represents a qualitative leap in the design and deficit and importing grains, private market channels execution of public works, a commitment which has for grains were very under-developed, the financial and been matched with massive resources. While still in an banking infrastructure had minimal rural penetration, earlier stage of development, the RSBY health insurance growth was consistently low, technologies available for program for BPL households is path-breaking in its program administration were rudimentary, and only design and has pioneered approaches to delivery which around one fifth of the population lived in urban areas. provide a model for other public programs. There has In much of the country, the scenario is quite different also been overdue but increasing recognition that some to the 1970s, which creates new opportunities to parts of the SP system have serially under-delivered and realize significant equity and efficiency dividends from need fundamental overhaul, as seen for example in the SP reforms. reforms of SGSY under the aegis of the National Rural Livelihoods Mission and the ongoing debate on the PDS while poverty has gradually reduced in recent in the context of the Food Security Bill. decades, vulnerability remains high, new sources of vulnerability have emerged, and the diversity of however, while the momentum of sp reform has been needs among the poor has increased. Since the 1970s, positive in recent years, there is a strong need to the poverty headcount has fallen from around one take the lessons of success from individual programs half of the population to closer to one quarter, though and from the experience of states as a basis for more poverty rates remain significantly higher in several fundamental transformation of the sp system in its lagging states. With increasing urbanization, the share relevance, efficiency and welfare impacts. Despite of the urban poor in total has also increased, particularly recent progress, India is not getting the "bang for the in small and medium towns. In addition, a large rupee" that its significant expenditure would seem to proportion of households, both poor and non-poor, warrant, and the needs of important population groups may be vulnerable to poverty even though they may remain only very partially addressed. This has several not be currently poor. Variability of incomes and coping elements. Firstly, PDS continues its long term pattern of with recurring shocks are a common feature, especially consuming large resources with huge inefficiencies and among poor households. Despite these developments, leakage, and "promotional" SP programs (e.g., SGSY) SP spending remains focused on programs to alleviate have not performed as expected in much of the country. chronic poverty, and focused overwhelmingly on rural Secondly, the needs of the growing number of urban areas. Divergence in income and social indicators across poor remain inadequately addressed, and initiatives and within states has also increased the diversity of such as JNNURM have not proven effective in reversing social protection needs in different parts of the country, the situation. Similarly, the system is largely unprepared including growing needs among the urban poor.2 Basic to address the needs of mobile populations, a group subsistence needs and services remain the priority which is likely to continue to grow as economic reforms in some areas, while other areas are facing second deepen. Thirdly, identification of the poor people who generation challenges of expanding SP instruments to the SP system seeks to prioritize remains problematic, deal with economic modernization, and the new risks with major issues in design and implementation of the and vulnerability it brings. BPL system. Fourthly, execution is a perennial challenge the authorities at central and state levels recognize and one where progress is slow. Innovations in delivery the need for a more effective and relevant social systems in terms of ICT use, increased social and protection system, and there has been considerable community engagement, and other features remain innovation on sp policy and delivery systems in piecemeal, both in program terms and geographic recent years. The period since the mid-2000s has been penetration. However, this variable SP program 1 Saxena(2006)documentstheevolutionofprogramssincethe1960s. 2 SeetheWorldBank'sIndiaPovertyAssessment(2011)forevidenceondivergenceacrossthecountryinkeyindicators. xv Social Protection for a Changing India: Volume II implementation experience across states offers valuable lessons for the future. B. The FRAmewoRk FoR it thus seems timely to look at where india's sp system looking AT The SP SySTem has come from, where it stands, and where it might the report organizes the discussion of sp policies and be expected to move in the coming decade. The report programs according to terminology widely used in deals with two key elements of India's social protection india.5 the three main pillars of sp programs in this system: safety net policies and programs, and social typology are: security for the unorganized sector.3 It has been prepared "promotional" measures, which aim to improve at the request of Government of India, which requested incomes, both in the short to medium term a review to assess which SP programs work best, which (through livelihood interventions) and in the are less effective, identify gaps and overlaps, and longer run (through human capital interventions). suggest reform options. In response, the objectives of In the context of this report, the key programs the report are to: (i) review trends in poverty, inequality in this area are SP interventions to support and vulnerability in India; (ii) evaluate SP program investments in human capital (e.g., stipends; performance ­ including both safety net programs and midday meals; conditional cash transfers), and social security for the unorganized sector - according targeted credit and livelihood programs for the to: (a) a coherent social protection policy framework; poor (see Chapter 4). Public works programs (b) empirical indicators of performance, including can be viewed as a hybrid of promotional and awareness, coverage, adequacy and benefit incidence, and preventive measures. (c) financing, administrative, and institutional systems; and "preventive" measures, which seek to avert (iii) provide recommendations for improving the ability to reach the poor with more effective public spending, deprivation prospectively by supporting private participation and stronger administration. households to manage different risks and shocks ex ante. The main focus in the report (Chapter 5) the report draws on existing and new data sources, . in this pillar is on public social insurance programs including analysis of: (i) administrative data; (ii) several for the unorganized sector. rounds of the National Sample Survey (NSS) data; "protective" measures, which provide relief (iii) the 2004/05 Human Development Profile of India against deprivation ex post to the extent that the survey (undertaken by NCAER and the University of other two sets of measures fail to do so. This could Maryland), which included a World Bank-financed address those falling into poverty as a result of safety nets module and was representative nationally shocks, and/or for the chronically poor. The main and for major states; (iv) a social protection survey public programs within this pillar in the report (SPS) undertaken for this report in 2006 in rural areas of are PDS, social pensions, and targeted housing Orissa, Madhya Pradesh and Karnataka; (v) dedicated programs for the poor (see Chapter 3). In the surveys on social pensions in Karnataka (KSPS) and private arena, such strategies would include sale Rajasthan (RSPS) in 2005 and 2006 respectively; of household assets, reduction in consumption, (vi) a living standards survey conducted in Jharkhand running down savings, or taking children out in 2005 (JLSS).4 In addition, the report incorporates of school. a rich body of secondary sources on SP program performance and impact by national researchers and this typology is similar to the social risk management Government agencies. (srm) framework commonly used by the world bank 3 Whilelabormarketpoliciesareanimportantelementofsocialprotection,theyarenotthefocusofthisreport.Forarecentstudyonlabormarkettrendsand policiesinIndia,seeAhmedandNarain(2010). 4 Annexes 1 and 2 give details of the HDPI and the SPS surveys. The HDPI survey data is publicly available at http://www.icpsr.umich.edu/icpsrweb/ICPSR/ studies/22626,seealsoAjwad(2006)andDevetal(2007). 5 DrèzeandSen(1989);Guhan(1994);SeealsoGentilini(2005)foradiscussionoftheseconcepts.Inthefollowingdiscussion,"socialprotection"referstopoliciesand programsunderallthreeofthesepillars;"safetynets"refertoprotectiveandpromotionalmeasureslargely;"socialassistance"referstoprotectiveprogramsonly; and"socialinsurance"and"socialsecurity"areusedinterchangeablytorefertoinsurance-basedprograms. Executive Summary xv and familiar internationally.6 The SRM approach . is based on the insight that individuals, households C. key FinDingS and communities are exposed to multiple risks, both while india's range of social protection programs is covariate and idiosyncratic, and that the poor are impressive for a developing country, the sp system particularly vulnerable to the manifestation of these risks in spending terms and priorities remains strongly in the form of shocks. Social risk management strategies focused on protective programs to mitigate chronic can be grouped into three broad categories, each of poverty, and on rural areas. This can be seen in which involves a mix of reliance on public interventions, Table 1. While a primary focus on protective interventions informal arrangements, and market-based approaches: is understandable, the policy mix of the SP system (i) risk prevention strategies, which seek to reduce the has not to date responded to the evolution of living probability of adverse shocks occurring.Such strategies standards as much as might have been expected.There may be at the macro level, such as labor policies are three areas where the evolution of SP programs and which increase employment growth, or investments spending has been less and/or later than one might at household level in human capital or livelihood have expected: promotion; (ii) risk mitigation strategies, which help to reduce the impacts of shocks ex ante when such shocks insurance-based interventions remain in their are unavoidable. The classic example of this is insurance infancy in terms of coverage of the unorganized against low income in old age, or health insurance sector, though RSBY is an exciting and rapidly in the inevitable event of serious health events; and expanding initiative which can provide a way (iii) ex post "risk coping" strategies, which seek to mitigate forward. Expanding informal sector coverage the impacts of shocks after they have occurred, which of social insurance is a challenge that India has would classically be social assistance programs, whether had relatively limited success with until very in kind or cash. recently, despite a series of central and state- Table 1: major central social protection schemes, 2009 Allocation (2009/10) Primary objective; Scheme Type Description (` crore) benefit (cash/food) Public Distribution Subsidized Subsidized wheat and rice, plus 42,490 for food; Protective System (PDS) food and fuel kerosene and sugar in most distribution states. Level of subsidy varies 2,866 for according to whether APL, BPL, kerosene/LPG AAY or Annapurna household (see Chapter 3). Food/fuel Mahatma Gandhi Self-targeted public Unskilled and low skill public MGNREG: Protective (and National Rural works works. MGNREG guarantees 100 preventative) Employment days employment per rural HH 30,100 Guarantee (MGNREG) per year in all districts. SGRY had aimed for 100 (non-guaranteed) days in rural districts, with a cash MGNREG: Cash and food component. After 2006 (formerly food and cash SGRY was limited to non-MGNREG under SGRY) districts, but was discontinued in 2008 (see Chapter 4). Swarnajayanti Gram Targeted credit Subsidized lending from banks 2,115 Promotional (short to Swarozgar Yojana scheme for to groups of BPL people, with medium term) (SGSY) groups and some allowance for some individual individuals lending, and small APL share in groups (see Chapter 4). Cash (credit) 6 SeeHolzmannandJorgensen(2000)foranelaborationoftheSRMframework. xv Social Protection for a Changing India: Volume II Allocation (2009/10) Primary objective; Scheme Type Description (` crore) benefit (cash/food) Indira Gandhi National Non-contributory Monthly cash benefits for BPL 5,109 Protective Old Age Pension social pensions elderly (originally called NOAPS Scheme (IGNOAPS) Annapurna until renaming to IGNOAPS and expansion in 2007), and state schemes for widows and disabled Social pensions: Cash people (since 2007 also included in Annapurna: Food the central Indira Gandhi pensions program) (see Chapter 3). Indira Awaas Yojana Targeted rural Subsidies for rural BPL with 7,920 Protective (IAY) housing inadequate housing for housing construction (see Chapter 3). Cash for housing Midday meals School feeding Hot meal for children in grades 1-8 8,000 Promotional (long term) program in government and aided schools (see Chapter 4). Food Schools stipends Stipends for school Some central and some state ­ Promotional (long term) enrollment schemes for various target groups, including SC, ST, some categories of girls, disabled Cash Rashtriya Swasthya Subsidized and Subsidized health insurance for 350* Preventative Bima Yojana (RSBY) targeted health hospitalization for BPL households insurance (and MGNREG) in 26 districts (see Chapter 5). Cash Aam Admi Bima Life/disability/ Free insurance covers natural ­ Preventative Yojana (AABY) accident insurance death, disability and accident for for BPL rural landless households (see Chapter 5). Cash Social insurance for Subsidized social Variable by state, though central 270 Preventative unorganized workers insurance welfare funds for select sectors (e.g. beedi workers) (see Chapter 5). Mixed Specific urban anti- Targeted urban Housing construction and 535 Protective poverty programs housing (VAMBAY) upgradation for slum dwellers, and employment and wage and self-employment for the poor (SJSRY) programs for unemployed or underemployed urban poor (see Chapter 3). Cash Source: See Table 2.1, Chapter 2. * RSBY allocation refers to 2008/09. specific schemes. This has been a product of poor promotional interventions in the public sector design, inadequate attention to institutional continue to receive relatively little emphasis, and implementation arrangements, and a "start- particularly given the continuing challenges stop" approach to new initiatives. Of course, such in improving human capital outcomes. expansion is not easy, and many developing Overall, safety nets in India remain primarily countries have struggled to expansion social "nets" rather than "ropes" or "ladders" which insurance coverage in the face of large informal seek to promote sustained movement out of sectors.7 poverty.8 7 SeeO'KeefeandPalacios(2006)foradiscussionofinternationalexperience. 8 SeePritchettetal.(2002)foradiscussionofthesedistinctions. Executive Summary xv otherthanPDS,SPinterventionsinurbanareas be hoped for a variety of reasons. In household are negligible and even more strongly biased terms, in 2004/05, total spending allocated per poor towards protective interventions. With the rural household nationally on major CSS SP programs combination of urbanization and slower rates was significant, at around `9065 (about 40 percent of urban poverty decline, the urban poor are a of the annual rural poverty line) Despite this, impacts growing share of all poor people. Despite this, on the poor as measured by survey-data appear to be per capita SP spending on the poor remain much lower. A first reason for this is that the absorptive heavily biased towards rural areas, and the capacity of poorer states is typically low. They are not promise of JNNURM as a vehicle for transforming able to utilize all funding available given limitations the situation in this regard has to date failed to in administrative capacity. SP CSS allocation formulae be realized in any significant measure. are redistributional, however, providing more funds there remains a strong emphasis on food-based to states which have higher poverty, but which have support which is subject to major governance the lowest capacity to spend effectively. This results and implementation problems and which is of in pro-poor allocations with regressive actual releases questionable relevance for many among the of SP funds across states (Figure 1). Such patterns are poor whose non-food spending needs are an then exacerbated at the sub-state level in many cases, increasing source of pressure. At the same time, and finally at the household/delivery level through food programs remain important for certain misidentification of beneficiaries, and a range of groups, in certain chronic food deficit areas and implementation problems. at certain times in other areas. These issues are in programmatic terms, the main driver of poor cost currently being debated in the context of the effectiveness and impacts of india's safety net is its Right to Food legislation. largest program ­ the public distribution system (pds). While it consumes almost 1 percent of GDP and has wider (a) Safety nets coverage than other safety net programs - between 20-25 percent of the population in the mid-2000s based while india spends significant resources on its core on actual drawing of grains by beneficiaries, and closer safety net programs ­ over 2 percent of gdp in to 40 percent based on administrative numbers on BPL recent years ­ the returns to spending in terms of households - its impact on the poor is very limited in poverty reduction have been much lower than could many states, particularly a number of lagging states. Figure 1: Correlation between major CSS SP program share in total allocations (LHS) and releases (RHS) by state and state share of total poor/poverty rate, 2004/05 60 25 50 20 Poverty share (%) 40 15 HCR 30 10 20 5 10 0 0 0 2 4 6 8 10 12 14 16 18 20 0 2000 4000 6000 8000 10000 12000 14000 Allocation share (%) Central release per poor rural HH Source: GoI, various years for program expenditure; Sen and Himanshu (2007) for poverty estimates. Notes: Programs included: PDS, SGRY/NFFW, SGSY and IAY. HCR refers to head count poverty rate. xv Social Protection for a Changing India: Volume II This is due to a combination of high leakage of grains guarantee act (mgnrega) in 2006. In philosophy, this is a (estimated by the Planning Commission to be around major innovation in taking a rights-based and demand- 58 percent nationally in the early 2000s and even driven approach, guaranteeing all rural households up higher based on estimates using NSS data), a range of to 100 days public works employment per year at the demand and supply side issues in program design and agricultural minimum wage. The program also has a host implementation, and considerable leakage of subsidies of implementation innovations (e.g., social audits by to the non-poor. Although many of the shortcomings of communities of performance; a structured role for PRIs PDS and its very poor performance have been known in implementation; closer attention to the staffing needs for some time, it continues to absorb substantial public at lower levels of the system) which provide a model for resources with limited benefits for the poor. This limits future reforms of other SP programs. Administrative the fiscal options for more effective SP interventions in data and field studies suggest both significant successes the absence of PDS reform (See Chapter 3). At the same relative to previous public works programs (e.g., high time, there have been a range of state-level initiatives, coverage of rural households, with impressive inclusion national-level pilots, and proposals to improve the of SC/ST and women workers) and many challenges in functioning of PDS which have merit. In recent years, translating improved program design into outcomes for there has been increasing recognition of the need for the poor, such as, matching demand for work with the reforming the PDS as evident by the findings of the administrative demands of opening worksites; aligning Wadhwa Committee report and the ongoing debate piece rate compensation with the minimum wage around the Right to Food legislation and the recent requirement; strengthening mechanisms for community proposal in the 2010/11 Economic Survey in favor of participation in works identification and oversight. The direct subsidy (through food coupons) as opposed to diversity of implementation experience across states the current indirect subsidy. provides a rich opportunity for cross-state learning to achieve better and more transparent impacts. A number the most heralded reform of sp programs in of innovations in MGNREG design ­ including social recent years has been the introduction of the audits ­ seem worthwhile to mainstream in other anti- mahatma gandhi national rural employment poverty programs. Figure 2: Diversion and leakage of BPL PDS grains (as % of total) by state, early 2000s All-India WB UP TN Rajasthan Punjab Orissa Maharashtra MP Kerala Karnataka HP Haryana Gujarat Bihar Assam AP 0 20 40 60 80 100 % of BPL grains diverted and leaked Source: Planning Commission, GoI. Executive Summary xx (b) Social security for unorganized in addition to the rsby, a number of important changes to the social security landscape appear to be workers unfolding both in terms of program design as well as insurance and pension programs are less developed delivery. The 2010 Budget includes a budget to finance relative to safety nets and cover fewer than ten a matching contribution of `1000 per annum to those percent of the labor force. Despite repeated attempts workers that voluntarily choose to contribute at least to introduce new programs for unorganized sector that amount. This incentive, combined with the kind of workers over the years, insurance against life's major outreach typical of the RSBY program, could generate shocks such as health, old age, disability and death has a significant increase in pension coverage for the first not been available to the vast majority of the population. time in India. Recognition of this public policy gap led to the passage of framework legislation in December 2009 and the creation of a Social Security Board. (c) Administration and delivery of social protection programs the most promising effort in this area is the targeted health insurance scheme known as rashstriya most safety net and social security programs in most swasthya bima yojana or rsby. This program now states are characterized by a range of problems provides insurance for hospitalization for more than which also reduce their poverty reduction impact. forty million poor people and is growing rapidly. The There are different experiences with implementation RSBY may be the first social sector program in India across SP programs and states. This includes programs that simultaneously takes into account the perspective which have wide coverage but are plagued by leakage of the poor, focuses on getting the incentives of the of subsidies that limit the impact on the poor (e.g., various players that have to deliver the benefits right PDS), others which are well targeted and increasingly and encourages changes over time based on evidence. well designed but face a range of implementation The experience of the first year or so has been positive, challenges (e.g., public works - MGNREG), and still others particularly in terms of demonstrating that the target which appear to be well designed and with systems for population can be reached (despite the poor quality better implementation (e.g., RSBY). Summary indicators of the BPL list). However, in no country has such a large of program performance are presented in Table 2, health insurance scheme operated without the oversight Chapters 3­5 provide discussion of programmatic of a specialized agency. Setting up this agency is the key performance and Chapters 7 and 8 examine some short run challenge for the RSBY. drivers of outcomes. Table 2: Summary of SP program performance (all-india) Coverage Bottom Benefit Spending (% of quintile as incidence evidence allocation relevant % of total (% to bottom of leakage hh rank population; beneficiaries quintile) outside Targeting Program 2008/09 2004/05) (2004/05) (2004/05) Awareness beneficiaries mechanism (1) (2) (3) (4) (5) (6) (7) (8) Public 23.3% of all 27% 29.7% High High BPL Distribution 1 HH (APL and System (PDS) BPL) mahatma 2 33% of rural ­ ­ Substantial ­ Self-targeting gandhi HH national Rural (2008/09) employment guarantee (mgnReg) xx Social Protection for a Changing India: Volume II Table 2: Summary of SP program performance (all-india) Coverage Bottom Benefit Spending (% of quintile as incidence evidence allocation relevant % of total (% to bottom of leakage hh rank population; beneficiaries quintile) outside Targeting Program 2008/09 2004/05) (2004/05) (2004/05) Awareness beneficiaries mechanism (1) (2) (3) (4) (5) (6) (7) (8) Sampoorna 5 1% of rural 43.4% 34.2% Low ­ Self-targeting grameen Rozgar HH yojana (SgRy) Swarnajayanti 7 0.8% of rural 32.9% 7.7% Low Intermediate BPL gram Swarozgar HH yojana (SgSy) indira gandhi 6 8.3 % HH 32.8% 31.1% Substantial Low Mixed national old with elderly Age Pension (noAPS) Annapurna ­ 1.7 % HH 51.9% 37.9% ­ Low Mixed with elderly widow pension ­ 6 .2% HH 43.2% 32.8% Substantial Low Mixed with widows Disabled ­ 14.1 % of HH 30.7% 27.6% ­ Low Mixed pension with disabled indira Awaas 4 12.8 % of 28.6% 24.6% Substantial High BPL yojana (iAy) rural HH midday meals 3 ­ ­ High ­ Government/ aided schools only Schools ­ 9.6 % of HH 22.6% 11.7% Low Low Various; stipends with children strong SC/ST school age focus Source: Ajwad (2006) based on the 2004/05 HDPI, except for PDS based on Bank staff estimates from the 2004/05 NSS 61st round. Awareness based on 2006 SP survey in three states; Leakage based on HDPI survey and NSS (for PDS) and GoI and CAG reports. Notes: l Col. 2: GoI budget data for spending allocation rank (see Chapter 6) l Col. 3-5: Coverage and targeting indicators are based on national household survey data, namely the 2004/05 National Sample Survey for PDS and the 2004/05 Human Development Profile of India survey (NCAER-University of Maryland) for all programs except PDS (see relevant tables in Chapters 3-5). The only exception is coverage rates for MGNREG ­ these are based on administrative data for 2008/09. See also Ajwad (2006) for coverage and targeting indicators based on the 2004/05 HDPI. Note that coverage refers to the share of beneficiaries in the relevant population. For instance, coverage of old age pensions are estimated for households with at least one elderly (65 years and above) member; PDS coverage for all BPL and APL households; MGNREG coverage for all rural households, etc. l Col. 6-8: These are broadly summarized based on the findings of several GoI and other research studies, including those commissioned for this report. For instance, assessment of awareness among households of various SP programs draws on the Social Protection Survey in three states, the Jharkhand Living Standards survey, the Rajasthan social pensions survey, etc. Awareness levels: High = 70% + of population aware; Substantial = 40-70%; Low = < 40% aware. Assessment of leakage is based on comparisons of household survey data against administrative data as well as findings of CAG reports and other studies. These are documented in the relevant chapters (Chapters 3-5 for leakage; Chapter 7-8 for awareness and targeting mechanisms). l Population coverage: For NOAPS as share of HH with member. 65+, for Annapurna, for HH with member 60+; for widow pension, for HH with a widow; for disabled pension, coverage as share of census PWD rate in state; Awareness levels: High = 70% + of population aware; Substantial = 40-70%; Low = < 40% aware. despite such spatial and cross-program lack of coordination and overlap in delivery diversity, a number of common challenges emerge, of programs (both within and across levels of including: government), reduces accountability of those Executive Summary xx responsibleforSPservicedelivery.Lack of financial investment. The RSBY program indeed provides a and human resource capacity - most notably at "state of the art" example of how a well thought- the block and GP level but also at higher levels out and executed delivery model has the potential in terms of functions like strategic planning - to be transformative and provide a backbone for compounds the challenges of service delivery. other parts of the SP system to consolidate around As in many areas of service delivery in India, SP robust systems (Chapter 7). programs suffer from incomplete alignment of for a number of programs, expansion of and the "3 Fs" (functions, funds, and functionaries) for innovation in the private sector has created a variety of reasons. Even where there has been possibilities for new modalities of Public Private progress in specific programs (such as MGNREG) Partnership (PPP) program delivery which have or states (such as Kerala), there remains a huge yettobeexploredfullybythepublicsector.Most unfinished agenda in "squaring the circle" in a SP programs remain dependent on the traditional sector that is dominated by centrally sponsored mode of publicly financed and delivered benefits. schemes operating in a broader environment of While for some programs (and/or in some constitutionally-mandated decentralization of geographic areas) market failures will continue service delivery (see Chapter 6). to make this the appropriate delivery mode, the maintenanceofa"onesizefitsall"SPprogram expansion of private sector players makes mixed andpolicymixfromthecentredoesnotrespond delivery options more feasible than in the past. to the growing spatial diversity in living Some programs such as MGNREG have successfully standards.While the insight that "India is a big partnered with civil society and communities for and diverse country" is a truism in public policy, program execution and oversight. Similarly, RSBY the CSS which continue to dominate SP policies has partnered with private sector providers for give states limited flexibility to tailor central program delivery (see Chapter 5). subsidies and programs to their diverse needs. a poorly designed and executed household While there has been progress in allowing targeting mechanism (the so-called BPL 2002 states flexibility at the margin in adapting methodology). The BPL method does not implementation specifics of some schemes, reflect good practice in design of proxy means- their overall SP policy mix remains largely testing mechanisms, and as a result in its design determined on a uniform basis by the centre, misidentifies almost half the poor as non-poor, more so in lagging states where own-resources and conversely almost half the non-poor as poor are limited. (see Chapter 6). (Figure 3). In addition, empirical evidence on the basic "nuts and bolts" of program performance indicates that it typically performs administration and procedures in most states worse in targeting terms than other methods. In are far below the standards that could be contrast, methods such as self-targeting (in public possible given India's technological and human works), mixed methods of identifying the poor capital capacity. The report outlines a range (as in social pensions) have notably better of challenges where India has struggled to targeting efficiency and inclusion of the poorest, modernize SP program administration, including while some states rely on community wealth program outreach and awareness raising, the ranking and verification (see Chapter 8). applications process, public expenditure and financial management, record keeping and the emerging experience with some programs payment systems, and monitoring and evaluation suggests that problems in sp service delivery can as well as public-private partnerships. At the same be overcome. Despite the significant policy and time, the progress made in modernizing program implementation challenges facing SP programs, administration in states such as AP and under experience with programs (such as MGNREG and specific programs such as MGNREG suggests that RSBY) and/or in specific states across programs (e.g., a modern SP delivery system is an achievable goal AP, Kerala, Gujarat, TN) suggests that sufficient will and for India with sufficient commitment and modest attention to incentive structures of different service xx Social Protection for a Changing India: Volume II Figure 3: Share of poor household by state misclassified as non-poor by 2002 BPL method WB UP TN RA PU OR MA MP KE KA HP HA GU BI AS AP 0 10 20 30 40 50 60 70 80 90 100 % of poor HHs misclassi ed as non-poor Source: Jalan and Murgai (2008). delivery agents can improve the implementation of programs. While the above problems have D. SoCiAl PRoTeCTion FoR A characterized the safety net system for some time, the ChAnging inDiA last decade has witnessed much greater innovation and experimentation by states in addressing perennial the report finds that significant reforms are needed challenges. The core challenges are therefore to with both respect to the medium term social protection create space for innovation by states and a feedback strategy and in implementation arrangements of loop, via the systematic collection of evidence on current programs. India needs to consider both the implementation, to the design of central SP policies vision and policy mix of social protection programs, and programs from best practice in the states. and how to improve implementation of its current and future programs. It suggests that India could significantly in recent years, increasing attention is rightly being improve and modernize its social protection system - in given to implementation, delivery and tracking both safety nets and social security ­ and in so doing help of benefits. A Committee has been charged with reduce chronic poverty and assist both poor and non- improving the flawed methodology for determining the poor households to manage risk better. A more effective BPL list and at least one state has already implemented social protection system would allow India to promote a new round. The Government of India is sponsoring an greater equity, i.e., to ensure the benefits of growth are unprecedented scale of biometrically based, national ID shared by the poor. It could also contribute to growth, through the Unique Indentification Authority of India by allowing individuals to make risk/return choices that (UIDAI) which could facilitate everything from financial contribute to higher productivity, enhancing dynamic inclusion for direct payments of benefits to verification of efficiency. Someofthekeymessagesare: identification for receipt of public transfers or MGNREGA wages. In parallel, many states are attempting to tighten verification and tracking of benefits in major programs (a) Policy reform in social protection such as PDS, NOAPS and MGNREG through the use of biometric identification and the creation of a back-end there is a need to deepen the ongoing policy database that can track what happens on the ground. reorientation of the indian social protection system The challenge facing the country is to coordinate all of to meet the changing and increasingly diverse needs of these good initiatives towards minimizing duplication, its population. Marginal changes alone will not deliver converging on national standards for interoperability the kind of safety net which a changing India needs for and allowing for portability of benefits. its poor and for its economy.This would involve several Executive Summary xx elements: (i) a rebalancing of the policy mix across poor.The "big elephant in the room" in this respect different types of public SP priorities; (ii) consolidation is fundamental reform of PDS. The poor long of the large number of central and state schemes to run performance of the program in many states a core set of flagship programs; (iii) in the context of suggests that the medium term vision of a reformed consolidation, introducing an element of choice and PDS for most groups should be cash-based, though flexibility for states in the specific program mix of this would face substantial resistance in light of centrally-supported schemes that they operate; and the ongoing debate around the Right to Food (iv) in some areas and for some programs, actively legislation. A reformed PDS could still provide exploring the possibilities for leveraging the role of food-based support for specially vulnerable private players (both non-governmental and for-profit) groups (consistent with Supreme Court orders), in delivery of interventions. and in specific areas facing chronic or acute food shortages, but for most areas and most people, a in terms of reorienting the policy mix, the report cash-based social assistance system seems a more suggests several directions: efficient and transparent means of providing an increasingtheemphasisonpreventiveprograms income floor. An intermediate solution currently which help the poor and those vulnerable to being mooted in the 2010 Economic Survey is to poverty to manage risks and shocks better. This transfer the subsidy directly to households (rather implies a significant expansion in coverage of than the PDS store owner) through food coupons different social insurance instruments, though with a lumpsum entitlement that can be exchanged in a phased manner consistent with institutional at any PDS store. and fiscal capacity. Experience to date suggests in the face of demographic change and slower that phasing would benefit from: (i) starting with rates of poverty reduction in urban areas, starting simpler-to-administer insurance products such as toaddresstheneglectofurbansocialprotection life and permanent disability, while continuing to policy. While some of the needs of the urban pilot and evaluate experience with more complex poor are common to their rural counterparts, products such as health insurance; and (ii) for the possibilities (and constraints) of the urban reasons of ease of worker mobilization and to environment suggest that simple mimicking of control transactions costs, focusing initially on the rural models of SP programs and service delivery "low hanging fruit" of unorganized workers who mechanisms is unlikely to be an adequate are members of groups (e.g., MFIs, cooperatives, response. For example, the options for "voice trade union and worker associations, SHG accountability" of service providers which can federations) that could play an intermediary be mobilized in rural areas through collective function between workers and the state/insurers. community action are likely to be less possible rethinking programs which seek to promote in urban areas, while the possibilities for "choice movement out of poverty in two ways: firstly, accountability" (through income enhancement moving from administratively driven subsidized and offering options in service providers where credit to public financing of a more diverse range possible) are likely to be greater. The JNNURM of livelihood promotion approaches better suited program offers a base for larger reforms affecting to the labor market conditions of individual states urban policy, however closer consideration may as is currently being proposed under National be needed regarding support to specific SP policy Rural Livelihoods Mission (NRLM); and secondly implementation. considering the options for use of safety net the specific proposal of this report is that central transfers (see next bullet) to leverage participation sp programs over time aim for a "3 +block" strategy. in core education, health and possibly nutrition This would involve 3 core CSS SP programs or "pillars", services, in order to promote long term movement combined with an SP block grant from which states out of poverty. could finance other SP programs - or supplement movingtomoreconsolidatedandmorecash-based benefits under the core pillar programs - more tailored social assistance programs for the chronically to the poverty and vulnerability profile of the individual xxv Social Protection for a Changing India: Volume II state. This would also involve significant expansion in vary according to state-level priorities, and include urban areas. This could promote both a more coherent interventions such as livelihood support of different and less duplicative SP system, but also give states more forms, targeted housing, interventions to incentivize use leeway to adapt the SP policy mix to the needs of the of basic social services, nutrition and/or early childhood poor in individual states in light of available economic care (e.g., through conditional cash transfers as being opportunities. The three core pillars proposed are: piloted in some states), specific urban SP programs, a major social assistance program. The obvious or other options as proposed by states. A secondary candidate for this is a significantly reformed PDS, benefit of such an approach would be strengthening of merged for specific groups with existing social complementarities between CSS and state-sponsored pension programs.9 Chapter 3 give more details schemes in order to control unnecessary duplication. on the options for reform of PDS proposed by A common core national SP system under the three this report, with a preference for a predominantly pilars could promote portability of basic entitlements, cash transfer approach. and be increasingly useful as mobility of workers and households increases. a public works program, for which MGNREG would be the building block, as well as piloting in terms of promoting both more effective spending on expansion in urban areas. There are several sp, the "3+block" proposal would allow more cross- benefits of a reliable public works program at this program flexibility to states ­ or possibly districts - stage of India's development: (i) by its demand- in deciding their sp expenditure priorities, while still driven nature, it can be responsive to shocks in a maintaining a common national core sp system. It way that longer run programs typically can not. would also allow for greater adjustment in light of In this way, it functions as a "quasi-insurance" poverty levels and key vulnerabilities. This could be done program for the extended period during in a variety of ways, possibly using a menu approach to which more structured insurance is expanded SP programs, and a flexible form of social protection to the unorganized sector; (ii) the positive block or matching grant which consolidates resources targeting outcomes of self-targeted works; and from existing SP CSS.10 This is an approach which has (iii) the potential for multiplier effects from received increased attention in India in recent years as asset creation and community mobilization policymakers seek greater impacts from SP spending. distinguish public works from other SP programs. Given current financing channels from the centre to See Chapter 4 for a discussion. states in India, a more flexible granting mechanism for abasicsocialsecuritypackageforthoseoutside SP programs to states could take different forms ­ as a the formal sector which could be expanded more "bundled" anti-poverty CSS, as Additional Central in terms of coverage and scope of benefits Assistance (ACA) along the lines of programs like as institutional capacity and fiscal space is Pradhan Mantri Gramodaya Yojana (PMGY), or through a developed. The core types of insurance which more fungible realignment of Gross Budgetary Support GoI aims to expand include life, disability, old age shares between the states and central Ministries which pension, and health, and the RSBY program is control CSS.11 already an important step forward in this regard. the above approach would require close consideration Chapter 5 gives suggestions on institutional, of the policy design of existing programs. The financing, sequencing and other aspects of an findings of this and other reports suggest that the expansion strategy. PDS require fundamental consideration of its current beyond the three "pillars", states could receive an design, SGSY is currently undergoing significant change additional transfer and implement state-specific and attention to implementation under NRLM, while sp interventions. How this is programmed could two others warrant experimentation to see whether 9 ThiswouldbesimilartotheChineseurbanandemergingruralsocialassistancesystem,whichisbuiltaroundthe"dibao"programwhichprovidescashbenefits tothepoor,andhasadditionalprovisionforspeciallyvulnerablegroupssuchasdisabledandunsupportedelderly.SeeWorldBank(2009b). 10 SeedeNeuborg(2002)foradiscussionofthestrengthsandweaknessesofdifferentblockandmatchinggrantmechanismsinthecontextofSPprograms. 11 SeeSaxena(2006)forahistoryofcentraltransfersforanti-povertyprogramssincethe1960s. Executive Summary xxv innovative approaches can yield better outcomes than a second overarching need in such policy reorientation seen to date ­ IAY and school stipends. The other major will be greater willingness to experiment in program programs ­ public works and social pensions - would also design and base reforms on results of evaluations. benefit from policy improvements but these are more This will require changes in two tendencies of Indian in the nature of incremental policy reform which can be SP policy since the 1970s: firstly, what CAG has called expected in the normal course of program evolution. "rechristening and revamping" of programs at the Urban programs remain small, but pre-conditions expense of genuine experimentation and innovation, suggest that merely transferring rural SP models to and secondly a limited willingness on the part of the urban settings will limit potential impacts, and equally central government to give states (and in some cases, the that the SP system needs to explore stronger linkages sub-state level) a freer hand in adapting their policy mix with the livelihood opportunities available to the poor among programs, by allowing flexibility in adjustment in urban areas. of specific programs to suit their diverse circumstances. The experience of a number of developing countries a number of cross-cutting issues in policy evolution including Bangladesh in recent decades provides a of the sp system will also be important. These positive example of the social benefits of experimentation include cross-program convergence and consolidation, in SP policy. Closer to home, there is growing innovation willingness to experiment with new types of programs at the state level in India which demonstrates the value and modes of delivery, and more structured efforts to of such an approach, and the increased buy-in among build understanding of the benefits of reforms and politicians and administrators for innovations which are manage the political economy of the reform process. "home grown". An additional challenge will be ensuring increased attention in SP policies to the needs to the urban poor. a third overarching theme of policy reform is that These are discussed in turn below. "government cannot do it alone", and programs would benefit from appropriate partnerships with firstly, central and state-level policymakers will the non-government sector. This partnership could need to accelerate cross-program convergence and be in both policy formulation and the specifics of policy consolidation in policy and administration to make design. The "non-government sector" in this respect the system more understandable to the population, could range from communities themselves (in the form realize economies of scale in policy development of SHGs and other forms of CBOs), to the NGO sector, and administration, and simplify ongoing planning to the for-profit private sector in specific programs and execution of sp programs. This will require and functions. The design of MGNREG is a promising enhanced efforts of institutional coordination within example of such a reorientation of policy formulation and across levels of government and administration. and program design, with its clear roles for community Both at central and state levels, there seems a need for and NGO actors, and willingness to bring in private formation of an inter-departmental Task Force for Social sector expertise and research institutions on areas Protection which would promote coordination (and such as M&E. But there is room for much more active possibly reduce duplication) across targeted programs, engagement with the commercial private sector also, and promote more coherent strategy development including in areas such as public grain distribution, on the medium term policy mix and priorities. This targeted credit and livelihood interventions for the poor, could in turn assist the transition from the current and low-income urban housing. program-driven approach to SP to thinking in terms of a social protection systemwhich is animated more by poverty outcomes and less in terms of scheme-based (b) improving implementation of SP target fulfillment. Such institutional reforms have been important elements of successful SP reforms in programs a range of developing and developed countries. Some even if the necessary reorientation of the sp policy Indian states, notably Kerala and more recently Delhi, and program mix can be achieved, it will not improve are also moving towards an integrated approach to outcomes for the poor unless accompanied by a SP delivery. thorough overhaul of sp program administration, xxv Social Protection for a Changing India: Volume II including institutional arrangements. Whatever smart cards and other innovations), building on the evolving mix of SP policies, there will be several improved rural banking infrastructure to overhaul key elements of administration and institutional payment systems, and building a strong culture of arrangements which will need to be confronted if India M&E. Recent reforms in RSBY and to a lesser extent is to achieve the poverty reduction outcomes that its MGNREG, together with a number of state-specific significant spending on SP warrants, including: program pilots, provide many lessons in this area, and it is hoped that they can be systematically delineating clear lines of accountability incorporated in other SP programs over time. accompanied by adequate staff and finances. Delineating appropriate institutional overhauling existing targeting mechanisms, responsibilities for all links of the SP service both at the household level and geographically. delivery chain, and aligning the division Any social protection system needs to be able of functions with assignment of personnel to identify who are the poor with a reasonable and allocation of resources for program degree of accuracy. Innovations already operating implementation will be critically important for in India and good practice from other developing improved implementation of SP programs. This countries offer a range of options for significant will require first and foremost greater proactivity improvements in targeting mechanisms. These on the part of states to approve policies and include: (i) development of "poverty maps" at put into practice the PRI/ULB decentralization a sub-state (probably block) level which would provided for under the 73rd and 74th constitutional allow more precise geographical allocation of SP amendments. This would need to be followed funds to poor areas; (ii) overhaul of the BPL 2002 by a process-intensive reconciliation of central methodology in line with good practice in design guidelines, state-level stances on service delivery of proxy means-tests (PMT) in other developing decentralization, and capacities at sub-state countries, including allowing for cross-state levels to perform the required implementation and urban/rural variations in the PMT formula; functions in SP programs. Achieving this goal will and (iii) in rural areas, continued piloting and require coordination and gradual convergence strengthened evaluation of community-based across the many departments of government beneficiary identification for SP programs with an responsible presently for different programs, and eye to convergence with a reformed BPL system, commitment to strengthening implementation and possibly ­ as has already happened in AP ­ its capacities at the lower levels of the system, in replacement in appropriate settings. While the particular at the block and GP levels. It will also new BPL methodology proposed by the Saxena mean building on innovations in institutional Committee improves upon the 2002 system in roles to promote greater accountability among several ways, several drawbacks of the previous SP service providers, as exemplified by social method remain. Piloting the proposed methods audits in MGNREG. and subjecting this methodology to the test rapid and substantial improvements in the basic using NSS data as done with the 2002 BPL method "nuts and bolts" of program administration would yield interesting insights into how well the and procedures. The detailed suggestions in new dejure targeting design would perform. this regard are outlined in Chapter 7. Broadly, in the area of social security for unorganized they would involve overhauling a range of workers, past experience suggests that direct bureaucratic procedures which impede funds public provision, financing and administration flow, strengthening processes for administrative is neither feasible nor desirable. It seems more and social accountability of service providers, feasible to partner with existing non-governmental a through modernization of program record entities (for-profit, NGOs, and membership - keeping and reporting arrangements (including based organizations) and restrict the role of computerizing systems and taking advantage of government to: (i) providing targeted subsidies; India's ICT prowess to look for "technology leap- and (ii) regulating these entities and setting basic frogging" opportunities such as introduction of standards. This model already exists in India in Executive Summary xxv several forms, such as the JBY scheme operated broader societal understanding of the need for and by Life Insurance Corporation of India (LIC) for benefits of reforms. life insurance, and more importantly the RSBY some of the political economy challenges that the program. In addition, many other schemes falling sp system confronts if it is to become more coherent into the community based or micro-insurance and more effective as a tool for promoting poverty category could be incorporated under an umbrella reduction and inclusive growth include: program that provided matching contributions or premia but set certain standards in terms of as in many areas of policy reform in India, benefit targets, eligibility conditions, investment consolidation and reform of the SP system policy and recordkeeping, among others. within a coherent strategy will run counter to the past experience of scheme-driven initiatives if such a reform agenda can be delivered, the benefits by a plethora of Ministries, and the observed for the poor of india could be substantial, and make tendency of each new government at both growth significantly more inclusive. In addition, there central and state levels to want new SP programs are likely to be positive impacts on growth itself from an clearly distinguishable from their predecessors. SP system which more effectively addresses a range of Reducing these natural bureaucratic and political market failures which result in poor and unproductive tendencies will be very challenging. A first step is citizens. The traditional view of social protection obviously having an integrated SP strategy which systems and the redistributional objectives underlying is driven by the top politicians and bureaucrats them was that there was a clear growth versus equity at central and state levels, with strong inputs trade-off. However, empirical evidence increasingly from civil society in its development, including highlights that a well-designed and implemented opposition parties. However, even if such a SP system provides dynamic efficiency gains to the strategy process can be developed, it will be economy through positive impacts on productivity, and important for it not to become a "one shot" as an important tool for governments in managing the exercise, but to have institutional coordination impacts of reforms in the wider economy.12 mechanisms in place which explore program duplication and exploit synergies. (c) Political economy of SP reform giving states a more flexible hand in use of central SP resources will be a challenging the political economy of sp policy reorientation transformation both for central administrators is complex, and will require intensive efforts to (whose past tendency has been to define build consensus on reforms. In particular, it will be the parameters for use of central funds quite important to ensure that the interests of perceived tightly) and politicians (who not unexpectedly "losers" of SP and broader economic reforms are taken seek political attribution for centrally-financed into account. While reforms that involve expanded coverage or new types of interventions are unlikely to schemes implemented by states). The first of be controversial, there are strong interests in preserving these challenges is perhaps easier to address the status quo in SP programs among a range of actors, through development of more outcome-based including administrators, politicians, contractors and monitoring systems. The second is more difficult others. Simply cutting programs or excluding certain in a democracy. groups of beneficiaries or institutional players is in a number of programs, there are presently therefore unlikely to be successful unless incentives significantrent-seekingopportunitiesforarange for institutions and households which will be affected ofactors.Such opportunities are facilitated by the by reforms can be part of the reform package. The current complexity of the SP program mix, but political economy of SP reform is challenging in all also by the number of intermediaries who often countries, and governments in India would benefit are involved in the interactions of poor people from more innovation in their efforts to create a with the SP system. The generic identity of such 12 SeeWorldBank(2004). xxv Social Protection for a Changing India: Volume II official intermediaries and unofficial middlemen National Rural Employment Guarantee Act, (now is generally well-understood, but minimizing the renamed the Mahatma Gandhi Rural Employment potential avenues for their continued roles has only Guarantee Act) was passed in 2005. Other similar recently become a more explicit goal of SP policy Acts include the Right to Education Act (2009), design. While it is too early to say, even apparently the Scheduled Tribes and Other Traditional naïve blanket bans on certain actors in legislation Forest Dwellers (Recognition of Forest Rights) Act and/or guidelines (such as the ban on contractors (often called simply the "Tribal Rights Act", 2006) under MGNREGA) do appear to help. However, a while legislations such as for food security are on more comprehensive approach will require a more the anvil. thorough modernization of SP business processes. a consequence of the above is the increasingly Examples where such approaches already appear prominent role of communities and civil society to be making headway include greater reliance on in promoting more effective poverty reduction direct transfers to beneficiaries through banking outcomes from spending. In this respect, the and postal systems, and innovations in use of ICT strengthening of the "authorizing environment" in SP program delivery. for communities in SP service delivery in recent it is increasingly important to understand how years is encouraging. This is both cross-cutting, increased decentralization of responsibilities through reforms such as the Right to Information to panchayats generates different patterns of Act, and program specific, such as the anticipated contention, cooperation and collusion between role of social audits in MGNREGA, and new roles newly elected panchayat officials and traditional for community groups such as SHGs in delivery of loci of influence among administrators and some SP services (e.g., running Fair Price Shops). higher level politicians such as MLAs and MPs. An However, there is no guarantee of "trickle down" essential first step in promoting decentralization to citizens in terms of awareness of their emerging of SP service delivery as a tool for contestability entitlements. The role of NGOs, media and other and hence accountability will be getting a better actors in this respect can not be under-stated, as empirical understanding of the diversity and various political and administrative actors at local evolution of experience. This would include how level may not have strong incentives to promote the gradual increase in the role of panchayats is such citizen-based accountability mechanisms. proceeding (and what factors ­ such as limited Notable examples such as Mazdoor Kisan Shakti control of resources and very low capacity), Sangathan (MKSS) in Rajasthan demonstrate the and the extent to which panchayats effectively potential impacts of such partnerships. promote accountability in SP service delivery are captured by local social, political and administrative elites. e. The STRuCTuRe oF The a more nascent, but powerful, element in the politicaleconomyofSPreformisthepolicyshift RePoRT towards a rights based approach. Government the structure of the report is as follows: .It starts of India is increasingly operationalizing such with setting the context within which the SP system an approach in a number of areas through is operating, with respect to the evolution of poverty, legislation and specific policies and programs. For inequality and vulnerability in India in recent decades, instance, the Right to Information Act was passed and insights on the patterns of risks, shocks and coping in 2005 and mandates the government to release mechanisms of households. There is then a discussion timely information demanded by citizens. It has of key performance indicators of the major SP programs been widely hailed as one of the most important in India, both protective, promotional and preventive, drivers of governance reform and transparency in exploring coverage, adequacy of benefits, incidence India. The Right to Food and Right to Livelihood of benefits across the income distribution and social movements are led by civil society, but have groups, and determinants of the programs outcomes, managed to influence government so that the and in the case of preventive programs, looking at Indian Executive Summary xxx experience and strategies for expanding coverage options for SP programs, in particular the BPL (Below of different types of social insurance to unorganized Poverty Line) targeting method that acts as a backbone workers. The following chapters then explore different targeting mechanism for several key programs. aspects of program financing, program administration, The concluding chapter attempts to summarize and institutional arrangements that help to understand recommendations with respect both to the overall the observed outcomes. This is followed by a more SP system, and the specific aspects of policy and detailed discussion of targeting mechanisms and implementation discussed in individual chapters. xxx Social Protection for a Changing India: Volume II CHAPTER Poverty, Inequality and Vulnerability in India 1 Setting the Context for Social Protection Chapter­1 Poverty, Inequality and Vulnerability in India Setting the Context for Social Protection it is increasingly clear that the social protection vulnerability such as natural disasters continue to remain policy debate needs to broaden its focus to the important, new sources have emerged with the changing currently poor as well as those vulnerable to external environment. An increasingly important part of poverty. Fifty years ago, nearly half of India's population this story relates to shocks experienced by households, was poor, with high levels of chronic poverty. Today, both idiosyncratic and covariate. As a result, volatility this figure is closer to a quarter of the population. of household incomes and exposure to risks remains However, there are growing concerns that the figures a serious concern, even in better-off areas and even as of the official poor understate the true extent of the chronic poverty continues to decline. problem and Government is making serious attempts to this chapter provides an overview of poverty, revise these figures.13 A changing external environment inequality and vulnerability in recent years in india and improved living standards over time suggest that in order to provide the social context within which the focus of attention needs to move beyond the social protection programs operate. In doing so, the currently poor. chapter draws on a vast literature on poverty and on the in fact, there is considerable movement in and out analysis of poverty and inequality trends in the World of poverty in rural india and even households not Bank's India Poverty Assessment (2011) as well as analysis currently poor can be in danger of falling into of various rounds of the National Sample Surveys (NSS). poverty. Households are exposed to risk and insecurity The analysis of household level shocks and informal in many different ways. While perennial sources of coping mechanisms draws primarily on two household 13 TheExpertGrouponpovertyestimation,ledbyS.Tendulkarsubmitteditsreportwitharevisedmethodologyfortheestimationofpovertylinesandpovertyratesto thePlanningCommission.Thecommittee'srecommendationswereacceptedbyGovernmentshortlybeforethepublicationofthisreport.Asaresult,theanalysis anddiscussioninthischapterandelsewhereinthereportreliesonpreviousofficialpovertystatisticsandnotthoseproposedbytheTendulkarCommittee. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection surveys: (a) the 2006 Social Protection Survey (SPS) related) while richer households suffer from covering 1,356 households in the states of Karnataka, more covariate shocks such as crop failures and Madhya Pradesh and Orissa undertaken specially for livestock epidemics. this report; and (b) the 2005 Jharkhand Living Standards Coping strategies adopted by households in the Survey (JLSS) covering 2000 households in Jharkhand.14 face of shocks vary notably across states and several key points emerge from the analysis: income groups, and according to the nature of shock experienced. There is some evidence that Poverty has significantly reduced over time and these informal coping mechanisms may have gradualprogressonpovertyreductioncontinues. adverse long-term impacts and are unlikely to As a result, the officially poor are closer to one provide adequate protection. quarter of the population than the one half of an earlier period. However, a large proportion of the The structure of the chapter is as follows. The first population remains chronically poor. section presents poverty and inequality trends for However, inequality has risen since the 1990s. India. Section B shifts the focus to vulnerability and While economic inequality remains moderate poverty transitions. This section presents evidence by international standards, sustained growth on household shocks that often drive vulnerability has increased the focus on the inclusiveness of and poverty, as well as findings on informal coping growth and its distributional outcomes. mechanisms adopted by households in the face of shocks. Section C concludes with implications for social There are significant disparities in poverty, protection policies and programs. inequality and vulnerability across the country, and across groups. As a result, there not only remain a large number of chronically poor A. PoveRTy AnD inequAliTy 15 people, but the factors that underlie their poverty may in some cases be less amenable to india has made steady progress in poverty reduction a simple growth-driven approach. and about a quarter of the population is poor There is a growing urbanization of poverty. today. This improvement in living standards is Though the majority of India's poor continue to evident not just in official poverty estimates but also reside in rural areas, poverty rates in small and in community perceptions of poverty and well-being. medium towns are comparable, and even higher, However, inequality has started to rise during the than rural poverty rates. 1990s, particularly in urban areas. Structural inequalities also continue to persist, leading to the exclusion of Povertyandvulnerabilityislikelytobefarmore certain groups from the development process. Inter- widespread than the official poverty figures state disparities in poverty and human development suggest. This is both because communities' outcomes also remain significant. Thus, rural poverty perceptions of poverty and what it means to be is concentrated in the seven poorest states of Bihar, poor are typically higher than official estimates Chhattisgarh, Jharkhand, Madhya Pradesh, Orissa, and due to the large proportion of households Uttarakhand and Uttar Pradesh. At the same time, that remain vulnerable to poverty. urban poverty is starting to become more significant A significant majority of households experience and poverty rates in small and medium towns are serious crisis events. Health problems are the comparable or even surpassing rural poverty rates. most frequent idiosyncratic and natural calamities There is a growing perception that poverty is likely to be the most frequent covariate shock affecting more widespread than the official figures suggest and households. Poor households experience that a far larger proportion of the population is either more idiosyncratic shocks (primarily health- currently poor or in danger of falling into poverty. This 14 SeeDevetal.(2007)andBalachanderetal.(2009)fordetails. 15 This section draws extensively on the analysis of poverty and inequality trends by the World Bank's recent India Poverty Assessment report. See World Bank(2011). Social Protection for a Changing India: Volume II is both because communities' perceptions of poverty percentage of population with consumption expenditure are typically higher than official estimates and because too low to purchase the basket of commodities (food of the large proportion of households that remain and basic necessities) as measured by the poverty vulnerable to shocks and hence, to falling into poverty. line - has fallen in both rural and urban areas.16 In 2004/05, 28 percent of people in rural areas and 26 percent of people in urban areas lived below the poverty line, down (a) Trends in poverty and inequality, from 56 percent and 49 percent respectively in 1973/74. early 1970s to mid-2000s The depth and severity of poverty declined even faster, suggesting that incomes gains leading to poverty india has made steady progress in poverty reduction reduction were experienced even by those well below the since the 1970s (see figure 1.1). Poverty incidence ­ i.e., the poverty line. Consistent with the decline in consumption Figure 1.1: Evolution of poverty since the early 1970s Poverty incidence 60 % Population below the poverty line 50 40 30 20 1973-74 1977-78 1983-84 1987-88 1993-94 2004-05 Rural Urban National Poverty gap - depth Squared poverty gap - severity 20 10 Squared poverty gap (SPG) Poverty gap index (PGI) 15 10 5 5 0 0 Rural Urban Rural Urban 1973-74 1977-78 1983-84 1987-88 1993-94 2004-05 1973-74 1977-78 1983-84 1987-88 1993-94 2004-05 Notes: Based on uniform recall period consumption aggregates and official Planning Commission poverty lines. Source: GoI official estimates, World Bank (2011). 16 Officialestimatesofpovertyarebasedonthe"thick"roundsofthenationallyrepresentativeNationalSampleSurveys(NSS).Changesinthequestionnairedesign ofthe1999/2000NSSroundrenderedconsumptionmeasuresfromthisroundnon-comparablewithprevious(andlater)rounds.Thisgeneratedconsiderable debateonthepreciseextentofpovertyreductionduringthe1990s,thoughmostanalysisconfirmsacontinuedgradualreductioninpovertyeveninthisperiod.See DeatonandKozel(2005)forasummaryofthisdebate.ThemostrecentNSSround2004/05iscomparablewithpre-1999NSSrounds,enablinganalysisofpoverty trends.TheTendulkarCommittee'sre-estimationofthepovertylineandrevisedpovertyestimatesfor2004/05isthemostrecentcontributiontothisdebate. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection poverty, communities also report improvements in well- points) higher than that obtained using official poverty being or declining perceived poverty.17 lines. Spatial comparisons of poverty are more difficult and ambiguous using community-defined poverty however, poverty in india is perceived to be more lines as these typically measure relative poverty rather widespread than that suggested by official poverty than absolute poverty. Nonetheless, these findings do figures to date. Several studies show that communities suggest that until recently official poverty estimates did perceive the official poverty line threshold to be too not adequately reflect people's notions of what it means low. Participatory methods including techniques such to be poor (see Box 1.1 for a discussion of groups that as the ladder of life and community wealth rankings are regarded as particularly poor or vulnerable). have been used in a number of studies to elicit people's views about what it means to be poor and to specify in contrast to the steady reduction in poverty incidence, a community poverty line--the threshold beyond other indicators of human development suggest much which the community would consider households to slower improvement. For instance, the pace of progress no longer be poor.18 For instance, the Moving Out of in improving child nutrition in India has been slow, and Poverty study (World Bank 2008b) finds that two-thirds much slower than would be expected given India's or more of the sampled communities in the four states pace of GDP growth. In 2005/06, 43 percent of Indian covered in the study--West Bengal, Assam, Andhra children (aged below five years) were underweight, Pradesh, and Uttar Pradesh--felt that the official 48 percent stunted and 20 percent wasted (based on poverty lines were inadequate to meet basic needs. In WHO standards). India's under-nutrition figures are these states, the levels of poverty based on community- among the highest in the world, making slow progress defined poverty lines are much (10 to 20 percentage all the more troubling.19 While outcomes are decidedly Box 1.1: lost in translation ­ invisible populations under - served by public programs Certain groups may be particularly disadvantaged in some way or systematically excluded from the development process and/or public programs and services. For instance, low educational attainment, poor employment prospects and stigma mean that persons with disabilities and their households are notably worse off than average. While official estimates of disability (Census and NSS) are low (around 2 percent), there is growing evidence that people with disabilities comprise at least 5-8 percent of the Indian population. Even on the very conservative official definitions, about 6 percent of urban households reported having a person with a disability (PWD) compared to 8 percent of rural households in 2002. Estimates by WHO and others are far higher.1 While programs exist for PWDs (notably disability pensions ­ see Chapter 3), these require a certification of disability in excess of 40 percent severity, leading to many being excluded from government programs. In urban areas, this could also include the `floating' population, such as the homeless, street and pavement dwellers, commercial sex workers, and working children. Working children constituted about 12.5 percent of the child population (aged 5-14 years) in 1995, of which about 21 percent were in urban areas.2 Urban working children are more likely to work outside the household rather than as helpers in family work as in rural areas and are also more likely to be at risk of exploitation in an unfamiliar urban setting. Within urban centers, those living in slums (particularly non-notified or unrecognized slums), temporary settlements and "invisible slums" bound to specific industries (such as construction sites, leather, waste handling, jute, etc.) are likely to be poorer than the general population. Localized studies also suggest a high degree of homelessness in urban centers, with a large proportion of the homeless staying that way for long periods (over ten years).3 The information base on these groups is extremely weak. In fact, it is likely that the scale of urban poverty is underestimated due to the exclusion of the poorest groups in the conventional household surveys. These groups form a part of the "invisible population" that are largely unrecognized and un-served by public programs and services. Seasonal migrants are also among the most vulnerable and least visible groups as far as public policy is concerned. Migration is often also viewed as a problem and a potential threat to social and economic stability rather than an important livelihood and coping mechanism for the poor. In the destination villages or cities, migrants typically remain without an identity and hence are unable to claim state resources for education, health care, water and sanitation and other basic services. Sources: 1. World Bank (2009a). 2. NIEPA (2001). 3. Mander (2009). 17 Improvementsareseenintermsofanincreaseinincomesandpurchasingpower,ineducationandhealth,andanincreaseinpersonalfreedomandchoices (relatedtoreduceddependenceonpatronsinruralareasandgreaterenterpriseinurbanareas).BasedonfindingsreportedinPraxis(1999),Jayaramanand Lanjouw(1999),Krishna(2004,2006),Swaminathan(1995),KozelandParker(2005),WorldBank(2008b). 18 Seefootnote17forsources.SeealsoChapter8ontargetingofsocialprogramsforadiscussionoftheselocalindicatorsofpoverty. 19 InternationalInstituteofpopulationsciences(IIPS)andMacroInternational(2007). Social Protection for a Changing India: Volume II worse for those below the poverty line, the burden of expenditure was reversed in the 1990s as inequality under-nutrition is not confined to the consumption poor. started to rise. This increase was particularly sizeable within urban areas with the Gini coefficient rising from as a result, there are divergent views on the extent of 0.34 in 1993/94 (and 1983) to 0.38 in 2004/05 (see poverty in india. For instance, the Sengupta Commission Table 1.1). Given the growing divergence between rural on the unorganized sector estimated that about and urban consumption levels and rising urbanization, 77 percent of India's population lives on less than ` 20 this is reflected in a rise in national inequality. per day and could be considered poor. The World Bank's international poverty line of $1.25 per day suggests that poverty incidence in 2005 is as high as 42 percent Table 1.1: india inequality trends in 2005.20 More significantly, the recent re-estimation Rural urban national of poverty lines by the Tendulkar Committee implies 1983 0.32 0.34 0.32 a significant revision of poverty rates in the country to 1987-88 0.30 0.35 ­ about 41.8 percent in rural and 25.7 percent in urban 1993-94 0.29 0.34 0.30 areas. This proposed methodology and consequent 2004-05 0.30 0.38 0.33 poverty headcount rates have recently been accepted Notes: Gini coefficients calculated using NSS data. by government, these estimates of poverty would be Source: World Bank (2011), Himanshu (2007). used for policy and planning, particularly the allocation of fiscal resources across states (see Chapter 6). however, it is likely that the rise in inequality is understated. This underestimation is likely due to poor households typically have few resources measurement issues such as the understatement of and face greater insecurity. Analysis of various rural-urban gaps and increases in the incomes at the top- rounds of the National Sample Surveys suggests that, end of the distribution in the NSS household surveys. In predictably, households with high dependency ratios, addition, estimates of economic inequality in India are low educational attainment, low physical resources, likely to underestimate the true extent of inequality, such primarily dependent on casual labor, and belonging as that due to social stratification and segregation by to scheduled castes or tribes are more likely to be caste, religion, gender or other sources of exclusion (see poor.21 Poverty profiles developed using participatory below). Though the rise in inequality appears modest, techniques also identify similar correlates of poverty.22 popular perception is of rapidly increased inequality. This Thus, communities identify poor households as those perception is possibly driven by the observation that rich that are engaged primarily in casual work, have few Indians have benefitted enormously from the economic assets and/or high debts, are unable to meet basic needs boom of the 1990s. Though these movements in the very (particularly food), and are illiterate or poorly educated. top end (99th percentile and above) of the distribution An interesting point of departure from poverty profiles are unlikely to drive movements in the overall income derived from quantitative survey data is the emphasis distribution, such changes in the tails of the distribution on volatility of incomes and insecurity in these local are critical for societal perceptions of inequality. definitions of poverty. Thus, communities identify the poor as those facing insecurity of tenure (in urban areas) other structural inequalities continue to persist and insecurity of work and income; those that lack and gender, caste and tribal identity remain critical economic support (such as widows and elderly); and markers of well-being. Despite highly impressive those that have suffered health-related shocks especially growth rates and low levels of income inequality, India's illness, disability or death of the bread-winner. exclusionary social structures continue to place obstacles for several groups in access to key opportunities. While inequality in india is moderate by international welfare indicators for scheduled castes and scheduled standards but has increased during the 1990s. The tribes are improving, the gap between them and the secular trend in declining inequality of consumption general population is large and persistent.23 Today, 20 Downfrom60percentin1981(ChenandRavallion,2008). 21 SeealsoJayaramanandLanjouw(1999)forsimilarfindingsbasedonareviewofvillagestudies. 22 See,forexample,Krishna(2004,2006),Swaminathan(1995),WorldBank(2008b). CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection scheduled tribes still experience levels of poverty seen in Figure 1.2: Poverty rates - rural, small urban centers and the general population 20 years earlier; while scheduled large cities castes lag 10 years behind the general population. 50 Amongst the Scheduled Tribes and Scheduled Castes it is 40 often women who experience the greatest disadvantage. Though there are positive signs of dynamism within 30 the caste hierarchy, caste remains a potent indicator of 20 social status and gaps between SC/STs and general caste groups persist with respect to educational attainment, 10 earnings, poverty rates and other indicators of welfare. 0 STs in particular remain the most disadvantaged and 1983 1993-94 2004-05 the remoteness of tribal habitations creates problems Rural Small/medium towns Large cities for service delivery.24 Notes: Percentage of population with monthly per capita consumption expenditure below the poverty line, 2004/05. States sorted on overall urban poverty rate. (b) urban poverty Source: Staff estimates using NSS. World Bank (2011). though the majority of india's poor continue to reside in rural areas, there is a growing urbanization to nearly 30 percent of people living in small and of poverty in india.25 Urban and rural poverty rates medium towns. By 2004/05, the poverty rate in small/ have been converging in recent years. Though rural medium towns had surpassed that in rural areas. As the poverty rates remain higher, the depth and severity majority of urban population (about 70 percent) is also of poverty in urban areas is higher than in rural areas concentrated in these smaller urban agglomerations, (see Figure 1.1 above).26 At the same time, India's urban the bulk of India's urban poor (about 85 percent) are also population is growing faster than the rural population. to be found in these towns.28 In 2001, almost 28 percent of India's population the comparative picture of relative urban and rural resided in urban settlements, and this is expected to poverty across states is also of interest.29 Urban reach around 40 percent by 2020, of which about one- poverty is estimated to be higher than rural poverty third are likely to be slum-dwellers and squatters.27 in the more urbanized and industrialized southern This indicates a growing urbanization of poverty, albeit and western states, while rural poverty remains higher at a relatively slow rate compared to the historical in northern and eastern states (Figure 1.3). Some of experience of other developing countries. these states ­ such as Andhra Pradesh, Maharashtra, Tamil Nadu ­ also report the highest number of towns size matters: a larger proportion of the urban with slums in the 2001 Census. Approximately 23 percent poor are concentrated in small and medium towns of the urban population in these states lives in slums. compared to large metropolitan cities. Poverty rates in The national pattern of higher inequality in urban areas, large cities (with population of one million or more) are relative to rural areas, holds for all states. dramatically lower than that in small and medium towns (see Figure 1.2). Thus, in 2004/05, only about 15 percent this suggests that greater policy attention would need of people living in large cities were poor compared to be paid to the social dimensions of urbanization.30 23 Thesetermsarederivedfromagovernmentscheduleforpositivediscriminationandarethecommonlyusedinadministrativeandsurveydocuments.Inthis report,wefollowthisconventionanduseSCandST.Inrecentyears,DalitsandAdivasisareincreasinglybeingusedinstead. 24 WorldBank(2011). 25 Thisisacommonpatterninthedevelopingworld(seeRavallionetal.2007). 26 WorldBank(2011). 27 WorldBank2007. 28 WorldBank(2011). 29 However,thereareseriousconcernsabouttheurbantoruralpricedifferentialsimplicitintheofficialpovertylinesthathampercomparisonsbetweenruraland urbanpovertyrates. 30 SeetheIndiaUrbanPovertyReport(GovernmentofIndia2009),India2025:InputsforanUrbanStrategy(WorldBank2007)andPerspectivesonPovertyinIndia: StylizedFactsfromSurveyData(WorldBank2011)forarecentanalysisoftheseissues. Social Protection for a Changing India: Volume II Figure 1.3: Inter-state disparities in poverty (Rural, % below poverty line) 60 40 20 0 AI PU HP HA AP OTH GU KA RA KE AS MA UP MP TN WB BI OS JH UT CH 1983 1994 2005 (Urban, % below poverty line) 60 40 20 0 AI HP AS PU OTH HA WB AP RA MA GU KA KE OS TN UP MP BI JH UT CH Notes: Poverty rates for the newly created states of Chhattisgarh, Jharkhand and Uttarakhand only reported after states were formed. For comparability over time, poverty trends reported for Madhya Pradesh, Bihar, and Uttar Pradesh in all three years are for the divided states. States sorted in ascending order of 1983 poverty rates. Source: World Bank (2011) for poverty estimates. indian policy-makers have traditionally emphasized rural (c) Disparities in poverty and human poverty alleviation in the past and though increasing urbanization and concomitant pressures has brought development across states urban issues into focus, funding for urban poverty still lags there are wide disparities across states and spatially, behind the magnitude of the problem. The level of funding poverty is concentrated in the northern and central for anti-poverty programs in urban relative to rural areas states. Urban poverty rates vary dramatically across was only 1:35 in the late 1990s, against a share of urban states, ranging from a low of 3-4 percent in Assam and to rural poverty of 1:3.5.31 In recent years, Government Himachal Pradesh to over 40 percent in Chhattisgarh, of India has initiated a comprehensive national urban Madhya Pradesh and Orissa (see Figure 1.3). In rural reform through the Jawaharlal Nehru National Urban areas, the seven poorest states--Bihar, Chhattisgarh, Renewal Mission (JNNURM) and the Urban Infrastructure Jharkhand, Madhya Pradesh, Orissa, Uttarakhand and Development Scheme for Small and Medium Towns Uttar Pradesh--have poverty rates between 33 and (UIDSSMT). While social sectors are included as possible 47 percent. The least poor rural states such as Punjab, areas of support under these schemes, city development Haryana, Kerala, Himachal Pradesh and Andhra Pradesh plans - the first step for cities to access the grant resources have poverty rates that are a third or a quarter of those under JNNURM - typically have very little on social sectors. in the poorest states.32 Thus, in 1983, a group of the It is increasingly clear that an understanding of the four poorest, and most populous, states ­ (undivided) spatial distribution of urban poverty and urban-specific Bihar, (undivided) Madhya Pradesh, Orissa and strategies are needed for strengthening social protection, (undivided) Uttar Pradesh ­ accounted for 50 percent but equally that such distinct thinking from Governments of the rural poor. By 2004/05, this share had increased has been lacking to date. to 61 percent.33 31 LakdawalaCommitteeReportandNIUA(1993).Inaddition,publicserviceprovisioninurbanareashastraditionallyfocusedoneconomicserviceslikeinfrastructure, transport,housing,sanitationandwatersupplytotherelativeneglectofsocialprotection. 32 WorldBank(2011). 33 WorldBank(2008a). CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection these lagging states and even some regions in leading result, a large proportion of households, both poor and states (such as eastern maharashtra, northern non-poor, may be vulnerable to poverty even though karnataka and andhra pradesh) are being left out they may not be currently poor. Vulnerability is often of the growth process. In particular, poor economic thought of as "the existence and the extent of a threat performance (except for Madhya Pradesh) may have of poverty and destitution; the danger that a socially exacerbated these disparities in household consumption unacceptable level of wellbeing may materialize."35 If expenditures incomes and human development the poverty line is taken as the threshold below which outcomes. Households in these lagging regions lack the level of wellbeing falls below a socially acceptable access to adequate infrastructure, markets and social level, then vulnerability can be conceptualized as the services, with corresponding fewer opportunities threat of future poverty (see Box 1.2 for a description of approaches to assess vulnerability empirically).36 As a for employment and income diversification. While result, volatility of household incomes and exposure to permanent migration, particularly across state borders, risks remains a serious concern, even in better-off areas remains low, a common coping strategy is seasonal and even as chronic poverty continues to decline. migration. Localized studies suggest that seasonal migration is particularly high from some of these a far larger proportion of the population than those distressed areas that have emerged as significant currently poor is in danger of falling into poverty. exporters of unskilled labor to urban centers and to The literature on poverty dynamics and vulnerability relatively richer states (see Section B (a) below).34 in India suggests that there is considerable movement in and out of poverty. In other words, regardless of the current poverty status of a household, households may B. vulneRABiliTy fall into poverty after exposure to large, severe and/ or frequent shocks, especially when clustered close it is increasingly clear that the policy debate needs to to the poverty line. Figure 1.4 shows a large majority focus not just on the incidence of poverty, but also of rural households clustering within one standard on vulnerability to poverty. Households are exposed deviation of the poverty line, while urban and mega- to risk and insecurity in many different ways. As a urban households show greater dispersion (reflected Figure 1.4: A large proportion of the population is clustered around the poverty line 1993-94 2004-05 Density of per capita expenditures Density of per capita expenditures 1 1 .8 .8 .6 .6 .4 .4 .2 .2 0 0 -2 -1 0 1 2 3 4 5 6 -2 -1 0 1 2 3 4 5 6 Log consumption per capita, normalized by poverty line Log consumption per capita, normalized by poverty line Rural Urban Mega-urban Rural Urban Mega-urban Notes: Distribution of log monthly consumption per capita, normalized by the poverty line. Poverty lines in 1993/94: Rural areas = ` 205.84; Urban/ Mega-urban areas = ` 281.35. Poverty lines in 2004/05: Rural areas = ` 356.30; Urban/Mega-urban areas = ` 538.60. Source: Jalan and Murgai (2008); Staff estimates using NSS. 34 WorldBank(2008a). 35 Dercon(2005),pp.3. 36 RigorousanalyticalworkontheseissuesislimitedbythelackofpaneldatainIndiatotrackconsumptiondynamicsofhouseholdsovertime,withthenotable exceptionoftheICRISATsurveysandthelargerNCAERARIS-REDSpanelsurveys. 10 Social Protection for a Changing India: Volume II Box 1.2: Approaches to assess vulnerability empirically As the notion of exposure to risk and insecurity underlies all definitions of vulnerability, incorporating this into an empirical assessment of vulnerability usually requires panel data with detailed information on exposure and response to shocks. There are very few panel datasets in India that track consumption dynamics of households over time, with the notable exception of the ICRISAT surveys and the NCAER ARIS-REDS panel surveys. vulnerability as expected poverty (Chaudhuri 2000) This approach defines vulnerability as the probability of a household falling below a specified welfare threshold. An attraction of the approach is that the econometric methodology does not require panel data with observed consumption distributions over time. However, this must then necessarily rely on stringent assumptions on how shocks evolve over time and space; the model also does not incorporate covariate risk. This approach develops a model to predict realized welfare, typically measured as consumption, over a specified time horizon. These predictions are used to generate the distribution of expected consumption and the predicted probability of being poor in the future. Households with probability of future poverty above an arbitrarily chosen vulnerability threshold, analogous to the poverty line, are classified as vulnerable, yielding an incidence measure that is analogous to the incidence of poverty. vulnerability as low expected utility (ligon and Schechter 2003) This approach is similar to the first in that it also defines vulnerability in the context of expected future welfare. However, this approach also takes into account risk faced by households. Specifically, vulnerability is measured as the difference between a threshold level of certainty-equivalent consumption (analogous to the poverty line) and the household's expected utility of consumption. The attraction of this approach is that it decomposes overall vulnerability into a poverty term and two terms capturing covariate and idiosyncratic risk. As a result, it is possible to identify the link between vulnerability and poverty status (due to low resource base, poor returns to assets, unfavorable location, etc.) and between vulnerability and risk. One disadvantage lies in the fact that the unit of measurement of vulnerability are not easily amenable to policy dialogue. vulnerability as uninsured exposure to risk (Tesliuc and lindert 2002) This approach explicitly focuses on household's exposure and response to risk. Unlike the previous two approaches that are essentially assessments of exante vulnerability to falling below a specified welfare threshold, this approach assesses the ex post realization of a negative shock that results in a household deviating from its expected welfare and leading to a socially unacceptable level of welfare. There are drawbacks in this methodology in that changes in consumption are treated as symmetric regardless of the household's position in the income distribution. Similarly, the impact of positive and negative shocks are assumed to have symmetric effects. These have largely been addressed in subsequent extensions to the original model. However, the data requirements are stringent ­ this methodology ideally requires several waves in the panel as well as information on both income and consumption. vulnerability as the threat of future poverty (Calvo and Dercon 2005) More recent work has formulated a concept of vulnerability that implies both the likelihood of falling into poverty in the future (as in the first approach mentioned above) and the severity of poverty in that case. These measures estimate individual household-level vulnerability using an axiomatic approach. In all these approaches, there are problems in aggregating these household-level estimates of vulnerability that have not yet been addressed in the empirical literature. Sources: See also Christiaensen and Subbarao (2005), Dercon (2005) and Hoddinott and Quisumbing (2003) for a methodological review. in the higher inequality figures in Table 1.1).37 A further typically incorporate notions of insecurity and risk. This implication of this clustering relates to the difficulty qualitative work finds that variability of incomes and of achieving fine-tuned targeting in social protection coping with recurring shocks are a common feature, programs in India (see Chapter 8 for a discussion). In especially among poor households.38 In addition, this fact, as noted previously, poverty profiles based on exposure to risk leads households to adopt ex-ante community perceptions of what it means to be poor coping strategies that may mitigate risk but often 37 SeeWorldBank(2011)foradetailedcomparisonofconsumptionandothersocio-economiccharacteristicsfordifferentgroupsoftheexpendituredistribution. 38 SeeWorldBank(2008b)andothersourcesinfootnote17. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 11 imply lower returns, while ex-post coping mechanisms there is little evidence of widows being predominantly can often have adverse long-term impacts, sometimes in poor households in India (or only when economies leading to poverty traps. of scale are incorporated).42 Participatory research reveals that communities also consider widows as poverty and vulnerability are similar but distinct among the most vulnerable groups. In fact, most state concepts. Households that are currently non-poor can governments have long-standing social assistance be vulnerable to the risk of falling into poverty as a programs for widows (see Chapter 3 for a discussion). result of a crisis event. At the same time, poverty and vulnerability can reinforce each other and poverty can be a source of vulnerability.39 An application of the first (a) emerging sources of vulnerability method described in Box 1.2 on earlier rounds of NSS data for 1993/94 and 1999/2000 illustrates this point.40 a range of factors contribute to household level This method defines vulnerability as the probability vulnerability, some of which are macro- and some more of future poverty and estimates vulnerability as the household specific. Perennial sources of vulnerability, probability of being poor at least once in the next such as climatic conditions, natural disasters, poor three years. As noted in Box 1.2, this method relies on public health environment and household-level shocks, several stringent assumptions on how shocks evolve continue to be critical to household welfare. In recent over time and space. Several empirical studies have decades, new sources of vulnerability have also emerged adopted this methodology to assess vulnerability in which are both significant in themselves and as factors developing countries.41 Evidence from these studies which exacerbate perennial sources of household suggest that this methodology, though imperfect, is and community vulnerability. While by no means a useful in predicting the ordering of groups with respect comprehensive list, some of these old and new sources to vulnerability to poverty. This analysis is illustrative of vulnerability include: of the patterns of vulnerability across socio-economic Weakening traditional support networks groups in India in the 1990s. Declining common property resources in most cases, the profile of vulnerable households Natural disasters mirrors that of poor households. For instance, SC/ST Increasing stress on agriculture households were found to be among the poorest and the most vulnerable groups. There was, however, one notable High and rising indebtedness exception. Households headed by widows or with one Poor public health environment or more widowed members were found to be highly Rising urbanization vulnerable to poverty relative to other households, even though these households were not necessarily currently Increasing casualization of the labor market poorer. Widows represent about 6.5 percent of the total Financial, food and fuel crises female population in India. Despite legal provisions, Mobile populations widows find it difficult to inherit property, have limited freedom to remarry, and are forced by social custom to traditional informal support networks are being reside in their husband's village. They also receive little placed under strain in the face of factors such as economic support from the community or extended declining inter-generational co-residence. Several family, except possibly in the form of co-residence with studies stress the inadequacies of familial support for an adult son. As a result, the common perception is that elderly.43 Similarly, there is some evidence that social widow-headed households would tend to be poorer family, kinship and community networks are under than the general population. Empirically, however, strain. Rising migration and urbanization put further 39 Essentially,povertyistheexpostrealizationofastatewhilevulnerabilityistheexanteprobabilityofahouseholdbeinginthatstate.Whilepovertyisafunction ofthemeanlevelofconsumptionanditsdistribution,vulnerabilityisafunctionofpovertyaswellastheinter-temporalvarianceinconsumption. 40 SeeDutta(2006)fordetails. 41 Seeforexample,Chaudhuri(2000)forChina,IndonesiaandPhilippines;delNinnoetal.(2006)forPakistan. 42 DrèzeandSen(2002),DrèzeandSrinivasan(1997). 43 Agarwal(1990)andPal(2004). 12 Social Protection for a Changing India: Volume II stress on such networks. There is also some evidence earthquakes, waves/surges (such as the 2004 Tsunami) of a decline in the traditional jajmani or patron-client and wind storms. Historically, it was droughts and slow relations (e.g. between the 1960s and early 1980s in onset disasters like famines that caused the highest Rajasthan). death tolls in the region (including in India). In recent decades, however, it is these rapid onset disasters that common resources are being lost as increased demand lead to large casualties.48 Recent research suggests for land led to the conversion of forests and natural increasing frequency and intensity of natural disasters habitat into land used for agricultural, industrial such as cyclones, floods and droughts. A common or urban use. This has costs with respect to the loss response to such disasters is displacement and short- of common property resources (CPRs), imbalances in term migration. In addition, the strain on eco-system- the local ecosystems, and low yields as such marginal dependent livelihoods, such as fishing and subsistence land is usually not very productive. Several studies farming, could contribute to long-term migration. have documented a decline in the access to CPR in This adverse impact is likely to be higher for poor rural areas, with the proportion of total geographical communities, who tend to be located in relatively area under CPR declining between 1981 and 1991 in all high-risk areas rely on such livelihoods and also have states, by between 4 and 32 percent.44 In addition, an limited resources to cope with disasters.49 estimated 41 percent of India's forest cover has been degraded to some degree in the past several decades.45 there is increasing stress on agriculture in the face of At the same time, informal access rights to uncultivated rising seasonal climate variability, soil degradation, private land have also declined during this period in all depleting water table resources and breakdown in states primarily due to human and livestock population traditional systems of water management. There is pressure and increased demand for land.46 The effects of still a high dependence of agricultural on rainfall as only this decline and rising degradation of common property about 40 percent of net sown area is irrigated.50 Rising resources have disproportionate impacts on poor land fragmentation, unpredictable supply of rain or households with limited alternative resources, especially irrigation water can increase the vulnerability of resource- tribal communities. The latter depend on forests for constrained farmers and the near landless.51 However, livelihoods, seasonal subsistence and food security, though irrigation facilities have reduced variability in so that infringement of these traditional rights and the annual growth rate of foodgrain output, there is a institutions could substantially increase vulnerability.47 rising problem of depleting ground water resources as 62 percent of irrigated area comes from this source.52 india is particularly prone to natural disasters. Falling water tables, in turn, force farmers to deepen Between 1990 and 2007, 249 natural disasters, wells frequently. This is a costly and often unsuccessful leading to 102,313 deaths, were recorded in India. exercise if the groundwater level is very low, thereby This accounted for 37 percent of all natural disasters pushing farmers into long-term debt traps.53 (and 25 percent of deaths) in South Asia during this period. The most common types of disasters in India there is high and rising indebtedness among rural includes floods, wind storms and epidemics of various households. By 2002, more than a quarter of rural kinds. However, the highest death tolls are from households were in debt, increasing by four percentage 44 Chopra(2001).SeealsoJodha(1995)forRajasthan. 45 WorldBank(2005c). 46 Chopra(2001). 47 WorldBank(2011)andWorldBank(2005c).Jodha(1986)foundthattheshareofincomefromtheseresourcesaccountsforabout20percentoftotalannual incomeforpoorhouseholds(andonlyabout1-2percentfornon-poorhouseholds)inRajasthanintheearly(1980s). 48 Heltberg(2007)usingEM-DATdata. 49 CIESIN(2009),IPCC(2007). 50 MSSRFandWFP(2004). 51 Brugereetal.(2003).Thereissomeevidenceofgreaterinstabilityinfood-grainsoutputinthepost-GreenRevolutionperiod.SeeRaoetal.(1988)andRamaswami etal.(2003)forareview. 52 Ramaswamietal.(2003). 53 SeeWorldBank(2006a). CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 1 points since 1991 even as the share of credit from increased urbanization is likely to translate into institutional sources declined. As a result, the bulk increasing pressure on basic amenities and services, of rural debt is from non-institutional sources and particularly for the urban poor. India's urban population largely for household expenditure, particularly for poor is growing faster than the rural population, while rural households. Farmer households are even worse off, with and urban poverty rates are converging. In 2001, about 28 nearly half reporting indebtedness - over 80 percent percent of India's population resided in urban settlements. of these are marginal and small farmers owning two This is expected to reach around 40 percent by 2020, of hectares or less of land. In absolute numbers, five states ­ which about one-third are likely to be slum-dwellers and Andhra Pradesh (as high as 82 percent), Madhya Pradesh, squatters.59 Low-income urban households often cite Maharashtra, West Bengal and Uttar Pradesh - account factors such as insecurity of land tenure, inadequate or for half the indebted farmer households.54 Farmer lack of shelter and poor access to amenities as significant suicides in some parts of these states (notably AP and problems. Potentially weaker social networks, especially Maharashtra) have been widely reported in recent years, in non-slum settings, may make it harder for urban bringing the underlying issues of indebtedness, agrarian households to cope with shocks, insecurity and threat distress and vulnerability to the fore.55 In response, the of crime. The greater commercialization of goods and government announced a farmer debt waiver program services, implying a greater dependence on cash income, (for institutional debt) in February 2008. may further exacerbate vulnerability. Across the world, growing urbanization has usually been accompanied public health environment remains poor, implying by rising deprivation in urban areas.60 As mentioned greater exposure to health risks. Despite significant earlier, Indian policy-makers have traditionally paid little improvements in recent years, improving access to basic attention to urban poverty and vulnerability. amenities (including basic sanitation and adequate housing) remains a challenge. For example, three- the labor market is becoming increasingly informal. quarters of rural households reporting no toilet facilities In addition to the persisting dualism in the Indian labor even in 2005/06.56 Village-level surveys in four southern market (approximately 90 percent of the labor force is states provide strong evidence of capture of sanitation employed in the informal labor market), recent trends have services along political and caste lines and a complete lack contributed to increasing vulnerability at the household- of awareness of the public health benefits of sanitation.57 level. These include rising contractual and part-time Urban areas are further characterized by traffic congestion, subsidiary employment, slower wage growth and rising lack of green spaces and high levels of air and noise inequality during the last decade.61 Household incomes, pollution. It is the urban poor who are primarily affected particularly those of poor households, are closely linked as they are unable to afford good quality housing in to the labor market. As a result, the nature of the work better served and safer localities. As a result, vulnerability contract - with respect to regularity, security, benefits and to health risks remains high. A national study finds that remuneration - would be expected to have significant self-reported morbidity is lower among households that bearing on household vulnerability.62 In addition,several use clean fuels (compared to biomass fuels), have piped studies suggest that access to jobs, land and other indoor water and flush toilets. However, only 7 percent livelihood opportunities are strongly influenced by of Indian homes have access to all three amenities while gender, caste, religion and neighborhood networks and nearly two-thirds do not have any of these.58 are often mediated through local power brokers.63 54 NSSO(2005a). 55 SeealsoDeshpande(2002). 56 InternationalInstituteofpopulationsciences(IIPS)andMacroInternational(2007). 57 Banetal.(2008). 58 Desaietal.(2009). 59 Supritietal.(2002). 60 Haddadetal.(1999),Ravallion(2002). 61 SeeAhmedandNarain(2010),UnniandRaveendran(2007)andPais(2002)fordetailedanalysis. 62 SeeDrèzeetal.(1992)andPalandKynch(2000)forananalysisforagriculturalworkers.Similarly,Harrissetal.(1990)findthat53percentofhouseholdsdependent onirregularandunprotectedemploymentarepoorinurbanareas. 63 Seeforexample,Iversenetal.(2009),MunshiandRosenzweig(2006). 1 Social Protection for a Changing India: Volume II the world has recently witnessed a sequence of food, Table 1.2: india migration rates (%) fuel and financial crises. India has to some extent year Rural urban been relatively cushioned from the impact of these 1983 20.9 31.6 global crises, with the food and fuel crisis probably 1987/88 23.2 32.9 more important in the Indian context. Growth declined from an annual peak of about 9.7 percent in 2006-07 to 1993 22.8 30.7 5.8 percent in the fourth quarter of 2008-09. The Indian 1999/00 24.4 33.4 economy stabilized fairly quickly and growth has picked Source: NSSO (2001). up to 7.9 percent in the second quarter of 2009-10.64 Inter-state migration was much higher in certain states However, the impact of these crises was felt intensely, (Punjab, Haryana and Maharashtra) than the national at least in specific sectors and geographical areas. Thus, average. The largest exporters of rural migrants were in India, job losses occurred primarily in the gems and Uttar Pradesh, Madhya Pradesh, Bihar, West Bengal jewellery, transport and the auto industry, textiles (to a and Maharashtra, while Andhra Pradesh, Maharashtra more limited extent) and mining. Adjustment to such and Uttar Pradesh were the largest exporters of urban shocks could take several forms: open unemployment, migrants. However, the bulk of migration is within rural underemployment, sectoral shifts in employment, areas - only 5 percent of urban migrants are from rural declining wages and earnings, changes in labor force areas. This relatively slow pace of rural-urban migration participation, migration, and increased informalization. suggests that constraints to urban living could possibly For instance, anecdotal evidence indicates that workers be serving as disincentives to this process.67 in mining and in ancillary jobs suffered, leading to an increased dependence on the Mahatma Gandhi however, these figures are likely to understate the National Employment Guarantee Scheme (MGNREG) extent of seasonal and circular migration. a growing in some (mineral-rich) parts of Orissa.65 Both falling real number of micro-studies have established that wages and reallocation of labor are likely to come at a seasonal migration for employment is growing both high social cost, especially as these crises are likely to in terms of absolute numbers but also in relation affect groups that may not necessarily be poor and are to the size of the working population as a whole.68 typically not covered by safety net programs. The National Commission on Rural Labour (NCRL) puts the number of circular migrants in rural areas alone at migration and increasingly mobile populations: Over a around 10 million (including roughly 4.5 million inter- quarter of Indians were migrants in the early 2000s, more state migrants and 6 million intra-state migrants). so in urban than rural areas.66 There has been a marginal These localized studies also highlight the importance increase in the extent to which the Indian population of earnings from migration ­ ranging from a sixth (in has become more mobile over time (see Table 1.2). selected villages in Andhra Pradesh) to half (in selected For men, the predominant reason for migration was villages in Madhya Pradesh) to over 60 percent (in selected employment ­ 30 percent of male migrants in rural and villages in Rajasthan) of the annual income of migrant 52 percent in urban areas migrated in search of work. households.69 However, the quality of employment is In contrast, women tended to migrate for marriage or typically poor as these seasonal migrants are usually when the entire household moved. Among migrants, engaged in casual work in agriculture, construction, the overwhelming majority (61 percent) was intra- urban informal manufacturing and services.70 district migrants, with about a quarter migrating across districts and only 13 percent across states (again, this temporary migration for employment in rural areas figure was higher among urban migrants at 20 percent). is higher among poorer quintiles whereas the pattern 64 CSOstatistics(seehttp://mospi.nic.in/press_note_gdp_2ndqr_30nov09.pdf). 65 Das(2009). 66 Thefiguresare27percent(asperNSS1999/2000)and31percent(asperthe2001Census)ofthenationalpopulation. 67 KozelandParker(2005). 68 Breman(1985,1996),Rao(1994),Rogalyetal.(2001),DeshingkarandStart(2003),Haberfeldetal.(1999). 69 DeshingkarandStart(2003),Haberfeldetal.(1999). 70 Dev(2002),basedontheNationalCommissiononRuralLabour(NCRL)study. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 1 is more diffused in urban areas.71. This suggests that studies confirm the importance of poverty transitions. the poorest may not migrate as a minimum level of For instance, between 1975 and 1985, 88 percent of material assets may be required to make the investment households in the ICRISAT villages experienced poverty for migration.72 There is considerable debate on the in at least one year, even though only 22 percent contribution of rural-urban migration to urban poverty, constituted the chronic poor.75 Since then, living based on the common perception that migration is the standards in these villages have improved substantially last resort for the rural poor. This is certainly true for those so that 57 percent of households moved out of poverty who migrate as a result of "push" factors-- escaping debt between 1984 and 2001, while only 3 percent fell into traps, feuds or oppressive patron-client relationships. poverty.76 Similarly, the larger NCAER ARIS/REDS panel These people are likely to be among the poorest and of 250 villages indicates that 47 percent of households most disadvantaged in urban areas. They are also likely moved out of poverty between 1970/71 and 1981/82 to drive down urban wages and thus increase urban while another 26 percent fell into poverty. Half the poverty in the short run. However, there is another households in the sample remained chronically poor class of migrants who are driven by the "pull" factors during this decade.77 These findings are substantiated of more remunerative work or greater entrepreneurial by qualitative work across four states ­ Assam, Andhra opportunities. These migrants tend to be better off Pradesh, Uttar Pradesh and West Bengal.78 compared to the previous class and they often maintain the factors associated with emerging from poverty social and economic links with their villages that enable are distinctly different from factors associated them to cope with shocks better. These migrants can be with falling into poverty. Thus, factors such as viewed as rational actors responding to better economic income diversification, information, aspiration and incentives and could contribute to poverty reduction.73 empowerment are seen as critical for escaping poverty. these perennial and emerging sources of vulnerability On the flipside, health shocks, social obligations and remain critical for household welfare. Poor associated indebtedness are the most commonly cited and nearly-poor households are likely to be more reasons for falling into poverty. vulnerable to these factors, potentially resulting in households that succeed in escaping poverty are impoverishment. often characterized by a favorable initial resource base (with respect to assets, literacy, declining family (b) Poverty dynamics size) and a diversified portfolio of activities. Though diversification of economic activity into non-agricultural though a substantial subset of rural households activities might enable some households to escape from remains chronically poor, there is considerable poverty, the majority of agricultural labor and cultivator transition in and out of poverty in india.74. These households may not have the skills, knowledge and estimates of chronic versus transient poverty vary widely resources to take advantage of such opportunities depending on the location, time period, number of so that mobility out of agriculture remains low.79 The years and sample size of the panel data. However, all chronic poor, on the other hand, stay poor largely as a 71 NSSO(2001). 72 DeshingkarandStart(2003). 73 DeHaan(1999),Loughheadetal.(2001),KozelandParker(2005). 74 Rigorousanalyticalworkontheseissuesislimitedbythelackofpaneldatatotrackconsumptiondynamicsofhouseholdsovertime.Asaresult,thisliteratureis largelylimitedtomicrostudiesandresearchbasedonavailablepaneldata.ThisincludestheextensiveliteratureonthesixICRISATvillagesinsemi-aridruralSouth Indiaoriginallybetween1974and1985andrecentlyrevisitedbetween2000-2004(seeforexampleBadianietal.2006a;GaihaandImai2004;WalkerandRyan 1990)andtheNCAERARIS-REDSvillages(seeBhideandMehta2004;Gaiha1988)aswellasvillagestudies(seeJayaramanandLanjouw1999forareview).More recently,therehavealsobeenmicrostudiesbasedoncross-sectionaldatawithrecallquestions(seeforexampleKrishna(2004and2006).SeeMehtaandShah (2003)forareviewofsomeofthesestudies. 75 GaihaandDeolalikar(1993). 76 Badianietal.(2006a). 77 BhideandMehta(2004).SeeGaiha(1988)forsimilaranalysisofearlierroundsofthepanel. 78 WorldBank(2008b). 79 Drèzeetal.(1992),PalandKynch(2000). 1 Social Protection for a Changing India: Volume II result of their low resource base (including no or bad adopted by households is an important step towards quality land owned, low levels of skills, high dependency understanding the processes underlying vulnerability. burden, illiteracy) and other structural factors (including the incidence of shocks in rural india is high and social exclusion). These are typically casual agricultural many households, even non-poor households, are laborers, small agricultural households, and scheduled vulnerable to poverty as a result. Findings from caste and tribe households. The chronic poor also have surveys in four states ­ Jharkhand, Karnataka, Madhya a strong subsistence orientation which in turn implies Pradesh and Orissa - indicate that between one- and that they are likely to have lower risk bearing capacity two-thirds of rural households report significant crisis and are less inclined or able to undertake high-risk high- events in recent years. Households in rural India are return activities that allow them to escape poverty. Thus, vulnerable to idiosyncratic shocks that impact particular explicitly incorporating risk attitudes of households is households, such as ill health, death, loss of livestock, an important step towards understanding the processes etc., as well as covariate shocks that impact the entire underlying vulnerability to poverty.80 community, such as natural calamities, epidemics, etc.84 the single most important factor driving households health problems are the most frequent idiosyncratic into poverty is the negative impact of shocks. These and natural calamities the most frequent covariate are chiefly health and social obligations (e.g., marriages shock affecting households. Other common covariate and funerals) and severe crop shock, where the typical shocks include severe crop shock (arising from a major household response is greater indebtedness. Aggregate pest attack or bad seed quality) in all four states and shocks such as drought (including both rainfall and epidemics (human and livestock) for the three states irrigation failure) are also often associated with entry where this question was asked. Figure 1.5 presents into poverty.81 Qualitative work using life histories also the incidence of particular shocks across households. finds that coping with recurring shocks are a common In addition, there is some evidence that even health feature, especially among poor households.82 We turn to shocks may not necessarily be idiosyncratic - a small the incidence of shocks and informal coping mechanism study in rural Andhra Pradesh finds that 29 percent adopted by households next. (4 percent) of households report health (death) shocks as covariate, suggesting a link between health risks and (c) vulnerability to shocks83 community infrastructure. Similarly, the widespread prevalence of malaria in Orissa could explain the fact A range of sources contribute to household level that the incidence of idiosyncratic (health) shocks is vulnerability and can trigger shocks, some of them similar for poor and rich households. covariate impacting entire communities and others idiosyncratic impacting specific households. Uninsured unsurprisingly, poor households experience more shocks are costly to households in terms of loss of health shocks while richer households report income, asset depletion, or reduced consumption that shocks arising from crop failures and livestock can perpetuate poverty. In addition, the threat of these epidemics (see Table 1.3). In general, the incidence of . adverse events could lead households to resort to options covariate shocks is much higher for the top quartile that mitigate risk but at the cost of income gains. Thus, (i.e., the richest households). Only households with examining the role of shocks and the coping strategies some cultivable land are likely to be impacted by a 80 Forinstance,theMovingoutofPovertystudy(WorldBank2008b)emphasizestheroleofattitudestoriskandentrepreneurshipamongpoorhouseholdsthatare successfulinrisingoutofpoverty. 81 Badianietal.(2006b),WorldBank(2008b). 82 See,forexample,Praxis(1999),JayaramanandLanjouw(1999),WorldBank(2008b),Krishna(2004,2006),Swaminathan(1995),KozelandParker(2005). 83 Thissectiondrawsonprimarilyontwobackgroundpapersforthisreport:Devetal.(2007)andBalachanderetal.(2009).Thesedrawontwohouseholdsurveys: the2006SocialProtectionSurvey(SPS)whichincludedruralareasofOrissa,MadhyaPradeshandKarnatakaandthe2005JharkhandLivingStandardsSurvey (JLSS).Thedearthofpaneldataconstrainsanalysisoftheroleofshocksinexplainingvulnerabilityatthehouseholdlevel.Instead,werelyoncross-sectionaldata toexaminetheincidenceofshocks. 84 Shocksleadingtoconsumptionshortfallscanbecategorizedonthebasisoftheirlevelofimpact,frequencyandseverity.SeeDercon(2004)forareviewofthe literatureonrisksfacedbyruralhouseholdsindevelopingcountriesandthestrategiesusedbyhouseholdstocopewithconsumptionshortfallscausedbyshocks. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 1 Figure 1.5: Health problems and natural calamities are the most common shocks affecting households Orissa Madhya Pradesh Others, 3 Others, 7 Epidemic, 11 Epidemic, 5 Health, 20 Health, 28 Crop failure, 17 Crop failure, 13 Death, 5 Death, 8 Natural calamity, 39 Natural calamity, 58 Karnataka Jharkhand Others, 2 Epidemic, 5 Others, 2 Health, 19 Social obligations, 2 Crop failure, 8 Crop failure, 7 Death, 7 Health, 16 Natural calamity, 0 Death, 5 Natural calamity, 38 Source: Dev et al. (2007) for the first three states (2006 Social Protection Survey - SPS); Balachander et al. (2009) for Jharkhand (2005) Jharkhand Living Standards Survey ­ JLSS). Notes: 1. The data for Jharkhand is not strictly comparable for the other three states. The two surveys ­ JLSS and SPS ­ have differences in recall periods (three years preceding the survey in JLSS compared to one year in SPS); list of shocks (e.g., JLSS does not include a separate question on human or livestock epidemics, but does include one on social obligations unlike SPS); and comparable information on household welfare (SPS survey does not have household consumption or income information). 2. The figures in the graphs are the incidence of the particular shock in the state (i.e., not just among households that report a crisis event). negative crop shock. These households are also likely Table 1.3: high incidence of shocks among rural to be in higher quartiles. Landless households, on the households other hand, will be in the bottom quartile and will madhya experience only the indirect effects of such shocks karnataka Pradesh orissa through changes in the demand for labor. In contrast, q1 q4 q1 q4 q1 q4 the incidence of idiosyncratic shocks (primarily health Natural calamities 50 38 34 89 18 48 shocks) is higher for the bottom quartile (i.e., the poorest Crop failure 3 18 1 30 2 21 households), affecting them both directly through Health, death 42 23 31 17 35 37 out-of-pocket expenses and indirectly through loss of earnings. In the case of Orissa, the incidence of health Livestock epidemic 2 4 0 7 6 11 shocks is similar across quartiles, possibly due to higher Other 3 3 8 6 2 2 overall poverty and/or the widespread prevalence of Source: Dev et al. (2007). Notes: Households are grouped into quartiles on the basis of their wealth ranking (as information on household consumption malaria in the state. or income is not available). 1 Social Protection for a Changing India: Volume II The above findings are consistent with the literature a considerable financial burden on households and on sources of risk and shocks in India.85 These studies may even push households into poverty. Estimates of find that the incidence of shocks is high in rural India, the impoverishing effect of such expenditure in India the most common being health and natural disasters. range from an overall poverty increase of 3.5 percent Other significant sources of shocks include loss of to 6.6 percent in rural and 2.5 percent to 5 percent in assets, crop failure, loss of livelihoods (due to adverse urban areas (depending on the methodology and market conditions, environmental degradation, death survey used).91 There is also some evidence that a large of livestock, etc.) and social obligations (including portion of farmer debt also tends to be for meeting social marriages and funerals). While there is some evidence obligations including marriages and ceremonies.92 that urban households face similar sources of vulnerability,86 a significant gap in our knowledge relates to the household-level incidence of shocks in (d) informal coping mechanisms urban areas. a common coping strategy in response to shocks is such shocks can have an overwhelmingly negative borrowing; but poor households cope somewhat impact on household welfare, often leading to differently from better-off households. Besides indebtedness and poverty. For instance, in the ICRISAT borrowing, other common responses include increasing villages, crop failures result in a higher proportion of household labor supply (as in Karnataka and Madhya households falling into poverty, while consecutive crop Pradesh), asset depletion (as in Jharkhand and Orissa), shocks may result in chronic poverty.87 Communities and reliance on familial and other social networks also cite health shocks or chronic ill-health as one of (as in Jharkhand), depending on the local context the main reasons for becoming, and often for staying, (see Table 1.4). For instance, the relatively low reliance poor.88 Ill-health implies expenditure on treatment, on labor market adjustments in response to shocks potential loss of work and often indebtedness. The poor in Jharkhand and Orissa could be a reflection of the in particular have little access to formal insurance and relative lack of employment opportunities compared informal networks have only a limited ability to protect to Karnataka and Madhya Pradesh. However, poor households against health risks. The coverage of health households cope somewhat differently with shocks. insurance in India, particularly among rural households, Where labor supply adjustment is a significant coping is extremely low with less than 3 percent of households strategy, a greater proportion of households in the reporting any form of health insurance. Coverage is poorest quartile respond to shocks by increasing their negligible for the poorest at 0.4 percent compared to labor supply compared to the households in the richest 7 percent of households in the richest quintile.89 As a quartile (see Table 1.4). In contrast, richer households result, out-of-pocket private expenditure by households tend to resort to borrowing or asset depletion.93 Similarly, is estimated to be as high as 76 percent of total health ease of access to credit may influence the extent to which expenditure.90 This out-of-pocket expenditure imposes poor households can use borrowing as a strategy.94 For 85 Thesestudiesaretypicallybasedonlocalizedsurveys.ForselectedstudiesinruralareasseeSubbaraoetal.(2007)forruralBihar;GaihaandImai(2002)forthe ICRISATvillages;Duflo(2005)andKrishna(2004)forRajasthan;Krishna(2006)forAndhraPradesh;WorldBank(2008b)forAndhraPradesh,Assam,UttarPradesh andWestBengal. 86 Amis(1997),Benjamin(2000)andSupritietal.(2002)forBangaloreandNoponen(1991)forChennai. 87 GaihaandImai(2004). 88 Seeforexample,WorldBank(2008b). 89 Ajwad (2006). Note that this estimate is for 2004/05 before the introduction of the Rashtriya Swastha Bima Yojana aimed at providing subsidized health insurancetothepoor.SeeChapter5foradiscussion. 90 ReportedbyBermanetal.(forthcoming)usingWHO2008statistics.SeealsoGovernmentofIndia2005.Infact,healthexpenditureconstitutesabout6percent ofpercapitahouseholdexpenditure,butislikelytobehighlyvariableandlumpy(especiallyincaseofsurgery).SeeBermanetal.(forthcoming)forananalysisof healthexpenditureusingNSSdata. 91 SeeGargandKaran(2008),Gupta(2009)andBermanetal.(forthcoming). 92 WorldBank(2008b). 93 Similarly,inBihar,inadditiontothecopingmechanismsnotedabove,migrationisalsoanimportantstrategy,especiallyforpoorhouseholds(14percentofthe bottomquintileasopposedto7percentofthetopquintile)(Sen,2008). 94 BhandariandShresth(2003)alsofindthattheurbanpoorrelychieflyonborrowingtocopewithhealthexpenditures,whiletheurbanrichrelyonpastsavings. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 1 Table 1.4: The poor cope differently with shocks than the rich Jharkhand karnataka madhya Pradesh orissa All q1 q4 All q1 q4 All q1 q4 All q1 q4 Debt 27 18 22 34 33 35 17 13 25 25 27 26 Family support 35 34 36 3 4 0 5 3 4 9 11 8 Asset depletion 20 23 22 4 3 5 7 0 10 10 4 10 Labor adjustment 2 5 2 30 32 24 36 45 28 8 14 5 Other 6 7 7 5 2 13 12 21 7 12 6 15 None 11 13 11 25 26 23 23 18 27 36 38 36 See sources and notes to Figure 1.5. instance, in states like Karnataka and Orissa that have findings are also consistent with the literature on risk a strong presence of self-help groups, households in coping mechanisms in developing countries.96 Ex-ante the poorest and richest quartiles are equally reliant on coping mechanisms intended to reduce exposure borrowing. In Jharkhand and Madhya Pradesh, however, to risk include diversification of income sources and a greater proportion of households in the richest quartile adoption of low-risk (though potentially low-return) resort to borrowing. A large proportion of households activities. The most common ex-post mechanisms report no particular coping mechanism. One possible to cope with shocks and smooth consumption are explanation, at least for Jharkhand, is that this entails the following: (a) indebtedness often at high-interest cutting back on consumption (as these households also rates; (b) asset depletion in the form of spending report no financial cost incurred due to the shock).95 from savings, reducing grain stocks, selling livestock, land and/or jewelry; (c) labor market adjustment by the most common coping strategy in response to increasing number of hours worked and/or increased health shocks is reliance on social networks or participation by women and children; and (d) support borrowing, while the most common response to crop from family and/or community networks. failure is to adjust labor supply. Differences in the cost implications of these different types of shocks may there is some evidence that such informal coping be part of the reason for the differences in the strategy mechanisms are unlikely to provide adequate adopted. Small localized studies suggest that health protection. There is some evidence from the ICRISAT shocks are frequent but relative low cost compared to studies that consumption adjusts to income fluctuations shocks to livelihoods (such as crop failure). As a result, indicating a less than perfect insurance function, social networks of friends and relatives may be able to especially among the poor and landless.97 These help one another with low-cost shocks but may not be mechanisms provide some measure of insurance, mainly able to cope with high-cost shocks or with covariate in the face of idiosyncratic shocks. However, with the shocks that impact the entire group. Similarly, borrowing possible exception of high-interest debt, these strategies may not be feasible as the single coping strategy when are likely to break down in the face of covariate shocks. the cost implications of a shock are high. Additionally, there are possible additional long-term adverse implications of these strategies, e.g., withdrawal though the evidence presented here is based on data of children from education.98 Rising indebtedness in from four states, these findings are consistent with rural areas is a significant source of vulnerability among those of other studies in different parts of india, households, especially among marginal and small giving us some confidence that the broad patterns farmers. As much of the debt incurred is for unproductive outlined here might be more widely applicable. These purposes (health, household expenditure and social 95 SeealsoBehrmanandDeolalikar(1990). 96 SeeDercon(2004)forasurveyofcommonriskcopingmechanismsadoptedbyhouseholdsindevelopingcountries,andRamaswamietal.(2003)forreviewofthe literatureonIndiadrawingonICRISATandotherlocalizedsurveys.SeealsoDuflo(2005);Krishna(2004,2006)formorerecentstudies. 97 SeeMorduch(1999,2004);RavallionandChauduri(1997),andTownsend(1994).SeealsoDercon(2002). 98 JacobyandSkoufias(1998),Kochar(1995). 20 Social Protection for a Changing India: Volume II obligations among others) and is from non-institutional needs for health, education and other services, sources with very high rates of interest, it is likely to be one would expect that the relative balance unsustainable.99 As a result, even transitory shocks can between food and cash in SP interventions would have permanent effects (sometimes leading to poverty shift in favor of the latter. traps), especially if coping mechanism are inadequate. In addition, progress on some key human development outcomes, particularly nutrition, has been poor. At the same time, there is a significant C. imPliCATionS FoR SoCiAl share of spending on SP interventions. This begs PRoTeCTion PoliCy the obvious question on whether India is using this spending as effectively as it could to leverage the trends in poverty, inequality and vulnerability better human development outcomes in the way reviewed in this chapter have significant implications that developing countries internationally and in for social protection (sp) policies and programs. To the region have done in recent years. the extent that SP policies and spending are responsive to this evolution, this analysis gives rise to some critical what is the appropriate role of states versus the centre debates about potential trade-offs in policy making, with respect to sp policy and delivery? There remain wide including the following: inter-state disparities with respect to poverty and human development indicators. Poverty remains concentrated what is the appropriate mix of sp policies and programs? in the northern and central states ­ 61 percent of India's The analysis in this chapter indicates that while poverty rural poor were accounted for by (undivided) Bihar, has fallen steadily, deprivation along other dimensions (undivided) Madhya Pradesh, Orissa and (undivided) and as perceived by communities remains high, Uttar Pradesh. The human development index in these particularly among certain groups. At the same time, states is also lower than other regions in the country. households are increasingly vulnerable to the danger of As a result, there is an increasing divergence of state- falling into poverty. This indicates a need for a debate level needs, arising due to differences in demographics, on the appropriate mix of SP policies and programs that poverty incidence, vulnerability and other factors, as well could address these perennial and emerging concerns. as differences in institutions and capacities. However, Three issues emerge as significant: there remains a dissonance between the continued The primary function for SP policy and programs centralization of funding and (on paper, and, to a more continues to be poverty mitigation for the limited extent, in practice)decentralization to district and chronically poor. A key concern would be to gram panchayat levels that has occurred in recent years. improve design and/or implementation of existing At the same time, this very diversity of endowments programs to reach particularly excluded groups. and outcomes would seem to strengthen the case for An increased emphasis on SP policies and central financing from an equity perspective. These interventions that promote exante management issues suggest greater flexibility would be expected in of household risk would also be expected. This the policy/spending mix and program design features would typically imply an increased emphasis available to states under centrally-sponsored social on insurance-based interventions. In particular, protection programs (see Chapter 6 for a discussion on given the especially high importance of health financing and institutions, including the relative role of shocks as a source of vulnerability for the poorest states in SP policy and delivery). households, efforts to assist the poor in managing what does this mean for targeting of resources? these shocks would be desirable. The halving in absolute poverty rates since the 1970s The aggregate shift from food deficit to food imply a shift from SP policies which are untargeted surplus, and the extension of reliable food markets to policies and programs which attempt to focus into many more parts of the country over time, spending more on the still-significant share who remain combined with increased household spending in absolute poverty. At the same time, there is a high 99 NSSO(2005b),Duflo(2005). CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 21 number of the "near poor" and vulnerable households, move out of or fall into poverty in the period between with considerable churning around the poverty line. any form of identification exercise (e.g., the BPL This suggests a re-examination of the basic rationale censuses). Practically, there is a problem of database for targeting, particularly with respect to geographic management when updating is decentralized with no allocation of resources. In general, poverty estimates provision for feeding into a centralized database of at the state level (through NSS poverty estimates) and the updated list. See Chapter 8 for a discussion of the sub-state level (through the primary administrative complexities and challenges of targeting in India. household targeting tool ­ the Below Poverty Line or what are the emerging areas for sp intervention? BPL census) are used for fiscal allocation of resources The analysis also highlights several areas that do not across and within states. The basic principle is to increase currently have a significant place in India's SP system the concentration of anti-poverty spending in those but are increasingly demanding greater interventions states and regions which have fallen further behind in in light of India's changing economic conditions. Many terms of poverty reduction. The question is whether countries have faced or are facing similar challenges this purpose could be better served through other with respect to urbanizing and increasingly mobile means. For instance, improved fiscal allocations could populations. potentially be achieved through improved systems A gradual increase in programs and spending for more fine-tuned geographic targeting that take into account differences in endowments and targeted to the poor and vulnerable in urban opportunities across and within states.100 However, it areas, particularly small and medium towns, is not clear what the political economy ramifications couldbeexpectedtoincreaseovertimeasIndia would be on such potential over-concentration of urbanizes. The urban context suggests that anti-poverty spending combined with historically low it is not sufficient to simply transplant rural execution in some of the lagging states in terms of institutions and policies for effective service delivery in urban areas. Instead, given the specific funds utilization. challenges and opportunities, a general theme Similarly, the identification of households that of urban development should be to address how qualify as beneficiaries of various SP programs public programs can leverage poor households could be achieved through various targeting tools. in accessing opportunities for livelihoods and International experience suggests a mix of targeting human capital acquisition. tools, most commonly in combination with some form The increase in long duration migration rates of geographic targeting, is most effective at targeting since the 1980s combined with a continued households. However, this is not a straightforward reliance on seasonal migration as coping or exercise for India given the clustering of households livelihood strategies for households suggests around the poverty line and the similarity in various the need for a significant departure from the characteristics of poor and near poor households. current SP system. Essentially, even though However, there is sufficient diversity of experience people are becoming increasingly mobile, the across states and it would be useful to draw on SP system as it currently stands is designed for the lessons learnt from this range of experience a static population. In the destination villages of identifying poor households or areas through or cities, migrants typically remain without an alternative methods. Not all methods would be equally identity and hence are unable to draw on their relevant across states (for instance, community-based entitlements for SP programs or to claim state methods may not be appropriate in highly fragmented resources for education, health care, water and communities). Another important consideration is sanitation and other basic services. A critical to examine options for building dynamism into the challenge for SP policy is to design programs that targeting system. This is necessary both for conceptual offer portable benefits and can cater effectively and practical reasons. Conceptually, households may to this increasingly mobile population. 100 Therecentlyproposedmethodologyforthe2009BPLcensusbytheSaxenaCommitteeposesasimilarquestionbyindicatingthatnearlyallhouseholdsindistricts with80percentormoreofthepopulationbelowthepovertywouldbeconsideredpoor(seeChapter8foradetaileddiscussion). 22 Social Protection for a Changing India: Volume II the following chapters assess the current status and increased focus on backward districts in anti-poverty trends in the indian sp system to explore the extent to and area development programs; a partial but notable which it is effectively responding to the challenges recent shift in the spending mix towards cash benefits; presented in this chapter. The rest of this report and a sustained drive to provide subsidized health explores more fully some of these policy trade-offs insurance to the poor. However, some of the core paying particular attention to fiscal constraints, political challenges remain to be met and policy development economy considerations, administrative and capacity to this end is at best just beginning and in some cases constraints. The objective is to assess the most binding under-developed. These include, for example, the need constraints and examine the experience of delivering for new urban SP interventions; the need to further SP programs to derive lessons on striking these trade- expand insurance-based interventions for the poor; the offs most efficiently and equitably. Naturally, every desirability of making centrally-sponsored anti-poverty public SP system has inertia and political economy spending more flexible and responsive to increasingly challenges in reform. India is no exception, and indeed diverse needs of different states and social groups; the political economy of reform in the Indian SP system and the requirement for effective targeting tool(s) at is considerably more challenging than most. the household level. Many countries have faced or are the subsequent chapters find that certain elements facing similar policy issues and challenges as India. In of the reform needs outlined above have happened this context, the lessons of other countries in dealing or are in the process of happening. These include, with some of the trade-offs described above may be for example, the move to targeting of the Public informative, as would lessons within India from well- Distribution System, the largest single SP program; an performing to lagging states. CHAPTER-1: Poverty, Inequality and Vulnerability in India: Setting the Context for Social Protection 2 CHAPTER 2 The Social Protection Policy and Program Mix Chapter­2 The Social Protection Policy and Program Mix india has a vast array of anti-poverty programs number of poorer states, where css are often the only which have been subject to periodic ­ and only sp schemes of any significance. With the exception of partly successful ­ efforts at consolidation and/ PDS, a number of the core SP programs had their origins or convergence. There are hundreds of anti-poverty as central schemes in the 1970s.102 The momentum for programs in India, both Centrally Sponsored Schemes, major expansion of schemes and spending came from and state-specific programs.101 Defining precisely where concern about spreading the benefits of the Green "area development programs" ends and "anti-poverty Revolution, an issue taken up by Mrs. Gandhi under programs" start is not a science. This report therefore the slogan of Garibi Hatao (eliminate poverty). While focuses on what can be considered the core anti-poverty the names and some design elements of schemes have or social protection programs which operate nationally. changed since the 1970s, the key elements of the main While state-specific programs are often significant in SP programs have not changed dramatically, though spending terms, they are typically less important in some new national programs such as midday meals and lagging states, and in other cases supplement central social pensions, have added to the policy mix.103 schemes by expanding coverage with state resources. on the face of it, the policy mix in social protection centrally sponsored schemes (css) dominate the is impressive for a developing country. "On the books", social protection program mix, particularly in a there is an impressive mix of SP programs which includes 101 CSShavebeensubjecttoperiodicexpansionandconsolidation,increasingfrom65in1969to190in1978-79,beforebeingreducedto75in1980andincreasing againto201by1985.DuringtheNinthPlanperiod,thisroseto360,beforefallingagainby2005tojustover200(seeSaxena2006preparedasabackground papertothisreport). 102 ThoughpublicworksinIndiadatebacktothe12thcenturyfromleaderslikeSherShahSuri,andwereusedinthe19thcenturyunderfaminereliefcodes,they continuetobesignificantsourceofspendingparticularlyindrought-pronestates. 103 SeeSaxena(2006)ontheevolutionofRuralDevelopmentanti-povertyprograms(APPs)sincethe1970s. CHAPTER-2: The Social Protection Policy and Program Mix 2 traditional welfare/social assistance interventions and on rural areas. While a primary focus on ex post which support ex post poverty mitigation (e.g., PDS; coping interventions is understandable, the policy mix social pensions), those which seek to mitigate risks for of the SP system has only recently begun to respond to households ex ante (e.g., social insurance), those which the evolution of living standards and needs in directions try to promote movement out of poverty in the short that might have been expected. There are three areas and long runs (e.g., SGSY and school stipends), and where the evolution of SP programs and spending has hybrid tools (public works). The main SP schemes dealt been less and/or later than one might have expected with in this report are outlined in Table 2.1.104 in the face of significant poverty reduction and gradual looking at the policy mix across major programs in urbanization. They are: light of their objectives and spending rank (table 2.2), it insurance-based interventions remain in their is clear that the sp system remains largely focused on infancy in terms of coverage in the unorganized programs which deal ex post with poverty mitigation, sector,thoughtheRSBYprogramisasignificant Table 2.1: major central social protection schemes, 2009 Scheme Type Description Public Distribution System (PDS) Subsidized food and fuel distribution Subsidized wheat and rice, plus kerosene and sugar in most states. Level of subsidy varies according to whether APL, BPL, AAY or Annapurna household (see Chapter 3). Mahatma Gandhi National Self-targeted public works Unskilled and low skill public works. MGNREG guarantees 100 Rural Employment Guarantee days employment per rural HH per year in all districts. SGRY (MGNREG) and SGRY had aimed for 100 (non-guaranteed) days in rural districts, with a cash and food component. After 2006 SGRY was limited to non-MGNREG districts, but was discontinued in 2008 (see Chapter 4). Swarnajayanti Gram Swarozgar Targeted credit scheme for groups and Subsidized lending from banks to groups of BPL people, with Yojana (SGSY) some individuals allowance for some individual lending, and small APL share in groups (see Chapter 4). Indira Gandhi National Old Age Non-contributory social pensions Monthly cash benefits for BPL elderly (originally called NOAPS Pension Scheme (IGNOAPS) until renaming and expansion in 2007), and state schemes for widows and disabled people (since 2007 also included in the central program) (see Chapter 3). Indira Awaas Yojana (IAY) Targeted rural housing Subsidies for rural BPL with inadequate housing for housing construction (see Chapter 3). Midday meals School feeding program Hot meal for children in grades 1-8 in government and aided schools (see Chapter 4). Rashtriya Swasthiya Bima Subsidized health insurance Subsidized health insurance for hospitalization for BPL Yojana (RSBY) households in selected districts, rolling out to national coverage by 2013 (see Chapter 5). Aam Admi Bima Yojana (AABY) Life/disability/accident insurance for Free insurance covers natural death, disability and accident BPL for rural landless households (see Chapter 5). Other social insurance for Subsidized social insurance Variable by state, though central welfare funds for select unorganized workers sectors (e.g. beedi workers) (see Chapter 5). Specific urban anti-poverty Targeted urban housing (VAMBAY) and Housing construction and upgradation for slum dwellers, programs employment programs for the poor and wage and self-employment programs for unemployed or (SJSRY), now merged under JNNURM underemployed urban poor (see Chapter 3). 104 TheterminologyforcategorizationofprogramsfollowslargelytheSocialRiskManagement(SRM)framework:i.e.,riskmanagementstrategiesthatmitigatethe impactofashock,reducethechancesofashockoccurringandhelpcopewiththeimpact(seeHolzmannandJorgensen,2000).TheSRMcategorizationisquite closetocategorizationwidelyusedinIndia.Theconceptsof"promotional"and"protective"effectsonlivelihoodsstemfromDrèzeandSen(1989).Guhan(1994) addsathirdconcept,"preventative".Whileoftenblurredinpractice,thesetermshavedistinctanalyticalfeatures:"promotional"measuresaimtoimprovereal incomes;"preventative"measuresseektoavertdeprivation;and"protective"measuresprovidereliefagainstdeprivationtotheextentthattheothertwosetsof measuresfailtodoso.SeealsoGentilini(2005)foradiscussionoftheseconcepts. 2 Social Protection for a Changing India: Volume II Table 2.2: major SP programs by objectives and spending rank Program ex post risk coping ex ante risk mitigation Promotional Spending rank PDS ** * 1 MGNREG + SGRY ** * 2 SGSY *** 6 IAY *** 4 Social pensions *** 5 Midday meal * * * 3 RSBY *** 8 AABY *** 7 Soc. insurance unorganized ** * 9 Urban poverty programs ** * 10 Source: Bank staff estimates and Government of India budget documents, 2008/09 (revised estimate) positivestepinthisregardandnewlegislationin informal sector, this is an area where an increasing 2008commitstofurtherexpansionofcoverage.105 number of developing countries ­ including in The ex ante mitigation arm of the policy mix the South Asia region ­ have had successes in is therefore weak. Expanding informal sector promotional interventions which contribute to coverage of social insurance is a major challenge human capital development through conditional in many developing countries, and one that India cash transfers (CCTs). has had relatively limited success with until very other than PDS, SP interventions in urban areas recently, despite a series of central and state- are negligible and even more strongly biased specific schemes.106 Given the increased social towardsexpostriskcoping.Given the increasing importance of vulnerability in recent decades, importance of vulnerability in urban areas, this it would be reasonable to expect significantly raises questions about the appropriateness of greater attention to ex ante mitigation policies the current policy mix in urban areas. In resource over time. terms, Table 2.2 indicates that the situation has promotional interventions in the public sector not changed dramatically from the period of continue to receive relatively little emphasis, Lakdawala report in 1993, which found a rural to particularly given the continuing challenges in urban poverty ratio of 3.5:1, but a rural to urban improving human capital outcomes. Overall, anti-poverty scheme spending ratio of 35:1. While safety nets in India remain primarily "nets" it is not suggested that an appropriate ratio would rather than "ropes" or "ladders" which seek to be parity, the differences are nonetheless stark. promote sustained movement out of poverty.107 While designed with such an objective, SGSY another important aspect of the sp policy mix is the and predecessors such as the Integrated Rural balance between food-based and cash interventions. Development Program (IRDP) have generally Food remains a very important element of SP programs been one of the weaker performing SP programs, in India, with the PDS still the single largest program, and attract limited resources. The most important and the sharp recent increase in midday meal spending initiative in this area is the major expansion of (see below). In addition, up to 75 percent of SGRY midday meals in the early 2000s, the impacts compensation to workers could be in food (and the of which suggest significant under-exploited estimate for this alone was ` 6,750 crore in 2008/09), potential for promotional/demand side SP and smaller programs such as Annapurna are also food- interventions. Unlike social security for the based. An important countervailing shift in recent years 105 SeeChapter5.TherelevantlegislationisTheUnorganizedWorkers'SocialSecurityAct,2008. 106 SeeO'KeefeandPalacios(2006)forshortreviewofdevelopingcountryexperiencewithexpansion. 107 SeePritchettetal.(2002)andDFIDontheterminologicalandpracticaldistinctions. CHAPTER-2: The Social Protection Policy and Program Mix 2 is the policy reform from SGRY to MGNREG, which relies manner to realize the possible nutritional benefits of entirely on cash for wage payment to participants, and food programs. Equally, international evidence indicates the introduction and likely future expansion of RSBY. that most developing countries have transitioned away Overall, however, the share of food in total SP spending from a strong emphasis on food in their safety nets as remains high by international standards, and more so they reach lower-middle income levels. when one takes account of state level supplements on the last several years have seen several positive the central subsidy and coverage of distribution costs. developments in terms of evolution towards a policy With the 2009 proposals for Right to Food legislation, mix in sp interventions which better matches socio- the spending share on food could increase further. economic developments, and is somewhat closer to the debate in india on the relative merits of food what one would expect in a lower-middle income versus cash in safety net programs is at times intense, country. At the same time, it remains unclear what will with a strong emphasis on direct food security as be the impact of a Right to Food Act on this evolution. a key objective (rather than indirect food security The key trends include (see Chapter 6 on financing for through increasing household incomes). This is an details): (i) substantial increases in most SP programs important debate, but one on which hard evidence in recent years, with an overall increase in terms of for India is almost absent on the relative impacts on share of total central government spending; (ii) a real household welfare and food security of food versus reduction in the mid-2000s and then a sharp increase in cash.108 Despite this, the issue is assuming greater spending on PDS in response to the food and fuel crisis; profile with the commitment of the Government to a (iii) a sharp increase in midday meal spending (which is Right to Food Act, which would enshrine in legislation a step in leveraging improved human capital outcomes various commitments on food and nutrition programs, from SP interventions), and (iv) massive expansion of including PDS. Evidence on relative program targeting public works through MGNREG. While the initiatives and leakage performance in the following chapters to expand social security coverage for unorganized suggests that ­ whatever the in-principle merits of food sector workers are in their early days in terms of versus cash ­ India has struggled to operate its food- expenditure levels, they would be expected to grow based programs in a sufficiently efficient and transparent notably as RSBY is rolled-out nationally in coming years. 108 AnimportantstudywhichshouldshedlightonthisquestionisunderwaybyateamofIndianresearcherswithfundingfromtheAustraliangovernment.The studyseekstocomparetheeffecsofPDSwiththecash-basedMGNREGonhouseholdnutritionaswellasotheraspectsoftheprogramsinthreestatesofIndia. SeeJhaet.al.(2010)andGaihaet.al.(2010). 0 Social Protection for a Changing India: Volume II CHAPTER 3 Protective Programs Chapter­3 Protective Programs this chapter looks in detail at the design and the following sections look at program participation implementation of protective social protection and its determinants, targeting, and benefit incidence programs and suggests a range of reforms which for major protective sp programs, across states as would be desirable for india's poor. The programs well as major wealth and caste groupings. These are covered include, the Public Distribution System (PDS), classic social assistance programs that include Public social pensions for elderly, widows and disabled people, Distribution System (PDS), social pensions, and Indira and rural and urban housing schemes (Indira Awaas Yojana Awaas Yojana (IAY). In addition, it discusses operational and Valmiki Ambedkar Awas Yojana respectively) for the features which may be contributing to the observed poor. Overall, the need for reform appears to be most outcomes for several programs. acute in the PDS, but more in the nature of incremental improvements for social pensions. For targeted housing programs, in particular urban housing programs, the A. PuBliC DiSTRiBuTion SySTem options within the current paradigm are more in the nature of incremental reforms, but market-based (PDS) AnD AnTyoDAyA AnnA yoJAnA (AAy) 109 innovations in models of low cost housing for the urban poor suggest that experimentation with new paradigms would be well worth considering. The recommendations the public distribution system has been subject to for individual programs are discussed in Section C. significant policy changes over the past 15 years, 109 Thissectiondrawslargely on two background papers: (i) Ajwad (2006), whichusesHumanDevelopmentProfile of India surveydata for 2004/05 (seeAnnex 1 for details); and (ii) Umali-Deininger and Deininger (2006), which uses National Sample Survey (NSS) data, together with updates from 2004-05 NSS 61st round)databyC.Ravi.ItalsodrawsfromtheextensiveliteratureonthePDSbyIndianresearchers,includingRadhakrishnaandSubbarao(1997),Duttaand Ramaswami(2001),SwaminathanandMisra(2001),Devetal.(2004),JhaandSrinivasan(2001),aswellasthePlanningCommission2005. CHAPTER-3: Protective Programs moving from a universal entitlement scheme, first which was a key plank of Congress's electoral manifesto. to a geographically targeted supplemental subsidy, The key features of the current draft legislation are also and since 1997 to a targeted approach based on outlined in Box 3.1, together with some of the civil household welfare levels that is independent of society views on the consultation draft. location. Major developments in the evolution of PDS in the past 15 years, and key features of the Targeted Public Distribution System (TPDS) are outlined in (a) Access to PDS ­ ration card holding Box 3.1. The evolution is set to take another turn with the In relevance to the importance of Below Poverty Line commitment of the new Government to a Food Security (BPL) cards for PDS participation, and for beneficiary Act intended to enshrine in law as the right to food identification in several programs which are BPL Box 3.1: Policy evolution of the PDS The pre-reform PDS which was in operation till 1992, was criticized for its failure to effectively reach the poor, urban bias, substantial leakages, poor quality of grain supplied due to deficient inventory management and relaxed specification for procurement, lack of transparent and accountable delivery systems as well as negligible coverage and low off-take in states with high concentration of poor due to non-availability of stock. In 1992, GoI introduced the Revamped PDS (RPDS), which focused on giving higher subsidies primarily to drought prone, tribal, hilly and remote areas. Despite this, it is estimated that leakages at the national level during 1997-98 amounted to 31 percent for rice and 36 percent for wheata. The RPDS was replaced in mid-1997 by the Targeted PDS (TPDS), which used household poverty rather than location as the basis for targeting. The PDS is the largest safety net program in India (and indeed the world in terms of population covered), and operates by providing a price subsidy to consumers for essential commodities. The most important of these are rice and wheat.b TPDS supplies these commodities at prices below the Food Corporations of India's (FCI) "economic cost", which equals the sum of FCI's procurement, storage, and distribution costs. FCI is the implementing arm of the Government of India's food grain policy. TPDS has a 2-tiered pricing structure for Below Poverty line (BPL) and Above Poverty Line (APL) households. In addition, GoI introduced the Antyodaya Anna Yojana (AAY) in December 2000, as a sub-scheme to benefit the poorest of the poor. AAY provides a larger price subsidy that received by BPL households. Each AAY household was made eligible for 25 kg of food grain per month at a Common Issue Price (CIP) of ` 2 per kg for wheat and ` 3 per kg for rice. In April 2002, the AAY and BPL household food grain allocation was increased to 35 kg per month. The AAY scheme has since been expanded to cover BPL households headed by widows or terminally ill, disabled or elderly with no assured means of support (2003-04) and all households at risk of hunger (2004-05). The total number of households identified under AAY was 243 lakhs in 2009. The total number of BPL households earmarked for PDS coverage is 652 lakhs ­ though several states cover a significantly higher number than the centrally determined number of BPL households. At inception, TPDS targeted the price subsidy exclusively to the poor, though this was later adjusted to provide a small subsidy to APL households. The allocation of rice and wheat stocks to APL households is linked to average offtake figures of previous years. The shift to the TPDS was a significant milestone in GoI's food security and social protection strategies. TPDS is operated under the joint responsibility of Central and State Governments. The Central Government is responsible for procurement, storage, transportation and bulk allocation of food grains, while state governments are responsible for distribution to consumers through the network of Fair Price Shops (FPS), numbering nearly 500,000 nationally. Operational responsibilities including allocation within the State, identification of families below poverty line, issue of ration cards, supervision and monitoring the functioning of FPS, rest with the State Governments. Under TPDS, the states were requested to issue food-grains at a difference of not more than 50 paise per kg over and above the Central Issue Price (CIP) for BPL families. The CIP is the price at which TPDS food grains are issued or "sold" to state governments. More recently, however, States have been given flexibility in fixing the retail issue price for TPDS food grains, except for AAY. The evolution of PDS is set to take a major turn with the proposed Food Security Bill. In the 2009 electoral campaign, Congress committed to right to food legislation, promising 25 kg of rice or wheat per month to all BPL households at ` 3 per kg, as well as subsidized community kitchens in all cities for homeless people and migrants. The main features of the Concept Note posted by the Ministry of Consumer Affairs in July 2009 for consultation are: the number of BPL families is based on poverty figures of the Planning Commission. While currently based on the 1993-94 poverty estimates, there are discussions for it to be based on 2004-05 estimates. the note acknowledges that there are state specific variations in the number of BPL ration cards issued, variations in issue price as well that in entitlements (in quantity and commodities) provided. It states that this variation should be eliminated and there should be a central order that should bind the states in the above issues. it proposes that there should be no sub-categorization of the BPL households into the AAY households. other schemes such as Annapurna, schemes for welfare institutions and hostels etc. may not be continued separately to avoid multiplicity. TPDS will cover beneficiaries of these programs using BPL cards. Social Protection for a Changing India: Volume II Box 3.1: Policy evolution of the PDS (Contd...) The BPL ration cards will be valid for 5 years, after which it automatically expires. the note estimates that with a central issue price of ` 3/kg of rice and wheat, the food subsidy bill would be ` 40,380 crores and annual requirement of food grains would be 277 lakh tonnes. states would pay food security allowance in case of failure to offer entitled quantities to eligible BPL families through their bank/ post office accounts. for ensuring transparency and efficiency, it is mandatory for all state governments to computerize/digitize ration card database as well as TPDS transactions. states need to set up grievance redress mechanisms. food security tribunals should be set up at the tashil/taluka level for expedite adjudication of case against violation of this law. Reactions to the consultation draft from civil society have been rather critical. Several critics (e.g. Khera (2009); Himanshu (2009) point to the fact that AAY households would have lower entitlements and that in eight major states the current price of BPL grains is below ` 3/kg. They also note the much wider coverage of subsidized rations in southern states already. On the face of it, there does not appear to be restrictions on top-ups from state resources, but this is not clear and is important in assessing the validity of these arguments. Beyond that, critics argue that the right to food goes well beyond PDS rations and should include other entitlements such as nutritional support to children, social assistance, and special provision for urban areas, in line with Supreme Court orders in the right to food cases. The Right to Food campaign has alternative legislation framed which incorporates this wider range of entitlements. On the other side of the argument, others argue just as strongly for transforming PDS from a food-based program to a direct cash transfer (e.g. Kapur, Mukhopadhyay and Subramanian (2008); Panagariya (2008). The 2010 Economic Survey of India proposes a move to direct subsidies to households through food coupons, with a lumpsum entitlement that can be encashed in a Fair Price Shop of their choice. There are also mixed views on the feasibility of the required grain procurement in drought periods such as presently being experienced. Sources: Department of Food and Public Distribution (http://fcamin.nic.in), GoI Economic Survey (2010), Panagariya (2008), Kapur, Mukhopadhyay and Subramanian (2008), Khera (2009); Himanshu (2009), www.righttofoodindia.org and http://www.righttofoodindia.org/data/concept_note_on_rtf_act_ food_ministry_040709.pdf for the concept note from the Department of Food and Public Distribution. Notes: a. Ministry of Consumer Affairs, Food and Public Distribution (2000). b. The program also supplies sugar nationally, and other commodities such as edible oils and coarse grains in some states. Kerosene is also provided through Fair Price Shops (FPS), but is not the focus of this chapter. A useful paper on LPG and kerosene subsidies and usage under PDS is Gangopadhyay et al. (2005), which finds the LPG subsidy to be poorly targeted, while kerosene is notably better targeted. targeted, Table 3.1 presents findings on ration card better-off states are in better position to access possession. These findings are complemented by more subsidized rations. detailed distributional analysis of ration card holdings therearevariationsacrossstatesintheshareof in Chapter 8 on targeting of social protection programs. households possessing an AAY card relative to Several observations emerge: the share with BPL cards, confirming differential states having among the highest poverty rates efforts to identify their centrally-allotted AAY arealsothosewiththehighestratesofnoration quotas.The reasons for this varies, with states like card,with more than one third of households in Andhra Pradesh having low AAY holding due to Jharkhand not having cards, a third of households very high BPL holding, while others such as Bihar in Chhattisgarh, and around 30 percent of and Assam had not done the AAY identification households in Bihar. process thoroughly (e.g., as of late 2005, Bihar had possession of BPL/AAY cards across states also identified only 40 percent of its AAY quota, against shows a clear pattern in terms of lagging states a national average of close to 75 percent).110 reporting BPL/AAY card rates which are in most looking at rural and urban areas separately by casesbelowtheirpovertyrates,andricherstates state, holding of any ration card is somewhat havingrateswellabovetheirpovertyrates.There higher in rural than urban areas, at 85 and just are exceptions to this general pattern, such as under 79 percent respectively.111 The difference Orissa and Rajasthan, but in general people in may in part reflect challenges in establishing 110 SeeSaxena(2006)forstate-wisedataasoflate2005onstateidentificationofAAYrelativetoquota. 111 ThismoredetailedanalysisbyurbanandruralareasisinAjwad(2006). CHAPTER-3: Protective Programs Table 3.1: Possession of ration cards by type and state, 2004/05 (% of households) State Any card BPl card APl card AAy card Andhra Pradesh 76.8 64.5 11.7 0.7 Assam 85.6 25 59.8 0.7 Bihar 67.4 33.5 32.4 1.4 Chattisgarh 69.4 37 23.1 9.3 Delhi 75.2 20.6 54.5 0.1 Gujarat 84.2 39.5 44.4 0.2 Haryana 93.6 16.4 76.4 0.7 Himachal Pradesh 96.6 17.3 72.4 6.9 Jammu & Kashmir 87.7 27.8 58.9 1 Jharkhand 62.1 37 22.7 2.4 Karnataka 72.4 52.8 16.6 3.1 Kerala 94.7 36 58.7 0 Maharashtra 90 25.8 61.9 2.3 Madhya Pradesh 75.8 26.6 44.9 4.2 Orissa 78 51.6 23.6 2.9 Punjab 90 4.6 85.3 0 Rajasthan 95.6 22.7 68.9 4 Tamil Nadu 93.8 47.7 45.9 0.1 Uttar Pradesh 83.2 18 60.2 5 Uttarakhand 92.1 27.9 57 7.3 West Bengal 94.1 24.9 66.7 2.6 Other 74.7 28.2 45.3 1.2 Total 83.3 33.7 47.1 2.5 Source: Ajwad (2006) based on 2005 IHDS data. proofofresidencyinurbanareasamongmigrant 2004-05 are broadly comparable with analysis from the households. More noticeable is the share of 2005 IHDS data (both are nationally representative), households with BPL or AAY cards, which is around which found that 26 percent of the households purchase 42 percent in rural and 22 percent in urban areas. PDS grains in the same year, and also found similar patterns across the states.112 (b) household purchases of PDS grains however, the more interesting part of the story is the difference in trends between rural and urban areas in purchasing pds grains at the national level, across the decade. For urban areas, overall coverage between one fifth and one quarter of households in 2004-05 was half that of 1993-94, with the decline in (number varies depending on the data source) coverage accelerating in the second half of the period. reported purchasing pds grains in the mid-2000s, which In contrast, in rural areas, while there was a sharp rise in represents a significant drop since the late 1990s. This coverage followed by a decline over the period, coverage can be seen in Table 3.2 using National Sample Survey was still one third higher than it had been in the early (NSS) data from various rounds, which shows that there 1990s. In relative terms, rural coverage went from only was a significant increase in household-level offtake or 65 percent of urban in 1993-94 to over 180 percent by usage from the early to late 1990s, followed by a reversal 2004-05. For a program which had been subject to urban in overall access by the mid-2000s. The NSS figures for bias, this is indeed a remarkable turnaround.113 112 SeeAnnex1andAjwad(2006)fordetailsonIHDSdataanalysis.SeeBox3.2forterminologyspecifictothePDSliterature. 113 SeeHowesandJha(1992)onurbanbiasinthePDS,priortoTPDS. Social Protection for a Changing India: Volume II Table 3.2: household-level offtake of PDS grains - All households and rural/urban, 1993/94, 1999/00 and 2004/05 (% hh) All households Rural households urban households State 1993-94 1999-00 2004-05 1993-94 1999-00 2004-05 1993-94 1999-00 2004-05 Andhra Pradesh 45.7 55.9 54.6 47 63.3 62.5 41.9 37.2 31.6 Assam 15.1 32.9 8.4 13.7 34.9 9.2 25.9 20 2.5 Bihar* 3 8.4 2.8 2.8 8.7 2.9 4.2 6.2 2.4 Gujarat 29.4 37.2 24 32.4 46.5 32.6 23.7 19.2 7.9 Haryana 3.3 2.9 4.4 2.7 3.2 4.1 4.8 2.1 5.3 Karnataka 44.8 63.8 47.1 41.7 71.2 59.8 51.8 46.5 21.8 Kerala 65.2 72.6 36.7 62.2 75 39.3 74.3 66.1 28.6 Maharashtra 26.1 36.3 21.1 27.3 48.7 30.5 24.4 17.3 7.9 Madhya Pradesh* 8.5 16.7 21.4 6.7 19 24 14.1 9.2 12.1 Orissa 4.7 47.7 19.4 2.9 50.2 21.6 16.6 35.5 6.7 Punjab 1.1 1.8 0.4 0.8 1.7 0.3 1.5 1.9 0.7 Rajasthan 12.6 5.6 10.1 13.4 6.3 12.8 10.2 3.3 2 Tamil Nadu 55 71.4 68.2 52 77.5 80 60.2 60.5 49.8 Uttar Pradesh* 2.9 8.3 6.5 2.6 9.2 7.5 5 5.2 3 West Bengal 14.3 20.8 12.7 7.6 20.8 15.1 32.3 21.1 6.6 Others (incl. UTs) 36.1 36.9 23.2 33.3 39.3 34.9 39.3 33.7 10.6 All India 22.6 31.6 23.3 19.9 34.1 26.6 30.6 24.9 14.7 Source: Umali-Deininger and Deininger 2006, estimated from NSS 1993-94 and 1999-2000, and report team estimates from the NSS 2004-05. Notes: For comparability, 2004-05 figures for Bihar, Madhya Pradesh and Uttar Pradesh are for unsplit states. several factors are likely to have contributed to this previous month for 1993-94, 1999-00 and 2004-05.114 trend. The first is obviously the targeting of PDS, and The findings are supported by an earlier study of PDS how that has played out in the early 2000s. In addition, food grains availability which found that Tamil Nadu there are a range of factors, including grain quality issues, and Andhra Pradesh were the only states where more higher transactions costs associated with procuring than half of PDS users reported "regular" purchases of TPDS grains, increased Government of India (GoI) PDS grains, while the figure for Uttar Pradesh and Bihar distribution of grains through other welfare schemes as was 1 percent.115 This is probably due to the fact that well as good rice and wheat harvest in 2003-04 that have Tamil Nadu and Andhra Pradesh are states where more likely influenced household participation rates in substantial state fiscal resources supplement the 2004-05. The factor which would require more detailed national subsidy in order to support state policies of research to further understand this trend is the impact more widespread access. of rising incomes in this period and possible impacts on an important explanatory factor in cross-state preferences for lower quality PDS grains. variations (and aggregated demand) in drawing of pds household offtake or usage in some states ­ most grains by households is likely to be the gap between notably bihar, uttar pradesh, rajasthan, and market prices of grains and the price of pds grains madhya pradesh - has remained consistently very for different ration card categories. The importance low, particularly in several very poor states. This of this factor has been highlighted in previous analyses can also be seen in Table 3.2, which reports the share of PDS and confirmed in the Planning Commission of household by state that purchased PDS grains in the evaluation of the program.116 The Planning Commission 114 TheNSSfindingsarebroadlyvalidatedfromthe2005IHDSdata.SeeAjwad(2006). 115 ReportedinSaxena(2006). 116 SeeforexampleRadakrishnanandSubbarao(1997)andPlanningCommission2005. CHAPTER-3: Protective Programs analysis indicates that in most states (with exception of majority of the poorest households were not Tamil Nadu, Himachal Pradesh and several North Eastern accessing pds grains in any of the years surveyed. states) the BPL grain price moved close to the market Even with the major reform of PDS, it has continued price between 1999 and 2001, contributing to a general to fall substantially short of its stated aim of providing demand disincentive. The report also notes that states subsidized grain to the poor. It is difficult to disentangle such as Bihar and Jharkhand with lower BPL offtake the mixture of supply and demand factors in this were those with among the lowest price differential situation, but the end result is clear and has been between BPL and market grain prices, though many consistent over time. other inefficiencies are at work in those states which reduce offtake. Given the sharp increase in open market the differentials in access to pds grains between the prices of grains in recent years, it will be important to rich and poor are far more pronounced in urban see how the increased PDS/market prices differential for than rural areas. As can be seen in Figure 3.2, the both BPL and APL has affected household offtake. variation in access to PDS grains across the distribution is far greater in urban areas, in addition to the much lower overall coverage rate. While the ratio of access (c) Distributional patterns of PDS of poorest to richest quintiles in rural areas was around grain offtake by households 2.6:1, in urban areas this was 8.9:1, probably reflecting much lower interest among better off people in urban with respect to coverage of pds across the income areas for the lower quality grain supplied through distribution, there was substantial improvement in the PDS. targeting performance with the introduction of tpds. This can be seen in Figure 3.1. All households except the looking at the distributional patterns of pds offtake richest quintile had higher coverage rates in 1999-00 across states for 2004-05, there are not clear patterns than 1993-94, while only the bottom two quintiles had across groups of states in the coverage rates among sustained higher coverage by 2004-05. The targeting the poorest relative to the richest. This can be seen in reform of PDS in the late 1990s was thus unusual. The Table 3.3, nonetheless, a few groupings emerge: distributional pattern of public spending improved the states which did very poorly on average even though coverage among the non-poor overall offtake/usageratestendtodosovirtuallyacross increased, except among the richest. By 2004-05, this thedistribution,pointingtoageneralized"system had become a more typical targeting outcome of lower failure".They include Bihar (including Jharkhand), coverage rates among all the non-poor. Uttar Pradesh, Rajasthan, and West Bengal. From survey-based data on relativities between PDS despite the improvement in coverage among the and market prices of rice and wheat, this appears poorest after the introduction of tpds, the large to affect demand, with relative FPS rice and wheat prices in this group of states notably below all- Figure 3.1: Household-level offtake of PDS grains by India relativities.117 quintile, various years 40 Figure 3.2: Household-level offtake of PDS grains by 30 quintile and rural/urban, 2004/05 Q1 40 Q1 Q2 20 Q2 30 Q3 10 20 Q3 Q4 10 Q4 0 Q5 1993/94 1999/00 2004/05 Q5 0 Source: Umali-Deininger and Deininger 2006, estimated from NSS for Rural Urban 1993-94 and 1999-2000, and report team estimates from NSS for 2004-05. Source: Report team estimates from NSS for 2004-05. 117 SeeAjwad(2006)fordetailedFPStomarketpricesbystate. Social Protection for a Changing India: Volume II Table 3.3: household-level offtake of PDS grain by quintile by state, 2004/05 quintiles States q1 q2 q3 q4 q5 Poor/Rich ratio Andhra Pradesh 66.3 68.5 58.8 48.8 30.3 1.7 Assam 18.5 11.4 5.2 3.0 3.7 4.4 Bihar 5.8 2.5 2.9 1.8 1.3 2.7 Gujarat 42.3 30.0 23.3 18.8 5.6 3.0 Haryana 10.2 6.6 3.3 1.6 0.5 8.1 Himachal Pradesh 63.3 52.1 48.3 45.5 25.6 1.6 Jammu & Kashmir 49.7 46.8 41.5 29.4 19.8 2.0 Karnataka 69.5 56.9 43.5 39.9 25.9 1.9 Kerala 60.0 44.7 34.5 25.3 19.1 2.4 Madhya Pradesh 37.6 24.5 21.2 14.4 9.2 2.6 Maharashtra 36.3 24.1 20.6 15.2 9.6 2.4 Orissa 42.1 24.6 15.3 10.2 5.1 4.4 Punjab 0.6 0.9 0.1 0.5 0.1 2.4 Rajasthan 16.8 13.8 9.6 5.9 4.4 3.0 Tamil Nadu 85.1 80.9 74.8 62.7 37.2 1.7 Uttar Pradesh 13.0 7.8 5.1 3.9 2.9 3.1 West Bengal 22.6 13.8 12.4 8.6 6.3 2.5 Other States/UT 31.1 14.0 11.0 9.8 7.2 2.7 All India 35.6 28.0 23.2 18.5 11.5 2.1 Source: Report team estimates from NSS 2004-05. Notes: Bihar, Madhya Pradesh and Uttar Pradesh refer to un-split states; Poor/rich ratios of coverage is of bottom two to top two quintiles. in contrast, southern states sustain high offtake comprised 47 percent of household grain consumption ratesquiteawayupthedistribution,with around on average for all households that purchased PDS half the third quintile purchasing PDS grains in grain in 2004-05. Interestingly, the significance of PDS the group of states. grains is true even for those in the upper reaches of Tamil Nadu provides an interesting case due to the distribution who access the system, though that is its universal access policy for subsidized grains also likely in part to be a product of total food grains even after introduction of TPDS. It is interesting consumption having a lower share in both food and to note that the relative coverage rates between total consumption for the rich. It is also significant the richest and poorest quintiles are very close to that the share of TPDS grains in total among those the national average, indicating the power of self- purchasing grains was higher for all quintiles of the targeting in the PDS system due to inferior grain distribution in 2004/05 than in the late 1990s. quality and other factors. However, from a fiscal viewpoint, the very substantial state-level annual Figure 3.3: Share of TPDS grains in total household subsidy (of around just under ` 3,000 crore in grain consumption among households 2008-09) suggests that such high coverage rates accessing TPDS, various years 60 may not be fiscally replicable in most states. 50 40 Q1 for households accessing tpds, the share of pds 30 Q2 grains in total household foodgrain consumption 20 Q3 Q4 is substantial at around half, and increased in 10 Q5 0 the early years of the decade, reflecting in part 1993/94 1999/00 2004/05 the increased per household allocations in tpds in Source: Umali-Deininger and Deininger (2006), estimated from NSS for those years. This can be seen in Figure 3.3. PDS grains 1993-94 and 1999-2000, and report team estimates from NSS for 2004-05. CHAPTER-3: Protective Programs looking at benefit incidence of pds grains, the households relative to those from other social situation is somewhat better than for coverage rates categories. alone, due to higher average offtake in the lower there are also significant variations across states quintiles. This is shown in Table 3.4. At the same time, in the main reasons for non-usage of pds. For several the relatively similar average levels of offtake across states, the role of unspecified reasons (which could quintiles among those accessing PDS is also notable. include price/quality differentials between market and PDS grains) are very important, while in others (e.g., Table 3.4: Share of PDS grains captured by West Bengal and Orissa), the irregularity of supply is a consumption quintiles, 2004/05 more major issue. Chattisgarh is also worth noting as a Average offtake Share of total Share of state where physical access to FPS appears to be a major (kg) By those beneficiaries total offtake problem. In addition, while poor quality of PDS grains quintile accessing PDS (%) captured (%) is not a major reason for non-usage nationally, it is a Poorest 23.0 30.5 33.5 significant factor in several states, including Karnataka, 2 20.9 24.0 23.9 Gujarat and Bihar.As noted, the relative prices between 3 20.4 19.9 19.4 BPL and market grains differ sharply across states (even 4 19.2 15.8 14.5 not allowing for possible quality differentials between Richest 18.6 9.8 8.7 PDS and market grains), which is likely to be another important factor. Source: Report team estimates from 2004-05 NSS data. (d) why are people not accessing PDS? (e) Diversion and leakage in the PDS looking at reasons for not accessing pds in terms of the very low household offtake rates in some states national averages, unspecified reasons emerge as the naturally raise the issue of leakage and diversion of main reason for non-use, followed by irregular supply grains. GoI itself confirms that leakage and diversion of grains in fps. Interestingly, financial constraints are from the PDS are high, estimated in the most recent not cited as a major reason for not accessing the system, evaluation at 58 percent of BPL grains, though the though states such as Jharkhand and several north- data underlying that are from 2001.119 The total is a eastern states are an exception. The patterns across combination of outright diversion of grains (due to ghost social categories and income levels are not dramatically BPL cards, as well as due to diversion in the supply chain), different, though differences are notable:118 and APL households benefiting from grains subsidized at BPL prices. Some of the relevant terminology is asonewouldexpect,thepooresthouseholds(and outlined in Box 3.2. The results for major states from the SCandparticularlyST)reportfinancialconstraints Planning Commission evaluation are shown in Table 3.5, asamoresignificantreason,with 12.5 percent of also showing states with extremely high diversion such the poorest quintile citing financial constraints as as Bihar and Punjab, where almost 82 and 76 percent the main reason for non-use. Also as expected, of BPL grains respectively were estimated to have been "no time to use FPS" (which can be interpreted as diverted. These estimates are consistent with an earlier opportunity costs of the household time being study by Tata Consultancy Services in 1999 which was too high relative to the benefit of the FPS subsidy) endorsed by GoI in its Mid-Term Appraisal of the 9th Plan. is over twice as high for the richest households More recently, the report submitted by the Supreme as the poorest, and quality concerns are more Court appointed vigilance committee is extremely notable among the better-off. critical of the PDS. According to this report, the PDS irregularsupplyisasignificantlybiggerproblem system has "collapsed" in some states (e.g., Rajasthan), inruralthanurbanareas,andamongSCandST working very poorly in others (e.g., Bihar, Jharkhand).120 118 SeeAjwad(2006)fordetailedresultsbywealth,socialcategoryandrural/urban.Thesourceforthesefindingsisthe2005IHDSdata. 119 PlanningCommission(2005). 120 MediareportsontheWadhwaCommitteereport. 0 Social Protection for a Changing India: Volume II Box 3.2: Some PDS terminology Discussions of PDS use a range of terms to describe different elements of the delivery chain. These include: Allotment (or allocation) of grains is the amount which the Food Corporation of India sets aside for each State/UT in quantity terms for rice and wheat, and includes allotments for BPL, APL and AAY. State offtake or lifting of grains is generally used to refer to the actual amounts that State/UTs draw down on their allotments from FCI. This is expressed in FCI statistics as a percentage of their allotment. Offtake is typically less than 100 percent of allotment for all India (e.g. it was 85.6 percent in 2008-09), but can in principle be above 100 percent, as it was for example for several north- eastern states in 2008-09. household offtake or lifting of grains is a less official term but would generally refer to an individual household's drawdown against its monthly quota of grains. Buffer stocks are the reserve stocks which FCI and SFCs aim to keep in addition to the requirements of wheat and rice under the TPDS. These stocks form the so-called "Central Pool" which is kept to meet any emergencies like drought/failures of crop, as well as to enable open market intervention in case of price rise. The minimum stocks that are meant to be in the buffer stock are as follows (in lakh tones): Date Rice wheat Total 1 April st 122.0 40.0 162.0 1st July 98.0 171.0 269.0 1 October st 52.0 110.0 162.0 1st January 118.0 82.0 200.0 Central issue Prices. Wheat and rice are issued to the States/UTs from the Central Pool at uniform Central Issue Prices (CIP) for distribution under TPDS. CIPs of wheat and rice are fixed for BPL and APL families separately. The CIPs of wheat and rice are subsidized and have remained unchanged for BPL families since July 2000. leakage of PDS grains has a variety of meanings depending on the context. This could be due to factors such as ghost ration cards or false use of cards by others than the original owners. It may also in some contexts refer to when grains intended for BPL or AAY households are sold to households outside those categories. In the Planning Commission report, the term "leakage" includes both leakage in these terms and diversion as defined below when aggregate figures are provided. Diversion of PDS grains refers to grains which in effect disappear in the distribution channel, or are provided to illegitimate beneficiaries/claimants. The most worrying case of this is when grains disappear altogether from the PDS system, due to theft and other forms of illicit diversion. This could in principle happen at any stage of the delivery chain after grains are purchased by FCI, including within the FCI/SFC network, by distributors, by FPS owners et al. losses in handing and transport refers to 2 percent of grain loss allowed by the FCI to handlers in the distribution chain, including those who load procured grain into goods train. wastage results from storage for overly long periods or in sub-standard condition, often in FCI or SFC warehouses, but also potentially in Fair price Shops. Sources: MCAFPD and FCI websites (www.fcamin.nic.in; http://www.fciweb.nic.in); Planning Commission 2005. Table 3.5: Planning commission estimates of BPl grain leakage in PDS, early 2000s State Diversion (% BPl grains) APl share of BPl grains (%) Total leakage of BPl grains (%) Andhra Pradesh 20.6 37.0 57.6 Assam 41.7 12.0 53.7 Bihar 81.5 9.6 91.1 Gujarat 42.1 5.0 47.1 Haryana 55.7 11.0 66.7 Himachal Pradesh 31.4 14.5 45.9 Karnataka 43.4 27.5 70.9 Kerala 21.7 17.3 39.0 Madhya Pradesh 62.4 3.6 66.0 Maharashtra 26.5 8.0 34.5 CHAPTER-3: Protective Programs 1 State Diversion (% BPl grains) APl share of BPl grains (%) Total leakage of BPl grains (%) Orissa 23.4 13.0 36.4 Punjab 76.5 13.0 89.5 Rajasthan 32.0 3.0 35.0 Tamil Nadu 15.7 49.9 65.6 Uttar Pradesh 61.3 6.2 67.5 West Bengal 19.2 7.8 27.0 All-India 36.4 21.5 57.9 Source: Planning Commission 2005. Notes: Bihar, Madhya Pradesh and Uttar Pradesh refer to un-split states. both goi and nss data confirm widespread leakage and it reflect the leakage of subsidized grains to those not in diversion of pds grains, though also with substantial the relevant BPL or AAY category. There will for example variations between states. Using the 2004-05 NSS round generally be some grains used to replenish buffer and state-wise offtake data from the Food Corporation stocks, plus there will be some share of grains which of India, it is also possible to compare officially reported are released but in transit at any point in time (e.g., in wheat and rice offtake figures by state with reported September 2009, the latter figure was 1.3 percent of all purchases of PDS grains by households (Table 3.6) in grains according to FCI's website). There is also a 2 percent order to get a more precise estimate of the gap between allowance for "losses in handling and transport" which provides for grain lost along the distribution channel official grain releases from Food Corporation of India through whatever means. (FCI) and what households ultimately receive. It should be noted that this gap is not entirely explained by based on the nss data, almost 60 percent of the pds diversion of grains in the distribution channel, nor does grains released by fci do not reach households, with Table 3.6: household PDS offtake as share of official offtake by state, 2004-05 (%) Rice wheat All Andhra Pradesh 11.8 69.7 68.9 Assam 0.06 15.4 10.6 Bihar 6.9 14.1 9.1 Gujarat 43.0 44.1 43.3 Haryana 15.8 NA 15.8 Himachal Pradesh 55.2 90.0 72.7 Jammu & Kashmir 13.3 73.0 51.1 Karnataka 55.2 65.7 63.9 Kerala 20.8 94.4 70.7 Madhya Pradesh 41.2 62.2 49.2 Maharashtra 44.4 52.0 47.2 Orissa 1.3 26.8 24.2 Punjab 5.0 38.5 5.2 Rajasthan 40.7 51.8 40.7 Tamil Nadu 103.2 87.1 87.4 Uttar Pradesh 17.7 17.8 17.7 West Bengal 7.2 29.9 14.1 All Others (incl. UTs) 7.2 39.5 24.8 India 24.8 54.4 41.4 Source: Report team estimate from 2004-05 NSS data. Notes: Bihar, MP and UP refer to un-split states. State wise off take data for the matching period taken from Ministry of Consumer Affairs, Food and Public Distribution (http://www.fcamin.nic.in). 2 Social Protection for a Changing India: Volume II the figure variable between rice and wheat, with households in most states have to pay for their wheat around 45 percent, and rice a much higher entiremonthlyrationinoneshot,whichimposes 75 percent. Moreover, in several states ­ including financialstrain(though an increasing number of Bihar, Assam, Punjab, Uttar Pradesh, West Bengal and states have followed the lead of Kerala and allow Haryana ­ the gap between FCI releases and household for weekly lifting by households). Villagers are receipts of PDS grains is over 80 percent.121In contrast, also often poorly informed of when stocks will several states do considerably better than the national be available, and certainly not in advance. This averages, most notably Tamil Nadu (with 13 percent means that the poorest may not have sufficient leakage and Andhra Pradesh with 31 percent). Some of cash ready available when the foodgrains arrive the factors driving such poor outcomes, and the political in the shop. economy challenges of significant reform of the system, allocations from GoI are valid only for a month, are discussed at the end of this section. and if the state government is not able to lift within that time, its quota lapses. This timeframe could be increased to a quarter. (f) implementation issues in the PDS lowqualityoffoodgrains,anddifferencesintype apart from the generic problems of program of local consumption patterns. Earlier reports administration discussed in chapter 7, there are from 2000 found that half the stock of FCI is at a range of implementation problems in pds which least two years old, 30 percent between 2 to 4 contribute to the poor outcomes seen. These have years old, and some grain as old as 16 years.124 been known for a long time, and include:122 While the data above indicate that the situation inadequate storage capacity with FCI and State may have improved in the interim, there remain FoodCorporationsinanumberofstates(e.g., in issues of grain type and the fact that it is in Bihar and Jharkhand, there are godowns in less many states not of a type consumed (e.g., par than 50 percent of districts). This is one factor boiled versus raw rice) as they are procured from contributing to irregularity of supply to FPS in a distant states. number of states. weak monitoring, lack of transparency thepricingmarginsforFPSowners,andarange and inadequate accountability of officials of other factors such as transport costs from implementingthescheme.While this is improving godowns,needforupfrontpaymentongrainsin in a number of states with transfer of FPS to moststates,rentalcostsofpremisesetc.aresuch Panchayati Raj Institutions (PRIs), Self-Help Groups (SHGs) and cooperatives, State Food Corporation thattheyareinmanystatesfinanciallyunviableif (SFC) and Departmental officials in many states theyoperatetheshopwithoutdivertinggrains.123 exercise very weak oversight of the system. The Planning Commission estimates that only around 23 percent of FPS are financially viable, FPS in many areas do not open for more than and that they are concentrated in a few states 2-3 days in a month, and card holders are not (Andhra Pradesh, Tamil Nadu and Maharashtra). allowedtolifttheirquotaofpreviousmonths. In addition, a full 44 percent of FPS owner costs efforts to raise citizen awareness of their rights of operation nationally are taken by transport and underPDSthroughrelianceonaCitizen'sCharter rental cost, a burden that makes financial viability wereveryslowtotakeoff,andfieldworkindicates very challenging. thattheirimpactinmanycasesremainslimited. 121 NotethatBiharfromearly2007hasswitchedtoafoodcouponsystemforPDSwhichhasalsobeenimplementedinstatessuchasAndhraPradeshwhichhave donesopreviouslyhashelpedtocontrolleakagetoanextent.InitialreviewoftheBiharexperiencesuggeststhatthishasalsobeenthecasethere.SeeVashisht etal.(2009)andBox3.3inSectionC. 122 See Radakrishnan and Subbarao (1997) for a detailed discussion of many of these implementation challenges, most of which in most states are still very relevant. 123 SeePlanningCommission2005foradetaileddiscussion. 124 WorldBank(1999). CHAPTER-3: Protective Programs For example, the Planning Commission PDS widows aged 40-64 years and persons with severe or evaluation found that under 20 percent of GPs multiple disability aged 18-64 years, providing ` 200 were aware of a Citizen's Charter, and that only per person per month. There are known as the Indira 4 percent of APL card holders and under 1 percent Gandhi National Widow Pension Scheme (IGNWPS) and of BPL card holders were aware of the Charter.125 the Indira Gandhi National Disability Pension Scheme (IGNDPS). In addition, until the 2009-10 budget, nearly despite these major structural problems, a number all states have operated a similar social pension scheme of states are innovating in pds implementation, and for destitute widows and disabled people, though improved performance can be seen in some cases. States financed from own resources.128 Some states also operate have been implementing a nine point action plan for state-specific old age social pensions which often pre- improving TPDS since 2006. The southern states as usual date the NOAPS, and allow for expanded beneficiary have led the way on many reforms intended to address numbers beyond those financed by the central subsidy. the issues above, and increasingly even poorer states ­ In several states (e.g., Himachal Pradesh and Andhra e.g., Madhya Pradesh, Orissa and Bihar ­ have introduced Pradesh), the numbers under state-financed elderly changes in policies and implementation mechanisms to schemes are significant. address the problems of PDS. These are taken up further in the conclusions and recommendations section. in 2009, goi announced two new national schemes: the indira gandhi national widow pension scheme, and the indira gandhi disability pension scheme. The B. SoCiAl PenSionS former covers widows between 40 and 64 years of age, and the latter covers those with at least 40 percent as with the pds, social pensions aim to alleviate disability between 18 and 64 years old. The centre is to chronic poverty, though among specified social provide ` 200 per month for both groups, estimated groups, i.e., destitute elderly, widows and disabled to include around 4.4 million widows and 1.56 million people.126 The Indira Gandhi National Old Age Pension disabled people. States are encouraged to double the Scheme (IGNOAPS) is provided to applicants over 65 benefits from own resources. For orphans and children years of age who are below the poverty line, though with disabilities, the Women and Child Development several states provide pensions for those 60 and Ministry is expected to develop a scheme. above who fulfill the criteria.127 As of mid-2009, the central transfer was ` 200 per month. A number of until the switch to full cash payment under mahatma states top up the benefit level from own resources, gandhi national rural employment guarantee scheme with the maximum benefit being ` 1,000 as of 2009. (mgnreg), social pensions were the most significant Supplementing IG/NOAPS since 2000-01 is the cash-only social protection transfer, and remain Annapurna scheme, which covers destitute elderly a key benefit for specially vulnerable groups. They eligible for elderly social pensions but not receiving are also of interest in their reliance on a mixture of them. These households receive 10 kg of foodgrains categorical targeting with household level BPL criteria. per month free through FPS. Also supplementing this Social pensions have become more important as part of is the National Family Benefit Scheme, which pays a the overall SP program mix in recent years, as indicated lumpsum of ` 10,000 to BPL families on the death of both by the significant increase in central subsidies (from the primary breadwinner. NSAP was expanded in ` 75 to 200), and the expansion of coverage to BPL February 2009 with two new pension schemes for BPL elderly, disabled and widow households. 125 PlanningCommission(2005). 126 OnNationalOldAgePensionScheme(NAOPS),thatwasextantbeforeIGNAOPSin8states,seeORG-MARG1998forareviewofearlyexperienceandIrudayaRajan (2001)forausefuldiscussionoftheprogram.Fordetaileddiscussionoftheoperationofsocialpensionsinspecificstates,seeMurgai(2006)onKarnataka,and Dutta(2008)onRajasthan,bothcommissionedforthisreport,GovernmentofHimachalPradesh(2005),Nayaketal.(2002),andAlam(2004). 127 IGNOAPSwaslaunchedinNovember2007andreplacedtheNationalOldAgePensionScheme(NOAPS),whichwasintroducedin1995.Asmuchofthedata underlyingthissectionisfromthepre-2007period,theNOAPSacronymisusedatpointsinthefollowinganalysis.ThekeydifferencebetweeneligibilityforNOAPS andIGNOAPSisthatNOAPSwasbasedonadestitutioncriterion(variablydefinedbystate),whiletheIGNOAPSisBPLbased.Theformerstate-wiseallocationsto NOAPSwerebasedonaformulaofpopulationover65timeshalfthegeneralpovertyrateforthestate. 128 SeeWorldBank(2009)fordetailsondisabilitysocialpensions. Social Protection for a Changing India: Volume II relative to most safety net benefits, the coverage for social pensions, the survey-based coverage rates rates of social pensions among the target groups imply under-coverage according to both the central are significant and are likely to have expanded ceilings for allocations in place at the time of the ihds further with the recent policy and financing reforms. data, and administrative data on program coverage. Table 3.7 presents coverage rates for NOAPS, Annapurna, The IHDS survey-based all-household coverage rate of for widows and disability pensions both for 2004-05, 0.5 percent implies 5.1 million NOAPS recipients, which is relying on the IHDS data. Social pensions account for a comparable with a MoF allocation number under NOAPS more significant share than most SP programs among of just under 6.9 million, and an administratively reported their target groups (except for PDS), with over 8 percent beneficiary coverage of 6.1 million for 2004-05.130 In of households with at least one member over 65 years addition, the reliance on 1998 population estimates and receiving NOAPS (and just under 10 percent when mortality rates, under-estimates the size of the elderly Annapurna is added), 6.2 percent of all widows receiving cohort who are the target group for the scheme. Rough that pension, and around 14 percent of households with a estimates using 2001 census and assuming a 2 percent disabled member receiving a disability social pension.129 annual increase in the aged population, would suggest Table 3.7: Coverage rates of target group for noAPS, Annapurna, widow and disability pensions by state, 2004-05 % hh with median annual % hh with % hh with a % of hh with 65+ receiving noAPS benefit among 65+ receiving widow receiving disabled receiving State noAPS receiving (Rs) Annapurna wP DP Andhra Pradesh 18.7 1,840 3.2 4 5.6 Assam 3 900 0 0 NA Bihar 4.1 1,138 3.6 5 8.8 Chhattisgarh 9.3 1,870 0.3 8.8 14.9 Delhi 5.4 2,175 0 2.2 NA Gujarat 0.9 2,696 0.6 2 NA Haryana 67.1 3,088 0 16.4 37.2 Himachal Pradesh 13.2 2,690 0.3 25.7 42.9 Jammu & Kashmir 0.9 2,079 0 5.8 6.7 Jharkhand 2.2 1,047 0 8.3 12 Karnataka 6.4 1,218 0 13.9 44.9 Kerala 6.3 1,110 0.2 5 22.2 Maharashtra 2.3 2,111 2.4 2.3 6.2 Madhya Pradesh 6.4 1,758 0 6.3 21.4 Orissa 22.2 1,399 8.5 21.3 28.8 Punjab 14.8 1,645 0 4.3 17.2 Rajasthan 7.8 2,902 2.7 5.3 20 Tamil Nadu 2.9 2,254 0.8 5.3 3.8 Uttar Pradesh 5.2 1,943 1.6 4.2 14.4 Uttarakhand 8.4 1,358 0 11 8.7 West Bengal 2.6 1,113 0.9 1.8 4.3 Other (incl. UT) 20.8 2,979 1 5.5 NA All India 8.3 2,008 1.7 6 14.1 Source: Ajwad (2006) based on 2005 IHDS data. Note: Disability pension coverage calculated from survey-based coverage for whole population and census-based state rate of disabled people. 129 Thiscannotbecalculatedfordisabilitysocialpensionsfromthesurveysourceasdisabilityinformationwasnotcollected,butisbasedonthe(conservative)census estimateof2.13percentofallhouseholdshavingadisabledmember,and0.3percentagepointcoverageofthegeneralpopulationagainstthatrate. 130 SeeSaxena(2006)fordetailedadministrativedata. CHAPTER-3: Protective Programs an updated figure for allocation purposes of closer the exception of results from state-specific surveys to 9 million. There seems therefore a pressing need to in Karnataka and Rajasthan, which confirm generally revisit the population estimates upon which IGNOAPS progressive targeting.131 Nonetheless, it is clear that allocations are based. there remain significant inclusion errors, with not insignificant shares of elderly in better-off households there was also major variation across states in receiving NOAPS benefits, and widows in even rich coverage rates, with some states having quite households receiving widow pensions. Interestingly, the significant coverage among the elderly and widows, coverage of disability pensions varies very little across though most states with a much lower rate for welfare indicators, though this may be a product more disability pensions. Interestingly, this is not simply a of low coverage rates when measured against the whole rich and poor state story, as even some very poor states population base. such as Orissa and Chhattisgarh had high coverage rates for elderly and widows pensions in 2004-05, reflecting in in terms of benefit incidence, both elderly and part their national funding quotas, but also a clear state- widow social pensions are also progressive, though specific importance placed on social pension benefits. somewhat less so than coverage alone due to the higher average benefits of recipients in the richest nationally, the targeting of social pensions across quintile, and other castes. This is also shown Table 3.8. wealth levels and social category was progressive Nonetheless, in terms of concentration of spending in 2004-05, with particularly good performance in among the poorest quintiles, these can be considered relative terms among the poorest quintile and among among the better targeted programs in India. These sc and st people. In terms of targeting performance findings are strongly supported by evidence from the of social pensions, the smallish sample size allows Social Protection Survey (SPS) in three states (Orissa, for reliable estimates only on an all-India basis, with Karnataka and Madhya Pradesh) which found that Table 3.8: household coverage rates and benefit incidence of social pensions by wealth, location and social category, 2004/05 (%) Annapurna widows' noAPS - 65+ noAPS-benefit Annapurna benefit widows benefits DP benefit coverage incidence (%) coverage incidence (%) coverage incidence (%) incidence (%) Poorest 14.8 31.1 4.7 37.9 10.9 32.8 27.6 Q2 8.6 19.5 2.4 26.9 5.9 18.2 16.5 Q3 7.8 15.4 0.9 11.2 5.6 19.4 19.9 Q4 7.0 15.6 1.1 22.6 4.9 17.2 11.2 Richest 6.4 18.4 0.1 1.4 2.7 12.5 24.8 Rural 9.4 86.2 2.0 92.0 6.2 70.4 88.1 Urban 4.6 13.8 0.7 8.0 5.5 29.6 11.9 OBC 7.0 34.6 1.9 49.6 6.0 38.2 48.9 SC 15.0 34.4 2.8 25.3 7.7 25.5 21.7 ST 11.6 5.4 3.6 11.3 10.3 10.9 6.4 Other 5.9 20.8 0.6 13.7 3.7 20.7 19.3 All India 8.3 100 1.7 100 6.0 100 100 Source: Ajwad (2006) based on the 2005 IHDS data. Note: Coverage rate for NOAPS among households with member 65 and above, for widows as share of households with a widow, and for disabled, as share of all households. Note: Disability pension coverage calculated from survey-based coverage for whole population and census-based state rate of disabled people. Note that the other social group category excludes Brahmins. 131 SeeMurgai(2006)andDutta(2008)usingtheKSPS(KarnatakaSocialPensionsSurvey)andRSPS(RajasthanSocialPensionsSurvey)respectively. Social Protection for a Changing India: Volume II social pension coverage rates among SC households for other purposes, and in others in significant to be around 80 percent higher than for the overall paymentdelays.Even rich states such as Gujarat population, and statistically significant.132 have in the past used the NOAPS transfer for other purposes (and made no NOAPS payments at all in social pension benefits reported by households 2003-05), while poorer ones such as Jharkhand indicate a very low level of benefit leakage among and Orissa in the early 2000s ceased to make those receiving benefits. As the data in Table 3.7 social pension payments for whole budget years. reflect the level of social pensions before the 2006 Even states which have not used the transfer hike to ` 200, they are not straightforward to interpret, for other purposes frequently have had delays given the variable top-up by states over the ` 75 base in payments of 2-6 months, as reports of states that the centre financed for NOAPS at the time of the to the Supreme Court Commissioners in 2005 survey. However, for most states, the reported annual indicate for states such as Bihar, Jharkhand, West benefits are close to the level of social pensions being Bengal and Manipur.135 paid in 2004-05. These findings are also supported by the dedicated Rajasthan and Karnataka surveys of thereisnoregularverificationofbeneficiariesin social pensions, which find benefit payments over the some states. While regular verification may be previous year close to 100 percent of those due, and less necessary for widow or disabled pensions, Himachal Pradesh where beneficiaries report timely evidence from some states indicates that elderly and proper payment. pensioners continue to "receive" transfers after death, or are otherwise untraceable. A survey in limited work has been done on the administration Delhi for example found around 6.5 percent of of social pensions, and factors driving performance NOAPS "beneficiaries" were dead, and a further across states.133 Given the increased central resources 17 percent untraceable. In the more detailed being devoted to social pensions, more research is exercise conducted for Karnataka, around 6 needed on the factors driving program performance. percent of elderly and widow pension records were In the meantime, dedicated studies in Karnataka, found to be likely duplicates (with wide variation Rajasthan and Himachal Pradesh provide some across taluks), and 9 percent of pensioners could insights, as do information from the Supreme Court not be traced due to movement or death.136 Commissioners on scheme operation.134 Some features the documentary requirements for proving of implementation include: eligibilitysuchasbirthcertificates,areoftennot there has been limited ministerial ownership of availabletobeneficiariesordemandinginterms NOAPS at the central level, and weak reporting oftimeandsometimescosttoobtain.Survey and and monitoring by GoI with the transfer of field work reveal that proof of eligibility is often program implementation to states from 2002-03 cited as a burdensome process by potential (and hence change from a Centrally Sponsored beneficiaries with almost half of current social Schemes (CSS) to Additional Central Assistance pension beneficiaries in Rajasthan reporting (ACA) in budgetary terms). How this may change difficulties in documents and procedures in the with recent policy and financing reforms remains application process, and a third of those who to be seen. may have otherwise applied citing complex the transfer of NOAPS funds through state procedures as the main reason for not doing treasuries(incontrasttoearliertransferthrough so.137 While at the local level this is sometimes DRDAs),combinedwithweakcentralmonitoring, dealt with in a practical manner, this is often not hasresultedinsomestatesindiversionoftransfers the case. 132 Devetal.(2007). 133 ThoughseetheusefulworkofNayaketal(2002),andGovernmentofHimachalPradesh2005. 134 Saxena(2006),Murgai(2006),Dutta(2008),GovernmentofHimachalPradesh2005. 135 Saxena(2006). 136 SeeMurgai(2006)andBox7.4inChapter7fordetails. 137 Dutta(2008). CHAPTER-3: Protective Programs from field work, it seems that there has been in thepastalackofclarityinsomestatesonwhat C. TARgeTeD houSing constitutes"destitute"forsocialpensioneligibility PRogRAmS purposes. The centre and most states quite sensibly leave some degree of local discretion (a) Rural housing ­ indira Awaas in the determination of destitution. At the same time, field work (e.g., in Rajasthan), revealed ad yojana (iAy) hoc criteria being applied at local levels which may as seen in chapter 2, targeted rural housing programs unwittingly exclude some of the very poorest. The are a significant element of the indian safety net. The imposition of the BPL criteria under IGNOAPS has main targeted rural housing program is IAY, which has removed this local level discretion. been implemented as a stand-alone program since 1996, inanumberofstates,specificexclusioncriteria having been a sub-component of public works schemes havebeenoverlystringent,andoftenappliedin prior to that. Its main objective is to provide dwelling amechanicalmanner.A common example was units free of cost to the rural BPL population.138 As of that applicants for NOAPS and widow pensions April 1, 2010, grants are provided to beneficiaries with a were not allowed to have a living adult son, and ceiling of ` 45,000 for new construction in plain areas and this is often applied strictly, even where the son ` 48,500 for hilly/difficult areas. In addition, up to is absent or failing to support the parent (even if 20 percent of IAY funds are available for upgrading he is severely disabled and unable to work). Both construction up to ` 15,000 for each dwelling unit. the Karnataka and Rajasthan studies found that Beneficiaries can use up to ` 2,200 from Total Sanitation a high proportion of social pension beneficiaries Campaign funds in addition to IAY assistance to construct did not qualify according to a literal application sanitary latrines. Further, an IAY beneficiary can obtain a of all the exclusion rules, despite the fact that loan up to ` 20,000 under the Differential Rate of Interest they were considerably poorer than average. scheme from any national bank at an annual interest While nearly all satisfied the strictly demographic rate of 4 percent. The house should be registered in the criteria (of old age and widowhood), in Karnataka, name of the female household member, or jointly in the only 9 percent of old age and 16 percent of widow name of husband and wife. More recently, the program pensioners met all demographic (including has been dovetailed with the Rajiv Gandhi Gramin family support) and destitution criteria. The Vidhuyutikaran Yojana for rural electrification, allowing situation was not much different in Rajasthan, IAY beneficiaries to get a free electricity connection. but the additional BPL criteria (over-riding all it is estimated that on an average about 23 lakh other family and destitution criteria) allowed houses are built every year under various schemes, a much higher share to be eligible. Only 26 with iay accounting for two thirds of the total, percent of elderly pensioners and 9 percent of . and other assistance provided by Housing and urban widowed pensioners would have been eligible Development Corporation Limited (HUDCO), National in Rajasthan were it not for the BPL criteria. Even Housing Bank, State Housing Boards and Commercial where discretion is applied in sensible ways, Banks the remainder.139 IAY enjoys considerable support the necessity of such discretion naturally opens since it creates a valuable asset for beneficiaries, leading up possibilities of abuse. The findings strongly to improved economic and social status with minimal suggest a need in many states for an overhaul requirements on part of beneficiaries. of their social pension rules to reduce exclusion errors. These family criteria no longer apply under results on coverage of iay and other housing schemes new Indira Gandhi old age, widow and disabled are shown in table 3.9, along with median amounts pension schemes, but it is unclear whether these received as reported by households. Several points remain applicable for state pension schemes. are apparent: 138 SeeSaxena(2006)fordetailsandhistoryofthescheme,theComptrollerandAuditorGeneralevaluation(CAG2003a),andthediscussioninDevetal.(2007). 139 The2001Censusestimatedahousingshortageof149.6lakh,witheightstatesaccountingfor81percentoftheshortage.Theannualrequirementisestimatedat 30lakhhousestomeetthebacklog,andabout10lakhshelterlessareaddedeveryyear,suggestingatotalshortfallofabout40lakhhousesperyear. Social Protection for a Changing India: Volume II Table 3.9: Schemes to provide support for housing, latrine or chulha construction by state (rural only), 2004-05 median value of support received by households (`) % hh ever money to build materials to build State received benefits Built house house house Sanitary latrines Andhra Pradesh 28.6 20,236 19,048 14,688 1,836 Assam 1.4 14,667 22,849 8,833 1,964 Bihar 13.3 16,524 15,669 20,000 1,200 Chattisgarh 4 ­ 10,322 5,667 650 Goa 19.7 ­ 14,000 ­ 3,614 Gujarat 6.3 28,271 22,250 13,150 1,885 Haryana 2.9 31,200 14,625 19,344 1,775 Himachal Pradesh 8.5 18,000 14,804 8,240 1,338 Jammu & Kashmir 0.8 20,000 27,100 .. 1,000 Jharkhand 9.1 16,925 18,559 12,000 2,000 Karnataka 14.6 22,043 15,994 6,662 1,789 Kerala 16.6 25,169 18,738 5,528 1,874 Madhya Pradesh 5.4 10,500 10,710 6,760 872 Maharashtra 7.2 23,991 22,272 7,442 2,123 Orissa 17.4 18,955 15,203 5,368 785 Punjab 2.4 24,039 9,357 6,400 2,423 Rajasthan 2.6 ­ 14,428 3,601 1,191 Tamil Nadu 7.2 27,654 17,127 13,774 779 Uttar Pradesh 5.3 13,318 14,449 12,718 1,650 Uttaranchal 14.6 5,000 15,188 22,000 1,253 West Bengal 7.2 13,944 7,767 2,452 1,030 All India 9.6 21,604 15,861 11,822 1,731 Source: Ajwad (2006) based on the 2005 IHDS data. Note: Recall period is open-ended housing-related support is quite significant, with to a special central package announced in 2004), almost 13 percent coverage of rural households in despite much lower than average coverage on the sample.However, this is somewhat misleading, most SP schemes, and release to allocation in as the recall period on assistance in the survey was 2004-05 was almost double. The fact that several open-ended, so that this is a cumulative figure very low income states have healthy coverage representing in effect the share of rural households rates may also be a reflection of the housing who have ever received housing support. shortage criterion on scheme allocations, with as with other schemes, there is substantial eight states accounting for over 80 percent of variation across states in coverage, though housing shortage nationally, and Bihar alone somewhatlesssointermsofmedianbenefitlevels. nearly one third of housing shortage.140 Some of this is likely to be driven by state-specific while significant (as would be expected given schemes (e.g., Himachal's state targeted rural the program policy), reported amounts receive housing scheme), but in others by more active indicateleakagebetweentheprogram-sanctioned use of IAY funds. The case of Bihar is of particular transfersandactualreceiptsbyhouseholds.While interest, as coverage under housing schemes is it is not possible to compare the survey results notably above the national average (in part due very precisely with IAY allocations, the results are 140 Saxena(2006). CHAPTER-3: Protective Programs consistent with field research for this report which there are a range of problems in iay implementation, found a fairly consistent "commission" being paid caused in part by its design of large, unencumbered by beneficiaries to sarpanchs, MLAs or other grants.143 The lump sum payment is large enough to intermediaries of ` 4,000-5,000 on IAY in order to attract substantial "attention" from local politicians, access the program.141 who may view this as an important mechanism for targeting of rural housing transfers is mildly patronage for supporters and there is evidence of high progressive, though as with other programs a proportion of benefits being manipulated towards this significant proportion of households in the upper end.144 These machinations are natural, since the total range of the distribution receive support. This can allocation of IAY grants - although substantial - is small be seen in Table 3.10, which presents results by wealth, relative to potential demand based on number of poor social category and location. While coverage rates in the people without housing. top quintile are notably lower, they remain high even in safeguards built into the design of the scheme have often the fourth quintile. This finding is mirrored by those from been ineffective in practice. For example, payments for the three state SP study, which found similar coverage each stage of construction are to be made only when the rates for IAY across the bottom three quartiles, but also preceding stage has been completed, and individuals are coverage rates of SC and ST households which were required to make their own arrangements for construction. 66 and 27 percent higher respectively than the overall population, and more than double those of backward In particular, officials are not allowed to engage contractors and other castes.142 A second ­ and more worrying ­ on behalf of the beneficiaries. According to a Comptroller result is that the average level of benefits reported by and Auditor General (CAG 2003b) report on the scheme, the rich is considerably above that of the poor for most almost one-third of IAY funds were misused. Of this, housing-related interventions. However, administrative almost half was accounted for by depositing of funds by indicate that nearly two-thirds of IAY houses since 1996 state governments into current accounts, civil deposits, or have been allotted to women. treasuries outside the government account. The remainder Table 3.10: Coverage and median benefits of targeted housing programs by wealth, location and social category (rural areas only), 2004-05 median of the value of support received by households (`) Benefit % hhs ever incidence received money to build materials to (% benefits group benefits Built house house build house Sanitary latrines captured) Poorest 15.6 20,946 14,010 8,170 1,448 24.6 Q2 13.3 18,420 15,210 11,073 1,751 18.5 Q3 15.1 20,736 16,312 13,537 1,727 24.9 Q4 13.9 24,329 19,692 14,994 1,637 22.7 Richest 5.7 20,269 18,810 13,802 1,935 9.3 OBC 9.9 19,294 17,476 13,721 1,664 32.7 SC 23.1 21,943 15,494 11,832 1,719 44.2 ST 13.3 21,466 15,292 10,835 1,919 10.4 Other 7.7 21,225 15,142 12,032 1,779 12.7 Source: Ajwad (2006) based on the 2005 IHDS data. Note: Benefit incidence based on results on the question "Money to build a house"; Recall period open-ended. Note that the other social group category excludes Brahmins. 141 SeequalitativefindingsinDevetal(2007),discussioninSaxena(2006),andCAG(2003a)forIAYevaluation. 142 Devetal.(2007). 143 SeeSaxena(2006)foradiscussion. 144 SeePlanningCommission2003,andqualitativefindingsofDevetal.(2007)onthesystematicnecessityoflargebribepayments. 0 Social Protection for a Changing India: Volume II was due to misappropriation, unapproved works, and scaled back in recent years, and has in effect been wound unauthorized activities. Almost 20 percent of the audited up with the advent of Jawaharlal Nehru Urban Renewal money was spent on construction of houses through Mission (JNNURM). There are in addition various state- contractors, though the Ministry of Rural Development specific programs, e.g., Karnataka's Urban Ashraya Housing notes that this practice has stopped with the ban on Program, and Kerala's Mythri Housing Scheme. contractors. Over-inflated expenditure combined with While these interventions are not a significant focus poor quality of dwellings was a natural outcome. Another of this report, results in coverage of public housing example with implications for gender bias in the scheme program support by state indicate that 2 percent of all is that the mandatory provision for joint registration of urban households have received such support at some houses in the name of both husband and wife is flouted point. Across states, the situation in 2005 based on in most cases, and in many states, field-level functionaries analysis of the IHDS data was:146 are unaware of the existence of such a provision. coverageinalllaggingstates,withtheexception ofOrissa,doesnotexceed1percentoftheurban (b) Targeted urban housing population, and in several states is below half a percent (Uttar Pradesh, Rajasthan, Bihar, Madhya in addition to iay, housing support for the poor has Pradesh). traditionally been one of the few dedicated urban incontrast,andpartlyreflectingstateprograms, safety net programs, and around 2 percent of urban households nationally have received such support.145 southern states have higher (though still low) The main central scheme has been Valmiki Ambedkar Awas coveragerates, ranging from around 3 percent in Yojana (VAMBAY), which provided housing construction Tamil Nadu to over 10 percent in Kerala. These are and shelter upgrading for BPL urban slum dwellers, with a also states with greater urbanization. focus on vulnerable categories such as SC and ST. VAMBAY in distributional terms, urban housing support is was introduced in 2001, and presents a shift in emphasis progressively targeted, though coverage for all from slum resettlement programs. The program operated groups is relatively low. Given the focus on SC and on a 50/50 financing split between centre and states, with ST, this is not surprising. However, it is notable that GoI funding routed through HUDCO. For the household, the program does considerably better at reaching SC half of the GoI portion was provided as grant and half as households than ST households, for whom coverage is loan. However, financing for VAMBAY has been sharply around half the average (Table 3.11). Table 3.11: Coverage rates and median benefits of urban housing by quintile and social category, 2004-05 median of the value of support received by households (`) % hhs ever received money to build materials to build group benefits Built house house house Sanitary latrines Poorest 3.2 23,194 13,594 6,510 1,749 Q2 3.3 26,996 14,745 13,320 1,668 Q3 2.6 35,184 20,692 10,481 1,907 Q4 1.3 43,969 25,640 20,000 2,141 Richest 0.3 73,488 1,000 - 1,808 OBC 2.0 30,842 14,825 9,052 1,931 SC 5.1 30,020 19,982 14,138 1,792 ST 1.1 23,860 11,947 8,000 1,801 Other 1.0 27,319 14,577 6,298 1,882 Total 2.0 30,062 17,122 10,333 1,863 Source: Ajwad (2006) based on the 2005 IHDS data. Notes: Recall period open-ended. Note that the other social group category excludes Brahmins. 145 SeeBuckleyetal.(2005)foramoredetaileddiscussion. 146 SeeAjwad(2006)forastate-wiseanalysis. CHAPTER-3: Protective Programs 1 while more detailed analysis is required, available among the poorest most often. This is discussed in the assessments point to a range of implementation recommendations section. challenges with vambay, though state programs such as those of karnataka and kerala are somewhat better (though at best, only rates as "fair"). Based D. DeTeRminAnTS oF on qualitative methods, VAMBAY has been assessed as poor in terms of program efficiency, administrative PRogRAm PARTiCiPATion simplicity, and sustainability, and only fair in terms of this section presents results of multivariate analysis transparency.147 on the determinants of program participation for selected programs.149 The primary objective is to urban housing policy for the poor is in a dynamic explore the impact of various factors on the probability phase presently, though in programmatic terms is yet of program participation holding all other variables to consolidate new approaches, and it is unclear how constant using the 2004/05 IHDS data. While it is not coherent or strategic a vision underlies the current possible to carry out such analysis for all programs, programs. JNNURM is the umbrella vehicle of the GoI results are available for social pensions and Annapurna, for urban renewal and low-income housing is part of PDS and housing support. One or more of the following its coverage. At the same time, initial efforts from states are expected to impact the probability of program and cities have tended to focus on other types of urban participation: location of the household (urban or rural), infrastructure. The specific window of JNNURM under the position of the household on a relative welfare which housing for the urban poor falls is the Sub-Mission scale, caste of the household, household size, for Basic Services for the Urban Poor (BSUP) for cities characteristics of the household head, participation of over 1 million population, and Integrated Housing and the household in agricultural or animal husbandry, and Slum Development Program (IHSDP) for all other cities the state on which the household resides. Results are and towns. VAMBAY is subsumed within IHSDP except presented for each in turn. for ongoing projects pending their completion.148 Both BSUP and IHSDP target slum dwellers and the urban looking at the impacts of wealth on program poor. These beneficiaries are eligible for benefits up participation, the results are in the expected to ` 80,000 for building homes, improving water and direction, with poorer quintiles more likely to sanitation etc., with at least 12 percent of the housing participate than the reference richest quintile cost to be borne by the beneficiary, or 10 percent for SC/ (table 3.12). While this is encouraging, several ST/OBC/BC and other "weaker sections". additional observations can be made which strongly caution against complacency: In parallel, GoI in the 2009 budget announced the Rajiv Awas Yojana which aims at promoting a slum- for several programs (NOAPS, housing, ration free India in five years and would focus on according card usage by BPL households for purchasing property rights to slum dwellers. It is also meant PDS grains), the significance levels for the fourth to provide basic amenities such as water supply, quintile remain very high, and for all programs, sewerage, drainage, internal and approach roads, significance levels for the third quintile are high street lighting and social infrastructure facilities in or very high. Well targeted programs should slums and low income settlements adopting a `whole not exhibit such patterns, given the overall city' approach, including providing subsidized credit. It poverty rate. is unclear how this scheme will interface with JNNURM. for all programs, the strength of the effect in most The final important development is increased efforts in quintiles is not very strong. Across all programs, the private sector to cater to the low income housing the effect of being in the lowest quintile ranges market as a commercial proposition, though with a from only 1.3 percent (widows and Annapurna) to focus on households who are low income but not 8.5 percent (housing). 147 Buckleyetal.,op.cit. 148 Seehttp://www.indiaurbanportal.in/JNNURM/2yrs-JNNURM.pdf. 149 SeeAjwad(2006).Aprobitmodelwithfixedeffectswasused. 2 Social Protection for a Changing India: Volume II Table 3.12: Program participation determinants by wealth quintiles, 2004/05 BPl Ration card noAPS widow pensions Annapurna housing support used last 6 months Quintile 1 0.0341*** 0.0133*** 0.0126*** 0.0845*** 0.022*** Quintile 2 0.0279*** 0.0079** 0.0157*** 0.0988*** 0.0207*** Quintile 3 0.0172*** 0.0115*** 0.0116** 0.0683*** 0.0142** Quintile 4 0.0105*** 0.0037 0.0061 0.0453*** 0.0126* Source: Ajwad (2006) based on the 2005 IHDS data. Quintile 5 as reference group. ***=significant a 1 percent level; **=at 5 percent level; *=at 10 percent. there is variation across programs in the strength thestrengthofcastecategoriesonparticipation of the impact of being in specific quintiles. The isgenerallyweakevenwheresignificant. coefficients for quintiles across programs vary looking at other socio-economic characteristics, the between evenly progressive (though small) results on program participation are not strong, variations as one moves up the distribution (e.g., NOAPS and PDS usage) and others where the with the exception of rural location (table 3.14). Most gradient of coefficients between quintiles is less of the indicators which are significant are in line with smooth (e.g., Annapurna and housing, where the what would be expected, e.g., household size being second quintile is more likely to participate). relevant (though very weakly) for scholarships and social pensions. What is more remarkable perhaps is the looking at the impacts of caste on program "dog that does not bark", i.e., that both land and animal participation, the situation is more complex than for ownership are not significant correlates of participation wealth. This can be seen from Table 3.13, from which in most programs. In contrast, the results on rural several observations emerge: location are more significant, though rather weak in casteisasignificantdeterminantofparticipation terms of coefficients. in the programs listed for SC households, and finally, looking at program participation state-wise, for food programs and scholarships for ST it is clear that location matters as far as program households. This is a positive result, though the participation goes, with "location" reflecting a relatively weaker position of ST households is range of state-specific policy, institutional and other also a cause of concern, and consistent with field factors not directly captured in the survey. However, work finding specific challenges in access for ST households. At the same time, for most programs, the strength of the locational impacts is not great for being an SC or ST household is less significant most programs and states. Observations on the state- in statistical terms than wealth levels, though level results include: remains highly so for housing programs for SC there are noticeable variations across states and use of BPL ration cards to purchase PDS grain. in whether the impact on specific program At the same time, the positive feature of findings participation is positive or negative. This is by caste is that being from "other" castes and OBC of course to be expected, but the findings is generally not significant. present useful shorthand of institutional and Table 3.13: Program participation determinants by social category, 2004/05 BPl card used last 6 noAPS Annapurna housing support months Other caste 0.0033 ­0.0018 ­0.0193* 0.0132 OBC 0.0016 0.0014 ­0.0154 0.0156 SC 0.0126** 0.0036 0.071*** 0.0195** ST 0.0058 0.0148** 0.0169 0.0196** Source: Ajwad (2006) based on the 2005 IHDS data. Brahmin as the reference group (note that the other social group category excludes Brahmins). ***=significant a 1% level; **=at 5% level; *=at 10 percent. CHAPTER-3: Protective Programs Table 3.14: Program participation determinants by selected socio-economic characteristics, 2004/05 government BPl card used assistance with in the last 6 noAPS widow pensions Annapurna housing months Rural 0.0106*** 0.0041 0.0019 0.0645*** 0.0343** HH size 0.0007** 0.0007*** 0.0001 0 0.0012 Female HH head 0.0151*** 0.0907*** 0.0016 ­0.0018 ­0.001 Age of HH head 0.0009*** ­0.0003*** 0 0.0001 0 HH head literate ­0.004* ­0.0025 ­0.0015 0.0049 0.0017 HH head completed primary ­0.0027 ­0.0006 0 -0.0044 0.0023 education HH owns agricultural land 0.0000 0.0000 0.0000 0.0000 0.0002*** HH owns animals ­0.0033 ­0.0051** ­0.0034 -0.0025 0.0169* Source: Ajwad (2006) based on the 2005 IHDS data. ***=significant a 1% level; **=at 5% level; *=at 10 percent. policy factors in states on specific programs. when comparing across programs nationally, The findings ­ and several relative outliers ­ are state location ­ which acts as a proxy for various consistent with examples of high policy priority geographic, policy and institutional factors ­ has given in certain states to programs (e.g., Haryana highly variable impacts on participation, after offers a high elderly social pension due to controlling for all other household characteristics supplementation from the state; UP has a strong (figure 3.4). There is strong variation across programs in emphasis on school stipends; Tamil Nadu has a the importance of state location in explaining program universal PDS). participation. For most PDS-related variables, the state the within-state directions of the impacts on effect explains almost all that can be accounted for.150 participation vary notably across programs. For Other programs vary, but the effects of state location each state, the different directions of locational contribute substantially to the explained determinants impacts reflect in part different policy stances, of participation. There is also a general pattern of state but also the relative demands of specific program factors mattering more in programs which are demanding implementation. in terms of institutional delivery arrangements. Figure 3.4: Proportion of explained program determinants accounted for by state location, various programs, 2004/05 % of determinants explained by states School Life insurance Government and military Health insurance Ration card used in the Widow pensions holders holders Annapurna holders Government assistance scholarships Kisan card NOAPS Ration card used in the Bought PDS Bought PDS APL card BPL card AAY card grian in the grian in the grian in the used in the Bought PDS Ration card Source: Ajwad (2006) based on the 2005 IHDS data. 150 Thoughmuchremainsunexplained,andmoreworkisneededoninstitutionsandpoliciestolookfurtherintothe"blackbox"ofthelocationaleffect. Social Protection for a Changing India: Volume II looking at results from the poor states in particular, e. ConCluSionS AnD it is hard to argue that pds comes anywhere near ReCommenDATionS achieving its food security and poverty alleviation objectives. however, reform options for pds have the analysis of this chapter suggests a range of to be assessed with a clear acknowledgement of the reforms which would be desirable in protective context: technical, political economy, and possibly programs for india's poor. Overall, the need for legal.This context is of course strongly affected by also reform appears to be most acute in PDS, but more in the commitment of GoI to the Food Security Bill which the nature of incremental improvements for social would convert PDS from a scheme to a legislated right. pensions, which appear to be performing already These include: reasonably well (though it would be useful in the near future to assess the impacts of the changes introduced reformofPDSisinextricablylinkedtothepublic under IGNOAPS). For targeted housing programs, in procurement system. While the benefits of PDS particular urban housing programs, the options within to households are spread across India, the main the current paradigm are in the nature of incremental beneficiaries of public procurement of grains to reforms. However, market-based innovations in models feed the PDS are concentrated among farmers in of low cost housing for the urban poor suggest that a few states: Punjab, Haryana, some parts of Uttar experimentation with new paradigms would be well Pradesh, and Andhra Pradesh to a lesser extent. worth considering, and recent moves by GoI and Without some reform of the public procurement some states suggest that there is appetite for greater system ­ in terms of pricing policy with respect innovation. The options and recommendations for to the Minimum Support Price and the control individual programs are discussed in turn below. of government in grain procurement more broadly ­ there will continue to be large grain (i) TheBPLsystem stocks purchased each year which need to be drawn down. the results in this chapter and chapter 8 indicate that there is considerable scope to reform the current bpl there is a large internal bureaucracy running system, both the bpl census and ration cards. This is a PDSwhichislikelytoresistfundamentalreform crucial cross-program reform, and Government of India options which would undermine their role. The is currently discussing a serious overhaul of the BPL FCI alone employs around 450,000 people in system. Chapter 8 discusses in detail recommendations India, and SFCs a further number. This in itself is a for revamping the "backbone" targeting mechanism strong lobby which is likely to resist any changes for identifying poor households, offering both modest in the PDS which would imply either a potential reform proposals which would retain the existing role for the private sector in grain provision, or approach but with a range of possible improvements in more fundamentally a shift in use of the PDS design, and a more fundamental option for rural areas. subsidy from food purchase and distribution to cash for poor households. (ii) PDS despitetheshiftattheaggregatelevelfromfood clearly the "big elephant in the room" as far as deficittofoodsurplus,thereare­andwillcontinue safety net reform in india is concerned is tpds. This tobe­areasofthecountrywhichareperiodically is for a range of reasons which are presented in this and foodinsecure.This reality will continue to support other reports. Firstly, it remains easily the largest safety arguments for a direct food security role for the net program. Secondly, it has poor targeting outcomes, state. At the same time, penetration of private food and in many of the poorest states appalling levels of markets has increased sharply in recent years, so leakage which have persisted over many years. Thirdly, that the actual availability of food is there with the the procurement system on which it relies is becoming exceptions of some tribal and dry land areas, and increasingly strained in the face of gradual liberalization market infrastructure of well-developed and sub- of agricultural markets in India, both domestically and in markets are increasingly well-integrated. Thus, terms of openness to trade. even with the food crisis of recent years, in most CHAPTER-3: Protective Programs areas the challenge is not so much availability per anintermediatereformoption,whichwouldretain se as price and affordability. a food-based entitlement program but introduce the maturity of the PDS has naturally created private sector participation in grain procurement socialexpectationsthatitispartofthefabricof anddeliveryandamorefundamentaloverhaulof social policy. The results on awareness of PDS thePDSadministrationthroughuseofsmartcards. (see Chapter 7) are instructive in this respect, fundamental reform which allows for cash and the social constraints of any enforced transfers instead of food-based transfers, fundamental change in PDS should not be either when the state proves itself unable to under-estimated. fulfillitsfoodtransferobligationsorbyoffering theremayalsobelegalconstraintsontheability households the choice of grain or the cash ofthestatetowithdrawfromdirectprovisionof equivalentofthegrainsubsidy.153 grainsthroughPDS.The Supreme Court of India, the overall position of this report is that fundamental in response to Public Interest Litigation take up reform options for pds should be considered and by the People's Union for Civil Liberties in 2001 that offering households the option of a cash with Union of India, FCI and eventually all state transfer - while retaining the core food security governments as respondents, has clarified that and buffer stock functions of fci, and ensuring pds Article 21 of the Constitution on the Right to grains in areas where access is a genuine issue ­ is an Life also implies a right to live with dignity. The attractive option. While there is certainly potential to SC agreed that one aspect of this was a right to improve PDS performance (as the case of Bihar in Box 3.3 food, with the PDS a central plank of the right. shows), there remain a host of structural issues with The Court has made various orders subsequently a SP program so dependent on many intermediaries in efforts to improve the functioning of food- operating within such weak governance systems. In that based schemes.151 A point of note is that the initial light, the incremental approach represents a triumph order from 2001 indicated that "what is of utmost of hope over experience which cannot be expected to importance is to see that food is provided to the resolve the situation of PDS in much of the country. The aged, infirm, disabled, destitute women, destitute long term problems of PDS cannot be solved quickly men who are in danger of starvation, pregnant or with a single prescription for all states and all time. and lactating women and destitute children, However, offering options which allow households especially in cases where they or members of their choices in how they benefit from the massive public family do not have sufficient funds to provide food subsidy of PDS could allow for flexibility, greater for them".152 The implications are taken up below. efficiency, and accountability for the system. While the proposed Food Security Bill will potentially have a in light of this context, three reform options are major impact on the legally possible range of options, it presented for pds, which could themselves have is hoped that it is not too prescriptive, but leaves open variants borrowing from the approaches outlined. approaches which would appear to have the possibility Each approach is outlined in detail below.It is stressed to improve the welfare of poor households. In this light, that none of the options assumes any necessary the default provided for in the consultation draft of decrease in the aggregate level of public spending the Bill for provision of cash where the PDS system is devoted to social assistance for the poor: unable to provide adequate and decent quality grains anincrementalapproachtoreformwhichwould seems sensible. The system has so clearly demonstrated retain the current PDS model but with a host of its inability to do so for so long in so many places that improvementsinthepolicyandimplementation ruling out such an option seems likely to leave many systemstoincreaseefficiencyfromitsoftenvery poor households with a stronger legal right but no poorstate. better a real world situation. 151 Ausefulsummaryandbackgroundmaterialscanbefoundatwww.righttofoodindia.org. 152 SeeSupremeCourtOrder,of23rdJuly,2001. 153 ThisoptionhasbeentakenbyarangeofIndiancommentatorsovertheyears,includingmostrecentlyKapuretal.(2008)andPanagariya(2008). Social Protection for a Changing India: Volume II Box 3.3: PDS food coupon reform in Bihar ­ Better results in a poor state As noted earlier, Bihar has traditionally been one of the worst performing states in PDS, with very low coverage rates, extremely high leakage of grains and deep dissatisfaction among users. In response to this, the Government of Bihar in 2007 introduced food coupons in the PDS in an effort to generate greater awareness of rights, accountability and performance. It should be stressed that the program was not "state of the art", either in the technical design of the coupon or in the policy framework (e.g., in terms of linking coupons from one period to releases of FPS for the next period). Despite this, initial evidence from a World Bank-financed before and after evaluation indicates significant improvements in PDS performance. A pre-reform baseline was carried out with the Institute of Human Development in early 2007 and a follow up survey in 2008 just over one year later with a panel design. The survey was done in one urban district (Patna) and three rural districts (Aurangabad, Purnea, and Siwan) and the sample size was 1,692 households. Surveys were also conducted with FPS dealers and village figures, as well as key informant interviews and focus group discussions. The key findings, presented in Vashisht et al. (2009) were: overall, around 59 percent of eligible BPL households had received coupons, ranging from 74 percent in Aurangabad to 50 percent in Purnea. This figure was higher (84 percent) among AAY households, who were prioritized in all districts. information was critical, and the major sources of information on the coupon scheme were local leaders (30 percent), PDS dealers (28 percent), officials (15 percent) and friends/family (9 percent). prior to the reform, only 2 percent of rural BPL households were accessing PDS rice and around 60 percent of Patna BPL households. Following the coupon introduction, this rose to nearly half of rural BPL households. Those who did not have coupons continued to have very low access. the effective transfer value to those with coupons was just under 5 percent of median per capita income for BPL and 12 percent for AAY households, based on the PDS and market price differential. for those accessing PDS, the grains now represent around 26 percent of total monthly rice purchases for BPL and 48 percent for AAY households. around 90 percent of BPL households rated the coupon system an improvement over the previous system, mainly due to the sense of reduced cheating of the system by FPS dealers. there was a secondary market for coupons due to the lack of name and beneficiary ID on the coupon, and there were issues for illiterate beneficiaries. the coupon system per se is unable to address the underlying issue of misidentification of households in the BPL identification process but was seen to have increased public scrutiny of that process. delivery of grain by SFC remained erratic even with the reform and problems such as low dealer official margins are as yet unresolved. Source: Vashisht et al. (2009). the "incremental approach" to reform largely more days in a month. States such as Himachal follows the line of goi in its most recent evaluation Pradesh and Gujarat are already moving in this and the nine point action plan, and would include direction, which is supported by the Planning several key elements: Commission in its PDS report; (iv) ensuring that FPS have a minimum catchment area and introducing a range of measures to make turnover of grain which allows for sufficient scale honestly operated FPS financially viable, so that to ensure viability. The Planning Commission leakageisnotafinancialinevitabilityofthePDS estimates that substantial share of FPS fail to meet atthatlevel.The measures proposed ­ many of this benchmark. There is an obvious challenge which are in practice in some states - include: to enforce such a requirement in remote and (i) doorstep delivery of grains to FPS at the cost of inaccessible areas, which are likely to be those the state (currently being done in 17 states/UTs); most reliant on FPS. One option for "squaring (ii) provision of rent-free premises by panchayats the circle" in this respect could be the use of for FPS operations; (iii) allowing FPS to sell non- mobile FPS vans such as Himachal Pradesh is PDS commodities which will both increase their already using in a number of remote districts; and viability and increase the likelihood of being open (v) in line with Planning Commission suggestions, CHAPTER-3: Protective Programs increase the margin on grains for FPS to 2 percent household's allocation. The only constraint on of the economic cost. this is the FPS being open on a more regular deepening ongoing reforms in management of basis, but that should be addressed through the FPS.States ­ both rich and poor - are increasingly management reforms outlined above. A second moving from FPS operated by private operators measure which may be more challenging from a to operation by community-based institutions, bureaucratic perspective ­ but seems desirable ­ including PRIs, SHGs, and cooperatives (e.g., in would be allowing BPL and/or AAY households to Tamil Nadu in 2009, almost 90 percent of FPS carry over their monthly ration from one month are run by cooperatives). This seems eminently to another in months when they had less need sensible in terms of having operators who are for their full ration. Given seasonal fluctuations in PDS to market prices, this could imply some more accountable to their communities and more lumpiness. A third option would be allowing likely to operate the FPS on a regular basis. Results households a more flexible mix of grains and from the SPS surveys on safety nets are instructive other basic commodities up to the value of their in this respect. Despite kerosene distribution in ration subsidy, an approach which GoI plans to PDS being notorious for being high leakage, the pilot in the coming year (see below). incidence of kerosene from PDS in Orissa is far more progressive than that of grains. A simple strengthening monitoring of PDS operations. A explanation appears to be that kerosene has been number of measures to increase transparency the first commodity transferred to operation by in the operations of PDS are feasible within women's SHGs in the state.154 By early 2009, about the current framework and are indeed being 83,000 FPS (out of the 4.98 lakhs shops across implementing in several states. For instance, one the country) had been allotted to cooperatives, element of management reform may be more women's and other SHGs, PRIs, etc. direct community oversight in PDS operations, e.g., through community verification of PDS delivery decentralization of grain purchases. This would quantities where they are not already operating have several possible advantages, and a number the FPS and/or through the involvement of PRIs in of states are increasingly exploring this option in vigilance committees to monitor FPS. Some states their grain purchasing. Firstly, locally purchased have made some progress in implementing a grains are more likely to accord with local dietary system of monthly certification by PRIs/vigilance preferences, and hence be in greater demand. committees for delivery of food grains to FPS Secondly, local procurement could spread the and allocation to ration card holders on time. In economic benefits of the procurement subsidy addition, public display of BPL lists and of district from the current concentration in a few states and FPS-wise allocations of food grains for public (two of which are very well-off ) to farmers in scrutiny as proposed under the nine point action all states. Thirdly, the costs of transportation plan would improve transparency. The review of and storage of grains are likely to be reduced BPL/AAY lists in 14 states as part of this process considerably. A number of states have started to has already led to the elimination of 100.51 move in this direction, including lagging states lakh bogus/ineligible ration cards. Concurrent such as Chhattisgarh and Orissa, though how far evaluations of PDS performance by NCAER and this can advance before hitting resistance on the IIPA have been commissioned in recent years by supply side in traditional supply states such as the Department of Food and Consumer Affairs. Punjab and Haryana remains to be seen. strengthening the use of Information and makingoptionsforhousehold-levelofftakefrom Communication Technology (ICT) in the PDS at FPS more flexible. The most obvious reform in the beneficiary end in order to promote more this respect is obviously allowing, as a number robust identification and in reducing leakage. of states already do, for weekly drawing of the This can take a variety of forms, some ­ such as 154 SeeDevetal.(2007). Social Protection for a Changing India: Volume II in Andhra Pradesh already ­ simple biometric The system has some obvious risks, including identification but not fully integrated into a "smart" forgery of coupons. Another issue may be the system, others involving bar coding and other development of a secondary market for coupons applications. The range of options for greater ICT (though how much of a concern this is if poor use in the PDS system have been exhaustively households place more value on the cash than reviewed in a report for the GoI Ministry of Finance grains is a question). The Bihar experience with financed by the World Bank, which also conducted food coupons has been closely assessed through detailed feasibility assessments in Anand district a two round survey-based assessment (see in Gujarat and Thane district of Maharashtra in Box 3.3).156 While only around 60 percent of eligible 2007.155 In addition, the use of new technologies households had been provided with food coupons such as GPS for tracking movement of vehicles in the first year of implementation, access among transporting PDS commodities. Funds for piloting them dramatically increased as a result of the this technology in Chhattisgarh, Delhi and Tamil reform, rising from only 2 percent to around half of Nadu were sanctioned in 2007-08. In addition, BPL households. Bar coded coupons/ration cards computerization of TPDS operations have been have been introduced under TPDS in six states. initiated in several states. the above options can be considered a minimalist conductinganindependentreviewofFCI'spricing approach to pds reform. the option of allowing grain structure, with an eye to reducing the economic purchases with coupons from approved private food costs of grain procurement and handling. It retailers would represent a more bold form of pds appears that FCI is inefficient in its handling of reform which feasibility studies indicate is viable. grains, and the costs of this are naturally borne by Some of the potential attractions of such an approach the budget and the end beneficiary. Despite this, include competition between PDS and private traders previous reviews have not been very probing in (which could act as an accountability check and perhaps their efforts to look at possibilities for improved incentivize efficiency improvements in FCI and SFCs), efficiency in FCI. A truly independent review of greater frequency of availability in areas where FPS their cost structure with a prior commitment by GoI open irregularly, and possibly reduction in the transport to implement its findings would seem desirable and holding costs of PDS for the portion of grains (though the technical demands of "efficient" cost purchases from private traders. The United Progressive estimation in the face of such a dominant market position should not be under-estimated). Alliance (UPA) Government in 2007 announced plans to pilot such a reform in 20-50 districts, using coupons for another option is introduction of food stamps or beneficiaries which can be redeemed in neighborhood coupons,anapproachthatisprevalentinAndhra stores other than FPS ("kirana" stores), though there Pradesh and Rajasthan, and was introduced in appears to have been limited progress. In addition, Biharin2007.The current food coupon approach beneficiaries would be allowed to have a more flexible involves coupons which are redeemable only in form of ration, allowing for purchase of any mix of the FPS network, though there is no reason not to grains, pulses or other household basics up to the value allow for a coupon which would be redeemable of the coupon. This reform option was also mentioned also in approved private food retailers (see below). in the 2010 Economic Survey. While details remain to Coupons may have attractions as a tracking and be worked out, the initiative seems worth pursuing and leakage-reduction tool if FPS owners are able to evaluating.157 replenish their grain stock based strictly on the basis of redeemed coupons. In Andhra Pradesh, it a more technically advanced form of food coupons is estimated that the introduction of coupons has could be introduction of "smart cards" in the pds, reduced leakage in the PDS by up to 25 percent. which could be redeemable also at approved private 155 SeeCal2Cal(2007)foraPDSsmartcardfeasibilitystudyproducedatrequestofMinistryofFinance. 156 SeeVashishtetal.(2009)whichanalyzesurveysfromapre-reformbaselineandafollow-upsurveyjustoverayearaftertheintroductionoffoodcoupons. 157 TimesofIndia,July3,2007,"FoodcouponstoendPDSmess?",seehttp://timesofindia.indiatimes.com/news/india/Food-coupons-to-end-PDS-mess/articleshow/ 2165884.cms,lastaccessedonSeptember22,2009. CHAPTER-3: Protective Programs traders and/or fps.158. A smart card system can facilitate would introduce the option of cash transfers.159 two key functions: individual identification and remote The following paragraphs outline two variants of a transaction processing/storage. Smart cards have the fundamental reform option for PDS: (i) switching to a ability to store and record a large amount of program cash transfer for the poor, or offering options for poor and authorized biometric information (signature or households between food and cash; and (ii) switching fingerprint image) that can be matched to the actual to a conditional cash transfer, in order to leverage fingerprint or signature of individual involved in a improved human capital outcomes from the huge transaction. In the context of the food ration system, spending on PDS.160 Neither option would eliminate the the card can store information on the identity of the need for food buffer stocks, nor for the continued use of individual, eligibility for rations, quantity, price and food-based transfers for specific situations (e.g., where time intervals at which he/she could be supplied relief aid is needed due to climactic or other disasters) rations, etc. Point of Service (POS) terminals are simple or specific areas (e.g., remote or dry land areas), and machines that read the cards and have the capability possibly specific population groups (e.g., those noted of authorizing the transactions via phone lines or the in the Supreme Court order above, which largely internet or recording the transactions on the smart card conform to the AAY category presently). Despite the itself. Further information on smart cards is in Box 3.4. ongoing debate around the Food Security bill, Indian policymakers appear to be willing to explore such smart card systems are not new to india or to south options. For instance, Government of Delhi recently asia. Several experiments are already in operation, announced a "cash for ration" pilot in one district and the Rashtriya Swastya Bima Yojana (RSBY) health which would provide ` 1,100 per month to women in insurance program introduced by GoI in 2008 clearly poor households (an amount which would purchase demonstrates that smart card applications are very around 30 kg of rice at market price as of mid-2009).161 feasible for wide scale use even in poor and low The 2010/11 Economic Survey suggests that the GoI capacity settings, provided the right mix of institutional players is involved and there is rigor in standard setting is considering the introduction of food coupons for and implementation (see Chapter 5). Indeed, there is households with a lumpsum entitlement (rather than discussion presently of "piggy-backing" other programs specific amounts for rice, wheat, sugar, etc.) that can be on the RSBY cards, and this is technically possible. encashed at a PDS store of their choice.162 Another notable small scale private initiative is that of the elements of a transition from a food to cash Swayam Krishi Sangam (SKS) in operation in one of the based pds could be the following: poorest districts of Andhra Pradesh (in Medak). The card reform would have to start at the procurement is being used essentially as electronic passbook to record endofthesystemifitistobefeasible,andinvolve all the transactions for micro-finance loans. Recently, areorientationofFCI'sfunctions.This essentially GoI has provided in the 2008 and 2009 budgets funds involves a reduction in government controls for piloting of a smart card-based reform in Haryana and over grain markets and procurement operations. Chandigarh which would also allow access to authorized private dealers. Box 3.4 gives details on other South Even in such a reformed system, FCI could still Asian experiences. be expected to play important roles. Firstly, it could compete with private players in the market, the parlous state of pds as reported by goi, national relying on the economies of scale in operations, researchers and this report, suggests however that a existing infrastructure, distribution networks and more fundamental reform of pds is warranted which other advantages to be competitive. Secondly, 158 GoIhaslauncheditspilotofthisapproachinHaryanaandChandigarh.AndhraPradeshhasalreadygonesomewayinthisdirectionbyrecordingbiometric(iris) identificationofallBPLrationcardholdersinacentraldatabase.Thisallowedforsignificantreductionsinduplicateandghostrationcardholders.SeeCal2Cal (2007)foradetaileddiscussionofthetechnologicaloptionsandfeasibilityofsmartcardsinPDS. 159 See,forexample,RadhakrishanandSubbarao(1997). 160 Inessence,thisisthestrategythatMexicofollowedinitstransitionfromtheso-called"tortillasubsidy"totheCCTprogramProgresa(nowcalledOpportunidades). 161 See Hindustan Times, August 27, 2009: Pro-poor schemes get the thumbs-up from plan panel, see http://www.hindustantimes.com/Pro-poor-schemes-get-a- thumbs-up-from-plan-panel/H1-Article1-447365.aspx,lastaccessedonSeptember22,2009. 162 GoI2010. 0 Social Protection for a Changing India: Volume II Box 3.4: Smart cards in food programs existing systems in other developing countries make use of smart cards and PoS in the following way: the individual beneficiary receives the card from the welfare/Ministry office where it is "recharged" with the relevant information and the amount of rations he/she is entitled to. once the individual goes to the shopkeeper ­ public or private - with the card, the transaction is promptly recorded both on the card and on the POS device that each shop keeper maintains. the transaction is backed up on the handheld device (Point of Service - POS) for upload on the central database and for being transmitted to the bank to reimburse the shopkeeper. Such duplicate storage is to ensure that data is not lost in the interim period. Such mechanism allows for a decentralized processing of transactions when there is no online access to the main database as would be the case in much of India. once every week or fortnight, the shopkeeper would take the handheld device (or the information contained in it) to a local agency (e.g., a bank or the concerned Ministry office) to upload all the transactions to the centralized system and to be compensated for the transactions undertaken during that period, preferably transferring money to the shopkeeper's bank account. Smart card systems are not new to India or to the region. Several experiments are already in operation, and RSBY is a major case of apparent success. For example, ICICI has instituted a smart card system for health insurance, and the Army uses smart cards for its health cards. One notable small scale private initiative is that of Swayam Krishi Sangam (SKS) in operation in one of the poorest districts of Andhra Pradesh (Medak). The card is being used essentially as electronic passbook to record all the transactions for micro-finance loans. Andhra Pradesh is also piloting smart cards in its social pension program, Kerala operated a PDS smart card pilot in Trivandrum during the first half of the 2000s, and states such as Gujarat already use smart cards on a wide scale for purposes such as drivers' licenses. In neighboring Bangladesh, smart cards have been introduced by various institutions, notably Grameen Bank and BRAC, largely for channeling micro- finance activities, but also for other activities. In Pakistan, the National Identity Card is serving also as a smart card as it is used not only for casting votes, but for purposes such as scholarships. The information flow of smart card use in PDS is reflected in the graphic below from the Cal2Cal report: FPS OWNER's Smart Card (1) Update Records, Calculate Demand FPS OWNER (2) Generated Challan for Grain upliftment Database at Taluka (3) Payment to Bank FPS OWNER's Smart Card FPS OWNER's $ Smart Card (4) Permit Generation for Stock upliftment Bank (5) Card brought in for Grain upliftment Godown FPS OWNER's Smart Card Source: Cal2Cal (2008). (6) Card updated with stock issued and very importantly, FCI could retain a role as PDS grains, FCI could continue to perform that manager of India's buffer stocks of grain and distribution function in coordination with SFCs. perhaps their distribution in situations or areas if procurement reform could be achieved, where they were needed. Finally, to the extent substantialresourceswouldbefreedupforusein that specific groups or areas continued to receive acashtransferprogramforthepoor.In essence, CHAPTER-3: Protective Programs 1 public funds that are now spent on the large food reading of the SC order is that Government may subsidy bill of GoI would be freed up to be used for not be constrained if it provided sufficient cash direct cash grants to the poor. The analysis of this transfers to the identified groups to allow them report shows that regular cash grants are subject to purchase food. A more demanding reading to less leakage than food in nearly all cases, are far suggests that a legal obligation directly to provide easier to administer, and are highly valued by the food may apply to the groups noted. At the poor. There is also no solid or systematic evidence same time, the expansion of midday meals and from field work of the concern that cash grants Integrated Child Development Scheme (ICDS), are drunk or smoked by the men in recipient together with the existence of other programs households, though of course that cannot be for the destitute such as Annapurna (and of discounted as a possibility. course social pensions) goes some way towards meeting such an obligation if supplemented by there are several obvious challenges and issues in a cash transfer. Clearly, however, this would need such a proposal. They include: close consideration of the legal issues and their likely strong resistance from the stakeholders implications for a reform strategy. Another way of outlined above in the discussion of political approaching this may be to offer households the economy constraints. This is natural. The main option of grains or cash and letting them decide. way of managing this risk is to have a strategy for This would presumably meet the state's legal "reinvention" of the FCI along the lines outlined, obligations while offering choice to households so that it becomes an active player in the new in exercising their rights. system, with important residual functions like the existence of areas where food insecurity is buffer stock management, rather than simply a chronic. Despite the aggregate move to food "loser" in the reform process. surplus, there is a natural concern about ensuring resistance from FPS owner groups. Given the food security in these areas. It may be that existence of over 400,000 FPS owners, thought PDS continues in those areas until there is an would need to be given to how to manage this assessment of the impact of transition to cash in risk. The first element of a strategy would be other areas. continuing transfer of FPS functions to PRI/CBOs aggregate evidence indicates that caloric intakes who are not "sole purpose" organizations in the are falling in india, even among the very poor.163 way that FPS owners would be, and hence are less in light of this, moving to cash for pds may seem likely to resist changes that may be supported counter-intuitive. However, the counter-balances to by their communities. Secondly, a sub-set of FPS that in the above proposal would include: (i) buffer would continue to be needed for the continued stocks remain an important part of the system, PDS functions which would continue even in a possibly with a "buffer stock plus" to allow a more reformed system. Thirdly, the government could generous provision for grains for emergency and as necessary support transitional livelihoods special situations; (ii) the recent expansion of midday support for FPS owners who lost their businesses meals may for children provide a caloric floor (see as a result of reform (though this has not proven below); and (iii) it is proposed that AAY continue to necessary to date in states which have transferred provide food for the poorest, as would Annapurna FPS ownership from private individuals to for a share of the elderly destitute; and (iv) ongoing community groups). efforts to improve the performance of the nutritional legalimplicationswithrespecttoexistingSupreme program under ICDS could ­ if effective ­ provide Court decisions and the proposed Food Security improved coverage of poor infants and pregnant/ Bill.This is an important issue, as any reform must lactating women. Cumulatively, these by no means be consistent with the law of the land, more so suggest a wholesale withdrawal from direct provision when it flows from constitutional provisions. One of food for the poor. 163 SeeDeatonandDrèze(2008). 2 Social Protection for a Changing India: Volume II the second variant of a cash-based reform of pds pension administration on eligibility policies. would be introducing a conditional cash transfer Field work in several states finds either partial (cct) rather than a simple cash grant as outlined knowledge or problematic interpretations of above. The basic arguments for such an approach ­ and previous eligibility criteria on "destitution" (now the political economy and legal constraints ­ are similar modified to BPL) and even the reforms under NSAP to those for a simple cash grant, though conditionalizing are likely to face some continuing problems of transfers may face more legal complexities. The types identifying the poor. Following a review of policy, of conditions that could be considered would vary a more thorough awareness raising and perhaps according to household characteristics, but could development of implementation guidelines could include registration of girl births and school attendance help improve the situation. for households with children. For other household morebroadly,asprogressismadeonexpansion types, conditionality may not be appropriate, especially of social security to the unorganized sector, it those in the categories outlined above. The obvious willbecriticalforthosemanagingsocialpension attraction of a CCT approach relative to a simple cash policy development to be involved in thinking grant would be that PDS spending could be used to through options for pension expansion, and the leverage improved human capital outcomes, or other complementarities with social pensions as a socially desirable goals such as better treatment of girl possible "zero pillar" of any contributory system children. The additional challenges would include the fortheunorganizedsector. administrative demands of operating a CCT, but the centralmonitoringofNSAPperformanceshould Janani Suraksha Yojana (JSY) institutional birth grant be further strengthened, despite the reliance and other state-specific experience in India suggests on ACA rather than CSS financing mode. Social that such demands can be dealt with. pensions are one of the few major national in light of various constraints and given the programs for which simple consolidated fundamental nature of such a reform, it would be reporting on various performance indicators is advisable to experiment with a cash-based pds, perhaps not available in annual reports of the Ministry of focusing on more food secure states/districts first. It Rural Development. At a minimum, this would would also be sensible to ensure that the groups covered be desirable. A welcome recent initiative is the under the Supreme Court order continue to receive computerization of NSAP beneficiaries that grains, possibly in addition to a cash transfer. Such a significantly increases transparency and enables the central Ministry to monitor outcomes. States pilot would need careful monitoring and evaluation to are currently in the process of moving to a full- assess impacts and ensure that food security was not fledged MIS for NSAP.As importantly, the central compromised, but for poorer states in particular seems Ministry could act as a more effective conduit a more promising way of ensuring (albeit in an indirect for sharing of good state-level experience in manner) the right to food. different aspects of policy and implementation. (iii) Socialpensions experience suggests that the previous fund flow modelofsendingsocialpensiontransfersdirectly the evidence of this chapter suggests that social to DRDA level is preferable to routing through pensions are a reasonably effective part of the indian state treasuries, and that its reintroduction sp system, and their enhancement in coverage and should be considered. The strongest argument central financing in recent years is well directed. for this is evidence of even rich states such as In light of this, the reform options for social pensions Gujarat diverting social pension transfers for are more of an incremental nature, both in terms of other purposes, something that is less likely with relative spending/coverage priority, and in terms of DRDA routing. In addition, delays in on-payment improvements in current policies and implementation. by state treasuries could be avoided with such The main recommendations are: a model. However, this potentially entails a there is a need in a number of states for better policy decision to make the program a centrally educationofofficialsandotherinvolvedinsocial sponsored scheme. CHAPTER-3: Protective Programs theverificationprocessforcurrentbeneficiaries A second obvious improvement would be to bystatesneedstobemaderegular,andaprocess provide for direct transfers to households via the forremedialactiondeveloped.Presently, simple banking system or post offices. Such a reform in checks such as comparison of census and IAY in Bihar has been introduced recently, and field beneficiary information by district is typically visits indicate significant reductions in leakage of not done, and has proven very informative where funds due to the reduction in intermediation by has been done in identifying outliers on the up officials in the benefit payment process. and downsides (e.g., in Karnataka). This would conductadetailedstudyonpracticesamongthe allow a more targeted process of performance rural poor in housing improvement, to explore audit. In addition to state departments, there options for better matching scheme design with may be a potential role in such an enhanced needsandpractice. process for PRIs. a longer term option for targeted housing reform areviewofpaymentsystemsforsocialpensions may be exploring possibilities for more market-led wouldbedesirableinanumberofstates,aswould provision of housing, particularly in urban areas. athroughefforttoreconcilesometimesdivergent locallevelandstatelevelreportingonbeneficiary Experience in some metros may provide guidance here, numbersandotherinformation. and exploring the role of the private sector as financier and constructor but with subsidies to households seems (iv) Targetedhousingprograms warranted. There are a range of challenges in market- based solutions which are particularly acute for the while there are clearly some significant challenges informal and self-employed sectors of the workforce due in making the targeting and performance of targeted to difficulties in credit assessment, the need to mitigate housing programs more equitable and efficient, risks for lenders, and transactions costs. Nonetheless, specific recommendations on improvement are more small scale experiences of Micro Finance Institutions difficult to suggest. Nonetheless, some suggestions (MFIs) such as Swadhaar, Ujjivan and BASIX, and interest include: from small and medium housing finance companies attempt to develop and implement mechanisms indicate both interest in non-traditional market players in states for minimizing the significant rent- in serving the urban poor for housing upgradation seeking that appears to be happening in IAY. needs. Initial assessments indicate that market-based This is inherently challenging so long as benefits solutions may be possible for households with monthly involve large lump sums. However, the nature of incomes in the range of ` 4,500-8,000.164 In addition, rural housing needs and behavior among the poor there is a need to integrate housing delivery under suggest that the program can be better adapted IAY into a larger process of facilitating access to safe, to needs. sustainable and adequate housing. This could include access to credit in order to meet the funding gap ensure that safeguards in current guidelines between the IAY subsidy and the cost of housing as well are actually enforced. The recent introduction as access to environmentally sustainable technologies of "permanent waiting lists" based on the BPL and building materials and services to assist people in list and the requirement of displaying these construction. waiting lists in a public area (e.g., the wall of the panchayat building) is the first step in increasing the more fundamental question on housing transparency in the program. In this respect, programs for the poor is the extent to which they should options such as social audits seen in the Mahatma remain as standalone css, or might more usefully Gandhi National Rural Employment Guarantee be allowed for within an sp (or generalized) block Scheme may offer an additional mechanism. grant system.165 The JNNURM approach of having an 164 SeeMonitorGroup(2007)forausefulmarketassessmentofbothdemandandsupplysidesoflowcosturbanhousingoptions. 165 AproposalalongtheselineshasbeenmadebyKapuretal.(2008)usingIAYandSGSYfundsaspartofanenhancedPRIblockgrant. Social Protection for a Changing India: Volume II option within a larger program but requiring more (v) Conclusion pro-active proposal development and management on the part of local authorities in order to use central the recommendations above look at each protective funds for low-income housing seems worth exploring program in isolation, but would in a more strategic in the context of IAY also. This would be more framework also need to be considered jointly. consistent with giving states/cities greater local This overall interaction and framework for the social determination in use of CSS resources. Whatever the protection system is taken up in Chapters 6, 7 and 9. approach, it would certainly seem that innovation is In short, it stresses the critical importance of a social needed in the area of public subsidies for housing for assistance "pillar" of the Indian social protection the poor. It is also important to view these programs in system, but also the desirability of consolidation across the context of the larger housing and habitat policy for programs, probably with certain categorical top-ups on the country. benefits and possibly specific eligibility criteria. CHAPTER-3: Protective Programs CHAPTER 4 Public Works and Promotional Programs Chapter­4 Public Works and Promotional Programs In addition to social assistance programs, India has had experience across states provides a rich a long term emphasis on public works programs and opportunity for cross-state learning to achieve programs which aim to help poor households move better and more transparent impacts. A number sustainably out of poverty. In recent years, both have of innovations in MGNREG design ­ including been subject to increased political attention, funding, social audits ­ seem worthwhile to mainstream in and efforts to improve poverty reduction outcomes. other anti-poverty programs. This chapter looks at the experience with public works Like its predecessors, the Swarnajayanti Grameen in India. It also looks at three key promotional social Swarozgari Yojana (SGSY) remains a poorly protection programs: targeted subsidized credit for the performing program, and is currently being poor, school stipends, and the midday meal program significantly restructured under the National Rural in schools. Livelihood Mission (NRLM). The key findings of the chapter are: While school stipends are expanding, their Administrative data indicate high coverage of targeting performance remains modest, and there the Mahatma Gandhi National Rural Employment is a need to link their receipt more systematically Guarantee program (MGNREG) relative to to actual attendance rather than enrolment only. previous public works programs, with impressive The recent expansion of the midday meal inclusion of SC/ST and women workers. While (MDM) program is warranted by the findings of the design of MGNREG addresses some of the positive impacts on enrolment, but more work problems in previous public works programs, field is needed to understand their impact on both studies highlight several challenges in MGNREG other educational indicators such as retention, implementation. The uneven implementation and on nutritional outcomes. Government of CHAPTER-4: Public Works and Promotional Programs India studies also point to concerns on leakage Studies suggest that, at least prior to the National Rural common to other food-based SP programs which Employment Guarantee Act (MGNREGA), top-down suggest a need for more careful monitoring and planning continued to dominate, leaving little scope for evaluation of the delivery mechanism. community participation.167 the most recent, and fundamental, policy shift in 2006 has been from scheme-based provision of rural A. PuBliC woRkS PRogRAmS works to a legislatively-backed 100 day employment since the 1970s, public works have been an important guarantee under national rural employment component of the indian safety net, with a succession guarantee act (subsequently renamed mahatma of programs, both centrally sponsored and state- gandhi national rural employment guarantee act specific. The assessment below focuses on the major or mgnrega).169 The National Rural Employment centrally sponsored public works programs. Table 4.1 Guarantee Act was notified by the Indian Parliament describes the evolution of public works programs in in September 2005 and became operational from India since 1980. While changes in workfare programs February 2006. Initially it operated in 200 backward have been frequent, much of the development has districts, expanded to an additional 130 districts from been rebranding rather than fundamental reform.166 2007 (called respectively the first and the second phase In the 1990s, the most significant policy shift was of implementation).169 The remaining districts in the the increased role for PRIs. However, even this shift, country were notified under MGNREGA with effect which is clear in the guidelines of workfare programs, from April 1, 2008 (the third phase). Currently, the has been more mixed than policies might indicate. program covers the entire country with the exception of Table 4.1: evolution of public works programs in india, 1980 to 2006 year Scheme Primary objective Plan Period Allocation (` cr) 1980 National Rural Employment To employ unemployed and 6th 980 Programme (NREP) underemployed workers to build 7th 1,682 community assets 1983 Rural Landless Employment To provide 100 days of guaranteed 6th 500 Guarantee (RLEG) employment to one member from each 7th rural, landless household 2,412 1989 Jawahar Rozgar Yojana (JRY) Merged NREP and RLEG 7th 2,100 8th 15,434 1993 Employment Assurance To provide employment during lean 8th 5,340 Scheme (EAS) agricultural season 9th 5,660 1999 Jawahar Gram Samridhi Yojana (JGSY) To develop demand-driven rural 9th 6,267 infrastructure 2001 Sampoorna Gramin Rozgar Yojana (SGRY) Merged EAS and JGSY 9th 6,200 10th 21,440 2004 Food for Work Programme (NFFWP) To generate additional supplementary wage 10th 6,000 employment and create assets 2006 National Rural Employment Guarantee Act To provide 100 days of legally guaranteed 10th 11,300 (recently renamed Mahatma Gandhi employment to all rural households; National Rural Employment Guarantee Act create community assets and empower MGNREGA) communities 11th 111,000 Source: CSE 2008 and GoI Exenditure Budget. Note the 11th Plan allocation for MGNREG is till 2010/11. 166 CAGreferstothisprocessas"revampingandrechristening"(seeCAG2000). 167 Mehrotra(2008). 168 ThishasrecentlybeenrenamedtheMahatmaGandhiNationalRuralEmploymentGuarantee(MGNREG),tothisreportusesthisacronym,MGNREG,toreferto theprogram. 169 Duringthefirsttwophases,districtsthatwerenotcoveredbyMGNREGcontinuedtobecoveredbySGRYandFFW. 0 Social Protection for a Changing India: Volume II districts that have a hundred percent urban population. the guarantee makes mgnreg an innovative example MGNREGA guarantees up to 100 days of unskilled work of a rights-based approach to poverty alleviation. per rural household per year on public works projects. A description of some of the basic design features of The work is to be provided on demand and is paid on a MGNREG is provided in Box 4.1. There are several aspects piece-rate basis such that a normal worker can earn the of the scheme that distinguish MGNREG from earlier state-specific minimum wage rate. public works programs like the Sampoorna Grameen Box 4.1: key features of mgnReg 100 days of work per rural household per year guaranteed upon demand. Each rural household is entitled to a free job card with photographs of all adult members living in it. A job card holder may then apply for employment and the government is liable to provide the same within 15 days, failing which a daily unemployment allowance has to be paid to the applicant. Furthermore, work must be provided within 5 km of the applicant's residence or there is a 10 percent premium on the scheme wage. How the household distributes the 100 days among its members is entirely the household's decision. The scheme is therefore completely self-targeted and follows a demand-driven, rights based approach. State specific agricultural minimum wages used as scheme wage rate, but based on Schedules of Rates which are output-based. The central government in the Union budget of 2009-10 promised to provide a real wage of `100 per day for work done under MGNREG. Payment is based on a rural Schedule of Rates that depends on the amount of work done by a person. This SoR should be set such that an able-bodied worker should be able to produce the output which could earn him or her minimum wages. Wages are paid fully in cash; a departure from previous public works programs that typically had a food component. Wages paid directly into post office or bank accounts. In a move to counter corruption in fudging job cards and muster rolls and improve transparency in wage payment, the Government of India announced that all MGNREG wages, with effect from April 1, 2008, should be paid directly into worker bank or post office accounts that will be opened free of charge. The intention is that this would help separate the implementing agency from the payment mechanism thereby reducing room for fraud and harassment. Provision of adequate worksite facilities. The Act mandates provision of certain basic facilities at the worksite to facilitate manual labor. These include provision of shade; drinking water; crèches for women to leave their children; and first aid facilities to attend to the injured in case of an accident. Several provisions of the Act favor participation of women. For one, the Act mandates that one third of the workers be women. Second, it ensures equal wages for both men and women, with the latter typically having to do less work to earn the minimum wage as set in the Schedule of Rates. Finally, the scheme provides for childcare facilities for children below the age of six (if more than five such children are present at a worksite). This in addition to the provision of work locally (within 5 km of one's residence) makes MGNREG work a feasible option for women. Focus on labor-intensive rural development works, with a heavy focus on water/irrigation activities as well as connectivity. In addition to the list of specific types of works allowed under the Act, there is a provision for addition of other types of works based on consultations between the state and the central government. Overall the scheme maintains a 60/40 labor capital ratio. Contractors and machines are explicitly banned. PRIs and communities play a central role in MGNREG implementation. Unlike previous public works and SP programs, PRI leaders and communities are meant to identify the shelf of works through discussions in the Gram Sabha. PRIs also participate in the execution (at least 50 percent of works by value are to be implemented through the Gram Panchayat), supervision and monitoring of works (including through social audits). This represents a significant shift from earlier public work schemes such as the SGRY which revolved around central rules, with PRIs only involved in supportive implementation or distribution functions. A dedicated administrative structure for MGNREG implementation. Unlike previous public works programs, there is a specific provision of administrative costs, borne by the centre and supplemented by states. There is a dedicated cadre of MGNREG officials and functionaries at the district, block (program officer) and gram panchayat (employment secretary) level that are responsible for implementing the program. Emphasis on accountability through the use of ICT tools and by relying on communities and third party monitoring. Management information systems (MIS) have been established for administrative data on employment generated and assets created at the lowest level. Field based monitoring is supposed to be done through national field-level monitors reporting to the central Ministry as well as local vigilance and monitoring communities. Community-based monitoring through social audits is also happening in some states. In terms of total expenditure on the program, the Central Government bears 90 percent of all the costs. This includes wage costs and three quarters of the non-wage component (including materials and most administrative, subject to a maximum limit), working on an assumed 60:40 labor capital ratio. States are responsible for providing unemployment allowance should it fail to provide work within 15 days of it being demanded. This penalty creates a strong incentive for the State Government to provide work. Source: MGNREG operational guidelines (www.nrega.nic.in) and various government orders. CHAPTER-4: Public Works and Promotional Programs 1 Rozgar Yojana (SGRY) and Food-for-Work (NFFWP). First, National Rural Employment Guarantee is undertaken it recognizes the `right to work' as a legal right. The state using administrative and household survey data from government is legally bound to provide employment various rounds of the National Sample Surveys (NSS) to a household within 15 days of its demanding work. and the 2005 India Human Development Survey (IHDS) If it fails to do, the household is entitled to receive an survey. In the absence of nationally representative unemployment allowance. In this the MGNREG is survey-based data for MGNREG, an important source of different from the supply based approaches adopted by information on program performance is administrative earlier public work schemes and is closest in spirit to the data reported by the official website (www.nrega.nic.in). Maharashtra Employment Guarantee Scheme (MEGS) The chapter also draws on the large number of studies initiated in the 1970s. Second, wage payments under on policy design and implementation experience MGNREG are entirely in cash, unlike previous programs of the program in the last four years.171 This is not a that also had a food component. Third, MGNREG was the comprehensive review; instead the discussion here first social protection program that devolved significant focuses on performance and implementation challenges resources to the gram panchayat level. Another key based on the experience to date. Though the majority of innovation was the detailed operational guidelines at the these studies are small-scale and based on purposively centre and state defining procedures and clear division selected samples (typically poor and backward areas), of responsibilities for implementing the program (i.e., a surprisingly consistent story emerges with respect to unbundling safety net service delivery). Finally, the MGNREG's performance in the field. design of MGNREG emphasizes to a greater degree than in previous public works (and other SP) programs (a) Coverage of public works programs transparency mechanisms and accountability. prior to mgnreg (and its immediate predecessor - The discussion below focuses on the performance of nffw), administrative data indicate that provision of MGNREG as well as its main predecessor programs, i.e., public works employment under sgry had been well the Sampoorna Grameen Rozgar Yojana (SGRY) and the below the target 100 days for the rural poor.172 National Food-for-Work Scheme (NFFW). The discussion Aggregate employment generated by the main public examines the coverage and targeting of these programs works programs in the first half of the 2000s (prior to and presents key challenges with respect to program the advent of NFFW and MGNREG) was just 600 million implementation. The discussion draws on the extensive person days ­ around half the level of the mid-1990s.173 literature on public works in India, including a Ministry of While not direct coverage rates, Figure 4.1 presents total Rural Development sponsored study of the Sampoorna number of "persondays" generated in public works by Grameen Rozgar Yojana (CMD 2005) and concurrent state (using administrative data) relative to BPL and evaluations by the Planning Commission (2000) and agricultural households in 2003/04. The provision of the Comptroller and Auditer General CAG (2000) for the employment on average was below 10 days per BPL Sampoorna Grameen Rozgar Yojana, CAG (2007) for the household nationally and in most states, and less than National Rural Employment Guarantee).170 The analysis 5 days looking at all agricultural laborer households in of coverage and targeting in public works programs all states except Assam. There is also some cross-state prior to the introduction of the Mahatma Gandhi variation in administrative persondays under public 170 Inaddition,someoftheearlieststudiesfocustheMaharashtraEmploymentGuaranteeprogram,includingAcharyaandPanwalkar(1988),DevandRanade (2001),GaihaandImai(2002),Gaiha(1997)andRavallionandDatt(1995)ontheMaharashtraEmploymentGuaranteeprogram. 171 TheseincludefieldstudiesonbehalfoftheMinistryofRuralDevelopmentbytheProfessionInstituteNetwork(e.g.,CBGA2006forAndhraPradesh,Chhattisgarh, JharkhandandMadhyaPradesh;IHD2006,2007forBihar,CDA2006forGujarat,ISWSD2006andothers­availableathttp://www.nrega.net/pin/)andthe ComptrollerandAuditorGeneral'sreportonMGNREGinPhaseIdistricts(CAG2007).ThereisalsoanimpressivebodyofresearchbytheRighttoFoodmovement (availableatwww.righttofoodindia.org,seealsoDrèzeandOldiges2009,DrèzeandKhera2009,BhatiaandDrèze2006,Khera2006,Sivakumar2006)andothers publishedintheEconomicandPoliticalWeekly(e.g.,Mehrotra2008,Ambastaetal.2008,Shah2007,Jhaetal.2008,VijayShankaretal.2006,AakellaandKidambi 2007a,b,AiyarandSamji2006,Narayanan2008);andsomeotherinstitutionalstudies(e.g.,CSE2008,PRIA2006,2007,2008,Sharmaetal.2009).Seeforthe referencesattheendofthereportforfulllistoffieldstudiesandotherpapersconsulted. 172 Despitecommitmentstofullruralprogramcoverage,theshareofvillagescoveredbyemploymentschemespriortoMGNREGhasbeenfarlessthancomplete. Analysisfrom2002NSSvillagedataindicatesthatonly49percentofvillages(56percentofpopulation)reportedanypublicemploymentprograminthepastyear (O'KeefeandPalacios2006).referencesattheendofthereportforfulllistoffieldstudiesandotherpapersconsulted. 173 O'KeefeandPalacios(2006)basedonMinistryofRuralDevelopmentdata. 2 Social Protection for a Changing India: Volume II Figure 4.1: SGRY workdays per agricultural worker and per BPL household by state, 2003/04 30 25 Ag. workers 20 BPL HHs Days public works 15 10 5 0 AP TN HP UP Raj MP Ker WB Guj Har Oris Punj Karn J&K Uttar Bihar Jhark Maha Chhat Assam Source: O'Keefe and Palacios (2006) based on Ministry of Rural Development; Labor Year Book 2002-03; BPL from indiastat.com. works. These figures are pure averages based on household survey data confirm low and falling aggregate administrative figures, and so need a survey- coverage on public works from the late 1980s to early based source for more focused information on target 2000s (prior to mgnreg). Figure 4.2 presents results from groups of interest.174 However, they clearly indicate the last four large NSS rounds. At the all-India level, the that the program commitment of up to 100 days coverage rate is lower in each subsequent period, falling employment per rural poor household in SGRY was from 4.8 percent in 1987/88 to 1.9 percent in 2004/05. never close to being met. Though works reliance is likely to vary by year depending Figure 4.2: Household coverage rates of public works by state and all-India, 1987/88 to 2004/05 18 16 % HH participating in public works 14 12 1987-88 10 8 1993-94 6 1999-00 4 2004-05 2 0 BI KE AS AP KA RA TN HP PU UP OR HA GU MP WB India MA Source: O'Keefe and Palacios (2006) and report team estimates using various rounds of the NSS. 174 There are problems with administrative data with respect to employment generation and implied coverage of SGRY and NFFW. Calculation of "mandays" in administrativedatahashistoricallybeendonesimplybytakingtotalprogramexpenditure,assuminga60:40labortomaterialsratio,andbackingoutthenumber ofworkdaysgenerated.Surveyresults,andpreviousanalysisofworkfareschemesbytheComptrollerAuditorGeneral(CAG),PlanningCommission'sProgramme EvaluationOffice(PEO)andothers,indicatethatthisresultedinquitemisleadingadministrativeestimatesofemploymentgeneratedbypublicworks(e.g.,wage sharesintotalworksspendinginthelate1990sinBiharandOrissawere28and32percentrespectivelyaccordingtoPEO). CHAPTER-4: Public Works and Promotional Programs on exogenous factors, most states (except Maharashtra of its implementation and covering just 200 districts, and Orissa) record much lower employment generated 38 million rural households had been issued with a job in 2004/05 compared to earlier years. In a number of card (see Table 4.3). By 2008/09, this figure has more cases, these state-specific coverage rates exhibit far than tripled to 101 million rural households registered. more pronounced inter-temporal variance, sometimes In the same year, about 45 million households ­ a third with a different pattern to the all-India (see Figure 4.2). of rural households ­ were provided employment. In addition, the NSS until recently only reported those Assuming an average household size of five, the number households as covered where more than 60 days of of participating households was higher than the entire public works was undertaken in the previous year, population of Bangladesh.175 This is unprecedented scale which suggests that such coverage figures are very when compared to any similar program in the world. In much a lower bound in light of the much lower average the Indian context too, this is a major achievement with employment estimated. This has been changed from the respect to program coverage relative to prior public 62nd round in 2005/06 to a "how many days employed on works schemes. For instance, SGRY and NFFW together public works" type question. Results from other state- generated 1,116 million person days in 2005/06. In specific surveys in fact indicate a significantly better comparison, MGNREG had generated 905 million person coverage performance in poorer states (see Table 4.2). days in the first year of implementation alone, and with partial coverage of 200 districts. this trend has been reversed with the introduction of mgnreg which has emerged as among the largest administrative data indicate that employment safety net programs in the world. Within the first year generated under mgnreg has been much higher than in previous public works programs, though Table 4.2: Pre-mgnReg public works coverage rates still less than the 100 day guarantee. The average from state-specific surveys number of person days of employment generated per % hhs participating participating household has increased from 43 to 52 (48) orissa mP Jharkhand person days for Phase I districts (all districts) between SGRY 22.9 4.0 9.7 2006/07 and 2008/09 (see Table 4.3). Although just 10 percent of participating households exhausted their FFW 6.3 17.8 4.4 100 days entitlement in 2006/07, this figure rose to about Source: Dev et al. (2007) for Orissa and MP, based on the 2006 SP survey (the sample size of participating households in the Karnataka sample was too 14 percent by 2008/09. Interpreting these numbers is not small); Balachander et al. (2009) for Jharkhand, based on 2006 JLSS. simple as they are likely to reflect a certain rationing of Table 4.3: Coverage of mgnReg using administrative data, 2006/07 to 2008/09 2006-07 2007-08 2008-09 Phases i i ii i+ii i ii iii i+ii+iii Households issued job cards (mn) 38 44 21 65 47 25 27 100 Households demanded employment (mn) 21 .. .. 34 .. .. .. 45 Households provided employment (mn) 21 23 11 34 22 12 12 45 % Rural households provided employment 39% 43% 33% 39% 40% 35% 23% 33% Number of person days emp. generated (mn) 905 1087 350 1437 1131 509 523 2163 Number of person days per HH employed 43 47 32 42 52 44 45 48 Number of person days per rural household 17 20 11 22 21 15 10 16 % Registered households provided emp. 56% 53% 52% 52% 46% 46% 43% 45% HHs provided 100 or more days of emp. (mn) 2 3 1 4 3 1 2 7 % participating HHs provided 100 or more 10% 13% 5% 11% 15% 10% 18% 14% days of employment Source: MoRD administrative data from the state-wise Monthly Progress Reports (www.nrega.nic.in), last downloaded in January 2010. Data on number of rural households from 2001 Census. 175 DrèzeandOldiges(2009). Social Protection for a Changing India: Volume II employment as field studies report widespread unmet is "demanded" under MGNREG. While the scheme is demand for work. Secondly, state-level data indicates framed as an on-demand and rights-based one, in significant variation across states on employment days practice, worksites have to be opened in a locality generated per household and the share of households before people are likely to come forward for work. exhausting entitlements, e.g., nearly half the households This in turn requires a quorum of people expressing a in Rajasthan as against negligible shares in most states demand for work. This "chicken and egg" issue suggests (more on this below). However while the provision of the importance of the supply side even in such a 100 days of guaranteed employment remains a distant demand-driven program, a point reinforced by the fact goal, employment generation under MGNREG has been that provision of employment is very high once workers much higher than under any other wage employment have "demanded" employment in program terms. This is program in the past ­ it provided nearly 43 person days discussed further below. of employment per household in its first year (despite in addition, work provision in mgnreg, like previous limited coverage) compared to 26 person days generated public works programs, does not adequately reflect by the pan-India SGRY in 2005-06.176 seasonality. At the national level, aggregate annual employment generation in phase i districts is higher public works employment, prior to MGNREG, actually than that in the phase ii and phase iii districts, but ran counter to expectations, with lower employment has recently stalled. The Phase I districts are not only provision in years with lower than average rainfall for the among the poorest in the country (and hence one most part (though this should be interpreted cautiously, would expect higher demand for the scheme) but have as both rainfall and employment provision are highly also had relatively more time to build and establish variable locally). Evidence from the Maharashtra processes for the smooth functioning of the program.177 Employment Guarantee Scheme also confirms In consequence, a matter of some concern is the stalling employment drops sharply during the monsoons when of employment generation in Phase I districts between market-based work is also least likely to be available.178 2007/08 and 2008/09 at 21 person days per rural This seasonal pattern appears to have continued and household (see Table 4.3). employment in MGNREG is also seasonal, peaking roughly in February-June and reaching its lowest despite the rise in registration, there is a significant, usually in July-October (see Figure 4.3).179 This is partly and increasing, gap between those registered with because the majority of works allowed under MGNREG mgnreg job cards and those actually demanding (more on this later) cannot be executed during the work (see table 4.3). At the same time, the gap between monsoons. This is reinforced by budgeting practices those demanding and provided work is negligible. which concentrate disbursements for works in the final This indicates at least two likely features of demand quarter of each fiscal. manifestation. The first ­ and benign element of the gap - is that households may well register with no program outcomes have been quite variable across current intent to undertake public works, exercising an states and even across districts within states, implicit insurance/guarantee function of the program, reflecting both demand and supply side effects. which has value even if not exercised (e.g., by possibly Greater variability can be seen on the supply of actual reducing need for seasonal migration, or by allowing employment to those registered.180 States such as riskier but higher yield cropping choice). The second ­ Andhra Pradesh, Chhattisgarh, Madhya Pradesh and and more worrying - is the process through which work Rajasthan stand out as high performers with respect to 176 Mehrotra(2008). 177 DrèzeandOldiges(2009). 178 SeeAcharya(2005)andSubbarao(2003)forseasonalityintheMaharashtraEmploymentGuaranteeScheme. 179 Seewww.nrega.nic.inforadministrativedataonemploymentgenerationbystateandbymonth. 180 Similarly,inOrissa,employmentgeneratedperruralhouseholddeclinedbetween2006-07and2007-08afterreportsofembezzlementin2006-07.Thiscould havebeenbecausegreatervigilancepostsocialauditsresultedin"real"reportingasopposedtoinflatedfiguresthatwerereportedearlier(asupplysideeffect). Alternatively,thiscouldhavebeenbecauseemploymentactuallydeclinedwithcorruptionplayinga"dampening"roleondemand.DrèzeandOldiges(2009) suggestthatbothfactorsmayhaveplayedarole.Insomestates,lowdemandmaydrivelowparticipation;e.g.,Keralaprovidesoneexamplewithyouthturning upforMGNREGworkbutgoingawayoncetheyrealizedthatitinvolvedmanuallabour(JacobandVarghese2006). CHAPTER-4: Public Works and Promotional Programs Figure 4.3: Seasonality in MGNREGs work provision, 2006/07 to 2007/08 900000 Madhya Pradesh 800000 700000 Andhra Pradesh 600000 Uttar Pradesh No. of households 500000 average of 10 400000 major states 300000 200000 100000 0 Oct 06 Dce 06 Feb 07 Apr 0 7 Jun 07 Aug 07 Oct 07 Dec 07 Feb 08 Apr 08 Jun 08 Source: Ravi and Engler (2009). the percentage of participating households as a share of the provision of work rather than a lack of demand. rural households as well as person days of employment The contrasts are even sharper at the district level with generated per rural household (see Figure 4.4). However, Madhubani in Bihar generating only one person day per the program is yet to make a dent in states like Bihar rural household in 2006-07 in contrast to 111 person and Orissa where the incidence of poverty and hunger days in Dungapur, Rajasthan.181 Rajasthan, in particular, are high, indicating likely supply-side constraints in has easily the highest employment ratio among major Figure 4.4: Coverage of MGNREG by state, 2008/09 100 100 % Rural HHs provided employment 80 80 Person days per rural HH 60 60 40 40 20 20 0 0 Punjab Haryana Kerala Gujarat Maharashtra Karnataka Orissa Jharkhand West Bengal Bihar Uttarakhand Uttar Pradesh Tamil Nadu Assam Himachal Pradesh Jammu & Kashmir Andhra Pradesh Madhya Pradesh Chhattisgarh Rajasthan % Rural HH emp Person days per rural HH Source: MoRD administrative data from the Monthly Progress Reports (www.nrega.nic.in). Notes: States sorted in ascending order of employment generated under MGNREG. 181 DrèzeandOldiges(2007). Social Protection for a Changing India: Volume II states, reflecting partly its long history of public works terms.183 Like most programs, however, the considerably as part of famine relief, a massive public awareness higher average benefits among the richest make the effort on the part of civil society in the state and focus benefit incidence somewhat less progressive (though on enforcing transparency mechanisms. still the second best program on that indicator also - see Chapter 8 for a comparison of targeting outcomes across programs). Note that the two data sources are (b) Targeting of public works programs not directly comparable. As mentioned earlier, the NSS estimate only includes households where more than prior to mgnreg, the targeting of previous public works 60 days of public works was undertaken in the previous programs appears to be have been more progressive year. The IHDS estimate includes only households where than all other social protection programs (except at least one individual reported participating in public annapurna), with the coverage rate in the poorest works as one of their main occupations in the previous quintile and among st households high relative to year. As a result, both estimates are very much a lower other groups.182. Results from both IHDS and NSS for bound estimate of coverage. SGRY/FFW from 2004/05 are presented below. Analysis in Table 4.4 indicates very progressive coverage across the positive targeting performance of pre-mgnreg expenditure (NSS) and wealth groups (NCAER), though public works programs from national data is overall coverage is low even for the poorest. Nonetheless, supported by recent evidence from a three state study the high relative coverage rates in the bottom quintile and of sp programs in orissa, karnataka and mp. Particularly particularly among ST households places public works for the Food-for-Work program, coverage in the lowest as the second best program performer in distributional quartile was relatively high (around 57 percent higher than the population average), while coverage among ST households was well above the population average, Table 4.4: Coverage of SgRy/FFw by expenditure/ by almost 80 percent in the case of SGRY.184 Similarly, wealth and social group, 2004/05 Planning Commission reports find evidence of relatively % hh % hh Benefit incidence good targeting in these programs ­ e.g., a large share of group (nSS) (ihDS) (ihDS) SGRY belonged to households below the poverty line, Poorest 2.8 1.5 34.2 lived in kutcha houses and had received only a primary Q2 2.4 0.9 17.5 education or less.185 Q3 2.0 0.8 16.2 Q4 1.3 0.5 15.3 earlier analysis on public works schemes indicates that average benefit incidence (such as that presented Richest 1.1 0.2 16.9 above) underestimates the gains to poor households OBC 1.4 0.7 36.3 from increased spending on works.186 The literature on SC 2.5 0.6 17.1 public works schemes (and transfer schemes generally) ST 3.5 2.8 33.8 typically assesses targeting in terms of average Other 1.5 0.4 11.4 incidence. This may be misleading in cases where there Total 1.9 0.8 100 are marginal adjustments in budgets (as is happening Source: NSS: Bank staff estimates based on 2004/05 NSS; IHDS: Ajwad (2006) based on 2004/05 IHDS data. through MGNREG). Table 4.5 presents the results based Note: Estimates refer to coverage across quintiles based on monthly per on 1993/94 NSS. The marginal odds of participation are capita expenditure (NSS) and asset ownership (IHDS). Benefit incidence the highest for the poorest quintile, the probability of refers to the percentage of benefits captured, estimated assuming equal number of work days across quintiles. The nature of the works question in participation of those in the poorest quintile is higher NSS does not permit benefit incidence analysis. than those in the richest quintile. As a result, expansion of 182 This finding is consistent with international evidence of targeting performance of public works programs. For example, in Argentina's Trabajar program, 80percentofbeneficiarieswereformthepoorestquintile,whileinChilethesharewascloseto100percent.InBangladesh,around70percentofbeneficiariesofthe Food-forWorkProgramwereinthelowestincomebracket,whileinIndonesia,thepost-crisisthanadministrativetargetingmethodspublicworksprogramwas muchmorelikelytoreachhouseholdwhohadsufferedlargeshocks.SeeSubbarao(2003)anddelNinnoetal.(2009)forasummaryofinternationalstudies. 183 Notethattheaveragehoursworkedperdayandthemedianwageratesdonotdiffersignificantlyacrossthelowerfourquintiles. 184 Devetal(2009). 185 PlanningCommission(2000). 186 LanjouwandRavallion(1999). CHAPTER-4: Public Works and Promotional Programs Table 4.5: Average and marginal odds of participation participating SC and ST households was consistently in indian public works programs, 1993/94 higher than their share in the country's population Average odds of marginal odds of (about 16 and 7 percent respectively). More encouraging participation in participation in were numbers on participation of women ­ nearly half quintile workfare (A) workfare (m) the person days generated in MGNREG are for women. Poorest 1.23 1.16 Only in five states (Assam, Jammu and Kashmir, Punjab, Q2 1.13 0.93 Uttar Pradesh and West Bengal) was the share of women Q3 1.04 0.80 lower than one-third. This is a major improvement in Q4 0.86 0.92 female participation relative to SGRY, estimated at only Richest 0.83 0.55 12 percent all-India by a Government of India evaluation (similar to CAG findings on JRY and EAS of only around Source: Lanjouw and Ravallion (1999). 16 percent female beneficiaries).188 the program is likely to be pro-poor. This is a particularly evidence from field studies also indicates that the important finding in terms of the potential of MGNREG program is reaching vulnerable households, but for including the poor in a more intensive manner than further confirmation by national or state-level previous public works schemes. household surveys is required. Even in the initial stages of implementation in 2006, several studies indicate that mgnreg is a universal right rather than a targeted coverage rate among SC/ST households was high in program, but administrative data indicates impressive relative and absolute terms with respect to registration inclusion of sc/st households and of women relative for work. In their survey of 100 MGNREG worksites to previous public works schemes (see Figure 4.5).187 spread across six northern states, Drèze and Khera (2009) . The share of SC households in MGNREG employment find that a majority of those employed in the program increased gradually while that of ST households belong to the most disadvantaged sections of society. fell during this period. This may be on account of a SC/ST workers together accounted for over 70 percent concentration of STs in phase 1 districts and a gradual of their sample, 81 percent of their sample workers decline thereafter as the program spread to other lived in a kuccha house, 61 percent were illiterate and districts of the country. Nonetheless, the proportion of 72 percent had no electricity at home at the time of the survey. Shariff (2008) finds that holders of the Antodaya Figure 4.5: Participation of women, SC and ST workers (AAY) ration card are more likely to participate. Gaiha in MGNREG, 2006/07 to 2008/09 et al. (2010) find in their study of Andhra Pradesh, 60 Maharashtra and Rajasthan, that SC/ST and landless 50 49 workers were more likely to participate in MGNREG. 43 40 41 Despite this, benefit incidence in terms of days worked 36 31 29 on the program is regressive in AP and Maharashtra, i.e., 30 25 27 25 relatively better-off workers worked more days than 20 acutely poor workers. This is in contrast to Rajasthan 10 where poor participants benefitted disproportionately 0 more than others. These nuances in the targeting Percentage of Percentage of SC Percentage of ST performance of MGNREG point to the importance of women 2006/07 2007/08 2008/09 household-survey based information for examining these issues nationally. A recent study using 2006/07 Source: www.nrega.nic.in. Notes: This refers to the share in person days of employment provided for NSS data for Phase I districts in fact indicates that less women, SC and ST workers. than half the wages accrue to the officially poor.189 187 Notethatlikemostpublicworksprograms,thenatureofthemanuallaborinvolvedresultsinalowproportionofdisabledprogrambeneficiaries(lessthanone percentoftotalparticipantsin2008-09). 188 SeeCMDstudyforMinistryofRuralDevelopmentonSGRY(2005);PEOEASevaluation(2000). 189 Bhalla(2010). Social Protection for a Changing India: Volume II (c) implementation and transparency safeguards. Another challenge is that the program aims to not just generate short-term public works programs prior to mgnreg were plagued employment but also to create sustainable livelihoods by major implementation problems related largely for the future. This has implications for the quality and to lack of accountability in public works design, relevance of assets created. financing and management. Several issues emerged from government and external assessments of public progress in implementing mgnreg has been uneven works prior to MGNREG.190 These include issues such as and both administrative data and field studies note top-down planning of works, with little community or the highly variable roll-out pace and effectiveness PRI participation; poor quality of assets created; leakage across states. This variation is attributable, at least of funds and parking of significant funds in deposit in part, to differences in the level of preparedness and other accounts (parked funds show up in public and innovation by state governments. There has accounts as executed but in fact remain undisbursed - been considerable state-level innovation in the areas see Chapter 6); evidence of corruption in many forms of accountability and community-based monitoring (e.g., fudging of muster rolls, payment of lower than (e.g., social audits), use of technology for monitoring prescribed wages, presence of contractors, etc.); lack (e.g., a web-based Management Information System) of reliable records, in particular muster rolls; and lack and financial inclusion (e.g., biometric smart cards of both strong administrative M&E and absence of for payments through post office accounts). Many of "bottom up" monitoring through mechanisms such as these innovations have since been mainstreamed social audits. nationally into program design. States like Andhra Pradesh, Rajasthan and Tamil Nadu for instance have the design of mgnreg assuages several of these taken active steps to establish institutional structures concerns. MGNREG was the first program in India and systems for monitoring and accountability that had an exceptionally detailed institutional (see Box 4.2). architecture and implementation processes outlined in the operational guidelines issued by the Ministry a large number of field studies of implementation of Rural Development.191 This unbundling of experience in a number of states provide useful insights service delivery and clear assignment of roles and on the implementation experience of mgnreg to date.192 responsibilities is critical for effective implementation. PRIs and communities are given a central role in Demandforwork planning, execution and oversight of MGNREG, the notion of "demanding" and applying for work has signaling a significant shift in design. There is also a not yet been firmly established. In principle, people strong emphasis on transparency and accountability have to demand work in order for work to be provided. in MGNREG, a design feature that is highly relevant for In practice, very few job card holders formally apply other anti-poverty programs. for work while the majority tend to wait passively for despite this strong design, some of the challenges work to be provided. At the same time, there appears noted above remain, while the demand-driven design to be considerable latent demand for work ­ i.e., not of mgnreg has introduced new challenges in ensuring all people who demand work are provided work, while a right to work. This requires a high degree of awareness even those who are provided work would like more of rights and entitlements under the program and the days of employment. For instance, almost all workers process through which to "demand" and apply for work. interviewed at 100 MGNREG worksites in six northern This also requires funds, institutional structures with states were ready to work for 100 days in a year, whereas adequate and appropriately skilled staff, and strong M&E only 13 percent had actually done 100 days of MGNREG systems to ensure adherence to operational guidelines work in the previous year.193 Part of the reason is poor 190 SeeCAG(1997and2000);PEO(2000);Nayaketal.(2003),ARC(2006)andmorerecently,Mehrotra(2008). 191 Themostrecentversionoftheseguidelinesisavailableatwww.nrega.nic.in;stateshavealsoissuedstate-specificguidelinesfollowingthecentralguidelinesasaformat. 192 Seefootnote173andthereferencesattheendofthereportforadetailedlistoffieldstudiesconsulted. 193 DrèzeandKhera(2009). CHAPTER-4: Public Works and Promotional Programs Box 4.2: Some state-level innovations in implementing mgnReg Rajasthan According to official statistics, 90 percent of rural households in Rajasthan were provided MGNREG employment in 2008/09 of which 41 percent were provided 100 days of employment. While official figures may overestimate the coverage of the scheme, Rajasthan is also well known for implementation of various transparency safeguards ­ a fact corroborated by many localized surveys. Muster rolls are present at most worksites, daily attendance is taken in front of workers, wages are paid based on group work and measurement and job cards are updated at the time of payment. Awareness of entitlements and processes is high. An active civil society played an active role in the campaign for the right to work and the right to information and has strongly promoted community monitoring through social audits. In addition, the state government has put in considerable effort to improve worksite management practices. Mates (work site supervisors) are trained to maintain muster rolls; they assign tasks to workers; record their output; and ensure worksites facilities are available. All such activities help improve productivity and increase transparency, allowing the program to be scaled up. Some problems persist ­ e.g., Rajasthan continues to have a poor record on payment of minimum wages and has not yet revised its Schedule of Rates to make it MGNREG- friendly. On the whole, however, Rajasthan's success could be attributed to the long history of public employment programs in the state and more importantly to an active civil society that has for years worked on other related movements like the Right to Information. Andhra Pradesh The movement here is driven by the government which has undertaken significant efforts to innovate and improve the delivery of the program, setting up administrative machinery and systems in readiness for MGNREG. Some of the best practices adopted by the government include a focus on monitoring and accountability by using ICT tools and relying on communities. The former included the development of a web-based Management Information System with a local language interface that handles registrations, work estimates, muster rolls and wages to ensure correct and timely payment of wages and allows open access to reliable, timely and comparable information from the field. AP was the first state to institutionalize social audits as early as 2006, with a focus on follow-up action by the administration in the aftermath of such audits. AP also undertook several time and motion studies to revise the rural SORs to reflect the higher labor intensity and regional variation in MGNREG work. In addition, AP was among the first states to introduce payment of wages directly into workers' post office or bank accounts (opened in the name of the women within the household). Tamil nadu The state of Tamil Nadu has used administrative monitoring effectively to increase transparency under MGNREG. In some districts, a daily audit is done of the nominal muster rolls (NMRs). All NMRs are closed by 11 am each day and the information is passed from village to block and then to the district level monitoring office through an SMS. Inspection officers visiting worksites then check details with entries in the NMRs. Regular monitoring in this way has helped reduced the scope for `ghost workers'. Similarly, while only 50 percent of MGNREG works are mandated to be undertaken by the Gram Panchayat, Tamil Nadu makes it compulsory that all works are undertaken by the village body. Finally, the state (along with Kerala) has the highest proportion of women participating in MGNREG ­ about 4 out of every 5 beneficiaries are women, a number that is also confirmed by many social audits. However, lack of childcare facilities at the worksites continues to remain a problem. awareness of the process of demanding work.194 Some at the same time, very few instances of the unemployment studies also note that there is little awareness that this allowance being paid are recorded. Often applications is a universal right for rural households and not limited for work are not formally recorded, with the result that to specific groups. In consequence, potential applicants unemployment allowance is not paid out in the event can be excluded based on certain characteristics work cannot be provided within the specified period. (e.g., widowed, single women headed, elderly etc.) While failure to pay the unemployment allowance is or documentation (e.g., ration card). In some cases, perhaps excusable in the initial implementation period workers are deliberately mis-informed that work would when the mechanism for timely opening of worksites was be provided when available (i.e., not necessarily on finding its feet, this is not the case as the program matures demand). Awareness campaigns need to stress the fact into its fourth year of operation. This is important for two that all are eligible for the program and that workers reasons. The first and most obvious is that the "guarantee" need to demand work, in quorum or individually, in of the Act is not as intended unless the allowance is paid order to get it. in a timely manner. The second is that states ­ which have 194 Basicawarenessoftheprogram'sexistencedoesnotsuffice;itisinformationabouttheprogramentitlementsandtheprocessbywhichtoaccesstheseentitlements thatmatters.Thisiswhereawarenessislowest(seee.g.,DrèzeandKhera2009).Studiesthathaverevisitedsamplelocationsbetween2006and2007indicatethat awareness,thoughstilllargelylow,hasincreasedovertime(seeforexampleCBGA(2006,2007)andIHD(2006,2007)forChhattisgarhandBihar).Thisisdiscussed inmoredetailforallSPprogramsinChapter7. 0 Social Protection for a Changing India: Volume II to finance the allowance from their own budgets ­ will in the release of funds, partly due to failure to release have far stronger incentives to improve implementation the state's matching share and limited capacity related of MGNREG if the unemployment allowance provision is to work provision as well as financial management (e.g., made a reality. timely completion of utilization certificates, etc.). This in turn delays the opening of works, intermittent opening Fundsandleakage and closing of works and/or delays in wage payments. the utilisation rate for mgnreg funds is fairly high there is some evidence of leakage of funds. Several field relative to other sp programs, but field studies studies indicate evidence of leakage and report several indicate delays in the transfer of funds to gps. In ways through which accountability mechanisms are 2008/09, about 25 percent of MGNREG funds were being subverted (more on this below), including through unspent, but the ability of states to spend on MGNREG fudged muster rolls, misuse of job cards and account varies considerably, with fund utilization rates ranging passbooks, wage payments below the program wage, from 56 percent in Tamil Nadu to 89 percent in Rajasthan. etc. Though the CAG report on the 200 Phase I districts There are also significant variations across districts within highlights several such irregularities, it does not present states, with some districts (e.g., in Orissa and Karnataka) evidence of large-scale diversion of funds. However, a even reporting expenditure in excess of available funds.195 recent study compares the MGNREG administrative data Given the demand-driven nature of the program, funds with NSS household survey data (2006/07) for the Phase I are supposed to be released on the basis of state proposals districts and finds that 50 percent of funds allegedly spent (rather than pre-determined allocations). Each state is on wages are in fact some form of leakage.196 required to submit a state annual work plan and budget proposal (AWPB) to the Ministry of Rural Development. Staffingandcapacity The release of funds depends on the actual utilization of funds previously released (including the balance left over key administrative and technical staff needed for from the previous financial year). Like other CSS, funds scheme functioning have still to be appointed in a are released directly to the district. Additional funds can number of states. The sheer scale and the administrative- be requested only after 60 percent of the allocation to heavy design of the program197 ideally require a dedicated any administrative unit (GP, block or district) has been cadre of professional administrative and technical staff to exhausted. In practice, some field studies point to delays implement (see Box 4.3 for the structure proposed by the Box 4.3: mgnReg implementation agencies Administration (MGNREG) Local Government State Employment Guarantee Council District Coordinator [Assistant Engineer] District council (Zilla Parishad) Block Program O cer [Junior Engineer/PTA] Block council (Panchayat Samiti) Employment Guarantee Assistant Village council (Gram Panchayat) (Gram Rozgar Sewak) 195 AccountabilityInitiative2010. 196 Bhalla(2010). 197 Forexample,householdsmustberegisteredfortheprogram,everysubsequentworkapplicationmustberecorded,andpaymentofwagesislinkedtomeasurement ofworkdoneevery15days,andsoon.Similarly,theprogramcallsforparticipatoryplanningofworks,adetailedprocessofadministrativeandtechnicalsanction ofworks,andaheavyemphasisonreporting. CHAPTER-4: Public Works and Promotional Programs 1 operational guidelines).198 States were supposed to set staff, this in turn implies poor selection of works and up Employment Guarantee Councils and appoint these at least in the initial year of implementation, reliance key officials ­ Program Officer at the block, Employment on the shelf of works prepared for previous programs. Guarantee Assistant at the GP level as well as a dedicated In a radical shift, GPs are required to manage funds cadre of engineers at the district and block level. Many averaging ` 12 lakhs in 2008/09, with no previous states experienced significant delays in appointment experience. Training and technical support to PRIs, of these officers ­ e.g., the CAG audit in 2006-07 found particularly at the gram panchayat level, for better that these POs had not been appointed in 70 percent of financial management, participatory resource planning the blocks surveyed in 19 states. Instead, existing Block and project management functions are essential. Development Officers (BDOs) were appointed as POs. The PO is a key player in administration of MGNREG, Programwages being the lynchpin linking the community and higher a common challenge with respect to wages has been the levels of the system in terms of planning, financing and tension between the program commitment to payment other implementation aspects. The same report found of agricultural minimum wages and the reliance on that about 20 states had not appointed Employment piece rates for wage payment. As per the operational Guarantee Assistants (EGAs) in half the panchayats surveyed and had not constituted the mandated panel guidelines, wages are to be paid either on a time-rate of accredited engineers at the district and block level to basis or a piece-rate basis. The former are based on daily assist in estimation and measurement of work. While field minimum wages, where the implementing agencies studies report progress in the appointment of POs and are supposed to ensure productivity norms are met. EGAs in recent years, the lack of adequate technical staff The latter are based on the district rural Schedule of remains a critical constraint to opening worksites and Rates (SoR) that determine the wage based on actual contributes to delays in work measurement and wage work done. The guidelines stress that these piece-rates payments as well as poor quality of assets. The increase should be such that an average worker is able to do in the administrative costs borne by the center ­ from the work required to earn the daily minimum wage. In 2 percent initially to 6 percent ­ may ease the pressure practice, field studies note the prevalence of primarily on states in this regard. piece-rate system,200 with the actual wage payment often falling short of the official agricultural minimum in addition, the capacity of pris to conduct their wage. This is true even in otherwise well-performing intended functions is very weak. The generally low states like Rajasthan, where even administrative data capacity of PRIs to carry out a range of MGNREG indicated an average MGNREG daily wage of ` 59 as functions ­ including planning, execution and opposed to the official minimum daily wage of ` 73 oversight - expected of them is a more difficult challenge. (see Table 4.6). Note however that administrative data Differences across states in the degree of devolution to are more accurately described as "wage cost per day" panchayats are also a factor. Participatory planning by (i.e., total expenditure on unskilled wage divided by communities and facilitated by the panchayat officials total persondays generated) rather than the actual happens only rarely. In principle, gramsabhas have a key wage received by households. The latter are possibly role to play in identification of works to be undertaken lower than that indicated by the administrative figures under MGNREG. In practice, there is evidence that gram reported in Table 4.6. sabhas have difficulty carrying out these responsibilities in many states.199 As a result, the MGNREG shelf of works this discrepancy is due in significant measure to continues to be identified either by the Sarpanch or the exacting rural schedules of rates (sor) that key officials in the gram panchayat or by the block or call for high productivity levels. The existing SoRs district administration. Given inadequate technical assume a level of capital intensity and mechanization 198 See,forexample,Ambastaetal.(2008)fordetailedsuggestionsontechnicalandadministrativestaffrequiredtoimplementtheprogram.Infactpreviouspublic worksprogramssufferedfromaseriousshortageofqualifiedengineers/supervisorstomonitorpublicworksprograms,asaresultofwidespreadvacanciesand withoutthesanctionofadditionalpositionstomeetanincreasingvolumeofworks. 199 Seeforexample,Besleyetal.,PRIA(2007). 200 TherearesomeexceptionsinstatessuchasBiharwherethetime-ratesystemwasmoreprevalentIHD2008. 2 Social Protection for a Changing India: Volume II Table 4.6: Rural daily minimum and average agricultural wages and mgnReg average wage cost 2004-05 2006-07 2007-08 2008/09 Ag wage - Ag wage - mgnReg mgnReg notified mgnReg State male female min. wage "wage" min. wage "wage" wage "wage" Andhra Pradesh 37 28 80 86 80 83 80 83 Assam 30 16 66 67 76 73 80 77 Bihar 45 26 77 70 81 80 89 85 Chhattisgarh ­ ­ 62 62 68 68 75 73 Gujarat 55 30 50 56 50 64 100 68 Haryana 58 23 97 97 135 124 141 120 Himachal Pradesh 13 ­ 73 69 75 75 100 99 Jharkhand ­ ­ 77 79 86 82 92 90 Jammu & Kashmir 32 ­ 70 69 70 70 70 68 Karnataka 49 28 66 67 74 72 82 81 Kerala 56 28 125 121 125 118 125 120 Maharashtra 53 32 69 104 ­ 90 69 75 Madhya Pradesh 41 27 62 60 67 64 91 73 Orissa 45 14 55 53 70 77 70 90 Punjab 32 ­ 98 94 102 101 98 111 Rajasthan 44 9 73 51 73 59 100 88 Tamil Nadu 61 31 80 80 80 77 80 80 Uttar Pradesh 48 26 58 56 100 93 100 99 Uttarakhand ­ ­ 73 73 73 74 100 85 West Bengal 45 32 68 70 72 79 75 78 India 61 45 ­ 65 ­ 75 ­ 84 Notes: Average daily wage rate in agricultural occupations (based on 2004/05 NSS) from Mehrotra (2008); minimum wages from MGNREG website (www.nrega.nic.in/wages.pdf, http://nrega.nic.in/Min_wages_new.pdf ); MGNREG wage is calculated as "wage cost per day" (i.e., total expenditure on unskilled wage divided by total persondays generated) from Drèze and Oldiges (2009) for earlier years and report team estimates for 2008/09. not appropriate to a labor-intensive scheme such as age, etc.202 An additional problem commonly reported MGNREG. In addition, SoRs do not adequately allow by field studies is delays in wage payments (beyond the for differences across areas (e.g., with respect to soil, prescribed 15 days of work completion), largely driven typology or climatic conditions) or across groups (e.g., by the lag between work completion and measurement with respect to age, disability or gender). SoRs also often due to the shortage of technical manpower. do not revise rates in line with the agricultural minimum wage.201 With the current SoRs, workers (particularly the average mgnreg wage has risen steadily in most women, elderly and disabled people) can find it difficult states since the introduction of the program. In some to earn the minimum wages. There is an urgent need cases, this is the result of the revised time and motion to revise these SoRs for MGNREG. A number of states, studies that enable workers to earn at least the minimum including Andhra Pradesh, Bihar, Gujarat, Haryana, wage. In others, it is the result of increases in the official Karnataka, Orissa, Tamil Nadu and Uttar Pradesh, have agricultural minimum wage. Most recently, the 2009/10 indeed addressed this issue through revisions to the budget revised the MGNREG wage upwards to at least rural SoRs. These revisions are based on work, time ` 100 per day. In general, the official minimum wage, and motion studies that, in some states, differentiate and in most cases, the average MGNREG wage rate, productivity norms by region (soil, typology), gender, is much higher than the average agricultural wage 201 SeeAmbastaetal.(2008)andVijayShankeretal(2006)foradiscussion. 202 APforexampleundertookaspecifictimeandmotionstudyfordisabledworkers. CHAPTER-4: Public Works and Promotional Programs (particularly for women) (see Table 4.6).203 Typically, there is inadequate planning of works both with several of the better targeted public works programs respect to timing and relevance for local needs and set program wages slightly below the prevailing wage little attention paid to maintenance of the assets rate for unskilled labor (e.g., Chile and Argentina). This created. Inadequate participatory planning of works is because the program wage vis-à-vis the market above implies that works planned under the MGNREG are wage and the nature of the work are both commonly mostly not integrated with wider village level planning, used elements of the self-targeting design of public most notably under the Backward Region Grant Fund works programs. Indeed, evidence from MEGS also (BRGF). As a result, the works selected may have little suggests that the access to the program was heavily relevance or importance for the local community or rationed and the targeting performance worsened may well be redundant. Recently, there are some efforts when the minimum wage was doubled in the late underway to promote greater integration in village 1980s.204 However, MGNREG confers a right to work to planning. A key constraint to building high quality assets all rural households (and not just poor households). is the lack of technical support to communities as input As a result, the self-targeting element is limited to the to planning MGNREG works (e.g., through resource expectation is that the nature of the work (unskilled mapping exercises) as well as the shortage of technical manual labor) would attract mostly poor households. staff in designing and supervising works. A large In this context, higher real program wages are seen number of works, particularly those related to water as obvious advantages for participating households. conservation, remain incomplete, either due to lack of There would be cause for concern only in the event the technical support to GPs or the onset of monsoons. For supply of worksites is inadequate to meet demand for example, Orissa completed only 7 percent of MGNREG work, leading to rationing of employment in a manner works in 2008/09, while Uttar Pradesh completed 61 such that the poor are excluded. percent of works.206 In many cases, poor planning and lack of maintenance result in completed assets falling Qualityofassets into disuse.207 Realizing the objective of creating durable, productive rural assets will be key in ensuring broader a consistent concern on public works in india has rural development through increased agricultural been the quality of assets created; this continues productivity, with MGNREG essentially providing an to remain a concern for mgnreg. It is very difficult to insurance function. estimate the economic impacts from assets in Indian public works schemes. Given the importance of rates several of the works on the approved list of works of return on assets in assessing total program impact, are not universally suited for all geo-climatic zones this is unfortunate, though a common problem in and seasons. Works allowed under MGNREG are largely public works programs worldwide. A challenge for the related to water conservation and road connectivity. MGNREG will be developing cost effective methods In 2008/09, nearly two-thirds of activities undertaken for estimating rates of return on assets.205 Qualitative related to water and soil conservation activities (see evidence is available from Government of India and Figure 4.6). While well suited to arid and drought-prone other evaluations however, largely related to indicators areas where such works can be undertaken practically of quality of assets rather than rates of return. A common throughout the year, some states (e.g., Assam, Bihar) are criticism of public works is that they are "washed away attempting to develop works that are suitable for flood- the next monsoon". One factor is that the objective of prone and water logged areas. There is also a proposal asset creation runs a very distant second to the primary to expand the list to allow more forestry activities. Field objective of employment generation. studies also report the need for expanding the scope of 203 SeeforexampleGaihaetal(2009)andMehrotra(2008).However,thishasgenerateddebateattheminimumwageinvenuestatesinabove`100. 204 SeeGaiha(1997)andRavallionetal(1993). 205 EvidencefromtheFoodforWorkPrograminBangladeshprovidesagoodexampleofeffortstocalculatesuchratesofreturn.Thesefoundsubstantialbroader economicgainsfromassets(e.g.,themarginalproductivityofbothlandandlabourwasaround27percenthigherinvillageswiththeprogramrelativetovillages withouttheprogram). 206 AccountabilityInitiative2010. 207 CSE2008. Social Protection for a Changing India: Volume II Figure 4.6: Main types of works under MGNREG, 2008/09 Irrigation on SC/ST land, 20 Micro irrigation works, 5 Renovation of traditional water bodies, 9 Drought proo ng, 7 Land development, 15 Water conservation and harvesting, 21 Other approved activity, 1 Flood control and protection, 3 Rural connectivity, 18 Source: www.nrega.nic.in. approved works to those not capable of hard manual to increase the 60:40 labor to material ratio or to maintain labor, e.g., to include more women- and disabled-friendly it for each work as opposed to an average across works type of works.208 In addition, in areas where such rural undertaken.210 The latter step could potentially restrict infrastructure is already well-developed, there is a case the range of productive assets of varying labor intensity for expanding the permissible list of works to include that could collectively meet village-level infrastructure social infrastructure development activities (these are gaps. Higher labor intensity naturally increases the also "softer" work options for vulnerable groups) like transfer to the poor in the short run, but an appropriate related to provision of social welfare or community balance needs to be struck in order to ensure the services, etc. Work is also allowed on private lands of SC/ long-term returns of high quality assets. International ST and BPL households and some states have extended experience from Argentina and South Africa, suggests this to small and marginal farmers. Some new types of that large variations in labor intensity across sub- works, such as social forestry, have recently been added categories of works are not uncommon (e.g., in South to the approved list and discussions on converging Africa, the share of labor ranged from 6-86 percent MGNREG with other government programs (e.g., Sarva according to the type of works undertaken, and in Shiksha Abhiyan, ICDS, watersheds) are ongoing. Argentina from 30-70 percent).211 mgnreg has done better than previous programs with Adherencetoguidelines respect to maintaining the mandated labor intensity of works. In 2008/09, approximately 69 percent of evidence from the field suggests poor adherence to expenditure was on wages and 27 percent on materials transparency safeguards. The operational guidelines (the remaining 4 percent went towards administrative provide for several transparency safeguards, including costs). There are wide variations in the labor share maintenance of muster rolls at the worksite, payment across states, ranging from 54 percent in Jharkhand to of wages in public in the first two years and through 100 percent in Tamil Nadu (not including administrative worker bank or post office accounts since (with pay slips costs).209 In fact, there is anecdotal evidence of attempts to be distributed in public), and community oversight 208 NotethatthiscanalsobeachievedbyrevisingtheruralScheduleofRatesthroughtimeandmotionstudiesfordifferentsub-groupsofthepopulationasdiscussedearlier. 209 AccountabilityInitiative2010. 210 Seeforexample,IHD2006,IWDS2006. 211 Subbarao(2003). CHAPTER-4: Public Works and Promotional Programs through village vigilance committees and social audits field studies largely indicate the absence of mandated (more on this below). In practice, unavailable and facilities at worksites. The guidelines provide for facilities fudged muster rolls continue to be a serious issue. Job such as first aid, shelter or shade, drinking water and crèches card entries are rarely made. In fact, job cards are not for children (provided there are at least five children below always in the possession of the household; instead the the age of six). The absence of the child care facilities at Sarpanch or other local official may hold it. The problems the worksite is a particularly critical gap as it potentially are exacerbated by low awareness of processes as well constrains the participation of women in MGNREG. Several as high levels of illiteracy among MGNREG workers. studies note that women are discouraged from working on MGNREG as they either have to leave their children at in an attempt to reduce corruption and improve home or unattended at the worksite. transparency in wage payments, from april 2008 mgnreg wages are supposed to be paid through household Monitoringandevaluation accounts with banks or post offices. Though a step in the right direction, several new implementation challenges The design of MGNREG makes huge additional emerge. The coverage of banks and post offices in rural administrative demands on the public works delivery areas is often patchy implying the need for banking system and requires a level of accountability which correspondent or similar models to bring banking services previous programs have failed to achieve. Intensive to the household (see Chapter 7 for a discussion). In many monitoring and evaluation of program outcomes is states, the delays in wage payments are exacerbated. therefore critical to learn whether the strengthened Some field studies also note that women may not have provisions have had the intended impact. access to their earnings if the bank account is opened in there is indeed a strong emphasis on monitoring the name of the male household head. Some field studies in mgnreg ­ a significant departure from previous report instances of corruption where village leaders, public works (and other sp) programs. MGNREG has officials, contractors or intermediaries manipulate even this system to siphon money from workers (e.g., through a management information system (MIS) for tracking all false entries in the account passbooks, withdrawing activities related to the program at the highest level of money directly from worker accounts, etc.). disaggregation, i.e., the household (www.nreg.nic.in). Besides the usual information captured in previous despite the ban on contractors, field studies indicate program reports (such as financial and physical the presence of contractors in several states. For progress), the MGNREG MIS captures details on workers instance, the Drèze and Khera (2009) study finds that more (i.e., registration, job cards, muster rolls, employment than a quarter of the 100 MGNREG worksites surveyed demanded and provided) as well as on works (shelf of in six northern states were managed by contractors; in works, work in progress). However, the large scale of the Jharkhand, this figure was more than half.212 The survey program, level of disaggregated information required, also found that workers, especially women, reported frequency of transactions to be recorded and limited lower harassment and greater adherence to the program's infrastructure and connectivity at the village level all transparency norms in contractor-free worksites. As impose formidable challenges for keeping the MIS a result, corruption levels tended to be higher where updated. In consequence, there are delays of as long contractors are involved, largely due to fudging of muster as six months in some states in updating information in rolls. On the other hand, contractors can also bring the MIS. In fact, in compiling the administrative data on with them project management expertise that could program performance for this report, we have relied on potentially result in better quality of assets. This project the monthly progress reports (MPRs) rather than the MIS. management expertise needs to be transmitted to the These are based on the consolidated data entered at the mates (worksite managers) that are appointed from the district level and are relatively up to date, though with pool of MGNREG workers under the program. less depth of information than the MIS. This backlog of 212 Evidenceontheextentofrelianceoncontractorsinpreviouspublicworksprogramsissomewhatmixed.AMinistryofRuralDevelopmentevaluationofSGRYCMD 2005foundcontractorswerereportedtobeinvolvedinonly14percentofassets.Thisissubstantiallylowerthanthatreportedbyotherstudies(Seeforexample, theAFCSGRYprogrammeassessmentfortheGovernmentofOrissa2005,PlanningCommision(2000),DeshingkarandJohnson(2003)forAndhraPradeshand WorldBank(2005a)Kerala,Karnataka,RajasthanandWestBengal).Inaddition,theAndhrastudyfoundthat,inthelargemajorityofcases,Sarpancheswerethe contractors­afindingofsomeconcernforMGNREG. Social Protection for a Changing India: Volume II information weakens the transparency function of the systems for field monitoring of the program using MIS and reduces its relevance as a project management national monitors as well as third party monitoring by tool for real time decision making.213 This backlog can independent agencies. A network of institutes has been be avoided using a transactions-based MIS as it makes established for independent concurrent monitoring of data entry mandatory at the same time as the process the program and has generated a body of research on (e.g., the job card is printed by the application as soon the performance of MGNREG in the field across states.215 the information if entered at the time of registration, However, there are no national or state-wide rigorous rather than a manual issuance of the job card, followed evaluations of impact (see Box 4.4 for potential channels by entirely separate data entry).114 MGNREG has also of impact of the program). Box 4.4: Potential channels of impact of mgnReg There is no rigorous national or state-level impact evaluation of the program, making it impossible to estimate the impact of MGNREG on key parameters such as poverty, labor markets, and the local economy. Absent rigorous evaluation, some findings from early simulations of likely scheme impacts estimated prior to introduction of the program are of interest to look at the potential of MGNREG to impact various social outcomes.a These simulations suggested that the program could potentially increase casual labor supply during the lean season if the program wage rate was fixed at just over ` 60 (in 2005 prices). The incremental labor supply impact was negligible at wage rates above ` 60. Note that these estimates were based on an assumption of a full guarantee ­ i.e., every household who wanted work got it subject to a 100 day limit. This finding is of interest given the rise in the program wage rate since the introduction of the program and the evidence of rationing (i.e., rather than a guarantee). The simulations also predicted significant positive lean season poverty reduction impacts of MGNREG, with progressive distribution of gains. More recently, an evaluation of the impact of MGNREG in one district in Andhra Pradesh also suggests positive impact on food consumption and savings of participating households (Ravi and Engler 2009). Another, also in Andhra Pradesh, suggests positive impacts in terms of reducing child labor among participating households (Uppal 2009). In principle, MGNREG could be expected to have an impact on wages and employment of participating households as well as non- participating households.b A priori, it would appear that the insurance and associated productivity impacts would be stronger if households perceived MGNREG as here to stay and if local MGNREG work was available at scale. If the program is seen as an unreliable source of employment, as field studies indicate is the case in some states, households would be less likely to change their labor supply decisions (e.g., with respect to migrating for work or doing other casual work). To the extent MGNREG does in fact provide a minimum floor for rural wages, it could also impact the private market for casual labor. This would imply gains/losses for those who do not participate in the program. However, in light of the widespread finding of excess demand and rationing of work, it seems likely that the bulk of these spillover effects to non-participants would be lost. Whether or not this impacts poverty would depend on whether it is mostly the poor who participate (on the assumption that the self-targeting mechanism works and only the poor turn up for work) and the opportunity cost of their participating in MGNREG (instead of doing some other work). Note that even among the poor, it is the able-bodied poor who will be in a position to participate. Also, depending on social norms and the extent households make joint labor supply decisions, impacts could vary between men and women. Anecdotal evidence from some field studies indicates some impact on local labor markets through rising casual wages and reduced migration. For instance, the IIPA (2006) study finds slightly lower incidence of migration among MGNREG households than among non- MGNREG households, more so in Bihar than in Jharkhand (see also Drèze and Khera (2009) and Khera (2006)). There is anecdotal evidence that this constrains labor supply and affects agricultural wages in Punjab and Haryana (see for instance a series of articles by P. Sainath (2008) in The Hindu based on field visits to various states). There also several potential ways that MGNREG could impact future livelihoods and rural governance, but the evidence from field studies suggest these channels of impact might be weak as yet. In addition to these direct income gains (or losses for those who hire casual labor), there are potential spill-over benefits through gains from the assets created, i.e., potential increases in future livelihood opportunities and incomes. Field reports of poor asset quality indicate that this second-order impact is unlikely to have made itself felt just yet. Finally, with its emphasis on empowering communities and local governments, MGNREG also has the potential to impact rural governance. In practice, however, this has been the hardest to operationalize, implying that it might be too early to see such impacts. Notes: (a) See Murgai and Ravallion 2005. The simulations are based on the 1999/2000 round of the NSS. (b) Evidence from the MEGS in the 1980s suggests a positive insurance function of public works, with attendant impacts on production decisions - income variability among landless agricultural households in EGS villages was significantly lower than that of non-EGS villages (Walker 1986, Scandizzo et al 2007). Similarly, there is some evidence of private casual wages rising in response to MEGS (Gaiha 1996). 213 Forinstance,DrèzeandOldiges(2009)pointouthowprogramofficersoftendonothavealistofactiveworksitesintheirblock. 214 SeeMehtrotra(2008)andHirwayandSingh(2006)foradiscussion. 215 ThisProfessionalinstitutionalNetwork(PIN)includesIndianInstitutesofManagement(IIMs),IndianInstitutesofTechnology(IITs),NationalInstituteofRural Development(NIRD),AdministrativeStaffCollegeofIndia(ASCI)andothers. CHAPTER-4: Public Works and Promotional Programs the use of social audits is one of the most interesting important development in strengthening such objectives innovations in mgnreg program delivery and can in India's safety net. In addition, there is a serious be replicated in other social programs, with the attempt to restructure the livelihood support programs potential to reduce fraud and corruption and to under the aegis of the National Rural Livelihood Mission make public delivery systems from accountable. Social (NRLM). This section looks at three programs which seek audits are mandated by the operational guidelines, but to promote movement out of poverty: (i) targeted rural few states, with the notable exception of Andhra Pradesh credit programs; (ii) school stipends/scholarships; and and Rajasthan, have made progress in implementing (iii) the midday meal program. these in the field. In Rajasthan the process was driven largely by high community-level activism that had its roots in the Right to Information movement.216 In (a) Targeted rural credit programs Andhra in contrast the government drove the process as with public works, india has an extended history by establishing an independent social audit society and of targeted credit schemes for the rural poor. This partnering actively with CBOs and NGOs to initiate audits began through the Integrated Rural Development in the field.217 These two states represent very different Program (IRDP), which was universalized from 1980, and successful models (see Box 7.5 in Chapter 7) that are of supplemented by various training and input schemes interest for other states attempting to institutionalize over the subsequent two decades. In 1999, all these social audits. Besides initiating the process of social schemes were merged into the Swarnajayanti Grameen audits, it is also important to move from "answerability to Swarozgari Yojana (SGSY), which is conceived as an accountability", i.e., the district and block administration integrated micro-enterprise development program needs to be responsive to the issues raised at the public for the rural poor, emphasizing self-help group (SHG) hearing at the end of a social audit.218 Unsurprisingly, formation, capacity building, and access to credit and there is evidence of resistance, sometimes violent, to markets. this process. An extreme manifestation was the murder while irdp concentrated on individual beneficiaries, of an activist while initiating a social audit in Jharkhand. sgsy lays greater emphasis on social mobilisation There have been sporadic incidents even in Rajasthan. and shg formation. SHGs are also given awareness There remains, however, a strong case to continue to training in the importance of regular savings and credit work towards greater transparency and accountability. discipline in an attempt to build up mutual support systems and a greater sense of self-confidence. However, the district administrative offices (DRDAs) responsible B. PRomoTing movemenT for administering the programme generally lack the ouT oF PoveRTy in The requisite skills in social mobilisation, and linkages with NGOs, which could have facilitated this process, have ShoRT AnD long TeRm also been weak in many states. In addition, lending as noted in chapter 2, india's safety net program mix by banks is a major problem area, with both credit to has historically exhibited a relatively low emphasis subsidy ratios considerably below program norms on interventions which seek to promote sustained (resulting in under 40 percent of the targeted credit movement out of poverty, either in the short run amount being mobilized), as have been average loan sizes estimated at ` 22,995 in 2007/08).219 (through livelihood support), or the long run (by incentivizing human capital acquisition). However, (i) Coverage the situation has changed markedly in recent years with respect to long run programs. The major expansion in the coverage of sgsy is very low, a reflection in part spending on midday meals in schools represents an of relatively low (and falling real) spending, and of 216 SeeforexampleSivakumar(2007)andseveralassociatedarticles(includingamanualforconductingsocialaudits)attheRighttoFoodwebsite. 217 SeeAakellaandKidambi(2007a,b).DetailedguidelinesareavailableonlineattheGoAPMGNREGportal. 218 SeeSen(2007)foradiscussion. 219 SeeSaxena(2006),RadhakrishnaCommitteereport(2009). Social Protection for a Changing India: Volume II a range of program design and implementation issues Table 4.7: Coverage of SgSy by wealth, social which have plagued the program and predecessors category and location, 2004/05 since inception. Analysis of nationally representative median hh benefits % of benefits data finds a coverage rate of 1 percent of households % hh (among SgSy hhs) captured nationally for SGSY, with only a few states (Rajasthan, Poorest 1.3 1,712 7.7 Himachal Pradesh, Maharashtra and Chhattisgarh) Q2 1.0 5,430 15.4 reaching 2 percent coverage.220 Madhya Pradesh is the Q3 0.9 3,685 10.4 only state that stands out as having notably higher SGSY coverage (at over 4 percent), though generally the Q4 0.6 10,025 16.7 poorer states have somewhat higher coverage rates in Richest 0.7 26,430 49.7 relative terms. Just as interesting is the high variability OBC 1.0 11,001 62.6 in reported benefits from the program, and the fact that SC 0.9 9,484 27.0 the median benefits are well below those suggested ST 2.5 1,560 5.2 by administrative data. The last point is consistent Other 0.4 4,576 4.7 with reports from qualitative work for this report and All India 1.0 7,681 100 other studies which have noted regular payment of Source: Ajwad (2006), based on 2005 IHDS data. "commissions" by beneficiaries for bank officials and others (see below). (SPS for Karnataka, Madhya Pradesh and Orissa and JLSS the low coverage rates of sgsy are confirmed by other for Jharkhand) find that SGSY is regressive in coverage. data sources. The state-specific surveys undertaken In Karnataka, MP and Orissa, the coverage rate for SGSY for this report (2005 SPS and 2006 JLSS) find that only in the top wealth quartile is over two thirds higher than 3.3 percent of households were covered by SGSY in for the poorest quartile, while more than 80 percent this Karnataka, Madhya Pradesh (the highest coverage state coverage rate was more than 80 percent higher in the nationally) and Orissa, and only 1.6 percent households highest compared to the lowest per capita expenditure in Jharkhand.221 Beneficiary numbers reported by the quintile. Of the range of safety net programs reviewed Ministry of Rural Development also indicate low coverage. in the study, SGSY was found to be the least effective Though nearly 31 lakh SHGs were formed in the first program in serving the poor, for a variety of reasons. ten years of the program, only about 22 percent were Given that SGSY is targeted to below poverty line or BPL able to obtain credit to take up economic activities.222 households, these results indicate that more research is Assuming they were all from separate households, this needed on the distributional aspects of SGSY. would be around 0.35 percent of rural households. the analysis also gives cause for concern with respect to benefit incidence across the distribution (ii) Targeting and social categories. Progressive coverage rates are evidence on targeting is more mixed, depending on the more than offset in terms of benefit incidence by the data source and whether coverage is examined across much higher average benefits reported by the rich, and the distribution of household wealth or per capita among OBC households. This can be seen in Table 4.7, expenditure. The IHDS data find coverage across wealth where the richest quintile captures almost half of the levels to be progressive (see Table 4.7), and SC coverage program benefits, while the poorest quintile captures notably higher (though still low). Earlier data from NSS under 8 percent. In caste terms, OBCs capture over finds that coverage was progressive in the late 1990s, 60 percent of the total benefits. This is interesting in light but very mildly so.223 However, the state-specific surveys of the 50 percent program reservation for SC and ST. 220 IntheNCAERsurvey,therewasacombinedquestionofSGSYoranyinsurancepayout.Giventheverylowcoverageofinsurance,thisistakeninattachedasaclose approximationofSGSYcoverage,andinanyeventisatbestanupperboundofSGSYcoverage. 221 SeDevetal(2007)andBalchander(2009). 222 RadhakrishnaCommitteereport(2009). 223 SeeO'KeefeandPalacios(2006). CHAPTER-4: Public Works and Promotional Programs (iii) Implementation224 because of indivisiblities in marketing costs and low risk bearing capacity. Low price received there are a range of fundamental implementation by poor for their products is also because of challenges faced by sgsy which suggest a serious interlocked output and capital markets, lack of assessment is needed of its potential in the indian value addition technologies, poor organisational safety net, and the current delivery modalities. The base and insensitive government policies. These scope of poverty reduction through SGSY is limited problems need to be attacked by the same both by the debt-capacity of the poor and by the high agency, which is not possible in government cost of appraising, monitoring and enforcing small loan system. Whereas NGOs could take initiative agreements. The first limitation is in theory offset by in some of these sectors, they cannot change subsidy, but it attracts rich borrowers to the scheme, exploitative marketing infrastructure. and thus creates political and administrative problems. Lack of repeated contacts between lenders The subsidy element has led to large scale corruption on the part of lower level functionaries who certify non- and borrowers. SGSY suffers from a basic eligible beneficiaries as being BPL. Even when the poor misconception that the provision of credit is a one- are selected there is often no intention on their part to time event rather than a continuing relationship create assets; the subsidy is the main attraction. It has between lender and borrower. It is unrealistic to also led to corruption on the part of bank staff, and on the expect the larger share of borrowers to "graduate" part of borrowers themselves, some of who sell off their just on the basis of an "injection" of credit (the assets and pocket subsidy, or who borrow by proxy for medical terminology is significant), even if non-target group borrowers. Because of the distortion provided in sufficient "doses" (which was until of objectives, lending policy under SGSY tends to be now not usually not the case). Most loans in SGSY driven by the availability of subsidised funds rather are a one-time affair, and the bank feels relieved than by the effective demand for credit. Problems in the when the file is closed. In the case of Grameen selection of viable activities and inadequate capacity Bank of Bangladesh, most borrowers start with building of beneficiaries combined with supply-side small loans, but as the relationship with the bank issues with the banking sector results in inadequate improves, more loan is given to the same person, flow of credit under SGSY.225 thus making their interaction and relationship last for a long-term, recovery and fresh loans often go apart from such "big picture" concerns, there are a hand in hand. This also improves the capability of number of other factors in design and implementation the poor to utilise the loan profitably. Their stake of sgsy which have, until recently, received inadequate in repayment is also higher. In India, since loan is attention in policy evolution of targeted credit taken only once in a life time by most borrowers, programs. they include: the tendency is not to pay and become defaulter. Lackofmarketsandinfrastructure.Unless credit Over-focus on asset formation. There is still an is accompanied with adequate infrastructural under-emphasis on activities which require support by way of extension, marketing, etc. the no fixed assets at all such as a large number of result is infructuous lending. SGSY is generally trading, service and even simple processing successful where infrastructural and institutional activities. Thus the mix of activities in India is support is available. These are the regions very different from other large micro-enterprise where many people even without the subsidy programmes in Bangladesh, the Philippines would have taken to entrepreneurial activities. and Nepal for instance, which finance a much However, the failure by the poor to use assets larger component of petty trading and service profitably stems from several factors, of which activities suitable to the poorest of the poor. Their control over markets is an important factor. The discouragement in India (due to a preoccupation poor are not able to secure economies of scale with asset formation) has restricted the type 224 ThissectiondrawsfromSaxena(2006). 225 Thisincludesinadequatebankingcoverageinruralareas,shortageofmanpowerinruralbankbranchesaswellaslackofprofessionalprojectappraisaland monitoringteamsinruralbankbranches. 0 Social Protection for a Changing India: Volume II of livelihood activities that can be financed, the old SGSY model. The NRLM aims to ensure greater and has led to overcrowding in those which are coverage among BPL households by mobilizing at financed (milch animals for instance). Instead of least one member from each BPL household into SHGs. responding to a demand-led pattern of lending Drawing on the experience of relatively successful opportunities, banks are still restricted to a list models such as the Kudumbasree model in Kerala and (albeit an expanding one) of "approved" activities. the Indira Kranti Patham (IKP) model in Andhra Pradesh, In practice the range of activities for which loans the NRLM proposes to federate SHGs at the village were available was quite narrow. On the demand and block level. NRLM also proposes enhancing the side, certain profitable enterprises were quickly subsidy under the program and to introduce an interest saturated because of too many loans for single subsidy for SHGs. Though the Radhakrishna report professions and their profitability declined. Since acknowledges the frequent misuse of both backend and indicators for monitoring are target and not front-end subsidies,227 the report emphasises the need achievement based, retention and profitability of for retaining the subsidy in principle, both to incentivize assets purchased are never monitored. poor households to enter the program and to insure Neglect of consumption and savings. SGSY against the associated risks of entering self-employment. completely neglects savings, on the mistaken belief The NRLM provides for an enhanced capital subsidy that the poor cannot save at all. The distinction made to cover the risk in the transition of poor households by banks between the acceptable use of credit for from wage to self employment. In addition, an interest productive purposes and its unacceptable use subsidy for loans up to ` 1 lakh is also provided to poor in consumption is an artificial one in the context households. The experience from SGSY indicates that the of poverty. About two thirds of the borrowing subsidy proved to be a major source of corruption and a of the poor in India is for consumption purposes disincentive to lend, on account of low recoveries. (all of it from the informal sector) of which three in addition to promoting self-employment, the nrlm quarters is for illnesses and household needs in also promotes wage employment for the poor through the lean season.226 In the absence of any system the introduction of placement-oriented skill training for encouraging even minuscule but regular model for youth as "special projects". An example savings a great deal of SGSY credit gets diverted to of such an approach is the training and job placement emergency consumption needs. program under the Employment and Marketing Mission Existingproblemsofmicro-enterprises.Designers in Andhra Pradesh. This is an autonomous government of SGSY do not look into the existing problems of agency which promotes employment for youth in remote the already established micro-enterprises. Thus, areas by linking them with formal sector employers and how can one empower the rural poor to become financing short-course training (see Box 4.5). Given that tailors, weavers, shopkeepers, or cattle owners the end objective of NRLM is productive employment if the present problems faced by those already opportunities, the proposal to direct a proportion of owning these assets are not looked into? Rather funds towards making young people employable in the than give subsidy to new enterprises it would growing range of employment opportunities provided be far better to solve the problems faced by the by India's economic growth is a welcome policy reform. existing units, whether these are in design, or in a multi­tier dedicated implementation structure is marketing or working capital. proposed with a national agency at the centre supported in light of these problems, sgsy is now being by agencies for implementation at the state, district restructured as the national rural livelihood and sub-district levels. The NRLM at the national level is mission (nrlm) for promoting rural livelihoods. The intended to serve as a financial and technical resource to Radhakrishna Committee report (2009) made several the state organizations and its mandate includes facilita- recommendations with respect to the design and tion of partnerships with civil society organizations and implementation that signal a significant departure from institutions, analysis and dissemination of best practices 226 MahajanandRamola(1995). 227 TheexperiencefromSGSYindicatesthatthesubsidyprovedtobeamajorsourceofcorruptionandadisincentivetolend,onaccountoflowrecoveries. CHAPTER-4: Public Works and Promotional Programs 1 Box 4.5: employment generation and marketing mission (egmm) The Employment Generation and Marketing Mission (EGMM) is an initiative of the Department of Rural Development, Government of Andhra Pradesh that strives to provide employment to economically and socially underprivileged rural youth in Andhra Pradesh. With an objective of "onejobforeveryruralfamily", EGMM provides skills training to rural youth and finds employment for them in the organized sector. EGMM was originally set up as the Employment Generation Mission in 2002 under the Andhra Pradesh Societies Act 2001 and was later rechristened as Employment Generation and Marketing Mission, to bring within its scope marketing support to the self-help groups, women entrepreneurs and the non-farm sector. The EGMM model works on a PPP mode with partnership between the public sector, private sector and the poor (see figure below). The state government mobilizes rural youth for general skill development, with some training programs tailored to specific industries or sectors. On completion of training, placement assistance is offered. These include the rural retail academy (with partners such as Reliance Fresh, McDonalds, Spencers, etc.), the security academy (partners include Group 4, Securicor, etc.) and a general training academy for English, soft skills and computer training. There are also models where private sector organizations partner in training as well as placement. These include Dr. Reddy's Foundation and Tally in service sector, among others. The training modules in the EGMM are carefully chosen based on industry demand and specifically designed for certain sectors that have growth potential and capacity to provide employment. supports mobilisation of rural youth The SHG provides infrastruc ture Network mobilises rural and training support youth 82000 rural youth trained and provides training privete sector EGMM 65827 provided empployment in 2007-08 provides placement assistance provides empployment provides funds State Government Provides infrastructure support The mission is headed by the district collector in each district. Each district also has a project director and the District Rural Development Agency (DRDAs) and Integrated Tribal Development Agency (ITDAs) are active partners. There is also a project management team comprising professionals recruited on a contract basis; this has a cell at the state level and a team of district managers for each district. At the grass root level, the mission is carried out by the district manager through Job Resource Persons (JRP) whose main job is to identify and motivate youth in opportunities from EGMM. Women Self Help Group (SHG) networks in the villages provide major impetus in this and often assist the JRPs in this process. The budget for EGMM is mainly from the state government Remote and Interior Areas Development (RIAD) budget and SGSY funds through the DRDAs. From 2005-06 to 2008-09, over 226,000 youths have been trained out of which 183,000 have been placed in various organized sectors. Annual salaries range from ` 45,000 upwards in metros and ` 24,000 upwards in semi-urban areas. Absent rigorous impact evaluations, studies that quantify changes in the lives of the participating youth and their families are of interest. One such study in the districts of Karimnagar, Anantpur, and Chitoor where youth were trained by the security academy and subsequently placed as security guards in cities indicated reduced dependence on money lenders and increased ownership of assets by their families. Source: Dutta (2008). See also EGMM website (www.egmm.ap.gov.in). and technical and institutional innovations, training and such social mobilization. Unlike under SGSY, PRIs are capacity building, monitoring and evaluation, promoting given an active role in the mobilisation of communities convergence with other anti-poverty programs and other and in exploring the possibility of giving the responsibility related functions. The state-level umbrella organizations for the execution of selected panchayat activities (such are seen as critical in the formation of and hand-holding as civil works, maintenance of common property etc.) support for SHGs, given that DRDAs are ill-equipped for to SHGs. 2 Social Protection for a Changing India: Volume II (b) School stipends for school stipends for major states, using household survey data from the 2004/05 IHDS data. These rates india has an extended history of providing school are estimated among those with school age children stipends from both state and central levels. (6-18), while the median stipend amount is the transfer most states have some form of school stipend received in the previous year among households in operation, at least for selected groups such receiving stipends. Coverage exhibits major variations as tribal children. This combines a number of across states, with several states covering up to one fifth centrally-financed stipend schemes (e.g., for disabled of children. Some patterns stand out and are consistent children, ST girls), and state-specific schemes (e.g., with national and state policies: Uttar Pradesh provides stipends to all SC, ST and several states with high tribal populations have OBC children in primary school). This school stipend program is intended to boost enrollment, attendance expectedlyhigherthanaveragestipendcoverage, and retention in school. thoughOrissaisanexception. in contrast, several poorer states have very low coverage of schools stipends is fairly high at almost stipend coverage, including Bihar, Orissa and 10 percent nationally, but with large inter-state Rajasthan. Uttar Pradesh is a notable exception variation. Table 4.8 presents results on coverage rates for a poor state with low tribal population, and this share is likely to have increased sharply since Table 4.8: School stipend coverage and receipts with the expansion of state-funded OBC stipends (rural and urban areas), 2004/05 in recent years. % hh with median benefits stipends stand out as one area where southern children aged (among hh with and richer states have lower than average State 6-18 years stipends, ` pa) coveragerates,in part perhaps because of higher Andhra Pradesh 2.0 4,043 private schooling rates but also driven by state- Assam 13.1 226 level policies. Bihar 0.6 2,563 thereisalsosignificantvariationinstatelevelsof Chattisgarh 29.0 1,788 stipendsreportedpaid.In contrast to the pattern Delhi 0.5 1,243 for many other programs, however, it is generally Gujarat 10.9 423 the poorer states (though with Madhya Pradesh Haryana 0.8 582 and Assam as notable exceptions) which report Himachal Pradesh 8.0 839 higher median stipends receipts. More specifically, Jammu & Kashmir 0.8 4,023 some of the more tribal states such as Jharkhand Jharkhand 10.7 1,556 and Chhattisgarh continue to have a healthy Karnataka 6.6 893 picture, and both J&K and Tamil Nadu standing Kerala 0.6 1,040 out with high median payments among those Maharashtra 5.0 1,666 receiving the stipend. Note that these numbers Madhya Pradesh 16.1 740 need to be interpreted with caution as the median Orissa 3.4 2,606 benefit amount does not distinguish between Punjab 1.6 308 households that have a single or multiple children receiving a stipend. Rajasthan 1.1 1,200 Tamil Nadu 2.3 17,638 looking at the same findings in distributional terms, Uttar Pradesh 31.4 759 stipend coverage is quite progressive in terms of simple Uttarakhand 28.3 691 coverage rates, but expenditure incidence exhibits West Bengal 1.8 400 a much more mixed picture, as the median annual Other 7.2 1,652 levels of stipends reported vary sharply across All-India 9.6 1,224 the distribution in a regressive manner in terms of Source: Ajwad (2006), based on IHDS data. both wealth and social category. This can be seen CHAPTER-4: Public Works and Promotional Programs in Table 4.9. In terms of benefit incidence, the poorest SC/ST stipends, the share of total benefits captured quintiles captured only 12 percent of total benefits, in by ST households is almost exactly the same as their contrast to the richest capturing 34 percent. Again, a share in the sample and for SC households it is less, caveat in the interpretation of these findings is that the while OBC households capture a higher share of total median amount received per child in the household stipend spending than their share in sample. However, is not known. Interestingly, despite the prevalence of this needs to be interpreted in the light of state-specific schemes, in particular the major expansion to OBCs that was occurring in Uttar Pradesh during the period Table 4.9: Coverage and receipts of school stipends by of the survey. wealth, social category and location, 2004/05 % hh with median benefits at one level, these school stipends can be considered children 6-18 for hh receiving Share of conditional cash transfers, though they differ receiving stipend total benefits notably from ccts internationally in having weak ­ or stipend (` annual) captured (%) in most cases, absent ­ enforcement of attendance Poorest 9.7 627 11.7 requirements, being based largely on enrollment Q2 14.4 883 21.6 in school in practice. In the past decade, there has Q3 9.6 730 12.5 been a growing reliance in developing countries on Q4 8.2 1,585 20.1 conditional cash transfers (see Box 4.6 for international Richest 6.5 3,592 34.1 experience with CCTs). These provide a cash transfer ­ Rural 11.5 983 70.8 often targeted to the poor or other specific groups ­ Urban 4.3 3,018 29.2 conditional upon households undertaking specified OBC 9.6 1,501 49.8 actions with respect to education, health, and in SC 14.6 860 24.3 some case nutritional interventions. CCTs rely on two ST 14.3 740 7.0 channels for affecting demand: (i) reducing the income Other 4.2 1,349 11.6 constraint on households; and (ii) providing a specific Total 9.6 1,224 100 incentive to undertake desired behavior. In this way, Source: Ajwad (2006), based on IHDS data. safety net transfers can become not only a short term Box 4.6: international experience with Conditional Cash Transfers (CCTs) The design of CCTs differs across countries, most notably in the scope of actions required by households to receive benefits. While several Latin American countries require education, health interventions and nutritional programs for children (and in some cases mothers and even elderly), others such as Bangladesh and Pakistan focus on educational attendance only. Despite the variation, the menu of actions from which countries choose their conditions for CCTs is similar, and includes: educationCCT components typically require school enrollment and attendance around 80-85 percent of school days in 1-3 month period. Most countries cover primary and secondary levels, while some countries also target specific groups (e.g., girls in the Bangladesh secondary school stipend). health/nutritionCCT components typically focus on children up to 2-3 years of age, though in some countries up to primary start age. Several countries also include pregnant and lactating women, and Jamaica goes further to include also 65+, disabled and destitute adults < 65. Households are required to undertake a regular schedule of health care visits (for child and sometimes other family members), commonly including growth/development monitoring, vaccination, health/nutrition education in some countries. With respect to targetingofbeneficiaries, approaches also vary, but most programs have a combination of geographic and household level targeting (though Turkey, for example is nationwide, and Honduras uses geographic targeting only). With respect to geographic targeting, two main considerations are typical: (i) some indicator of locality marginality (e.g., malnutrition data in Honduras, marginality index in Mexico); and (ii) supply side capacity to deliver (e.g., Mexico and Nicaragua ­ within certain distance of education and health facilities, Colombia requires adequate services, a bank and database). With respect to household level targeting, proxy means testing has been most common (e.g., Mexico, Turkey, Jamaica, Colombia). A challenge in cross-sectoral CCTs has been institutional coordination. Different countries have dealt with this in different ways, including: (i) Social line ministries: Brazil, Jamaica, Nicaragua, Kenya, Mexico; (ii) Presidency/social cabinet: Colombia; Dominican Republic; (iii) Specialized public agencies: Honduras, Sri Lanka, Pakistan; and (iv) Social Investment Funds: Chile, Turkey, El Salvador. Social Protection for a Changing India: Volume II Box 4.6: international experience with Conditional Cash Transfers (CCTs) Contd... A number of CCTs have been subject to rigorousimpactevaluations, and findings are as follows: on household consumption, the results are generally positive (especially where transfers are large), including: (i) in Mexico, households randomly assigned to the CCT program group had consumption that is 13-18 percent higher. In addition, households appear to invest about 25 cents out of every peso transferred by the CCT program in productive assets, so that, by investing transfers, beneficiary households increased consumption by about 24 percent after six years on the program; (ii) in Nicaragua, CCT households have 13 percent higher consumption, implying decreases in the extreme poverty rate of 15 percentage points, and in the poverty rate of 5 percentage points; (iii) in Colombia, CCT households have 15 percent higher consumption than matched set of comparison households; but (iv) in Ecuador, there was no significant program effect of the CCT on consumption, explained at least in part, by a massive 17 percentage point reduction in child labor on educationaloutcomes, evidence is generally positive with respect to improved utilization of services, though results on the end outcome of educational attainment remain very sparse. However, evidence from evaluations in 9 countries, finds: (i) positive program effects on enrollment, including those with the most robust evaluation strategies; (ii) impacts are larger among groups that had the lowest probability of enrollment at baseline: CCT programs may help reduce "inequality of opportunities"; (iii) Larger program effects in countries with lower baseline enrollment levels; (iv) Larger program effects in transition grades with high dropout rates; (v) with respect to learning outcomes, evidence is mixed, and draws primarily on Mexico, where the CCT increased years of schooling but did not lead to higher scores on standardized tests. on utilizationofhealthservicesandhealthoutcomes, evaluations have found increases in coverage of some services (for example, growth monitoring for children, preventive check-ups for adults), but not others (for example, immunization rates). Evidence from one country (Mexico) also suggests that CCT program reduced child morbidity and mortality, and improved adult health status on childnutritionalstatus,the evidence is inconclusive. For example: (i) in Mexico, evidence on impacts on child height is mixed, in part due to technical issues in comparison over time; (ii) in Nicaragua, the CCT program reduced stunting by 5.3 percentage points, but result is only borderline significant; (iii) in Honduras, there was no significant effect of the CCT on child height; (iv) in Colombia, the CCT reduced stunting among children younger than 2 years of age by 6.9 percentage points, but had no effect on children aged 3-7 years at baseline; (v) in Brazil, the CCT appears to have had a negative effect on child height and weight, perhaps because households believed that their children needed to be malnourished to be eligible for transfer A natural concern in considering CCTs is the supplyside, and the extent to which it is reasonable to condition transfer receipt on use of absent or poor quality educational or health/nutrition services. Several countries have avoided this problem by requirements of minimal supply side presence; others have attempted to address it by building additional infrastructure or finding alternatives for delivery. Others such as Bangladesh (for girls' secondary education) and Pakistan Punjab (for female school stipends in grades 6-8) have not, and despite this the effects on enrollments appear to be positive, with for example a net program impact of 9 percentage points in Pakistan Punjab. A further point of interest on the supply side is inclusion of only public schools in the program (as in Punjab) or both public and private schools (as in Bangladesh). Some countries, like Bangladesh and Mexico, have also sought to improve quality through incentives (pay-for-performance schemes for providers in parallel with the CCT). While strong universal conclusions are difficult on the CCT experience, several conclusions seem robust: (i) CCTs are well targeted to poor households and have helped provide a consumption floor; (ii) CCTs significantly increase the utilization of education and some health services; (iii) evidence on the impact of CCT programs on "final" outcomes in education, health and nutrition is less clear-cut; (iv) the institutional framework and administrative systems for implementation remains a big and country-specific issue; and (v) important knowledge gaps remain about "optimal" program design: Source: Fiszbein and Shady (2009); de Janvry and Sadoulet (2006); Chaudhury and Parajuli (2006). source of poverty mitigation, but also hopefully help as a concept, conditional cash transfers are not a leverage improved human capital acquisition, which new to india. In fact, the first CCT scheme was could contribute to longer term and more sustained introduced in India as early in 1994 (a Haryana poverty reduction. While most widespread in Latin government state scheme in 1994). Since then, America (where both middle income countries such many programs have been introduced both by as Mexico and Brazil, and lower income countries the centre (primarily to promote positive human such as Nicaragua and Peru use different forms development outcomes for the girl child) as well as of CCTs), CCTs are increasingly being tried or piloted states, some of which are listed Box 4.7. However, by other developing countries, including Bangladesh, in many ways, these programs function as cash Pakistan, Kenya, Cambodia, Turkey and India transfers rather than conditional cash transfers and (see Box 4.6). the implementation of these programs differ in many CHAPTER-4: Public Works and Promotional Programs Box 4.7: overview of Conditional Cash Transfers (CCTs) in india Conditional cash transfer schemes in India are more than 15 years old, with the first scheme "Apni Beti Apna Dhan" started by the Government of Haryana as early as 1994. Some of the programs introduced both by the centre (primarily to promote positive human development outcomes for the girl child) as well as states since then are listed below. key centrally sponsored CCT programs in india: IndiraGandhiMatrisahyogYojana was started in 2009 by the Ministry of Women and Child Development. This Conditional Maternity Benefit Scheme aims to provide cash to pregnant and lactating women in response to fulfilling specific conditions like registration of pregnancy, iron fortification during pregnancy, attendance at counseling sessions, registration of birth, immunization and fortification as well as exclusive breast feeding for the new born child. A budget allocation of ` 3.6 crores has been made for this scheme in 2009-10. Dhanalakshmi was launched in March 2008 by the Ministry of Women and Child Development and covers 11 blocks across seven states. The scheme provides cash transfers (and insurance coverage in certain cases) to the family of the girl child (preferably to the mother) on fulfilling certain specific conditionality's for the girl child: registration of birth, immunization, enrolment and retention in primary and secondary school and marriage after the age of 18 years. JananiSurakshaYojana was launched under the National Rural Health Mission in 2005. The main objectives are to reduce maternal and neo-natal mortality by promoting institutional delivery and for making available medical care during pregnancy, delivery and post delivery period. All women receive cash assistance if they have their baby in a government health centre or accredited private institution. Eligibility criteria and incentives in rural and urban areas differ across low and high performing states (with respect to institutional delivery rates). BalikaSamriddhiYojana was started in 1997 and covers girl children (born on or after 15 August, 1997) in BPL families in rural and urban areas. An eligible girl child is entitled to a post-birth grant amount of ` 500/- and annual scholarships for each successfully completed year of schooling as long as she is unmarried and attends school regularly until the Xth grade. Some state sponsored CCT programs Ladli was launched in August 2005 in Haryana and Delhi. Parents are given ` 2500 per child per year for five years, on the birth of their second daughter. The transfer is invested in Kisan Vikas Patras or similar savings scheme. In addition to being unmarried at 18, the child must be enrolled in school/early child care centres and fully immunized as per age for her to receive regular payments. DeviRupakYojana was introduced in Haryana. If parents undergo sterilization after the first or second child they are given a monthly pension for 20 years. The amount varies if the first child is a male or female. KanyaJagritJoti was introduced in Punjab in 1996-97. An amount of ` 5000 is invested in the name of the beneficiary with the Life Insurance Corporation. The child gets a scholarship of ` 100 per month at the age of 6 to 12 years. The rate of scholarship is doubled at the age of 12 to 18/21 years. At the termination of the scheme, the beneficiary gets a lumpsum amount of ` 5000 plus bonus accrued thereon, provided the parents continue to follow the two child family norm and also the child passes at least Matriculation. This scheme has about 8000 beneficiaries every year. KanyaVidyaDhanYojana was introduced in Uttar Pradesh in rural and urban areas. The scheme provides a sum of ` 20,000/- to each girl who passed intermediate examination from U.P Board and belongs to BPL families. ApniBetiApnaDhan was introduced in October 1994 in Haryana. This scheme pays mothers ` 500 within 15 days of the birth of a girl child, to meet her nutritional requirements. An amount of ` 2,500 is also invested, within 3 months, in an `Indira Vikas Patra' or other similar savings scheme in the name of the new born baby. Eligibility is restricted to socio-economically disadvantaged families belonging to SC/ST and BPL families, and only if they have three or fewer children. respects from the international experience noted (c) mid-day meals above. This is largely due to the lackluster tracking and enforcement of the conditionality associated with the midday meal school feeding program in schools the transfers. This in turn is partly due to supply-side has expanded rapidly in the past decade, and is an constraints and partly the result of poor monitoring increasingly important plank of educational and and evaluation systems. Another challenge in the broader social policy. while midday meals have been effective implementation of these programs is that delivered in some parts of the country from as early the institutional inter-ministerial framework for as the 1950s (in tamil nadu) and 1980s (in gujarat), administering such cross-sectoral programs is still the major expansion came only after 1995, and more being established in India. particularly 2001. While cooked midday meals were Social Protection for a Changing India: Volume II mandated from 1995 (with two years given to put the coverage of children as a proportion of those enrolled system in place), they remained the responsibility of in schools is over 90 percent in most states (see Figure states until 2001, and experience across states was very 4.7). This is however a likely overestimate of coverage mixed in terms of performance. A key step towards as enrolment as the administrative data reported by expansion was the 2001 Supreme Court order, which schools and enrolment is a primary criteria for eligibility ordered states to provide the meal with minimum to receive grains to serve schools meals. Information caloric content in all government and aided primary from household surveys might be more informative in schools.228 Although initially only primary students this regard. were covered, in October 2007, the scheme was further household data from the nss suggests much lower revised to cover children in upper primary (classes VI coverage rates, more so in urban areas, with wide to VIII) initially in 3479 Educationally Backwards Blocks variations across states (see table 4.10). Karnataka, (EBBs) and from 1st April, 2008 across the country. The scheme receives budgetary support from the centers as Tamil Nadu, Himachal Pradesh and Chhattisgarh perform a CSS, for which the nominal GoI allocation has increased relatively well with over 70 percent of rural households by more than six times between 2002/03 and 2008/09. reporting having received midday meals at-least once While the center finances the grains and transportation, in the last year. Bihar, Jharkhand, Rajasthan and Uttar states retain the responsibility for several elements of Pradesh however report less than one third the coverage implementation, including cooking infrastructure and (less than 20 percent) in these states. The situation in provision of cooks. urban areas across India and in a majority of the states remains far worse with less than half the coverage at present the midday meal scheme is the world's rates in rural areas. It is possible that household survey- largest school feeding programme and feeds about based data underestimates coverage as the estimates 118.46 million children in over 9.5 lakh schools are for households with at least one child aged 6-10 across the country. Administrative data suggests that rather than those where a child of that age is enrolled Figure 4.7: Coverage of midday meals according to administrative Data (among children enrolled in primary school), 2007-08 120 97 100 100 100 100 100 95 96 90 92 87 88 79 80 75 71 67 67 60 % 40 20 0 Gujarat Jharkhand Bihar Uttar Pradesh Madhya Pradesh Nagaland West Bengal Maharashtra Jammu & Kashmir Uttarakhand Assam Chhatisgarh Orissa Rajasthan Andhra Pradesh Delhi Source: Ninth Report of the Food Commissioner's appointed by the Supreme Court 2009. Note: States sorted in ascending order of coverage. 228 AspertherecentlyrevisednutritionalnormsforMDM,everyprimarychildshouldreceive450calories,12gmofproteinsperdayandadequatequantitiesof micronutrientslikeiron,folicacid,vitamin-Aetc.andupperprimarychildshouldreceive700caloriesand12gmofproteinswithadequatemicro-nutrients. CHAPTER-4: Public Works and Promotional Programs Table 4.10: Coverage of midday meals, 2004/05 Table 4.11: Coverage of mDm by per capita % hh among households with at expenditure and social category, 2004/05 least one child aged 6-10 years % hh among households with at State Rural urban least one child aged 6-10 years Andhra Pradesh 54.8 26.2 Rural urban Assam 32.9 7.7 Poorest 51.7 41.5 Bihar 17.7 7.9 Q2 48.7 35.9 Chhattisgarh 77.4 30.9 Q3 46.8 31.3 Delhi 3.5 10.7 Q4 42.9 28.0 Gujarat 62.6 22.4 Richest 30.9 9.9 Haryana 36.5 7.2 OBC 42.8 25.4 Himachal Pradesh 73.2 32.4 SC 49.2 28.7 Jammu & Kashmir 3.0 0.2 ST 55.0 26.9 Jharkhand 19.6 5.6 Other 43.0 14.9 Karnataka 77.0 34.8 Total 45.7 21.6 Kerala 56.1 41.0 Source: Staff estimates, based on 2004/05 NSS. Maharashtra 66.7 27.9 Madhya Pradesh 61.8 21.7 of children from households in the richest quintile is also Orissa 60.3 26.9 quite high, at least in rural areas at over 30 percent of Punjab 7.0 1.2 the richest households. Part of the reason for the sharp Rajasthan 36.3 8.9 drop in coverage of children from the richest quintile Tamil Nadu 77.4 44.6 in urban areas could be the fact that a much higher Uttar Pradesh 27.0 7.2 proportion of these children are enrolled in private Uttarakhand 52.7 11.7 rather than public schools. As expected there are state- West Bengal 63.9 31.8 wide variations. An analysis of the 3-state SP survey data All-India 45.7 21.7 finds that the proportion of participating ST households ranged from about 90 percent in Madhya Pradesh to Source: Staff estimates, based on 2004/05 NSS. 75 percent and 71 percent in Orissa and Karnataka. As far as the economically poor are concerned, the proportion in a government/aided primary school. However, in of participating households from the poorest quartile states with relatively low rates of private schooling, this was even more varied at 82 percent, 76 percent and is unlikely to be a major concern since it is commonly 59 percent in the three states respectively. known that children attending private school enroll in government school to avail of the MDM. One of the multivariate analysis of determinants of midday reasons for lower coverage in urban areas could be due meal coverage from the three state sp survey also to the relatively higher reliance on private schooling. confirms a range of factors which are positively associated with taking of school midday meals by though the midday meal (mdm) program is universal children. These include being from poorer households, in coverage (for all children enrolled in government being an SC household (though not ST), having higher and aided schools), coverage among poorer sc/st social capital, and women's participation in community households is relatively higher than that among life. This can be seen in Table 4.12, which presents richer and general caste households, particularly in findings from a study of Orissa, Karnataka and Madhya urban areas (see table 4.11).229 In urban areas, 41 percent Pradesh in 2006. of the poorest households reported children benefitting from midday meals compared to about 10 percent while systematic, reliable national studies on the of the richest households. Given that the program is impact of midday meals are not yet available, studies universal rather than targeted to the poor, the coverage from a variety of states consistently report positive 229 SeealsoGaihaetal(2007). Social Protection for a Changing India: Volume II Table 4.12: Determinants of participation ­ karnataka, from the PROBE survey in India, which found that school orissa and madhya Pradesh (2006) meals have a positive and significant effect on girls' hh characteristic Coefficient grade attainment. The probability of completing primary education was 30 percentage points higher for girls Wealthlevel: living in villages with a School Feeding Program than Poorest quartile 0.888** for other girls.232 Another study using the longitudinal Second quartile 0.484* survey data in 2002 and 2007 in Andhra Pradesh using Third quartile 0.399* propensity score matching methods finds that there Socialgroup: were significant impact of midday meals on test scores SC 0.629** of older children of older cohort (born between January ST 0.286 1994 and June 1995).233 Given the need to look over HH social capital index 0.687** longer time periods for more complex outcomes, it will Women's participation in 1.502** be important to continue studies of impacts on the meetings and elections above indicators. For instance, a study in Madhya Pradesh Household size 0.513** suggests that the enrollment impacts at grade 1 may Source: Dev et al (2007). not persist as strongly in higher grades, but generalizing Notes: ** = significant at 5% level; * = significant at 1% level. Reference from that experience would be unwarranted. categories are the richest quartile (for household wealth) and other castes (for social group). even less is understood on the net nutritional impacts of the midday meal program. While impacts impacts on school enrollments, with particularly on education outcomes have been the focus of most positive impacts on girls. Drèze and Kingdon (2001) studies to date, there are very few rigorous studies on using the PROBE survey data find that school meals are the nutritional impact of the MDM scheme. There are associated with a 50 percent reduction in the proportion two elements to consider on nutrition. The first is the of girls who are out of school. Also female school (initial nutritional content of the meal itself, over which there and current) enrolment is about 11 to 15 percentage has been much debate in recent years with respect to points higher when the local school provides a mid- the cost allowance per meal and the level of dietary day meal than when it does not.230 Equally, given variation provided for. The second is the extent to the universal coverage of MDM, it is difficult even in which school meals are complements to or substitutes dedicated studies to measure precisely the net impacts for home meals. The two studies that look at the latter on enrollments. Nonetheless, all studies which have question find significant substitution effects in the looked at the impact find increased enrollments which home for midday meals.234 However the net effect of are attributed to the program. These findings are having MDM might still be positive. Evidence from consistent with field work for this study. Madhya Pradesh suggests that school feeding reduces daily protein deficiency of participants by 100 percent while impacts on enrollments appear to be positive, and calorie deficiency by almost 30 percent, even after recent field reports also point to positive outcomes accounting for possible substitution effects at home.235 on other educational outcomes such as attendance, The same study also finds that the daily nutrient intake retention, and attainment. however, more systematic of a child increases by 49 percent to 100 percent of the work is still needed to understand these impacts.231 total transfers made. Another study using longitudinal An important study that looked at this aspect comes data in Andhra Pradesh in 2001 and 2007 finds positive 230 DrèzeandKingdon(2001). 231 SeeSen(2010)forasummaryofthefieldstudiesoneducationalrelatedoutcomesofMDM.Thesearelargelysmall-scalestudieswiththeestimatesofimpact typicallybasedontheperceptions(typicallyofparentsand/orteachers)oronschool-leveladministrativeinformation(suchasregisters,etc.)ratherthansurvey- baseddata.Asaresult,generalizationsaredifficultandpointtotheimportanceofrigorousassessmentsofperformanceandimpact. 232 DrèzeandKingdon(2001). 233 Singh(2008). 234 SeeBlue(2005)forstudyofUdaipur,andAfridi(2004)forMP. 235 Afridi(2007). CHAPTER-4: Public Works and Promotional Programs changes in weight-for-age and height-for-age between Table 4.13: infrastructure for mDm these two years attributable to the MDM program among % Schools % Schools % Schools the younger cohorts (born between January 1994 and with with with storage June 1995).236 This effect is particularly strong for child kitchen cooking for drinking groups whose families have been affected by drought. States sheds utensils water But there was no effect on nutritional outcomes among Andhra Pradesh* 19.9 ­ ­ the older cohorts for both drought and non-drought Assam ­ 27.6 ­ affected groups. This suggests that more work would be Bihar ­ 25.6 ­ desirable to understand the net impacts, and perhaps to Chhattisgarh 27.6 100.0 100.0 think through how the content of school meals can be Gujarat* 32.4 93.2 0.5 designed to take account of home feeding responses. Himachal Pradesh ­ 0.0 ­ The experience of states such as Tamil Nadu and Gujarat, which have provided meals over and above the Jharkhand ­ 12.9 ­ minimum requirements, could provide useful lessons in Karnataka* 54.5 ­ ­ this respect. Kerala* 88.2 100 ­ Madhya Pradesh* ­ 7.4 ­ the available studies and goi assessment of the Maharashtra* 3.9 8.0 70.9 program point to a number of ongoing challenges Punjab* 0.0 0.0 0.0 in ensuring effectiveness. implementation of mdm Rajasthan* 0.0 48.1 65.2 has experienced several challenges related to timely Tamil Nadu 96.3 80.8 83.7 and adequate procurement, transport, storage, Uttar Pradesh* 6.5 88.6 12.0 cooking and distribution of food.237 The 2006 revised West Bengal 0.0 92.9 68.4 guidelines have introduced greater flexibility for states with respect to distribution channels, e.g., through SHGs, Source: Annual Work Plan & Budget documents, 2007­2008. * AWP&B of 2006­2007, quoted in Seventh Report of the Office of the VECs, mothers' groups, etc.238 There is significant variation Supreme Court Commissioners. among states in terms of infrastructure like kitchen sheds, cooking utensils and drinking water.Overall good in Orissa.240 A World Bank study in Rajasthan also found performers are Tamil Nadu, Kerala and poor performers that 68 percent of sample teachers spent more than are Punjab, Mizoram, Tripura and Himachal Pradesh(see 1 hour or more than 17 percent of their allotted teaching Table 4.13). Despite a decade having passed since the time on preparing and serving the school meal.241 program was first introduced, a nationwide CAG audit found that in 11 Indian states, classrooms were still used to store grains and cook meals.239 Ensuring hygiene and quality is another major challenge. The CAG audit for C. ConCluSionS AnD quality test of grain also revealed that in three of the ReCommenDATionS four states surveyed, the rice served "was adulterated and not fit for human consumption". Of more concern both public works and promotional social protection is the finding that logistics of cooking and serving can programs have received increased political and be a source of distraction to school staff, leading to a budgetary priority in recent years. This seems a sensible reduction in teaching time. Government evaluations of effort to get greater leverage on poverty reduction and their own program estimate a loss of about 11-30 hours human capital formation from the safety net. Given that per week in 6 states and up to 41 percent of teaching time significant spending on these programs, particularly 236 Singh(2008). 237 SeeforexampleCAG(2008). 238 Priortothenewguidelinestoo,severalstates,includingpoorstatessuchasOrissa,hadexperimentedwithdifferentdegreesofallowinglocalcontroloveruseof publicfundstopurchasethefoodformiddaymeals,primarilytothispointinprocuringsupplementalitemstothebasicration,suchaseggsorsomevegetables. Thishasobviousattractionsfromanutritionalviewpoint,butalsopossiblyintermsofsomelocallinkagestolivelihoodactivities. 239 CAG(2008). 240 CAG(2008). 241 WorldBank(2007b). 100 Social Protection for a Changing India: Volume II MGNREG and mid-day meals, can be expected to there has been some evolution of IEC campaigns in continue, it will be increasingly important to ensure that terms of moving from the dissemination of rights the desired equity and public expenditure efficiency to an emphasis of the need to and the means by objectives are achieved. The following sections provide which individuals can demand that right. It is also some recommendations in this respect. necessary to make communities aware of the unemployment allowance provision in the Act for the guarantee function to be credible. It is vital (i) Public works to deepen awareness raising efforts on MGNREG many of the appropriate reforms of public works policy entitlements, in close collaboration with civil are already reflected in the guidelines of mgnreg, society and using strategies that are tailored to a which in several ways represents the most serious largely illiterate audience. effort to date to address many of the institutional Leapfrogging technical manpower constraints and implementation problems encountered in previous (e.g.,theshortageofengineersattheblocklevel) works (and indeed several other) programs. In many by developing detailed technical specifications states there has been greater political and institutional of MGNREG works for different geo-climatic commitment to trying to "make the program work". conditionsasapreparatorystage.This minimizes Examining the implementation experience of MGNREG the technical input required at the block and GP in the last four years suggests some lessons for improving level at the planning stage and while starting the delivery of the program in states that are not doing as a worksite. This can be done without the aid of well as others. It also identifies some areas that warrant technology (e.g., as in Madhya Pradesh) or with increasing attention as the program matures. technology as an integral part of a transactions- the variable implementation experience across states based Management Information System (e.g., as offers several valuable cross-state learning. One in Andhra Pradesh). of the biggest challenges in implementing MGNREG is Streamliningtheflowoffundsinvariouswaysso to match the expression of demand with the supply of astopreventfundingdelaystoconstrainopening worksites and employment opportunities. It is vital to of worksites or payment of wages. For instance, address the constraints ­ formal and informal ­ on this some states (e.g., MP) make available advance process. At the broadest level, this goes to the heart of funds (linked to the volume of MGNREG work) what a rights-based, demand-driven approach means in with GPs that makes it easier to open worksites practice. Some states have done better at establishing in response to demand. In addition, some states systems to improve the responsiveness of supply to the have reduced delays in payment of wages from demand for work. This has typically involved attention the GP to worker post office accounts (e.g., by to or innovations in the following areas: placing a "float" with post offices to make wage Establishing the implementation structure payments while waiting for funds transfer; and early in the game and ensuring adequate staff by mandating that the GP MGNREG account and with the appropriate orientation and skills. This accounts of MGNREG households to be in the includes serious and sustained efforts at building same branch). capacity at all levels of the delivery process, Revisingtheruralscheduleofrates(SoRs)through often in partnership with the State Institute for detailed time and motion studies for different Rural Development. Under the Act, a portion of localesandgroupstoenablea"normal"workerto funds are available for capacity building of those earntheminimumwageatMGNREGworksites. involved in MGNREG implementation, including Partneringwithcivilsocietyorganizationstowork PRIs. Using these funds effectively will be a critical assupportagencies on a variety of areas, including element of program success over time. orientation and capacity building of MGNREG staff, Generating awareness of the rights and awareness generation and mobilization among entitlements under MGNREG as the first step workers, promoting participatory planning of towards establishing a right to work. Over time, works, and enhancing accountability. CHAPTER-4: Public Works and Promotional Programs 101 Emphasizing the commitment to transparency be on the approved list of MGNREG works but are and accountability. While the design of part of the larger village development plan. MGNREG contains many safeguards in terms Revitalize Gram Sabhas and institutionalize the of transparency and accountability, actual directinvolvementofcommunitiesinidentification implementation on this front has been highly ofworksundertakenunderMGNREGinawaythat variable. This includes institutionalizing social isintegratedwiththelargervillagedevelopment accountability mechanisms, making the MIS up- plan.This has to date been the weakest element in to-date, and a continued emphasis on monitoring the chain, largely because gramsabhas are often and community mobilization. not held. If the Act's objective of empowering in addition, increased attention by policy-makers communities and strengthening grassroots and implementing agencies on improving the quality democracy is to be achieved, it will be important and relevance of assets created under mgnreg is to ensure that the role of gramsabhas anticipated warranted as the program matures. This is critical if under the Act is made a reality. This includes MGNREG is to have any long-term impact on the rural ensuring gramsabhas are held regularly, building economy and future livelihoods. In addition, the creation the capacity of gram sabhas in participatory of appropriate, durable and productive assets could planning, community oversight and other potentially help garner the support of non-MGNREG relevant areas and the mobilization of groups participants for the program and a stake in improving such as SHGs and CBOs to have their priorities program outcomes. There are many elements to this reflected through the gramsabha process. increased focus on the second objective of the Act: Establishsystemsforprovidingin-timetechnical Explore options for a wider range of works inputsforassetplanningandevaluation,beyond authorized under MGNREG to reflect variable the currently mandated technical supervision needsandtodovetailwithotherprogramssoas during asset creation. This includes providing tocontributetowardsacoherentvillagedevelop- technical inputs (e.g., through village-level resource ment plan. The list of eligible works needs to mapping to ascertain the technical feasibility be flexible enough to incorporate seasonality, of different types of works) to the gram sabha differences in geo-climatic conditions and the during the planning process and development needs of specific groups, particularly those not of shelf of works. Similarly, monitoring of the capable of hard manual labor. As experience quality and durability of assets created will grows, this would be worth looking at from increasingly become important. In this respect, several perspectives, some of which are currently MGNREG could draw on international experience under discussion. The first is the range of works (e.g., Bangladesh's Food For Work program) on that could be implemented in flood-prone, water- developing cost effective methods for estimating logged, heavily forested or mountainous areas. rates of return on assets. In addition, the use of The second is the range of possible works which technology such as GIS in both planning and could be offered to all MGNREG workers, and monitoring would be invaluable. whether some "softer" work options ­ e.g., related to provision of social welfare or community an additional issue for consideration is whether services for vulnerable populations ­ could any element of direct human capital formation can be introduced which have positive social be factored into mgnreg as it matures. Presently externalities. The third is looking at the specific there is no provision under MGNREG for skill formation needs of sub-groups for whom special efforts among workers. This may be something that could be may be needed to provide appropriate work, e.g., considered in due course with the view of enhancing disabled people who may not be able to carry not just current but also future livelihoods. One option out hard physical labor. The fourth is exploring to consider is the South African public works program options for MGNREG to finance the labor which provides for two days training per month component while the community (or other line of work for those undertaking public works. While departments) co-financing works that may not such an approach obviously requires a supply side 102 Social Protection for a Changing India: Volume II agency ­ probably on a contracted-out basis ­ which and within states varies sufficiently that a "one size fits can provide useful training, it seems a useful option to all" credit program seems increasingly inappropriate; consider in future development of MGNREG. (ii) the livelihood support needs of individual groups and poor households that go beyond credit are also a final issue is that public works for the poor remain diverse, and uniform CSS have failed to respond to this; restricted to rural areas. Recently, policy-makers (iii) the growing penetration of both commercial banks have started to discuss the possibility of designing a and other non-bank players such as MFIs make the self-targeted public works programs for the urban poor. challenge for the poor making them "bankable" rather Such programs already exist in the works schemes of a than being the sole source of formal credit. number of developing countries such as Ethiopia, Liberia, Colombia, and some other African and Latin American such an approach would require greater effort by countries. Drawing on the experience of these countries, states to develop state specific poverty reduction for example the role of urban communities in identification strategies for the effective use of central subsidies of beneficiaries, planning and execution of works, and for livelihoods promotion. The NRLM design gives oversight, would be useful in designing an urban public states a fair degree of flexibility in trying different works program for Indian cities and towns. In fact, one approaches to livelihood support; some further options of the north eastern states is planning to introduce an could be explored in this regard: employment guarantee for urban areas as a state-funded using reputable MFIs as a channel for credit scheme. Combining such a program with vocational or delivery where they have a presence. MFIs have technical training would enable young participants to a strong interest in developing a sustained upgrade their skills and also compete in the labor market. relationship with BPL clientele, and more diverse An example of such an approach is being developed in and community-based sources of information Kenya for youths living in urban slums.242 on borrowers, allowing for a more informed assessment of risk in the lending transaction. They are also more likely to enforce repayment, reducing (ii) Programs to promote movement rent-seeking opportunities seen presently, and out of poverty in the short and provide appropriate support services to their long run members. MFIs often have savings as an integral element of their relationship with clients, so that (a) Targetedruralcredit credit provision is reinforced by savings that can act as cushions against household shocks. At the the government has recently undertaken a significant same time, a significant issue would be the relative overhaul of the sgsy in the form of the national terms of SGSY lending and those of most MFIs to rural livelihood mission. The main reasons for SGSY's their existing clientele, and such an option would and predecessor programs' weak performance strongly demand some degree of harmonization between suggested the need for a fundamental reform that went MFI financial, reporting and other procedures and beyond marginal improvements. In many ways, the those of government. restructuring of the existing program into NRLM goes a where local labor markets are stagnant and/or long way in doing this. migrationisalreadysignificant,financingsupport the nrlm moves away from the precisely defined credit services for poor migrant workers which would program for the poor as exemplified by sgsy and irdp and reduce some of the economic and social costs of instead includes a range of livelihood support options migration, and increase its benefits. There are for the poor. this is consistent with the options in the already interesting small-scale examples of such sp block grant proposed within the "3+2 strategy" initiatives, e.g., in southern Rajasthan with workers outlined in the executive summary.The rationale behind migrating to Gujarat. Some of the services that such a restructuring and reorientation is necessary for a could be provided include: (i) reliable remittance variety of reasons: (i) the nature of labor markets across mechanisms; (ii) support for children of migrating 242 SeedelNinnoetal.(2009). CHAPTER-4: Public Works and Promotional Programs 10 workers either in situ or in destination sites to constraints in the provision of services, particularly in ensure that they do not drop out of school; and rural areas, could also play a role. (iii) provision of reliable information on labor market the arguments on introducing a demand side element conditions in destination areas, probably through to stipends through either expansion to private a contracted-out service provider contract. unaided schools or use of education vouchers (b) Schoolstipends redeemable in private schools are more complex. It is probably not possible to think about any wholesale while the evidence on school stipends suggests shift to systems like vouchers which would be that they are not as well targeted as many safety relatively new in the Indian context, and one where net schemes, their objective is distinguishable. the international experience ­ particularly in developing position of this report is that they will continue to countries ­ is far more limited. However, this should be an intervention worth expanding, though with not preclude piloting in states where the conditions serious thought about some design elements. The are appropriate, or perhaps in areas in selected states more pertinent concerns with school stipends seem where the participation in private schooling by poor to be two-fold: (i) in their current form, where stipends households is more pronounced (e.g., in urban areas are based on enrollment and not attendance and thus where the experience of a pilot in urban Delhi was with no assurance that the intended outcome is being fairly positive).243 promoted, stipends are less likely to achieve the goal of improved human capital acquisition among the poor; (c) Middaymeals and (ii) stipend schemes do not provide choice for households in schooling, being linked to enrollment only Like MGNREGA, the mid-day meals program is also in government and aided schools. This may reduce their conceived as a universal right rather than a targeted potential welfare impacts not only for the household, program. Mid-day meals are a universal entitlement for but also for the education system, by increasing the all children enrolled in government or government aided marginal costs of private schooling and reducing the schools. With the recent expansion, midday meals have accountability of the public system that might otherwise become an increasingly important part of the Indian be encouraged by households "voting with their feet". safety net. While the national level impacts of midday meals in educational, nutritional and other dimensions there seem solid arguments for exploring ways in which remains to be understood in depth, the evidence to make stipend receipt conditional on a specified level available suggests that MDM have had positive effects of actual attendance rather than simply enrollment, at least on enrolments. As a result, the MDM program which in itself may not contribute to human capital provides an interesting example of a major demand investments. The obvious challenge that such a transition side SP intervention which has great potential. In this would present is operating an effective system to record context, a stronger focus on monitoring and evaluation and verify school attendance. However, evidence from is warranted. This requires establish systems to monitor a growing number of countries which have introduced the performance in the field with respect to inputs conditional cash transfers based on a specified level of (such as already being done for example for school school attendance indicates a range of positive effects infrastructure for providing meals), outputs (besides on attendance, and other household welfare indicators. administrative data on children availing of MDM) and Though several central and state-specific CCT programs outcomes. The latter requires conducting more widely exist in India, these operate more as cash transfers representative studies on the various impacts of midday rather than conditional cash transfers. Key challenges meals, in particular nutritional impacts. In addition, in the effective implementation of these programs as while several concerns on coverage and implementation CCTs include the monitoring and enforcement of the have been addressed through the recent expansion and conditionality and a weak institutional framework for revised guidelines, more can be done to reduce the such cross-sectoral programs. In addition, supply-side variability in performance across states. 243 CMSSocial(2009).SeealsoShahandBraun-Munzinger(2006)foracriticalreviewoftheexperienceofelevencountrieswitheducationvouchers,withadiscussion oflessonsforIndia. 10 Social Protection for a Changing India: Volume II CHAPTER Social Security 5 Closing the Coverage Gap Chapter­5 Social Security Closing the Coverage Gap in addition to social assistance programs aimed at force. Closing the "coverage gap" is an important policy alleviating existing poverty, many social protection objective in India and many developing countries.247 systems insure against a range of shocks that can expanding coverage of social protection programs often lead to poverty.244 These social security schemes generally include provisions for old age, disability and was a major plank of the common minimum program death of the breadwinner under the umbrella term, (cmp) of the upa government and remains a priority. pensions.245 Many countries have also tied health In 2006, a major report was produced by a special insurance coverage to membership in social security commission under the auspices of the Ministry of schemes which require contributions or premia while Labour and Employment (MOLE). The Commission others have opted for general revenue financing.246 This provided a diagnosis of the situation of social insurance chapter is about the schemes that operate in India today coverage among unorganized workers, laid out possible for both organized and unorganised sector workers. In approaches to the problem, and set ambitious targets addition, based on recent experience, it outlines thinking for coverage over a multi-year period. Framework about how to expand pension and health insurance legislation was passed in December 2008. In the last coverage to the vast majority of India's unorganized labor few years, the MOLE has achieved notable success 244 Otherimportantsourcesofincomeprotectionsuchascropandlivestockinsurancearenottypicallycoveredbysocialinsuranceprograms,butarecriticalinterms ofprotectionagainstcovariateshocksintheruralsectorincountrieslikeIndia.Itisalsoworthnotingthatlumpyexpendituresforweddingsarealsoimportantin theIndiancontext. 245 Unemploymentandcashbenefitsformaternity(asopposedtomedicalinsurance)areoftencoveredbygovernmentsponsoredsocialinsuranceschemes.There aremoralhazardproblemsinbothcases,especiallywithregardtounemployment.Moreover,thedefinitionandmonitoringofunemploymentforunorganised sectorworkersisdifficulttoadminister.ESISoffersanunemploymentbenefit. 246 SeeWagstaff(2007)forasummaryoftheongoingdebatebetweenthesemodels. 247 Foradetaileddiscussionofextensionofsocialinsurancecoverageinthecontextofpensions,seeHolzmann,RobalinoandTakayama(2009). CHAPTER-5: Social Security: Closing the Coverage Gap 10 in expansion of some forms of social insurance, in thereareimportantlessonstobetakenfromthese particular for health insurance for the poor under the programswhichindicatethatdifferentapproaches Rashtriya Swasthya Bima Yojana (RSBY) program. RSBY areneededtoreachouttounorganizedworkers offers many lessons on expanding social insurance withsocialsecurity.For example, there is evidence coverage to unorganized workers. that unorganized sector workers are willing to contribute for their insurance and pension the structure of the chapter is as follows: section coverage. In addition, effective outreach has been a provides a brief description of the labor force in achieved usually as a product of collaboration the unorganized sector based on secondary material between Government, insurers, and intermediary and analysis of survey data. It focuses on the factors organizations of unorganized workers. More that are likely to be important for any attempt to reach study is needed however, especially with regard these workers through new programs, especially those to the role of intermediaries and the scalability of that may require workers to help finance the scheme different schemes. themselves. Section B then looks at pension and health insurance schemes in place prior to 2007. These can be large scale coverage expansion will require separated into those that are mandated for organized government subsidies to address the question sector workers, mostly public sector and large private of affordability and incentives for unorganized firms, and those that are targeted to the unorganized workers, and oversight in order to mitigate sector ­ mostly small firms and the rural labor force. fraud and/or mismanagement. However, the Section C then focuses on initiatives since 2007 to address government can partner with market and non- the "coverage gap" in social insurance, in particular the governmental actors to avoid creating new design and initial experience with the RSBY health layers of bureaucracy to implement schemes. insurance scheme for BPL workers. Conclusions and On the other hand, given economies of scale, recommendations follow. the need for portability of benefits and the exigencies of supervision, the government's role the key messages of this chapter are: can usefully include initiatives to create robust despite general agreement on the size of the and harmonized systems for identification of unorganized sector, there is less understanding beneficiaries, recordkeeping and benefit delivery. of its heterogeneous nature, especially as achieving critical mass and establishing the apply to the implications of these differences credibility of a new policy to expand coverage for initiatives to expand pension and health couldbefacilitatedinthefirststagebytappinginto insurance coverage. existinggrouparrangements(tokeeptransaction thelimitedevidenceavailableshowsthatcoverage costslow)andfocusingonhouseholdswithsome is closely correlated to income/consumption ability to smooth consumption (affordability) levelsaswouldbeexpected.Pension and health if provided access to the right instruments. This coverage has historically in India been extremely approach recognizes the complementary roles of low in the bottom half of the income distribution, safety nets and insurance programs. with life insurance moderately higher and higher theRSBYprogramispath-breaking,notonlyfor overall than other types of insurance. social insurance expansion, but for the health therearemanyschemesdirectedatunorganized system and social protection programs for the sector workers operating at present including poor more broadly. While it is relatively early national and state government programs and days of implementation, the RSBY program has an increasing number of non-governmental numerous design features and a learning-by- programs, including microinsurance schemes. doing approach to implementation which reflects However, there is no overarching policy lessons of Indian and international experience. framework nor ­ with the notable exception The program has the potential to help poor of RSBY - systematic monitoring of their households mitigate the impacts of serious health performanceorimpact. shocks and the approach that has been taken with 10 Social Protection for a Changing India: Volume II regard to a variety of implementation challenges sector but are still eligible for benefits at some point can be harnessed to improve other programs. in the future. Finally, both in schemes for civil servants and the major formal sector schemes, recordkeeping has always been problematic. Aside from the problem A. SoCiAl PRoTeCTion that this causes for individual members, it makes it very difficult to assess trends in coverage or the impact of CoveRAge AnD The nATuRe new policies. oF The unoRgAnizeD survey data provide evidence that unorganized sector lABoR FoRCe workers are heterogeneous in ways that will directly influence any attempts to expand social security definitions of the informal or unorganized sector coverage. Table 5.1 shows that earnings vary widely. vary but there is some consensus that approximately This information is important for policymakers seeking nine of ten workers in india belong to this category. to expand coverage by showing that workers fall roughly The report of the National Commission for Enterprises into three categories: in the Unorganized Sector (NCEUS) cites two definitions that result in estimates of the informal sector ranging the first group is individuals with high incomes ­ from 85.8 to 91.3 percent of total employment. The professionals, large landowners etc. ­ that should World Bank has cited contrasting definitions yielding a be able to avail themselves of voluntary savings range of 86 to 93 percent.248 Jhabvala et. al. (2003) note and insurance schemes already available in the that the System of National Accounts (SNA) definition market (typically with tax preferred status). results in significantly lower estimates of around the second group is the one for which anything 75 percent of the labor force. other than very marginal expenditure on premia or contributions would likely be unaffordable and social security coverage is also difficult to measure could actually reduce current welfare unduly by and varies depending on the type of risk covered. lowering consumption levels below subsistence. For example, there many more accounts open at the Employees' Provident Fund Organization (EPFO) than finally, there is a third group that may be in a there are contributors in any given month. Some of position to benefit from insurance and pension these workers have changed jobs and are being double coverage and could afford to pay for a significant counted. Others have died and no one has closed their portion of the required premia. The first group is accounts. Still others have moved into the unorganized not a public policy priority while inclusion of the Table 5.1: Distribution of unorganized sector workers aged 20-50, by earnings decile, 2004/05 income Decile income Range (Rs) Share (%) Cumulative Share (%) 1 <=11000 11.8 11.8 2 >11000 & <=16800 11.0 22.8 3 >16800 & <=21600 11.6 34.4 4 >21600 & <=27000 11.6 46.1 5 >27000 & <=35000 11.6 57.7 6 >35000 & <=42000 10.6 68.2 7 >42000 & <=54000 8.6 76.8 8 >54000 & <=72000 8.9 85.8 9 >72000<=105000 6.8 92.6 10 >105000 7.4 100 Source: Authors' calculations based on IRES survey data. Bank staff estimates. 248 Forexample,differentdefinitionsareappliedinJhabvalaetal.(2003),NCEUS(2006)andAhmedandNarain(2010).Thesedefinitionsrevolvedaroundvariantsof thelegaldefinitionoforganizedsectorwherebyenterpriseswith10ormoreworkerswerecoveredunderparticularlaborstatutesand,intheory,shouldbecovered bycertainsocialsecurityprograms. CHAPTER-5: Social Security: Closing the Coverage Gap 10 second group, while desirable and high priority, is the spontaneous demand for micro-insurance and likely to require public subsidies to finance social participation in lic, uti and state welfare schemes insurance. The third group would also be a likely (described in the next section) demonstrates that target of an attempt to expand insurance and some workers are both capable and willing to pension coverage, and the relative use of public purchase insurance and to contribute financially subsidies to incentivize participation is a key to their own social security. This revealed demand question.249 has been identified in other studies both for India and other countries with similar characteristics.250 On in addition to differences in ability to pay, the data the other hand, it is clear that formal sector schemes suggest other differences that could affect potential that are designed around stable wages that are coverage expansion plans, in particular membership of relatively easy to monitor and employer mandates unorganized workers in groups that could facilitate are not easily extended to these workers. A different program administration. For example, the data allow approach is required at least until other factors that for an estimate of membership in certain types of groups restrict the formalization of the labor markets are that can or already do offer some social security to addressed. their members (see Table 5.2). While the figures shown suggest perhaps only one in seven unorganized sector as illustrated in the next section, several recent workers belongs to some kind of group, it is also the case initiatives to extend coverage in india are intuitively that membership in MFIs and SHGs has been increasing based on a strategy of identifying those parts geometrically in recent years and would already be of the unorganized sector labor force that can higher today than these 2004 data suggest. Given practically be attracted into participating in social the low coverage base, the inclusion of these groups security schemes. Implicit in their design is the need within SP programs would be a significant increase in to consider factors such as transaction costs, incentives percentage terms. As discussed in the next section, for voluntary take up, and affordability. These one possible approach would involve encouraging initiatives are so far largely uncoordinated and are existing groups such as microfinance institutions not part of a national strategy. Nevertheless, these (MFIs), Self-help groups (SHGs), co-operatives or other innovations and experiments offer important insights organizations to `plug in' to a well designed system in into which approach may be saleable as well as order reduce transaction costs. negative lessons. Table 5.2: unorganized sector workers that are members of groups by earnings decile membership (% of Decile) income Trade Decile Trade union welfare Fund Association Coop Society Self-help group Total 1 0.10 0.10 0.23 0.78 8.27 9.49 2 0.33 0.08 0.64 1.25 4.56 6.87 3 0.93 0.24 0.95 1.91 2.54 6.57 4 1.05 0.11 1.05 1.71 2.77 6.70 5 1.50 0.24 1.08 2.69 2.53 8.04 6 1.83 0.58 1.39 2.56 1.89 8.25 7 1.53 0.25 1.53 3.75 1.25 8.32 8 1.47 0.34 1.09 3.90 2.19 9.00 9 1.59 0.68 1.50 5.21 2.49 11.47 10 2.29 0.69 2.52 5.91 1.22 12.62 Source: Bank staff estimates based on IRES survey data. 249 AsimilarcategorizationismadebyMadeshwaranetal.(2006). 250 ForIndia,seeMadheswaranetal.(2005a)andDror(2006).ForIndonesia,seeAngeliniandHirose(2004). 110 Social Protection for a Changing India: Volume II targeted subsidies for social insurance expansion as life insurance. The defined contribution scheme that aim to identify households falling in the bottom also functions as a forced savings mechanism for third of the income distribution can complement these other purposes ranging from children's marriage to initiatives. The RSBY program is a promising example of unemployment. The central government contributes such a program. However, the experience ­ described 1.16 percent of the covered wage bill for these formal in detail below ­ shows that good recordkeeping, the sector workers. Meanwhile, civil servants have a non- right incentives and other design details are crucial in contributory, defined benefit pension scheme that is determining program success. unfunded and has changed little in the last century (see Table 5.3).253 the problems with these schemes have been well B. SoCiAl SeCuRiTy SChemeS documented and the potential for expanding their oPeRATing in inDiA coverage significantly appears to be limited.254 Pension and health schemes for civil servants, for government-sponsored and administered programs example, are essentially occupational schemes that have dominated pension and health insurance have resulted in large unfunded liabilities. Meanwhile, provision in india. The most extensive mandatory the EPFO and ESIS have, among other problems, schemes for pension and health were both set up fifty serious issues with evasion and arrears of employers years ago. In health, the Employees' State Insurance that actually are mandated to participate. Like many Scheme (ESIS) is the largest insurer, followed by schemes designed for formal sector workers, neither schemes that cover central government, defense and is well suited to incorporate most unorganized sector railway employees.251 Public sector health insurers workers. In fact, coverage in the EPFO scheme has cover ten times as many people as their private sector increased slowly since it was established by a special competitors. In pensions, formal sector employers Act in 1952. Measured on the basis of affiliates, the with at least 20 employees are mandated to join the share of the labor force (including the informal or Employees Provident Fund Organization (EPFO).252 unorganized sector) covered has increased from Workers covered by this scheme have defined about one to five percentage points, or roughly one contribution and defined benefit pensions as well percentage point per decade. Table 5.3: key indicators of mandated social security programs workers covered* Payroll tax/contribution as Program Statutory coverage (thousands) Type of benefits % of wage EPFO Employees of registered firms 47,000 DC and DB pension, life, 12 employer with more than 20 employees disability, withdrawals for 12 employee other reasons ESIS Employees of firms using power 8,400 medical, sickness, maternity 4.75 employer with more than 10 employees unemployment, funeral or 20 without power 1.75 employee Civil service Civil servants at federal and 24,000 DB pension for those hired None in old DB scheme; 10% each pension state level before; DC for new hires for employer and employee * Figure is for 2009 from EPFO Annual Report; note that independent survey data suggest that the figure for active contributors (as opposed to open accounts) is much lower than the number of affiliates. Figures for civil service are based on survey data; Figures for ESIS reported taken from website. *** refers to total spending including administration for 2005-06. 251 GuptaandTrivedi(2005),Table2. 252 Employer-administeredprovidentfundsthatmeetcertainrequirements,includingbenefitsthatareatleastasgenerousasthoseofferedbytheEPFO,canreceive anexemption.Currentlyaboutonequarterofaffiliatesbelongtotheseexemptfunds. 253 A new, defined contribution scheme was introduced for federal government employees whereby workers contribute ten percent of basic wage plus dearness allowanceandthegovernmentasemployermatchesthiscontribution.ThisreplacestheolddefinedbenefitschemeforfederalcivilservantshiredafterJanuary 2004.SimilarschemeshavebeenadoptedbyseventeenIndianstategovernmentstodate,albeitwithdifferentstartingdates.ImportantexceptionsincludeWest BengalandKerala. 254 FortheESISsee,GuptaandTrivedi(2005).FortheEPFOandcivilservantpensionschemes,seeWorldBank(2001). CHAPTER-5: Social Security: Closing the Coverage Gap 111 the failure of mandated schemes to reach the vast discussed in the next section, the new RSBY health majority of workers is clear. Workers covered by insurance program targeted at BPL households has formal social security mandates for health insurance started to change this picture. and pensions represent less than ten percent of the life insurance coverage is also concentrated among estimated labor force in India. Life insurance, much of the better-off, being biased towards urban areas and it voluntary and subsidized through the LIC, is more with low prevalence among backward castes. Only prevalent with around one fifth of households reporting 3.6 percent of households from the poorest quintile report that they have some kind of policy.255 that a member of the household has life insurance, while in distributional terms, social security coverage 48 percent households from the richest quintile report is concentrated in the upper part of the income the same. Coverage rates in urban areas are more than double rates in rural areas. There is considerable variation distribution. As shown in Figure 5.1, pension coverage in life insurance coverage rates across caste groupings. is strongly concentrated in upper income groups and is only significant in the top half of the income in addition to the formal sector schemes, there are a distribution. Not surprisingly, it is also concentrated in number of central and state government initiatives urban areas and least prevalent among the backward aimed at the informal sector. These initiatives fall castes.256 In fact, survey data for 2004 shows that roughly into three categories ­ welfare funds, NGOs households from the richest quintile constitute more of various types and schemes offered by several large than half of all households with health insurance. As providers. There is some overlap between the last two Figure 5.1: Life insurance and pension coverage by income decile 70 60 50 Life insurance percent of workers covered 40 30 Pensions 20 10 0 1 2 3 4 5 6 7 8 9 10 Income decile Source: 2004 IRES Survey. 255 Thesourceofhealthinsurance(publicorprivate;providedthroughemployerorpurchasedindependently)isnotknownfromthesurvey.Thisisalsotruefor lifeinsurance,thoughtheGovernment-ownedLifeInsuranceCorporationofIndiaunderwrotearoundthreequartersofalllifeinsurancepoliciesatthetimeof thesurvey. 256 Ajwad(2006). 112 Social Protection for a Changing India: Volume II types since many micro-insurance schemes use the a parallel initiative based on the same defined partner-agent model for microinsurance.257 contribution model was activated in may 2009 by the pension fund regulatory development authority welfare funds include those administered centrally (pfrda). The PFRDA has established an infrastructure through the ministry of labor at the goi as well as a and rules which allow any adult citizen to make voluntary variety of state level schemes. The centrally-sponsored contributions to the New Pension Scheme (NPS). This schemes include welfare funds cover five occupational worthwhile initiative suffers however, from at least two groups including three types of miners, cine workers design problems. First, up front costs for recordkeeping and the largest group ­ beedi workers. They are financed (paid to the Central Recordkeeping Authority (CRA)) by a cess on the particular commodity produced. The are very high relative to the potential contributions of total number of workers actually covered is not known, most unorganized sector workers.260 Second, there is but a rough estimate would be around 4-5 million. The no obvious incentive to sacrifice liquidity and defer Ministry of Labour and Employment reported spending consumption until the specified retirement age of 60. in FY2004-05 of around one billion rupees.258 The The matching contribution schemes mentioned above schemes provide a variety of services including medical would help address this issue but so far, the state services, life insurance, education and housing.259 schemes have not been linked to the NPS.261 the state level schemes for unorganized workers vary a small scale version of the prfda scheme is already significantly both in terms of coverage and benefits operating in india in the form of the `micro-pension' provided and importantly, no aggregate estimates of product of the unit trust of india (uti). The first client membership exist. For example, during the last decade, in this partner-agent arrangement was SEWA, a well- state governments in West Bengal, Andhra Pradesh and known MFI that provides loans to low income women in Rajasthan have introduced pension schemes to provide the state of Gujarat (and is expanding into other states). income in old age to unorganised sector workers. The In spring 2006, around 30,000 women joined their DC first scheme was introduced in 1998 in West Bengal for pension scheme where contributions of around 200 landless rural workers followed by a scheme for certain rupees per month are invested in a balanced fund invested occupational groups in the urban areas with incomes in bonds and equities. Individuals must maintain a savings under 3500 per month in 2002. The schemes in Rajasthan account with SEWA bank. (There are around 200,000 and Andhra Pradesh are more recent having been such accounts opened at present). Contributions are established in 2008 and 2009, respectively. In AP, the collected following the standard mechanism for saving program is available to members of SHGs. The parameters account deposits. Appointed collectors (currently 75-80 of the different schemes vary. The match is 20 rupees per individuals) accept cash contributions, issue deposit slips, month in West Bengal, 30 in Andhra Pradesh and up to and make record in each members' deposit book. Three 1000 per annum in Rajasthan. There are other important times a month, a special script with standing instruction differences including the use of private asset managers gets triggered to record debit on the member's savings and market investment returns in AP and Rajasthan account and credit on the side of the retirement savings while government manages the fund and determines scheme.262 To simplify operation, subscription amounts the interest rate in West Bengal. The common feature may only be changed once a year. Corresponding data is that they are defined contribution schemes where with the new contributions and details of the accounts the subsidy from the government takes the form of a is then exported to an excel spreadsheet and provided matching flat deposit into an individual account. to the UTI-Ahmadabad office. Although an obvious 257 ThismodelhasbeenencouragedbytheIRDA,especiallythroughtheissuanceofitsmicroinsuranceregulationsin2005.SeeRothetal.(2005). 258 NCEUS2006,p.23. 259 Interestingly,theschemesdonotofferoldagepensions.However,therearespecialarrangementsforincludingbeediworkersintheEPFOschemebasedonakindof cessleviedontheindustry.Onestudyfoundthatthereweremanyproblemswithcompliancewiththisarrangementinpractice.SeeMadheswaranetal.(2005b). 260 Recognizingthisfact,thePRFDArecentlyrequestedthattheGovernmentofIndiacoverthesecostsintheearlystageoftheprogram. 261 Thecentralgovernmenthasrecentlyintroducedamatchingincentiveof`1000inthe2010-11bueget. 262 SEWAgetsacommissionof3percentofthetotalreceiptsfromUTI(whichsupposedlygetsnettedoutofthereturnintheprocessofunitpricecalculation).The commissionisnotlikelytobesufficienttocoveralloperationalexpensesofthescheme­althoughitwouldbedifficulttosegregatethecostoftheschemefrom otheradministrativeexpendituresofSEWA­anditbenefitsfromimplicitoperationalcross-subsidies. CHAPTER-5: Social Security: Closing the Coverage Gap 11 candidate, to date this scheme has not been integrated aimed at rural landless laborers. In both cases, there is into the NPS. an important subsidy element (half of the premium in JBY and the entire premium in AABY) that is financed the experience of kerala ­ the state with the highest from a central government grant to the LIC. A major level of social security coverage through a variety problem with these schemes is that they are generally of welfare funds ­ is also instructive. There are an run as `unnamed' policies with little or no direct contact estimated 55 such schemes covering a wide variety of with the beneficiaries. In one state-level survey of the occupational groups ranging from headload workers to JBY where BPL workers were supposed to be covered, cashew nut farmers with different packages of benefits the results revealed an extremely low awareness of the and contribution levels. Some have memberships that entitlement to these benefits and claims have tended to number only a few thousand. The success in terms of be much lower than what would be expected. coverage (estimated to be 54 percent of informal sector workers) contrasts with the high level of administrative in health insurance, the main initiatives until recently costs which, in some cases are many times the benefits were those of microinsurance and mutuals. The early paid out to workers. Even the larger schemes such as success of the Karnataka's Yeshashvini scheme265 ­ a the Labor Welfare Fund and the Agricultural Workers mutual with state government financial support ­ is one Fund spent 61 and 44 percent of contributions on of the largest rural health insurance programs in the administration.263 world with around 2 million members, has led to great interest in other Indian states and beyond. It appears an important set of programs to provide life to owe much of its success to having been able to link insurance to unorganized sector workers are offered to a network of reputable hospitals around the state. by the government-owned life insurance corporation In addition, a number of microfinance institutions (lic). The older of the two biggest schemes is called have set up such schemes including SEWA in Gujarat Janashree Bima Yojana (JBY).264A list of 44 occupational and SKS in Karnataka.266 Box 5.1 outlines the main groups, chosen to target those living near the poverty models of unorganized sector worker health insurance line, determines the potential universe of eligible schemes operating in India presently. As mentioned workers. The scheme pays rupees 20,000 in the case of above however, the scalability of disparate group and natural death, 50,000 in the case of accidental death or NGO-sponsored policies is questionable given their permanent disability and 25,000 for partial permanent geographic concentration and limited coverage. disability. There is also a scholarship of 300 rupees per quarter per child paid to workers who send their children experiences with health insurance in rural india have (up to two) to grades 9-12 for a maximum of four years. run into much greater supply side constraints.267 The package is financed by a premium of 200 rupees Some of the recent experiences with national schemes collected through `nodal' agencies, i.e., groups that are outlined in Box 5.2. Acharya and Ranson (2005), must include at least 25 workers. A number of groups, for example, point out that the state of Gujarat has ranging from SHGs to relatively small occupational more than three times the number of hospitals per groups, have signed up with JBY acting as `nodal population and twice as many beds compared to the agencies' and reflecting a growing tendency to rely on average in India. In the same paper, the authors review the partner-agent model in this area. A second scheme, four community-based health insurance schemes in the Aam Aadma Bima Yojana (AABY) was introduced in the state. Even in this limited sample, the wide range October 2007 with similar life insurance coverage but of models and products is evident. Two of the schemes 263 Themembershipofthetwofundsin2001wasroughly500thousandand1million,respectivelyaccordingtoIrudayaRajan(2004).Kannan(2002)alsoindicts manyofKerala'swelfarefundssayingpointedlythat"...asystemthatwasestablishedtoservetheinterestsofworkers...hasultimatelyturnedouttobeanexercise inself-servingforthebureaucraticinterests." 264 ForadescriptionseeILO(2006). 265 SeeRothetal.(2006)foracasestudy.ItisworthnotingthattheIRDAhasexplicitlypromotedruralinsurancethroughregulatorymandatesthatrequireacertain shareofinsuranceportfoliostobesoldinruralareas. 266 ILO(2005)foundthatadozeninsurancecompanies­publicandprivate­nowoffer46singleriskproductsand37riskpackageproductsto`disadvantagedgroups' inIndia.Thereportcitesawiderangeofproductsincluding42forlife,9forhospitalization,5forcriticalillness,29foraccidentaldisabilityand2forpension. 267 SeeDuflo(2005)foranexampleinthecaseofthestateofRajasthan. 11 Social Protection for a Changing India: Volume II Box 5.1: main models of health insurance for unorganized workers in india There are a few basic models of community-based social insurance: (Devadasan et al., gives a useful typology and overview, noting around 20 community-based health insurance schemes. See also Ahuja and Narang (2005). the insurer-agent model, where the NGO/MFI or other founder acts as intermediary between members and the insurer, in India typically one of the public insurance companies (though SEWA has in recent years also involved private commercial insurer in its social security scheme). This model has been relied on by large organizations such as SEWA and Buldhana, and much smaller ones such as Navsarjan in Gujarat and BAIF in Maharashtra. (Acharya and Ranson, 2005). the founding organization acting as the direct insurer, but is not the provider of the insured services. This applies both to some NGO schemes (Yeshasvini in Karnataka; Dhan in TN) and more occupationally-based programs (e.g., Tribhuvandas Foundation). (Kuruvilla et al., 2005, for a detailed discussion of Yeshasvini, and also ILO, 2005b). the founding organization is the direct insurer and the main provider of the insured services. Examples include ACCORD in TN, Kasturba Hospital scheme in Maharashtra; Students' Health Home in West Bengal. operate with insurance company partners while two run approach, much less coordinated efforts at the public the schemes internally. Reimbursement levels vary from policy level.268 ` 2000 to 15,000 and the services covered (in-patient versus outpatient etc.) differ. Premium levels vary the rajiv aarogya sri scheme in andhra pradhesh is widely as does the level of implicit or explicit subsidy. an innovative case of large scale health insurance Finally, two of the four schemes operate computerized coverage expansion by a state government. Started in databases and two do not. In short, these and a series 2007, catastrophic health insurance for a limited range of other published case studies illustrate that many of hospital procedures are covered for all BPL workers in group schemes are being introduced to meet demand the state. The results have been impressive, with 800,000 but that there has been little convergence on the operations documented and a nominal coverage of Box 5.2: Recent failed attempts to expand insurance coverage to unorganized workers in india In just the last five years there have been a series of attempts to expand coverage that have failed to achieve their targets or have been abandoned. In 2001, the Government piloted a new LIC program ­ Krishi Shramik Samajik Suraksha ­ in 50 districts. The scheme was operated through LIC and covered life, survivor and pension insurance for agricultural workers, based on a contribution from the worker of 1 rupee per day, with a ` 2 per day contribution from GoI. It aimed to achieve coverage of 1 million agricultural workers within three years, though had reached around a quarter of its target by the time of closure three years later. According to NCEUS (2006), the failure was in large part due to the unfunded mandate on local governments and lack of financing for administration. In health, a major initiative was launched in 2003. The Universal Health Insurance Scheme (UHIS) is a voluntary contributory scheme for BPL households, covering medical costs of hospitalization, loss of income during short term illness, and death. There is a contribution subsidy from GoI ranging from ` 200-400, so that the net contribution from the contributor ranges from `165 to ` 300, depending on household size. In parallel, a scheme for unorganized non-BPL households was introduced. Outcomes on UHIS appear also to have been limited in the initial phases, with only around 400,000 households covered in the first year of operation (less than 5 percent of them BPL) and a further 31,000 households up to January 2005. In 2004, GoI introduced a social insurance scheme for unorganized sector workers (excluding agriculture), intended to be piloted in 50 districts nationwide and targeting around 2.5 million workers. The scheme was to be managed by EPFO in collaboration with ESIS for provision of health services. It was voluntary and contributory for those with monthly incomes under ` 6,500 and provides for old age, medical, and accident insurance. GoI would contribute around ` 250 annually per worker. Premia for workers were ` 50 per month for workers to 35 years and ` 100 above that. Employers were meant to contribute a further ` 100 monthly. Workers cover the employer contribution themselves in addition to their basic contribution when no employer is identified. The scheme had negligible penetration, with an estimated enrollment in mid-2005 of less than 10,000. 268 Casestudiesinmicro-healthinsuranceexperiencesinIndiaincludeAcharyaandRanson(2005),Garand(2005),Radermarcheretal.(2005).Whilethecasestudies areusefulinthemselves,thediversityofthepackagesofferedbytheseschemesmakesthemdifficulttocompare.Theanalysistodatehasnotledtoresultsthat couldbeextrapolatedfornationalpolicyoranattempttoscaleuptheseprograms.Toourknowledgetherehavenotbeenpublishedcasestudiesofcontributory oldagepensionschemesorlifeinsuranceexceptdescriptiveaccounts(e.g.,thedescriptionoftheLICJBYschemeinILO,2005a). CHAPTER-5: Social Security: Closing the Coverage Gap 11 more than three-quarters of the population.269 The coordination or harmonization of schemes leads to scheme owes its success to direct leadership by the higher costs via duplication of administrative structures. State Chief Minister who has devoted significant Moreover, these schemes lack portability, an especially human and financial resources to the implementation important feature in the context of rural to urban of the scheme as well as strong incentives for one of the migration and an evolving labor market. country's most expansive hospital networks. this brief review reveals an uncoordinated set of insurance and pension schemes aimed at expanding C. ReCenT iniTiATiveS To coverage but lacking an overarching policy exPAnD SoCiAl SeCuRiTy framework. Table 5.4 presents a range of programs, by no means exhaustive, of programs aimed at the CoveRAge unorganized sector already. The total coverage of social security coverage has been low for decades in these schemes is not likely to exceed ten percent of the india, due largely to the more general phenomenon unorganized labor force for any given type of insurance, of informality, the causes of which are beyond the although presumably, it is growing. While some have scope of this report. What is clear, however, is that been studied, there is a surprising dearth of information any large scale social security system will have to cope on what factors lead to success or could be scaled up with the reality of the unorganized sector for decades under the right circumstances. Several things do seem to come. Recognizing this, the Government of India has clear however. First, demand exists. Second, lack of introduced a number of important initiatives that have Table 5.4: Selected insurance and pension programs of the unorganized sector estimated membership Type of risk name of program (000s)** covered Government Universal Health Insurance 1000 Health Central Welfare funds (5)* 4500 (est.) Health, education, housing, other Kerala Welfare funds (55) 4900 Varies, all kinds Karnataka Labor Welfare fund 675 Life, health Andhra Pradesh Labor Welfare fund 1000 Life Tamil Nadu construction worker welfare fund 631 Life, health, pension, other Tamil Nadu Voluntary Health Service 125 Health Maharastra Mahadi workers fund 150 Life, health, other West Bengal Provident Funds (2)*** 1800 Old age Non- LIC ­ JBY scheme 3570 Life governmental**** UTI ­ pension scheme 100 Old age Yashivini ­ Karnataka 2100 Health Karuna Trust ­ Karnataka 14 Health Spandana 386 Life Shepherd 15 Life SEWA Gujarat 100 Life, health, pension, other People's Rural Health Promotion Scheme 75 ASA 66 Life and Health * includes beedie, mica, limestone, iron ore and cine workers funds. ** memberships are double counted in some cases since both groups and providers are shown here. *** includes both urban and rural provident funds. **** list is not exhaustive and excludes, among others, life and health insurance for ICICI clients. Sources: Irudaya Rajan (2004); Roth et al., (2005); LIC 2006; NCEUS 2006. 269 Thisisduetotwofactors.First,inAPtheeligibilityforBPLstatusarelaxresultinginaveryhighproportionofthepopulationfallingintothiscategory.Second,BPL householdsareautomaticallyeligible,i.e.,thereisnoenrolmentprocess. 11 Social Protection for a Changing India: Volume II broad political support. In October 2007, Prime Minister premium set and a public insurance company was paid Manmohan Singh announced three major initiatives a lump sum amount based on the estimated number of aiming to address the coverage gap. These included a beneficiaries. Implementation details were mostly left to new life insurance scheme for landless rural laborers, state governments. This was an `unnamed' policy in that an expanded social pension scheme for the elderly, and there was no direct contact with beneficiaries and no list a new health insurance scheme for households living of policyholders. Not surprisingly, claims have been low. below the poverty line (BPL). This section reviews these Estimates suggest that only one quarter of potential initiatives with a special emphasis on the most promising claims were ever made. of the programs, the RSBY health insurance scheme. A very different and much more promising approach was the social pension scheme, a traditional cash taken with regard to the new health insurance program, transfer scheme, already existed and is discussed known as Rastriya Swasthya Bima Yojana271 or RSBY. The in detail in chapter 3.270 It is administered by state following section describes the RSBY, its design and the governments which apply their own eligibility criteria early experience of implementation. and partly financed by the central government based on an estimate of the number of poor elderly persons in each state. The 2007 initiative modified this formula (a) key parameters and operational thereby increasing the overall central budget allocation design of RSBy for the program. As of early 2010, 18 state governments were making the full matching contribution and taking the objective of rsby is to protect below poverty line advantage of this additional allocation. Eleven states (bpl) households from major health shocks that were contributing less than 200 rupees. involve hospitalization. This protection aims at making treatment more affordable so as to increase the amount the life and health insurance programs were new of health care that a household can purchase but also to programs, although there had been previous attempts limit the disastrous financial impact that households often (see box 5.2) and some parallel programs already experience through out of pocket spending.272 RSBY is a existed. In the case of the life insurance scheme, the classic demand side intervention using a sophisticated AABY followed the standard top-down approach used version of a voucher combined with targeting. Table 5.5 for other schemes: the benefit package was defined, the shows the main parameters of the program. Table 5.5: key parameters of RSBy Parameter Description Additional comments/caveats Benefits covered Cost of hospitalization for 700+ procedures at empanelled hospitals up Pre-existing conditions are covered; to 30,000 rupees per annum per household plus 100 rupees transport minimal exclusions; day surgeries covered; cost per visit up to 1000 rupees. Eligibility criteria Must be on the official state BPL list; limited to five members of the All enrolled members must be present to household including household head, spouse and three dependents be enrolled; infants are covered through mother Premium and fees 30 rupee registration fee per household per annum paid by household; Average premium for active districts is Per household premium payment determined through competitive around 580 rupees bidding process; Policy period One year starting the month after first enrolment in a particular district Enrolment period is four months Financing 75%/25% Government of India/State Government The ratio is 90%/10% in Northeast states and Jammu and Kashmir Sources: www.rsby.gov.in 270 TheNationalOldAgePensionSchemewasrenamedtheIndiraGandhiNationalOldAgePensionSchemein2007. 271 TheEnglishtranslationoftheHindititleis"NationalHealthInsuranceScheme". 272 AccordingtotheWHO,Indiahasoneofthehighestratesofoutofpockethealthspendingintheworldat78percentoftotalhealthspendingand94percentof privatehealthspending.Datafromthe60thNationalSampleSurveyshowthatamongthebottomtwoquintilesoftheruralpopulation,47percentwereforcedto borrowtofinancehospitalization. CHAPTER-5: Social Security: Closing the Coverage Gap 11 the benefits of the scheme are delivered through particular district can use his or her smart card at any a contractual arrangement between insurance RSBY empanelled hospital in the country. companies and state governments. The process is clearly the operation of the system involves three stages ­ spelled out273: State governments sign a memorandum of enrolment, hospital transactions and monitoring, understanding with the central government, advertise, discussed in turn below: set up technical and financial review committees to assess tender documents, select an insurance company Enrolment. State governments must provide an and sign the contract. The current policy period is one electronic list of eligible BPL households to the year, though it is under consideration whether a more insurer in a pre-specified format. The list should extended period for each tender would be desirable. be posted in each village prior to the enrollment and the date and location is publicized in advance. the insurer must be licensed to provide health Mobile stations are set up at local centers and are insurance by the insurance regulatory development equipped with the hardware required to collect authority (irda) and must meet several other minimum biometric information (fingerprints) of the criteria. The insurer must agree to cover the benefit members of the household and to print smart package prescribed by GoI through a cashless facility cards with a photo. The smart card, along with that in turn requires the use of smart cards which must an information pamphlet describing the scheme be issued to all members.274 In practice, this makes it and the list of hospitals, should be provided on necessary for the insurer to contract a qualified smart the spot once the beneficiary has paid the fee of card provider. 30 rupees. the insurer must also agree to engage intermediaries Aside from the beneficiaries themselves, the with local presence such as ngos, mfis, etc. in order insurance company's representative or contractor to provide grassroots outreach and assist members must be present. A district-level, state government in utilizing the services after enrolment. It must officer, known as a Field Key Officer or FKO must also provide a list of empanelled hospitals that will also be present and must insert his or her own, participate in the cashless arrangement. These hospitals centrally-issued, smart card to verify the legitimacy must conform to certain basic minimum requirements of the enrolment. (In this way, each enrollee can be (e.g., size and registration) and must agree to set up a tracked to a particular state government official). special RSBY desk with smart card reader and trained At the end of the day of enrolment, the list of staff. The list should normally include both public and households issued smart cards is downloaded private hospitals. from the FKOs card and centralized at the district level and eventually state level. This is the basis the financial bid is essentially an annual premium for the payment of the premium. per enrolled household which is determined Transactionsathospitals.The smart card entitles in a competitive bidding process. The insurer is compensated on the basis of the number of verified its bearer to a list of pre-specified in-patient smart cards issued, i.e., households covered. Each services in the month following enrollment. contract is specified on the basis of an individual So, for example, if enrolment in a particular district in a state and the insurer agrees to set up an district begins in February, the card can be used office in each district where it operates. While more from March 1 until the end of February of the than one insurer can operate in a particular state, only following year. one insurer can operate in a single district at any given The transaction process begins when the point in time. The hardware and software specifications member visits the hospital and her card is laid down by GoI imply inter-operability across districts swiped. If a diagnosis leads to a procedure, the and states. In other words, an RSBY member from one appropriate prescribed package is selected in 273 TheprocessflowdocumentcanbefoundontheRSBYwebsite.Italsolaysoutprocesseswithregardtootherdetailssuchasreplacementoflostordamagedsmart cardsandothercontingencies.Thesamesitecontainsdocumentationonthedifferentstepsinvolvedinimplementingthescheme. 274 ThesmartcardsarenominallypaidforbyandbelongtotheGoI. 11 Social Protection for a Changing India: Volume II the software menu and the procedure is blocked. beneficiary, for example, a number of design choices When the patient leaves the hospital, the card is have made the system attractive and easy to use. For again swiped and the pre-specified cost of the example, in contrast with most government schemes, procedure is deducted from the 30,000 rupee households are not generally provided choice as to total on the card. The transaction is recorded on who provides services, especially with regard to private the hospital's computer and on the card itself. A sector options. In addition, the `front end' of the system receipt is printed and provided to the member. is user-friendly. Enrolment is brought to the household's Transport costs are to be paid by the hospital to location even in remote villages. The thirty rupee cost the patient in cash. is nominal and only one visit is generally required to receive a working smart card. There are no complicated Monitoring. Data on enrolment is downloaded exclusions and pre-existing conditions are covered from the FKO cards and aggregated for each and the only eligibility condition is to be on the BPL district. Insurers are also required to submit data list. As far as hospitalization, the smart card allows for to the GoI in a pre-specified format on regular paperless, cashless transactions. The last point is crucial basis. Payments of premium to the insurer are since raising a large amount of money and then waiting based on the FKO data. Transaction data is also for claims to be reimbursed may simply not be possible submitted by the insurer which receives data from for many BPL households. each hospital. A separate set of pre-formatted tables are generated for the insurer and for the an important element of the design is the direct government respectively. This allows the insurer interface required between the insurer and the to track and settle claims. It also helps the insurer policyholder. Experience with other schemes in as well as the government to identify anomalies India with `unnamed' policies often result in low and patterns that may lead to on-site audits. The claims because individuals are not aware that they are state governments are able to monitor utilization covered. In this case, the need to obtain the household's of the program by members and to some extent, biometric information forces the insurer to make direct begin to measure the impact of the program. contact. In addition, the 30 rupee payment cannot be Periodic reports are to be made publicly available made on behalf of the beneficiary. This not only makes on the internet and through published reports. it necessary to have direct contact, it is also thought to provide some level of ownership, despite the small amount involved. (b) Rationale behind the RSBy design another important feature of the scheme is the design of the rsby is based on at least three key interoperability. By standardizing the specifications premises: for software and hardware, the smart cards issued by a Program design must address the practical particular insurer in a particular district can be used in situation on the ground from the perspective of any RSBY hospital throughout the country. In contrast, the beneficiary household. other schemes targeted to BPL families are state- specific and benefits are not portable across state lines The design details must be allowed to adapt or between rural and urban areas in the same state. This over time through "course corrections" based on is especially problematic for some of the poorest states evidence. This, in turn makes it necessary to have such as Bihar where a significant proportion of the timely data on key indicators and concurrent workforce is migrant labor. evaluation. ultimately, the true indication as to whether the Sustainability of the program depends on beneficiaries can effectively participate in the program incentives of the key players being aligned with will be reenrollment and utilization rates. At present, the objectives of the program. there are no data available on reenrollment rates, but these premises are reflected in everything from one post-enrolment survey data found that all RSBY the use of technology to the approach taken to members intended to enroll again. The second year contracting insurers. From the perspective of the of enrolment is about just starting in several districts CHAPTER-5: Social Security: Closing the Coverage Gap 11 and should be available by late 2009. Although it is members as a potentially major source of revenues. In premature to base any conclusions on early utilization principle, this can be the case even for public hospitals data, the incidence of hospitalization will be crucial which can create societies that allow them to retain toward assessing performance. a share of the revenues that they collect. Insurers, in contrast, have an incentive to monitor participating the approach to monitoring rsby is different than hospitals in order to avoid excessive claims through that applied to most centrally-sponsored social fraudulent or unnecessary procedures. programs, both in terms of intensity and focus. Monthly and sometimes weekly data on enrolment since the contracting process is repeated over and utilization flows from insurers to state and central time, insurers would have additional incentive government. Reports based on these submissions will to meet other performance criteria that will be soon be available on a public website with information considered in future bidding rounds. This behavior available at a highly disaggregated level. Types of can be reinforced by proactive state governments that medical procedures can be tracked by hospital and data use the information gathered through monitoring in are available down to the village level. Administrative subsequent tender processes. For example, surveys data is an important element of program monitoring. In of RSBY hospitals and post-hospitalization surveys addition, continuous feedback is collected through field of beneficiaries can be used to monitor quality and visits and spot audits, although levels of state government capacity of facilities over time. Other examples include engagement vary widely. Finally, with technical support adherence to guidelines for the enrolment process, from the World Bank and GTZ, GoI is also planning a the quality and timeliness of data submission and the series of surveys that will capture important patterns alacrity of claims payments to hospitals. not visible through the administrative data. to summarize, the rsby program has been designed to respond to the `ground realities' from the perspective perhaps the most important element of the rsby of bpl households, allow for evidence-based approach is the attention given to aligning the adaptation of program details and, most importantly, incentives of the stakeholders. The business model align incentives for stakeholders to ensure applied relies on the different actors to pursue their sustainability. The early implementation experience interests in a way that contributes to the sustainability has, to a large extent, validated this approach. At the of the program. same time, a number of problems were encountered the main players are the central and state and the program has not yet stabilized. governments, the insurance companies, the hospitals and the beneficiaries. As mentioned above, there is strong political support at all levels for the extension of (c) early implementation experience social security coverage. For state governments, the fact there was great uncertainty during the early months that the central budget covers three-fourths of the cost of the program. What would the insurers bid for the of the program is a significant incentive to participate. premium and would there be sufficient competition? Although some state governments considered the Would the smart card technology that the program relied implementation requirements to be onerous, the fact upon work as it should under difficult field conditions? that key processes had been worked out by the GoI Would state governments implement the scheme as helped make this more palatable. A vigorous campaign intended? What other problems might arise? to explain the benefits of the scheme to each state government also proved crucial.275 the level of the premium was especially important since the goi had set a cap on the subsidy. As it turned insurance companies have an incentive to maximize out, the winning bids in the first round were below enrolment since their market-determined premium this cap. In general, the software and hardware did is paid on the basis of the number of households operate as planned, although there were glitches such enrolled. Hospitals have an incentive to attract RSBY as the limited supply of smart card printers required for 275 SeeSwarup(2009)foradetaileddiscussion. 120 Social Protection for a Changing India: Volume II enrolment and delays in procuring enough smart cards have been able to enroll. On many occasions, this led to that met the required specifications. Another example disruption of the enrolment process and even violence. was the difficulty of printing smart cards in the extreme The problems encountered have led at least one state to summer heat of certain districts in Rajasthan. compile a new BPL list.277 some state governments were proactive and devoted despite the problems with the bpl list and other minor adequate human resources to the various tasks while problems, the first round of enrolment has either others were quite passive. There is some evidence that started or been completed in more than one hundred these differences were reflected in outcomes. Enrolment districts. Total enrolment as of June 2010 was reported rates tended to be higher in states where steps were at close to 17 million households or around 50 million taken to ensure the presence of district level officers individuals (See Figure 5.2).278 known as Field Key Officers (FKOs) ­ a requirement for enrolment software to operate ­ and in states obviously, enrollment is not the ultimate objective where there was good logistical coordination between of the scheme. The real objectives of providing financial government and insurers. protection for households affected by major health shocks and improving health outcomes can only be perhaps the most problematic aspect of the enrolment achieved if the enrolled avail of the services covered. process was the poor state of the bpl database. This was Early data on hospital claims is mixed and must be the list of potential beneficiaries of the program based interpreted carefully. According to data from the 60th on a 2002 census that collected information allowing round of the National Sample Survey, the incidence of state governments to determine whether a household hospitalization is around 2 percent of the population. would be considered BPL. The test that was applied in However, it the same data also show that there is this regard has been criticized and empirical evidence significant variation across states and, as would be has revealed large inclusion and exclusion errors (see expected, that the poor have lower utilization rates. It is Chapter 8).276 This classification is also used for targeting important therefore, to take these and other factors into benefits of a variety of social programs besides RSBY, account when comparing utilization under RSBY.279 but had never been subjected to the scrutiny required by the RSBY enrolment process. Errors were common it is too early to determine the impact of the scheme on and were caused by several factors including poor data utilization rates, but the preliminary evidence suggests entry, lost records and the lack of updates for deaths, that it is increasing among the insured. There are large births, marriages and migration. variations across districts and even villages within the same district and most districts have not completed a full reports from the field suggested that the flaws in policy period. In the districts that have at least one year the bpl list led to resentment at the village level of experience with transactions, utilization rates vary and also reduced enrolment rates significantly. widely. In the two districts where data were available The latter effect was the result of the need for rigorous for a full year, the rate of hospitalization among RSBY biometric identification based on the BPL list which members reached around 2 percent. could not legally be corrected in the field. It was also necessary for the insurer to have a clearly defined list of while these figures are slightly higher than those potential enrollees authorized by the state government reported by the national sample survey for the lower and approved by the GoI. In short, if the list did not quintiles of the income distribution, the interpretation include a BPL household, or more frequently, excluded of these initial numbers is not straightforward. To certain members, those people could not be enrolled. begin with, the monthly figures show that there is often For example, in one district half of the households very low utilization in the first months of the scheme. did not have any dependents listed. This effectively This could be due to inexperience of RSBY members excluded close to half a million individuals who should on the demand side and lack of preparation on the 276 SeeMurgaiandJalan(2008);andMehrotraandMander(2009). 277 ThestateofKeralaimplementedanewBPLcensusinMay2009. 278 ThesefiguresarebasedonreportsfrominsurancecompaniesandnottheinterimMIS.SeeSun(2010),foradetaileddiscussion. 279 HouandPalacios(2010)findpreliminaryevidencethatutilizationratesarehigherthancomparableNSSfigures,withseveralcaveats. CHAPTER-5: Social Security: Closing the Coverage Gap 121 Figure 5.2: Household enrolment in RSBY by month, February 2008 ­ August 2009 20000 18000 16000 14000 Thousands of households enrolled 12000 10000 8000 6000 4000 2000 0 Feb-08 Jun-08 Oct-08 Feb-09 Jun-09 Oct-09 Feb-10 Jun-10 Source: Administrative data from MOLE database. See www.rsby.gov.in for state level figures. hospital side. Second, modifications of the scheme that in sum, the rsby has demonstrated that it can have been made during this period should increase reach the target population and attract state utilization. These include a shorter enrolment period governments, insurers and hospitals to actively (from six to four months) and the addition of maternity participate in the program. Whether the program benefits to the list of procedures covered. Finally, some succeeds in the longer run however, depends on state governments are now becoming proactive by whether the millions that have enrolled are able to avail sponsoring health camps or coordinating with hospitals of the potential benefits in terms of higher utilization to reach out to RSBY members. and lower out of pocket costs. utilization alone will not deliver positive results unless hospital quality meets certain standards. This is a medium term challenge for state governments (d) implications beyond the RSBy and the GoI made more difficult by the dearth of if the rsby achieves the scale envisioned by the information on hospital quality and lack of experience goi, there would be important implications beyond in this area.280 Over time, a database that can be used the bpl population and even beyond health to monitor progress could be developed and state insurance. The RSBY approach can be used to cover governments could increasingly take the evidence workers beyond the BPL list. In several states non- produced into account in the bidding process or insist BPL households that are the target of government on certain thresholds for empanelment. To the extent programs such as construction workers, are receiving that RSBY generates a significant revenue stream for the same package as RSBY members. The 2010 both hospitals and insurers, there should be some budget reflected the decision to expand coverage leverage to promote better quality. to NREGA workers. The existence of processes and 280 Onlyrecentlyhastherebeenanefforttobuildahospitalaccreditationprocess. 122 Social Protection for a Changing India: Volume II technological infrastructure related to the smart the number of `ghost pensioners'. Similar applications card reduce the marginal cost of implementation could be used in other programs such as conditional to a level that makes these non-BPL group schemes cash transfer programs that require verification at the feasible. Given the known exclusion errors of the BPL provider end (including perhaps, private providers such list, providing coverage to non-BPL households is an as schools). For certain programs, the platform could important step.281 allow for portability of benefits for migrants who are otherwise tied to the place where they appear on the BPL the approach and platform of rsby may also be useful list until the next census. This feature is likely to become in delivering other social programs. An obvious more important in the face of massive anticipated candidate is the Public Distribution System (PDS) which urbanization in the next few decades. is one of India's largest and oldest schemes, but also one of the most corrupt. In principle, poor households are finally, it is important to note that there are efficiency entitled to essential commodities such as subsidized gains to be attained in converging on a single rice, grain, sugar and kerosene. Massive leakages have platform rather than duplication administrative been document with as much as two-thirds of the rice structures for the large number of social programs and grain never reaching the beneficiaries. There have that india extends to poor households. On the other been several proposals to reform the scheme including hand, trying to stack too many programs and functions shifting the huge budget for the scheme from in-kind on the RSBY platform itself could be counterproductive. to cash transfers. However, this proposal would run Piloting such ideas carefully rather than forcing the up against significant resistance from the bureaucracy issue too quickly seems advisable.282 and 400 thousand shop-owners that distribute these commodities. the rsby may be the first social sector program in india that simultaneously takes into account using the rsby approach in pds could lead to the perspective of the poor, focuses on getting the improvements through two channels ­ competition incentives of the various players that have to deliver and accountability. This could be achieved by the benefits right and encourages changes over time adapting the smart card technology used in the RSBY based on evidence. The experience of the first year or so program to serve an analogous function for PDS (see has been positive, particularly in terms of demonstrating Chapter 3). As in RSBY, smart cards would contain that the target population can be reached (despite the biometric information of the household in question poor quality of the BPL list). However, the real success in order to identify beneficiaries and would swipe the of the scheme will come only when RSBY members are card at a reader in the ration shop. The card would be shown to have increased access and affordability of debited each time as per the price and amount of the decent hospital care. This outcome depends, in turn, on commodity and data would be uploaded to a central central and state governments ensuring that insurers server. Instead of an insurer, the PDS distributor would perform and that obstacles are removed from the path be compensated by the relevant government agency of poor households that need hospitalization. It will be based on the uploaded records. These cards would especially important in the next few years to monitor work at any shop in the country, so beneficiaries of the progress in this regard through MIS data, surveys and program would have choice of provider, just as in the field reports. The oversight of a complex program like case of RSBY hospitals. RSBY with tens of millions of members, two dozen in other cases, the identification process embedded in state governments, a dozen insurers and thousands of the smart card could simply be used to verify that hospitals will require a strong institutional presence at a particular individual had received the intended the center, probably in the form of a stand-alone agency. benefit. For example, pension receipt under the NOAPS The faster this institutional capacity is developed, the could be biometrically verified periodically, reducing more likely that the RSBY will succeed. 281 Inatleastonestate,Kerala,thegovernmentisextendingtheschemetoallhouseholdswhileonlysubsidizingthoseontheBPLlist. 282 Inthecontextofthe2009electioncampaign,onestategovernmentattemptedtointroduceanewcashtransferprogramontoasmartcardalongwithaseparate healthinsuranceschemeafternothavingreceivedpermissiontodosounderRSBY.Theeffortwaseventuallyabandoned. CHAPTER-5: Social Security: Closing the Coverage Gap 12 in contrast, the nascent effort to expand pension D. exPAnDing SoCiAl coverage requires more thinking. Take up by SeCuRiTy CoveRAge in unorganized sector workers is not likely to reach the The Coming DeCADe magnitudes achieved by RSBY unless at least two of the latter programs distinctive features can be emulated. this section explores the options for india as it seeks First, workers must be provided with direct incentives to expand social security coverage. It looks first as to give up their liquidity and save for old age. Since current proposals for expansion, before examining tax incentives reach very few people, only matching the range of factors that are relevant in determining contributions can provide such an incentive. Second, a feasible expansion strategy. These include demand incentives for providers (in this case pension fund side issues, various aspects of program administration, managers) to attract clients is required. Group schemes and design elements of different types of social that already exist and state sponsored schemes could insurance. usefully be integrated with the NPS while district level contracting of identified and targeted households The RSBY scheme aims to cover sixty million households. could be attempted. In sum, while the NPS concept For a variety of reasons discussed above, this is is sensible, its implementation to date has not taken unlikely to happen in the five year period originally the incentives to beneficiaries and providers fully into envisioned. Nevertheless, at the current rate of account. increase, tens of millions of people that have never had health insurance will be covered in the next few In the case of workers whose risks could effectively be years, a remarkable achievement. As the program is insured, a target level of consumption smoothing could extended beyond the BPL population, its reach could be be defined, perhaps in relation to the poverty line. even greater. The RSBY scheme is likely to be the best Through actuarial analysis, the cost of this insurance opportunity for expanding social security coverage in could be calculated and revised according to actual the next decade. experience over time. Box 5.3 presents the example Box 5.3: illustration of a contribution based old age pension scheme The first step in designing a scheme is to set the benefit 12% target and eligibility conditions. For example, an inflation indexed annuity of 6,000 rupees per year would be higher 10% than the inflation adjusted national rural poverty line and slightly lower than the urban equivalent. In this example, 8% % of income we set the eligibility age for receipt of benefit at 60. (At 500 per month, the target benefit is 2.5 times the current 6% social pension paid by NOAPS and the eligibility age is five years lower.) 4% The contribution rate required to generate this pension level depends on the rate of return on the invested funds after charges, 2% the number of years of accumulation and life expectancy. Adapting the calculations of Shah (2005) for the case of a worker 0% contributing from age 24 to age 60 and investing conservatively, 3 4 5 6 7 8 a contribution of roughly 5 rupees per day or 1825 per annum decile would result in a high probability that the DC scheme would at least reach our lower target benefit. This calculation is based on historical returns for government bonds. A diversified portfolio that included other assets such as equities could yield a higher pension, but with correspondingly higher risk. (A maximum age of joining and vesting requirements could be incorporated.) The final step in the design is to determine the incentives to participate in the scheme as well as its affordability. The figure below shows the share of income that the required contribution would entail by earnings decile were the worker to make the entire contribution and where there was a 1:1 matching contribution. The share without the match reaches levels likely to discourage participation in the lower deciles but with a match, the range is between 2-5 percent. Combined with the high rate of return that the match implies, it seems feasible that the scheme could attract participation along with other factors, such as transaction costs and institutional trust. 12 Social Protection for a Changing India: Volume II of a defined contribution (DC) pension scheme that existing schemes are contributory, although in most targets a pension level of about 500 rupees per month cases, there is a direct or indirect subsidy to cover actual at age 60. In this example, the premium works out to expenses. It is also important to note that most of the about 5 rupees per day for a worker that contributes initiatives under way around India today involve groups regularly from the age of 26, but this assumption is only that are organized either on an occupational basis or on illustrative. The principle, however, is that the benefit some other basis such as MFIs, SHGs or cooperatives. target determines the stream of contributions that are In addition to acting as a credible intermediary with need to preclude any unfunded liability from arising as insurers, these organizations can keep transaction well as to ensure that the target benefit level is achieved costs relatively low.284 In short, the available evidence with a high degree of probability. A defined contribution suggests that individuals participating in groups and arrangement is also able to cope with variable incomes having some minimal willingness to pay contributions and the absence of a reference wage. It is also, at least are the best candidates for inclusion in any new in principle, portable since the balance could be shifted contributory schemes. into an EPFO account or even into the new DC scheme realistically however, group schemes are not likely that applies to new civil servants or vice versa. to reach the scale possible through state and an important design principle is that the premium central government programs. The RSBY and NPS are or contribution level should be affordable to a the two most important central government programs significant share of the target population in the for expanding social security coverage at the moment. unorganized sector. In the DC scheme example in In terms of reaching unorganized sector workers At the Box 5.3, the contribution level would represent between same time, Indian states are larger than most countries, 3 and 11 percent of the earnings of unorganized sector and several have undertaken important initiatives in this workers in deciles 3 to 8, according to the survey data. area. In health insurance, the Aarogyi scheme in Andhra In order to attract voluntary take-up and participation, Pradhesh has by far the largest coverage with the older some matching contribution or premia would be Yeshaswini scheme in Karnataka coming in a distant required. A 1:1 matching contribution would halve those second. Andhra Pradesh introduced its co-contribution percentages and simultaneously provide a significant pension scheme for ten million self-help group members incentive for voluntary participation in the form of an in 2009. implied 100 percent rate of return before investment clearly though, some households will not be able income is taken into account. Note that in each case, to afford anything beyond a nominal contribution the eligibility rules as well as the target benefit and and neither contributory insurance nor pensions is incentive levels would be set in relation to the `social relevant. For example, the recently released summary pension' parameters in such a way that did discourage participation. For example, the potential accumulation of the NSS 61st round consumption survey shows in the DC scheme would be taken into consideration that ten percent of rural households spend less than when determining the age at which an individual could 270 rupees per month on a per capita basis. Out of this participate in the contributory scheme as well as the meager sum, more than 90 percent is spent on food contribution amount. and other basic necessities. Furthermore, insurance for those already widowed or disabled or contributory available data, various studies and real world pensions for the elderly (or soon to be elderly) are experiences show however, that a large portion of irrelevant. For these households, well targeted transfers the unorganized sector in india is capable and willing are needed simply to survive from one day to the next. to contribute to their own social security.283 Many Improved targeting of safety net programs285 discussed 283Madheswaranetal.(2005a)findthatbetween67-75percentofunorganizedsectorworkerssurveyedwerewillingtocontributeforvarioustypesofinsurance. Onaverage,agriculturalworkerswerewillingtomakeamonthlypaymentof624,607and696rupeesforoldage,deathandsickness,respectively.Two- thirdsofthoseunwillingtocontributecitedpovertyasthemainreason.Dror(2006)foundthathouseholdswerewillingtopay,onaverage,559rupees peryearforhealthinsurance.Intheareaofpensions,`micropension'schemeshaveemergedsuchastheoneoperatedbySEWAortheforworkersindairy cooperativesinGujarat. 284 Ahuja(2005)makesthispointinthecontextofhealthinsuranceprovision. 285 Forexample,theNSAPprogramthat,inadditiontooldageandwidows,paysdeathandmaternitybenefitstopoorhouseholds. CHAPTER-5: Social Security: Closing the Coverage Gap 12 in Chapter 8 and expansion of coverage would help the individual member level was already taking achieve this primary objective. place. This is particularly important in the early stages of any national initiative since achieving a the other criteria relate to implementation; specifically, minimum scale and using existing recordkeeping efficient recordkeeping, asset management, insurance infrastructure will help keep administrative cost provision. A key policy decision in all three areas is the ratios low during the start up period. role of government versus private sector providers. One model would involve the creation of a new bureaucracy a second advantage of this model would be the that would administer the new schemes and perform all promotion of competition on the basis of cost three functions. This is what is done already in the case and quality of services. of pensions and life insurance for formal sector workers third, many unorganized sector workers have in the form of the EPFO. One potential advantage of such no experience or direct interaction with formal an entity would be economies of scale in administration. financial sector institutions so that groups can However, the experience of parastatal monopolies such serve as an effective intermediary.286 as the EPFO in Asia as well as other parts of the world is perhaps the greatest challenge for either approach, that they are not adept at managing assets or avoiding however, will be that of tracking participants and the creation of large unfunded liabilities that often lead the financial flows associated with each of them. to financial insolvency. Service standards and lack of Recordkeeping must be accurate, timely and reliable technological innovation in the absence of competition if the scheme is to gain the trust of participants are often cited as problems in these institutions. who can opt out of the scheme at any time (unlike Applying the current ratio of staff to active contributors the mandated schemes such as EPFO). Even a small in the EPFO to a new scheme for the unorganized sector number of errors that result in financial losses early with participation of say one of every five workers, would in the life of the scheme could irreparably damage the result in the expansion of the civil service by around image of the program. Recordkeeping clearly does 200,000 employees. involve large economies of scale and this has led some the alternative to direct public provision and countries to centralize this function while maintaining administration is to use existing non-governmental competitive and decentralized provision of other entities and restrict the role of government to: services.287 (i) providing targeted subsidies and (ii) regulating for several reasons, the challenge of tracking these entities and setting basic standards. This model flows of money and individuals will be particularly already exists in India in several forms already mentioned, daunting in india. The lack of a unique and universal with RSBY as the best example. In addition, many other national identification (ID) so far has been a challenge schemes falling into the community based or micro- for many public programs. Several existing (ID) systems insurance category could be incorporated under an operate in a disconnect, by nature of their mandate umbrella program that provided matching contributions provide only for a limited coverage, and would often or premia but set certain standards in terms of benefit not meet requirements sought for a modern public targets, eligibility conditions, investment policy and benefit program. For example, tax registration covers recordkeeping, among others. only income tax payers. Registration systems of there are several potential advantages to this various welfare funds are believed to be operationally approach: outdated. first, by harnessing existing groups including the scenario could improve dramatically in light SHGs, coops, MFIs, etc., transaction costs could of the high priority that has been given by the new be kept low, especially where recordkeeping at administration to implementation of the national, 286 InonestudyforKarnataka,morethansixtypercentofunorganizedsectorworkerschosetosavethroughSHGscomparedtoaround23percentinbanksand postoffices. 287 ExamplesintheareaofpensionsincludeMexicoandSweden.ItisalsoworthnotingthatthePFRDAActalsoenvisionedacentralizedrecordkeepingagencywhile allowingassetmanagementandannuityprovisiontobecontractedouttotheprivatesector. 12 Social Protection for a Changing India: Volume II biometric id.288 A unique ID would facilitate identification example, it seems that the infrastructure of the rural for a wide variety of social programs. Some universal Internet kiosks can provide a backbone for some public requirements could be introduced in respect to the data benefit programs. At the same time, in places where the collection as part of the registration process to ensure on-line connectivity is still beyond the reach, the smart minimal information requirements and compatibility of card technology can facilitate efficient off-line record member's data across agencies. management. computerized record management makes any operations of the cash collection and dispersed program more transparent and flexible. Still today systems in the unorganized sector pose serious many public programs in India rely on archaic systems challenges. Conventional financial intermediaries are of paper records that are generally very limited in often ill prepared to deal with mass transactions of their capacity to provide for, e.g., data aggregation very small amounts in remote, dispersed, and poorly and analysis, duplicate records checks, and effective educated communities. Penetration of commercial coverage assessment. Furthermore, cross-eligibility banks in the rural areas in India remains low. While checks for multiple programs may virtually be the India's extensive Postal system seems operates in impossible. Investment in computerized administration almost every corner of the country, the adequacy of its and processing systems would be necessary for the accounting mechanisms and capacity to assume new success of any large scale program. Still, with a highly tasks remains to be evaluated. In generally, operational sophisticated computerized core of operation, some subsidies, partnerships with local intermediaries, and elements of administration will probably continue to various innovations ­ all will help the effort of coverage operate with paper records (for example, management extension. of signature sheets for benefit disbursement). But both in india and abroad, a number of interesting even this is not a dogma anymore ­ the boundaries of and relevant innovations have been adapted to the technology today are being tested by introducing, e.g., needs of both public and private sector in extending paperless biometrics-based identification systems. the coverage of financial services. Low cost ATMs programs with systematic personalized record- developed at a fraction of cost of the conventional keeping provide much greater capacity for adding machines, equipped with finger print identification and new benefits and improving existing schemes. Examples long-life batteries facilitate cash distribution in remote of the schemes that would often not have updated areas (some interesting and quite competitive prototypes membership records include various micro-credit have been developed in India). Point-of-service (POS) programs (operating on a group basis) and conventional devices installed in local shops and gas stations can social security programs where employer is responsible use regular telephone lines to process on-line cash for premium payments (in some cases, the agency does disbursement transactions. Special branches on wheels not collect records and/or centralize of the covered make conventional banking mobile, enabling regular individuals until the time of their retirement). As long and schedules visits in the remote communities. Finally, as operations of such programs lack personification, mobile phone banking is an emerging phenomenon capacity to introduce new benefits or facilitate targeting that utilizes comprehensive penetration of the new will remain limited. communication networks. All these solutions offer immense opportunities but also come with limitations the access gap can further be reduced by means (e.g., mobile phone banking works when money has of various technological innovations. Several been deposited in a bank account, etc.). A combination mechanisms are being tested where conventional of various solutions would likely be required. infrastructure fails to provide for connection between the local communities and some remote centralized india's dynamic and growing economy will lead operational core. Opportunities of both the on-line to greater demand for solutions that allow full and off-line communications are being explored. For portability of benefits. It is not clear if and when 288 Afewmonthsafterthe2009elections,theGovernmentofIndiaappointedtheformerheadofInfosysandwell-respectedreformadvocate,NandanNilekanito headuptheeffortandallocatedsignificantbudgetaryresourcesfortheeffort. CHAPTER-5: Social Security: Closing the Coverage Gap 12 some centralized public record-keeping platform intervention. These programs are dealt with in Chapter 3. could or should be developed. At the same time, the At the same time, a large proportion of households would regulatory mandate can provide for opportunities of benefit from ex ante interventions that allowed for risk standardization in the product design. (In particular, pooling and were made more affordable through direct for the retirement programs, the plans with a defined subsidies. Among these households, a subset is already contribution benefit offer greater portability by the participating in a number of uncoordinated attempts generic nature of their design). Furthermore, some ­ mostly through group arrangements ­ to provide off-shelve software products and utilities could be protection against a range of life-cycle risks. Some are promoted (and provided to participating groups at low sponsored by government, such as welfare funds while or no cost) as basic record-keeping platforms facilitating others are NGO-based initiatives that often involve easy implementation of the regulatory standards. microinsurance. the discussion in this section then suggests that the lessons from some of the schemes that already exist any new, contributory scheme targeted at the in india should be distilled and applied in the process unorganized sector should be: of adopting a new national policy for extending social protection coverage. These experiences could 1. based on target consumption smoothing objectives defined by public policy provide policymakers with information ranging from the factors affecting voluntary participation to data 2. affordable to a significant share of the that could help with actuarial calculations of cost.289 unorganized sector workforce Implementation lessons might include estimates of the 3. flexible enough to handle volatile income costs of specific transactions, such as processing a claim streams or issuing a statement and the potential savings from the scale economies usually found in administration 4. financially sustainable based on the principle and recordkeeping operations. Comparing notes across that assets match liabilities existing schemes might yield useful information about 5. administered in such a way so as to minimize how to efficiently use technology or how to set up transaction costs without compromising on the processes to certify eligibility and verify claims. In some accuracy and timeliness of recordkeeping cases, there may be evidence that supports government 6. accessible and portable throughout the action through better regulation or direct provision of country (and ideally based on a unique iD). public goods. The case of unique identifiers, already mentioned, is likely to be one such case. many of the existing schemes use a form of the e. ConCluSionS AnD partner-agent model and this approach has several advantages over the alternative, the creation of a new ReCommenDATionS layer of government bureaucracy with potentially india's unorganized labor force is vast and lakhs of employees. In particular, the PA model has heterogeneous and this reality must be taken into become increasingly relevant in India in recent years account in any attempt to extend coverage of as financial sector reforms have resulted in competitive social security programs. One important distinction markets for insurance and asset management and have is between households that are either too poor to led to the creation of specialized regulatory institutions. contribute or have already experienced the shocks This favorable situation should be exploited in order to that could otherwise be insured through contributory reduce costs and preempt the problems observed with schemes. In these cases, social assistance i.e., cash monopoly, quasi-state entities that cover formal sector transfers on a means-tested basis are the appropriate workers. The key question is whether this `bottom up' 289 Unorganised sector workers and subsets therein will exhibit mortality and morbidity patterns that may differ greatly from the typical clientele of insurance companies.Forexample,thelifetablesusedbyLICarebasedonannuitantsmostofwhomarehigherincomeindividualswithmuchlowermortalityratesthan thegeneralpopulation.Adatabasethatsystematicallycollectedthisinformationcouldbeusedtoproduceactuarialtablesthatwouldserveasabenchmarkfor assessingcostswhendesigningtheschemeaswellassupervisingprovidersoncetheschemewasoperational. 12 Social Protection for a Changing India: Volume II approach can be successfully scaled up with government actuarial calculations that were adjusted over support. Answering this question could begin with a timetoreflectexperience.This would help ensure series of meetings that could bring together potential that unfunded liabilities (for example, those that group participants (SHGs, MFIs, coops etc.), insurance arise when guarantees are offered) did not arise companies, asset managers, regulators, government and compromise financial sustainability. It would ministries and researchers. also make any subsidy involved explicit and transparent. initially targeting existing groups such as shgs and mfis is recommended for several reasons including direct government subsidy is the third element for reducing transaction costs, achieving effective of successful coverage expansion. In order to risk pooling290 and ensuring a minimum critical mass encourage voluntary take up among low income of voluntary take-up in the early stages of the segments of the unorganised labor force, the program. However, there are tradeoffs involved in required premia and contributions would have to such a strategy. First, while the number of such groups be subsidized. This subsidy would have to be set has been increasing rapidly in recent years, coverage at a realistic level given budget constraints.293 is still relatively low and regionally concentrated the rsby provides a model of coverage expansion that (predominantly in the Southern states).291 Second, any effectively encompasses these elements to provide subsidies involved in such a scheme should not be catastrophic health insurance coverage to tens of expected to mainly reach the very poorest households. millions of unorganized sector workers and their Various studies have shown that membership in families. The scheme is notable for its scalability and groups tends to be concentrated in the second and relative transparency achieved through a combination third quintiles rather than the bottom quintile.292 As of standardized processes and innovative use of mentioned above, for many of these households, the technology. By contracting out service provision, it condition that would have been insured already exists has allowed competition to set the premium level (e.g., life insurance for widows). In these cases, ex- and enough flexibility to adjust this over time. It has post transfers in the form of social pensions seem the also defined a targeted subsidy which appears to be appropriate policy response. sustainable in the broader fiscal context. All of these at least three measures are required if these initiatives features make it a good practice example for further are to be successfully scaled up. progress in reducing India's coverage gap. first, in order to take advantage of economies achieving widespread coverage in an effective manner of scale, to ensure portability and to facilitate is not possible without the development of information supervision, some common standards would systems that allow the goi to track members of these have to be developed, particularly in the area programs and the financial flows (contributions, ofrecordkeeping.A good example is a universal premia and benefits) efficiently. Innovative use of standard for identification of covered workers technology may be part of the answer, but in order to that, in itself, would be a public good with many keep costs down, existing infrastructure such as post other uses. Another is a ratings system for SHGs offices and banks will have to be harnessed. In addition, and other groups based on objective criteria. and especially during the first phase of implementation, second, appropriate contribution and insurance the recordkeeping that already exists for groups such as premia would be calculated based on rigorous SHGs and MFIs should be utilized and, where necessary, 290 Thesegroupsare,forexample,inamuchbetterpositiontoensurethatallmembersparticipateinahealthinsurancescheme,reducingadverseselectionproblems. 291 AsGhate(2006)reportshowever,thegrowthinthenumberofSHGslinkedwithbankshasincreaseddramaticallyandregionaldisparitiesareshrinkingsomewhat. 292 Seeforexample,BasuandSrivastava(2005)fortherelationshipbetweenincomelevelandmembershipinSHGsinAndhraPradesh.Inthesamevein,Ghate(2006) concludesthat"Microfinanceisbestsuitedtoreachtheeconomicallyactivepoor,whichmayexcludesomeinthelowestdecileortwoofthepopulationthatsuffer fromoldage,illhealthordisability." 293TheGoIalreadysubsidizesmembersoftheEPFOwithanannualcontributionof1.16percentofthecoveredwagebill.Inaddition,favorabletaxtreatment ofsuperannuationproductsandthePersonalProvidentFund(PPF),amediumtermsavingsinstrument,resultsintaxexpendituresofanunknownamount thatmostlyaccruetohigherincomeworkerswhopayincometax.Itcouldbearguedthatamatchingsubsidyofthekinddescribedherewouldbebothmore progressivethanandhaveamorepositiveimpactfromapublicpolicyperspectivethanexistingsubsidies. CHAPTER-5: Social Security: Closing the Coverage Gap 12 upgraded so that it is possible to `plug in' large numbers finally, although not mentioned above, there are many of participants in a cost effective manner. Both disparate and disconnected initiatives in progress by governmental and external assistance could be made central and state governments in india that should available to groups willing to participate and meeting be brought under a consolidated national policy recordkeeping standards (including compliance with framework. The passage of the Social Security Act at the unique ID system) should be a condition for receipt the end of 2008 is a positive step towards establishing of subsidies. A centrally managed recordkeeping such a framework. It now requires implementation. The system is one option, but common standards that Social Security Board envisioned under this act can serve allow supervision entities to monitor effectively are as a coordinating body bringing various efforts under a minimum prerequisite. The time and resources one umbrella, spell out a national policy complete with required to design and construct a national system of target benefit levels, costing, subsidy and strategy for this kind should not be underestimated but getting the phased implementation. In light of the importance of implementation machinery right at the beginning will implementation issues and, especially recordkeeping, avoid bigger problems down the line. At the same time, a special advisory group to make recommendations India is a world leader in MIS applications so that state of on these aspects with the best available technical the art solutions are possible if policymakers accord the support from public - private sector partners in India issue the importance it merits. could be formed. 10 Social Protection for a Changing India: Volume II CHAPTER 6 Financing and Institutions for Social Protection Chapter­6 Financing and Institutions for Social Protection like other public interventions, effective social direction, though spending on insurance-based protection systems rely on a mutually supportive programs outside the formal sector and urban web of appropriate policies, capable and empowered programs remains remarkably low. Looking institutions for implementation, and transparent and at international evidence across developing adequate financing. This chapter provides an overview countries, insurance-based interventions could of financing and institutional aspects of SP programs, key be expected to be a major source of spending determinants of SP system effectiveness in India. It first growth on SP programs over time. explores financing sources, mechanisms, and patterns in despite clearly progressive allocation of SP recent years for the major centrally sponsored schemes of resources across states, actual state-wise social protection, before examining the rapidly evolving expendituresonSPprogramsareregressive.The institutional framework for SP program delivery. limited analysis on intra-state SP expenditures suggests that some states reinforce the regressive the key findings from the chapter are: pattern of transfers across districts. financingofSPinterventionsinIndiaissignificant, looking ahead, there are four major challenges at over 2 percent of GDP. If one adds formal for India in financing of its social protection sector pensions, the SP spending share rises to system: (i) sustaining roughly the current share around 4.3 percent of GDP. In international terms, of public spending as overall spending rises; financing of safety nets in India is adequate, but (ii) continuing to adjust the composition of there remains a major agenda in getting better spending to greater emphasis on ex ante poverty reduction impact from spending. risk mitigation and promotional programs; within SP programs, financing priorities have (iii) increasing the flexibility of funding from the shiftedsomewhatduringthe2000sinapositive centre to states in terms of how it can be used for CHAPTER-6: Financing and Institutions for Social Protection 1 different SP interventions, while strengthening the emphasis on spending outcomes; and A. FinAnCing oF SoCiAl (iv) deepening a range of administrative and PRoTeCTion PRogRAmS institutional reforms in SP service delivery which can contribute to greater expenditure efficiency (a) Aggregate spending on centrally (see Chapter 7 on administrative and public expenditure management issues). sponsored SP schemes institutional roles and responsibilities for SP several observations can be made in looking at service delivery have evolved rapidly over the central spending on sp interventions in recent years pastdecade.However, the evolving assignment of (see table 6.1): institutional responsibilities for SP programs has India spends a significant share of resources on typically not been informed by a clear alignment socialprotectionprograms.At 2.2 percent of GDP of functions, funds and functionaries, though in 2008/09 from the central level, SP spending is MGNREG provides a promising example of the more than double public spending on health (at way forward. around 1.0 percent of GDP). There is approximately Table 6.1: Spending on major SP programs, 2002/03- 2009-10 2009-10 Program 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 (allocated) Food subsidy ­ primarily for PDS 24,200 25,200 25,800 23,200 24,200 31,545.6 43,627.2 42,489.7 Kerosene/LPG Subsidies + freight 4,735 6,372 3,553 2,930 2,625 2,730 2,724 2,866 NSAP and Annapurna 680 618 1,055 1,183 2,481 2,892 4,442 5,109 Central Labour welfare funds 102.7 104.2 109.5 160.3 162.3 263.6 259.2 269.9 RSBY NA NA NA NA NA NA 204 308 IAY 1,552.5 1,710 2,607 2,475 2,625 3636 7,919 7,920 SGRY 8,642 9,640 4,590 7,650 2,700 3,420 6,750 NA SGSY 656 720 900 899.8 1,080 1,620 2,113 2,115 Welfare schemes for SC 741.6 766.9 803.2 951.9 1,169.6 1,582.3 1,690.8 1,674.2 Welfare schemes for ST 960.7 871.8 1,016 1,370.2 1,611.6 1,637.7 1,894.3 2,578.7 Welfare schemes for disabled 203.2 207.9 203.0 220.8 220.4 226.0 242.3 271.4 people Midday meal 1,237 1,375 1,507.5 3,345.3 5,348 6,678 8,000 8,000 Urban SP programs (VAMBAY, 361.9 333 402.6 342.6 313.4 358 530.2 535 JNNURM, SJSRY) Employment Guarantee 860 0 1,818 4,050 10,170 10,800 30,000.19 30,100 (MGNREG) +NFFW (till 05/06) Total central SP expenditure 44,933 47,919 44,365 48,778 54,707 67,389 110,396 104,237 Total central SP expenditure as 11.12% 10.10% 8.77% 9.59% 9.41% 9.50% 12.25% 10.21% % of total GoI expenditure Total central SP expenditure as 2.0% 1.9% 1.5% 1.5% 1.4% 1.6% 2.2% NA % of GDP Estimated state level SP 0.54% 0.61% 0.58% 0.51% 0.52% 0.55% 0.54% NA spending as % of GDP Total central plus estimated state 2.54% 2.51% 2.08% 2.01% 1.92% 2.15% 2.74% NA level SP spending as % of GDP Source: GoI Expenditure Budget, various years. GDP and state level total expenditure from RBI: Valmiki Ambedkar Awas Yojana (VAMBAY) was merged with JNNURM in 2008. Data for 2006-07 include Vambay, JNNURM and SJSRY. State level SP spending calculated based on 3 percent of total state level expenditure by year. Note: Spending data for all programs are revised budget allocations from the Expenditure budget documents of various GoI budgets and not actual audited expenditures. 1 Social Protection for a Changing India: Volume II an additional 0.54 percent of GDP in addition MGNREG), and some marginal increases for IAY spent from state level budgets, making for a total andsocialpensions.There was also an appreciable of 2.74 percent on safety net programs. If one adds increase in the share of midday meals by the mid- to safety nets spending that on public pensions (a decade which has largely been sustained since further 2.1 percent of GDP in recent years), total then. However, the most notable areas of low SP spending is around 4.8 percent of GDP and expenditure within social protection ­ urban around the same level as public spending on the SP programs and social insurance outside the education system. civil service and formal sector ­ have remained negligible proportions of total to date. while the nominal level of spending has increased rapidly during the second part of the 2000s most sp css require financial contribution from states, (having been fairly flat in the first half of the so that goi spending does not represent the entire decade), core SP spending as a share of total GoI picture even for central schemes. A typical financing spending (plan and non-plan) actually declined split on SP CSS has been 75:25, though the two largest in the early years of the decade before a partial schemes (PDS and MGNREG) have a much lower state recovery in the last two budget years, with a typical share, as does RSBY which is likely to increase in coming range between 9 and 11 percent of total central years. MGNREG requires only around 10 percent of total spending (and a tendency to be on the higher from states (i.e., 25 percent of non-labor costs, with an end of the range in years preceding national assumption of 60:40 labor and other costs split), and elections). It is also notable that the GDP share of PDS requires only transportation and other logistical SP spending has had upticks in the years leading costs from states, with commodities provided free of up to national elections, though in the case of charge to the states. In addition, north-eastern and hilly 2008/09 there was also the likely upward impact states have special treatment in most schemes, with of SP responses to the food and fuel crises. higher central shares in total SP spending. over the period shown, there has been an in addition, there are a range of state-specific sp appreciableshift­whichisalsoshowninFigure6.1 programs which in some cases account for non- awayfromPDSasashareoftotalGoISPspending negligible shares of total sp spending at state level. towardsanincreasedshareforpublicworks(under It is very challenging to put together a full picture of Figure 6.1: Share of main CSS in total central SP spending, 2002/03, 2006/07 and 2008/09 2002-03 2005-06 2008-09 1% 4% 1% 0% 4% 1% 0% 5% 0% 2% 2% 4% 3% 21% 7% 7% 24% 33% 2% 2% 3% 5% 42% 54% 2% 7% 64% Public works (SGRY/FFW/NREG) SGSY Midday meal NSAP/Annapurna IAY PDS Urban SP programs Central Labour Welfare Funds Welfare schemes for SC, ST, PWDs Source: GoI Expenditure Budgets, various years. CHAPTER-6: Financing and Institutions for Social Protection 1 state-specific SP schemes. Overall the situation is highly cases.296 This estimate needs to be treated with caution, variable across states, with some such as Orissa and but seems useful in assessing orders of magnitude of Bihar having until recently virtually no state-financed SP spending at the state level, and having a sense of SP schemes of any note in spending terms, while others aggregate SP spending in India. The three percent of have had major outlays on state-specific schemes in total state level expenditure would have accounted for addition to their CSS counterpart funds. Interestingly, ` 26,783 crore in 2008/09.297 This would be close to an while poorer states in general have a higher share of additional 25 percent spending above GoI expenditures, central spending in total SP outlays in-state, there are or around 0.54 percent of GDP SP spending from the several examples of state-specific schemes even in state level. Allowing for specifics such as TN and AP's poorer states (e.g., Jharkhand spends significant shares additional commitments on PDS, the figure may well on state-specific schemes which in some cases replicate be higher, and more work is needed to develop robust CSS; UP has an unemployment allowance for educated estimates of state-level spending. unemployed youth). Some of the motivation for state in international terms, safety net spending of around schemes ­ particularly where they mirror CSS ­ may be the desire for states to "take credit" for programs. 2-2.5 percent of gdp is above what most countries at india's level of income spend (figure 6.2). This is a given the variability of state-specific sp programs, reflection of the political importance that attaches to SP the demands of css counterpart funding relative programs in India. As the Figure indicates, India spent to total state expenditure, and incremental state- well above its South Asian neighbors on SP programs as specific policies on even css, it is challenging to a share of GDP in the early years of the decade (and the estimate a precise figure for states' own spending on share has been even higher in recent years), as well as sp programs. Additional spending on other programs other larger countries in the Asia region such as China is highly dependent on state policies and fiscal capacity, and Indonesia. At the same time, international evidence so that state SP spending as share of total is variable from countries at all different income levels (including (e.g., in the mid-2000s in HP, SP spending was around the OECD) suggests that safety net spending generally 1.9 percent of state expenditure, while Rajasthan spent tops out at 2-2.5 percent of GDP, so that it is probably 2.3-2.5 percent, Orissa over 3 percent, and Jharkhand not reasonable to expect safety net spending in GDP a much higher 18 percent).294 In contrast to the central terms to increase far beyond the current GDP share that government, own-spending on PDS is generally low one observes in India in recent years. in states. However, there are notable exceptions international evidence also suggests that the such as TN which in its 2008/09 budget allocated major driver of increased spending on sp systems as 2,988 crore for PDS due to it being universal and at countries get richer is expansion on social security/ a very low purchase price, and AP's PDS allocation insurance spending. This can be seen in Figure 6.2, in increased from 1980 crore in 2008/09 to ` 3,500 crore which the major distinction between middle and higher in 2009/10. Another highly variable spending item until income countries is their spending on social insurance recently has been social pensions, due to the benefit rather than safety nets. It would be surprising if India top-up provided by many states.295 does not follow a similar trajectory over the coming 2-3 based on the set of states for which sp spending decades, more so as current social insurance spending data was analyzed in detail, a crude (and probably is almost entirely concentrated on the formal sector, conservative) estimate is that about 3 percent of total which accounts for less than 10 percent of the labor state spending is allocated to the core sp programs force. The rapid expansion of RSBY health insurance for outlined above, with significant additional funds on the unorganized sector suggests that an expansion in area development and relief programs in a number of social insurance spending may already be underway. 294 SeeWorldBank(2006d),variousstateeconomicreports. 295 WiththeincreaseofcentraltransferforNOAPSin2007/08budgetto`200perelderlypensioners,andthesubsequentassumptionofcentralfinancingofwidow anddisabledsocialpensions,manystateshaveseenasignificantreductioninsocialpensionspending. 296 Usingfiguresfromthemid-2000s,thiswouldaccountforonlyaroundhalfofownRDspendinginallstates,whichseemsaconservativeestimate. 297 SeeRBIfortotalstatelevelexpenditurebyyearatwww.rbi.org.in. 1 Social Protection for a Changing India: Volume II Percentage of GDP 0 1 2 3 4 5 6 7 8 SSA SAR MENA LAC EAP ECA OECD Senegal Maldives 0 Philippines Bosnia & Herzegovina Lebanon Paraguay Pakistan China Uruguay Tajikistan Programs 2 Colombia Venezuela, R.B. De Chile Kyrgyz Republic Source: Weigand and Grosh (2008). Mexico Bangladesh Peru Jamaica Madagascar 4 El Salvador St. Kitts & Novis Sri Lanka Yemen. Republic of Poland Romania Nicaragua Ecuador 6 Vietnam Guatemala Bulgaria Albania (b) Statewise allocations for CSS SP Indonesia Jordan % of GDP Latvia important to look at both allocations and actual looking beneath the aggregate level, it is Mongolia 8 Serbia Brazil Macedonia, Fyr Argentina region, early 2000s (% GDP) Costa Rica Georgia Azerbaijan Table 6.2. Grenada Egypt. Arab Rep. of 10 expenditures Moldova Panama Dominican Republic Spending on social assistance and social insurance by Croatia of Russian Federation Morocco Montenegro sp Uzbekistan Bolivia 12 St. Lucia css Turkey Armenia Dominica Kazakhstan Figure 6.2: Spending on social assistance and social insurance by region, early 2000s (% GDP) India St. Vincent funds Tunisia 14 Czech Republic Kosovo Honduras Iran, Islamic Rep. of Ukraine across South Africa Slovak Republic Algeria CHAPTER-6: Financing and Institutions for Social Protection Malawi Ethiopia Djibouti Soc.Ins Soc.Ass 1 Mauritius All CSS have allocation formulae which should formulae for some major programs are outlined in states. in principle be redistributive. The allocation Table 6.2: Allocation rules for selected SP CSS Program Allocation rule across states PDS Based on state rural and urban poverty rates for 1993/94 adjusted for population growth, and subsequent adjustment SGRY Based on state rural poverty rates for 1993/94 adjusted for population growth, and subsequent adjustment MGNREG Initially 200 backward districts as defined by the Planning Commission based on merging of RSVY and NFFW districts, using agricultural productivity per worker, SC/ST share and casual agricultural wage rates by district. Now covering all districts of the country. SGSY Based on state rural poverty rates for 1993/04 adjusted for population growth and subsequent adjustment IAY 50% on adjusted state rural poverty estimates and 50% on housing shortage NOAPS 50% of state-specific NSS poverty rate for population x over 65 years population by state. Now 100% of state-specific poverty rate under IGNOAPS. Source: Saxena 2006. while the general base for many schemes is the state a result, a compromise was reached which ensured poverty share, this is subject to an adjustment that allocations according to NSS poverty rates would factor proposed by the lakdawala committee in the be no less than 15 percent below those using the Task late 1990s, which ensured that the reduction in the Force methodology. The adjusted formula is now used app allocation of specific states did not fall by for SGSY, SGRY and IAY (the latter with an additional more than 15 percent from the previous allocation factor of housing shortage), and for TPDS, both formula. Prior to 1997/98, allocations were based urban and rural poverty are taken in account. The on a methodology of estimating rural poverty than comparison of state-wise allocations according to was subject to considerable criticism (hereafter the the Task Force, Expert Group and Adjusted Shares are "Task Force" method). When this was dropped in shown in Table 6.3. While for most states the adjustments favour of NSS poverty estimates, certain states are rather marginal, there are some obvious "winners" would have had their allocations fall sharply. As in the adjustment process, including AP, Maharashtra, Table 6.3: Core funding shares by state for rural anti-poverty programs State Task Force 93-94 expert group 93-94 Adjusted Shares Andhra Pradesh 6.635 3.257 5.64 Assam 1.513 3.866 3.222 Bihar 19.265 18.476 18.476 Gujarat 2.11 2.547 2.123 Haryana 0.721 1.498 1.249 Himachal Pradesh 0.272 0.631 0.526 Jammu & Kashmir 0.191 0.781 0.651 Karnataka 5.011 3.934 4.259 Kerala 1.094 2.293 1.911 Madhya Pradesh 11.018 8.859 9.365 Maharashtra 9.905 7.922 8.419 Orissa 7.59 5.774 6.451 Punjab 0.262 0.728 0.607 Rajasthan 2.665 3.88 3.234 Tamil Nadu 5.546 4.987 4.987 Uttar Pradesh 20.365 20.333 20.333 West Bengal 5.479 8.601 7.169 North Eastern states 0.235 1.147 0.967 and Union Territories Source: Saxena (2006). 1 Social Protection for a Changing India: Volume II Orissa, and MP. There are also a few "losers" but the conversely, several lagging states get some impacts are not dramatic, including West Bengal, what less ­ in the case of Bihar/Jharkhand and J&K, and Kerala. UP/Uttaranchal considerably less ­ than formula would indicate. are these allocation principles adhered to in practice? The short answer is broadly yes, though Table 6.4 as would be expected, the geographic targeting of below - using allocation data from 2004/05 for major NFFW program results in a very different allocation SP programs - indicates that allocations in practice do pattern across states, whereby most lagging not precisely follow the norms. Overall, the observed states do considerably better than formula (with patterns suggest that in general states which perform the notable exception of UP/Uttarakhand). well in terms of funds/food offtake are rewarded at using actual allocations, despite the variations seen the margin in allocations, and vice versa for poorer above, an encouraging distributional picture emerges performers. Of note is that: across states when comparing allocations on general most southern states do better in terms of actual (i.e., non-geographically targeted) sp programs. This allocations than the formula alone would suggest, can be seen in Figure 6.3, which shows a strong positive with some such as TN and AP doing considerably correlation between the allocation shares for major better than formula. programs (PDS, SGRY, SGSY and IAY) and the share of Assam and "other" (primarily NE states) also do major states in total poor people in India. The figure uses considerably better than the simple allocation data from 2004-05, but the general picture is similar for formula would suggest. most of the decade. Table 6.4: Formula share for SP CSS and actual allocations (% of total by state), 2004/05 Adjusted Actual expenditures (% by state) allocation norm SgRy SgSy iAy PDS (BPl) 4 programs nFFwP Andhra Pradesh 5.64 5.22 5.31 7.31 6.97 6.58 6.05 Assam 3.22 7.20 7.20 7.55 3.27 4.70 8.24 Bihar (+ Jharkhand) 18.48 17.36 17.38 25.68 12.95 15.43 24.27 Gujarat 2.12 2.29 2.00 2.10 2.97 2.69 1.98 Haryana 1.25 1.21 1.18 0.71 1.02 1.03 0.14 Himachal Pradesh 0.53 0.51 0.49 0.31 0.66 0.58 0.15 Jammu & Kashmir 0.65 0.60 0.61 0.38 1.19 0.96 0.24 Karnataka 4.26 3.90 4.01 3.78 5.11 4.67 1.45 Kerala 1.91 1.75 1.80 2.34 2.55 2.33 0.27 Madhya Pradesh 9.37 9.21 8.81 5.61 8.76 8.51 12.99 (+ Chhatisgarh) Maharashtra 8.42 7.71 7.92 6.71 9.56 8.80 7.67 Orissa 6.45 5.91 6.07 5.88 6.93 6.57 11.04 Punjab 0.61 1.34 0.57 0.47 0.61 0.74 0.35 Rajasthan 3.23 2.96 3.04 1.98 3.00 2.88 1.75 Tamil Nadu 4.99 4.57 4.69 3.67 8.34 6.91 2.40 Uttar Pradesh 20.33 18.63 19.13 14.77 16.38 16.77 13.57 (+ Uttaranchal) West Bengal 7.17 6.57 6.74 7.89 7.24 7.15 5.67 Others 0.97 3.06 3.05 2.85 2.50 2.67 1.76 India 100 100 100 100 100 100 100 Source: GoI, MoRD and Food; Bank staff calculations. CHAPTER-6: Financing and Institutions for Social Protection 1 Figure 6.3: Correlation between major CSS SP program share in total allocations by state and state share of total poor in India, 2004/05 25 20 Poverty share (%) 15 10 5 0 0 2 4 6 8 10 12 14 16 18 20 Allocation share (%) Source: GoI; Sen and Himanshu 2007 for poverty estimates. (c) Statewise expenditure performance offtake for BPL/AAY in the period, with all but MP below the national average. At the same time, it is the middle for CSS SP programs income states which had a notably higher offtake than while the allocation formulae for sp css are broadly the higher income group. Clearly more factors are at in line with poverty rates, the more interesting work than simply state income levels, with absorptive distributional story lies in actual program expenditures capacity, need, price differentials between PDS and and their diversity across states. The story of state wise market grains, and other factors at work. expenditure execution on SP CSS is often characterized as looking at the next biggest program, workfare, a one of poorer states failing to use their full allocations at the somewhat clearer pattern emerges, with the poorer expense of better-off states. While this is correct to some and ne states clearly with lower average offtake extent, the situation is somewhat more complex. While in terms of both cash and food (though notably certain poor and low capacity states have indeed been worse on food offtake in relative terms). At the same perennial laggards in spending execution on programs time, the grouping is by no means uniform, as poorer (e.g., Bihar, Jharkhand, Chhattisgarh), this is not true for states such as MP, Orissa and particularly Rajasthan all poor states, nor is it true for all programs even among managed very good and in some cases excess offtake the laggard states (e.g., Bihar has executed relatively well relative to allocations, while mid-range income states on IAY in recent years). The situation on specific major like West Bengal were among the weakest performers. schemes is presented below, followed by the aggregate The significantly greater shortfall of the poorer states on picture across major programs in recent years. average for food than cash is also of interest, as this may be related to the higher administrative and logistical with respect to pds, and focusing on bpl/aay - the demands of food delivery relative to cash. (Figure 6.5) group of interest ­ it is clear that offtake is highly variable across states. In spending terms, the major looking across several of the major sp programs program remains PDS for which state-specific offtake combined for the most recent year for which full numbers are presented in Figure 6.4, with offtake for data are available (2006/07), we see that there were BPL/AAY and APL separately. The low income states large variations in expenditure releases by state per on average have a considerably lower average TPDS rural poor household.298 This broad pattern holds for 298 Notethatthisistotalspending,ofwhichaportiongoesonadministrativecosts,plusifsubjecttoleakagebetweenreleaseandreachinghouseholds.Theamounts shouldthereforenotbeinterpretedasactualamountsreceivedbyhouseholds,whicharereportedfromsurveysourcesinChapters3and4. 10 Social Protection for a Changing India: Volume II Figure 6.4: TPDS BPL/AAY offtake by state, 2002-05 India UP Raj Oris MP Jhark Chhat Bihar WB TN Keral Karn AP Punj Maha HP Guj 0 20 40 60 80 100 120 % of allocation Source: Ministry of Consumer Affairs, Food and Public Distribution, various years. Figure 6.5: SGRY offtake on cash and food, 2004-05 India NE UP Raj Oris MP Jhark Chhat Bihar Food WB TN Kerala Cash Karn AP Hary Punj Maha HP Guj 0 50 100 150 200 % of allocation Source: Saxena 2006. the earlier years of the decade also. This can be seen in non-MGNREG main programs, and approaching Table 6.5. A few points emerge, including: twice once MGNREG is included. southernstatesdoconsiderablybetterinreleases conversely,mostlaggingstateshaveconsiderably per poor rural household, with states such as lower per poor rural household releases, though AP doing particularly well, at around 2.5 times even within them there is variation, with states the unweighted national average. For the four such as Rajasthan notably better performers southern states overall, they average over twice and the poorest states such as Bihar and Orissa the national poor rural per household allocation for particularly low. CHAPTER-6: Financing and Institutions for Social Protection 11 Table 6.5: expenditure releases per poor rural household for major SP CSS, 2006/07 exp per rural poor on 4 progs + State Poverty head Count (%) exp per rural poor on 4 progs (Rs) mgnReg (Rs) Andhra Pradesh 15.8 17,701.7 21,823.8 Assam 19.7 18,639.6 24,218.1 Bihar 41.4 3,868.0 4,984.3 Chattisgarh 40.9 6,160.1 10,760.6 Gujarat 16.8 7,141.5 7,763.1 Haryana 14 8,793.2 9,623.6 Himachal Pradesh 10 22,244.7 27,565.1 Jammu & Kashmir 5.4 82,088.5 87,321.7 Jharkhand 40.3 4,169.9 7,717.5 Karnataka 25 12,095.1 14,899.5 Kerala 15 12,929.5 13,302.5 Madhya Pradesh 38.3 5,569.9 11,075.7 Maharashtra 30.7 7,156.2 7,613.8 Orissa 46.4 4,500.9 6,766.5 Punjab 8.4 7,395.2 8,216.3 Rajasthan 22.1 7,036.8 11,357.1 Tamil Nadu 22.5 14,221.8 14,915.2 Uttar Pradesh 32.8 6,080.9 7,014.9 Uttaranchal 39.6 6,042.5 6,934.3 West Bengal 24.7 6,392.7 7,461.0 India 7,091.1 9,065.8 Source: MoRD Annual reports, various years. Bank staff calculations. The 4 programs are PDS, SGRY, SGSY and IAY, with MGNREG added in final column. Poverty head counts from NSS 61st round 2004/05. the major north-eastern and hilly states do very survey sources in Chapters 3 and 4 (and below for well on average in terms of allocations, perhaps Orissa from administrative sources). reflecting the lower matching fund requirements. This holds even if one excludes J&K which is a the relationship between state poverty rates and major outlier on the upside. spending per rural poor household is negative, indicating that the ex post outcomes in sp css releases totalspendingperpoorruralhouseholdnationally are regressive across states. This can be seen in on these programs was significant, at ` 9,065 in Figure 6.6, which presents average spending per poor 2006-07, or over 40 percent of the annual rural rural household by state for 2006/07 for SGRY, SGSY, household poverty line. Assuming an average IAY, PDS and MGNREG combined. A similar pattern 5 person household, the per poor rural household holds for earlier years in the decade, and if one excludes spending on these programs was around MGNREG. 42 percent of the annual rural poverty line.299 Such a figure needs to be interpreted with caution, as another way of looking at the expenditure a portion of spending goes on administrative performance across states on the main css sp costs, plus is subject to leakage between release programs is to compare their adjusted allocation and reaching households. The amounts should norm of total central spending and the outcomes of therefore not be interpreted as actual amounts actual expenditure from the centre on the main css. received by households, which survey sources This is presented in Table 6.6 using data from 2006/07 indicate are far lower, and are reported from on the 5 main CSS programs, and the patterns largely 299 The2004/05monthlyruralpovertylinewas`356.3perperson,orannually`21,378perhousehold. 12 Social Protection for a Changing India: Volume II Figure 6.6: Average spending per poor rural HH on main SP programs and poverty headcount by state, 2006-07 30000 HP 25000 AS AP 20000 15000 TN KA KE RN MP 10000 CH HR IN PB GU MA JR WB UP UT OR 5000 BI 10 20 30 40 50 HCR Expenditure per rural poor household Fitted values Source: Spending from Annual Report 2007-08 of MoRD and www.nrega.in; HCR: Planning Commission (URP 2004/05); Rural poor households based on state- wise projected rural population in 2005 from 2001 Census. State-wise rural poverty rate in 2004-05 divided by the average household size (NSS 61st round) to obtain rural poor households. Table 6.6: Spending on the five main CSS for conform to the picture above, though there are some 2006/07 and adjusted allocation norms notably better lagging state performers such as MP. Adjusted the relative spending performance across states is State allocation norm 5 programs driven by implementation differences, as allocations % % are far more progressive. Differences in implementation AP 5.64 8.4 performance are driven by a range of factors, some Assam 3.22 6.0 related to state-specific policies and procedures, and Bihar (+ Jharkhand) 18.48 11.0 some to institutional capacity at different levels within Gujarat 2.12 2.5 states. These issues are taken up in the remainder Haryana 1.25 1.0 of this chapter, and on a program-specific basis in HP 0.53 0.8 Chapters 3 to 5. J&K 0.65 1.3 Karnataka 4.26 5.6 Kerala 1.91 2.3 (d) intra-state SP expenditure MP (+ Chhattisgarh) 9.37 12.4 performance Maharashtra 8.42 6.8 given significant heterogeneity of living standards Orissa 6.45 5.1 within states, it is also important to look at intra- Punjab 0.61 0.6 state expenditures on sp programs. In principle, some Rajasthan 3.23 4.2 of the regressivity seen in cross-state spending could be Tamil Nadu 4.99 7.6 offset by geographically pro-poor release across districts Uttar Pradesh 20.33 14.5 within states.300 Such analysis is not readily available in (+ Uttaranchal) consolidated form. The following discussion therefore West Bengal 7.17 6.4 focuses on a few states for which district level SP Others 1.37 3.5 expenditure analysis has been done, taking selected SP India 100 100 program spending for Rajasthan, Jharkhand, and Orissa. 300 SeeWorldBank(2006d),forgeneraldiscussionofintra-statevariationsinhouseholdlivingstandardsandaccesstoservices. CHAPTER-6: Financing and Institutions for Social Protection 1 The data suggest that there are significant variations Table 6.7: Average household spending on SgRy/ across states, across programs, and in some cases across nFFw, SgSy and iAy by region, 2002-05: time in their intra-state expenditure patterns. Looking at BPl and below the poverty line households the figures, one can see: Program Spending in Orissa, Table 6.7 shows a clearly regressive Annual spending per BPl pattern of spending across regions on the main nSS region household RD SP programs (SGRY/NFFW, SGSY and IAY), with Coastal (poverty rate = 32 %) ` 965 spending per poor household in 2002-05 roughly Northern (poverty rate = 50%) ` 1453 twice the level in richer coastal regions as in the Southern (poverty rate = 87%) ` 961 poorest southern regions. This is not to say that Annual spending per poor the southern regions are ignored, as the RLTAP household (nSS) program and other initiatives indicate. However, Coastal ` 1867 it confirms on an intra-state basis in a very poor Northern ` 1300 state that expenditures on these programs remain Southern ` 966 regressive in geographic terms. Source: NSS 55th round, Bank staff estimates, BPL and spending from GoO. in Jharkhand for 2001-04, Figure 6.7 shows that districts that were more backward typically spent less per person than less backward areas on the indicating highly progressive intra-state spending same SP programs. This trend may well be offset across districts. once tribal program spending is included, but overall, therefore, it appears that intra-state for SGSY, SGRY and IAY and area development spending patterns on key sp programs in some states programs, the geographical pattern at district are reinforcing the distributional concerns that are level is mildly regressive. evident on a cross-state basis, though more state- for Rajasthan, the correlation between SGRY specific analysis is needed to confirm the findings. spending and BPL share by district was positive As with the cross-state analysis, absorptive capacity but not very strong or improving between and other factors are at work, so that there is no simple 2001 and 2005, while the correlation for SGSY policy prescription for how to address the spending improved sharply from 2003/04 to a high level, differentials issue within states. Figure 6.7: Intra-state SP spending Jharkhand: Spending on APP by districts, 2001-04 Jharkhand: Spending on APP by districts, 2001-04 Rajasthan: Spending on SGRY/SGSY by district, Rajasthan: Spending on SGRY/SGSY by district, correlation with BPL share correlation with BPL share 0.9 0.9 *: Correlation between district APP Fig.Fig. *: Correlation between district APP expenditure and backwardness expenditure and backwardness 35 35 30 30 Backwardness Index Backwardness Index SGRY Spending SGRY Spending & & 25 25 0.6 0.6 BPL Share BPL Share 20 20 SGSY Spending SGSY Spending & & BPL Share BPL Share 15 15 10 10 0 0 100100 200200 300 300 400 400 500 500 600600 700700 0.3 0.3 District capita expenditure District perper capita expenditure 2001-02 2001-02 2002-03 2002-03 2003-04 2003-04 2004-05 2004-05 Source: GoK. programs = SGRY, SGSY, IAY and area development programs. Source: GoR for spending and BPL by district. Index from RD Dept, GoJ. 1 Social Protection for a Changing India: Volume II (e) Financing sources for SP programs a distortionary effect to the extent that program financing is done through changes in distortionary at both state and central levels, general revenues commodity taxes and subsidies (e.g., if financing remain the overwhelmingly dominant mode of is done through reduction in distortionary program financing. While the spending split between subsidies, this will improve second order effects, centre and states varies by program, in nearly all cases and vice versa if such subsidies are increased to both levels finance from general revenues. The most finance the program). notable exception is that of various central and state- level welfare and provident funds for specific groups of in conclusion, the evidence on financing of major unorganized sector workers. Most of the central welfare sp programs suggests that sp programs which are funds are financed through a sector-specific cess (e.g., designed on redistributional lines become mildly less for the Beedi Workers Welfare Fund, a cess on all beedis so when one looks at allocations across states, and actually regressive when expenditure releases are produced; for construction workers, a cess of 0.3 percent used. This is not a problem amenable to easy solutions, on the value of construction undertaken). At the state level, the financing patterns are more mixed, though as flaws in implementation are often driving some of many require a contribution from workers, governments the poor spatial distributional outcomes. However, the and employers in some cases.301 Apart from welfare conclusions section offers some suggestions which may funds, the notable major CSS SP program with mixed be worth considering. financing is the midday meal, which in principle benefits from the education cess levied by GoI. A final exception to the practice of general revenue financing has been the B. inSTiTuTionAl RoleS AnD famous Maharashtra Employment Guarantee program, ReSPonSiBiliTieS in SP which has raised funding from a small tax on all those employed in the organized sector. PRogRAmS in india ­ and the large majority of developing even with sensible policies and implementation countries ­ there has been very little analysis of the procedures, programs are only as effective as the general equilibrium effects of different financing institutions responsible for their implementation. sources for sp programs. In most developing This section looks at some overarching institutional countries, analysis of both equity and efficiency aspects issues in SP program implementation, while of cash transfers have tended to focus on the direct Chapters 3, 4 and 5 explore program-specific institutional distributional impacts, using partial equilibrium analysis. and implementation issues. Overall, institutional A second important dimension is the indirect effects of arrangements for SP programs are in a state of flux programs, which looks at the impacts of transfers and which results in a considerable degree of heterogeneity their financing modes on the level and composition of across states, sometimes significant gaps between de supply and demand in the economy. While such work jure and de facto roles of different institutional actors, does not yet exist for India, analysis from Mexico on the and divergences in institutional priorities, both between indirect effects of cash transfer programs provides a levels of administration and between administrative and useful example of CGE modeling of such impacts, and its elected institutions. implications for different financing models.302 Indirect sp programs in india are characterized by a vast range effects would through three main effects: of institutional actors. This has several dimensions: a redistribution effect from the patterns of tax to at central and state levels, programs are spread finance cash transfers across a range of Ministries and agencies, a reallocation effect, which will arise if the income including Rural Development and/or Panchayati elasticities of demand among those financing the Raj, Food Distribution, Social Welfare, Labor, programs differ from those receiving transfers Urban Employment and Poverty Alleviation, 301 SeeILO(2004). 302 SeeCoadyandHarris(2004). CHAPTER-6: Financing and Institutions for Social Protection 1 Human Resource Development, Tribal Welfare, given strengthened implementation functions to Food Corporations etc. GPs and gramsabhas. This shift is evident in the across levels of administration, central, state and Eleventh Plan document that seeks to use PRIs sub-state officials are involved, with the relative as the "primary means of delivery of essential importance of the three shifting in SP programs services that are critical to inclusive growth". over time in terms of different elements of despite the increased de jure role of PRIs, field implementation in SP programs. researchindicatesthatstatesandadministrative at sub-state levels, there are roles in SP programs channels/line departments continue to play a for administrative and elected officials which vary substantial role in program implementation, across states both de jure and in practice, and though this varies across states and programs. across programs. The de facto outcome of institutional roles is driven in large measure by the misalignment of community groups such as SHGs and cooperatives functions of PRIs with funding levels and controls have in a number of states been given roles in over resources, and the lack of capable personnel implementing SP programs, and/or monitoring for executing those functions. their performance. onamorestate-specificbasis,communitygroups there are several important institutional trends are assuming new roles in SP service delivery. with respect to major sp programs in recent years, Examples include operation of FPS by SHGs and including: cooperatives in several states; involvement of the increased reliance on CSS, combined with communities in social audits of MGNREG, often increaseduseofdirectfundtransferstodistrict with facilitation by NGOs; involvement of VECs level,hasincreasedtheroleofthecentrevis-à-vis and womens' groups in preparation of midday statesinmajorSPschemes.This has attractions in meals, and in some cases supply of additional terms of the equity function of SP programs, with food items for the meal. greater scope for cross-state allocation of funds. in a number of states, or specific programs, At the same time, it may reduce the incentives NGOs and CBOs have been given a promotional for states to monitor programs financially, and or facilitating role in program implementation. other functions. Examples of this trend include contracting of theincreasedimportanceofSPfundsflowsdirect NGOs for capacity building and other support to to district level presents some differences across SHGs seeking to access SGSY credit, facilitating statesintermsoftherelativerolesofZillaParishads roles in promoting community oversight and (ZP)anddistrictadministrationrepresentativesin social audits, and reliance on worker associations, DRDAs.While in principle the DRDA should be a MFIs and trade unions in mobilization of members channel for elected representatives to have a role for participation in social security schemes for in program planning and disbursements, the role the unorganized sector. Another recent example of ZP officials is highly variable. is the role of different grassroots organizations the emerging role of PRIs (in particular Gram in outreach for the RSBY program. While states Panchayats (GPs)) in aspects of SP program and researchers are generally positive on the implementation. While MGNREG represents the contribution of NGOs in program implementation, most thorough effort to date to give GPs a central there is also skepticism in some quarters on the role in SP service delivery,303 other programs have risks of "NGO capture" in cases where NGOs are co- in either central rules (e.g., SGRY) and/or state-level opted by the public system as part of its network implementation (e.g., PDS through handing over (and NGOs such as MKSS have deliberately FPS to PRIs; IAY in local beneficiary identification) avoided formalizing their engagement with the 303 InMGNREG,GPs(andatblockanddistrictlevels,panchayatsamitisandzilaparishadsrespectively)areexpectedtoidentifysub-projects,mobilizeandorganize beneficiaries,assistinopeningworksites,insomestatespaybeneficiaries,andplayaroleinmonitoringthroughthegramsabha. 1 Social Protection for a Changing India: Volume II state in order to retain a more independent voice the fact that the 73rd and 74th amendments are non- in program monitoring and mobilization).304 mandatory with respect to devolution of functions to pris has resulted in very variable devolution of sp while the situation is evolving rapidly with respect to institutional actors in sp service delivery, (and other) programs across states, and often lack of clarity within states over which institutions are this has increased the diversity of institutional empowered to undertake specific program-related arrangements across states. This has naturally resulted activities. Despite the specificity of the list of services in uncertainty ­ and in some cases resistance - within which can be assigned by states to PRIs, the situation states as new institutional modalities for SP service varies greatly on whether ­ and if so in what detail ­ states delivery mature. Overall, the experimentation that is have subsequently empowered PRIs in specific areas happening both within CSS institutional design, and in of service delivery.307 As the power of PRIs in service state-specific implementation modalities is a welcome delivery is derivative of states, such assignment by development in light of implementation problems that states is critical. The situation in different states can be have long plagued SP programs. At the same time, some described as follows: of the flux is less positive in that it appears to be driven by genuine lack of coordination and even confusion on most states have not subsequently assigned the part of policy makers with respect to institutional specificSP-relatedactivitiestoPRIsinacoherent assignment of responsibilities. The following paragraphs manner, so that PRIs have no specific mandate discuss these issues. in state legislation (though they may derive this fromcentrallegislationandprogramguidelines). a starting point in examining institutional roles in At the same time, program specific initiatives in sp programs is legal assignment of responsibilities. a number of states are increasing the effective Interestingly, neither rural development nor poverty delegation of activities to PRIs and community alleviation figures in any of the lists of central, state groups on a more piecemeal basis. or concurrent subjects under the Seventh Schedule, in a number of other states (e.g., Rajasthan Article 246 of the Indian Constitution. The State List includes Agriculture, Water, Animal Welfare, Land, and and Karnataka), functions have been explicitly Fisheries, whereas Economic and Social Planning is given devolvedtopanchayats,butinbroadtermswhich in the Concurrent List. The State List includes `Relief of result in concurrent responsibilities between the disabled and unemployable', but no general poverty states under existing legislation and PRIs. This alleviation function.305 results in lack of clarity on assignment of specific activities within the broad devolved heads, and in contrast, the 73rd and 74th amendments to the overlap of functions across levels of government/ constitution are more specific in allowing ­ though administration. not mandating - assignment of sp program functions onlyKeralahasdevolvedactivitiesinaninclusive to panchayats by individual states.306 For rural areas, manner in terms of subject areas and with a the SP areas where states may assign functions to PRIs reasonabledegreeofspecificityin terms of specific include rural housing, poverty alleviation programmes, activities and roles of different actors. This avoids public distribution system, social welfare, and welfare the issue of concurrent powers which many states of "weaker sections", particularly SC and ST. For urban encounter. areas, the 74th amendment allows urban local bodies to have responsibilities for urban poverty alleviation. as a result, pri capacity for implementing sp programs This de jure power is supplemented in practice by is highly variable across states. Given the large direct assignment of functions to panchayats under the volume of funds currently being administered by guidelines of central schemes such as MGNREG. panchayats under the Backward Region Grant Fund 304 SeeJenkinsandGoetz(1999)forausefuldiscussionof"statefostered"civilsocietyorganizationsandtheirroleinservicedelivery. 305 Saxena,(2006). 306 ThefollowingparagraphsdrawonWorldBank(2006b). 307 Stateshavebeenencouragedtooperationalizethe73rdAmendmentthroughsevenround-tablesorganizedbytheMinistryofPanchayatiRajaswellasregular StateofthePanchayatReports. CHAPTER-6: Financing and Institutions for Social Protection 1 (BRGF), MGNREGA and other central schemes, building program challenges such as lack of responsiveness of capacity to administer these schemes as well as general schemes to local needs and weak local accountability skills related to planning, financial management and due to oversight from a distance. procurement becomes critical. Recent initiatives such it is clear from the above that the assignment as the Panchayat Empowerment and Accountability of institutional responsibilities for sp programs Initiative (PEAIS) seek to address this issue. The PEAIS has often not been informed by a clear alignment includes providing training to functionaries and elected of functions, funds and functionaries, though representatives on roles and responsibilities as well as IT mgnreg is a promising case of efforts to improve the and FM systems to panchayats. situation.309 The dominance of CSS in SP with the need the importance of css in the sp sector adds further for state and sub-state level implementation creates complications to assignment of institutional challenges for developing an appropriate institutional responsibilities, as css guidelines increasingly spell framework. In this respect, it is important to have an out institutional responsibilities which may or may alignment of functions to be carried out by different not be consistent with assignments in individual actors (administrative and elected public sector, NGOs states. In addition to the lack of clarity within states on and communities themselves) which reflects ­ or assignment of institutional responsibilities for specific mobilizes ­ the capacity of functionaries at different levels service delivery tasks, the guidelines of CSS increasingly to undertake the assigned functions, and is supported introduce roles for PRIs in program implementation. The by sufficient funding and control over spending for each most thorough example is MGNREG, but other schemes level to perform its anticipated functions. such as SGRY, SGSY (to a modest extent), and IAY also unbundling the key activities that are needed in mandate a PRI role. Two points to note in this respect individual sp programs would therefore seem a sensible are: (i) that different central schemes presently have first step towards more appropriate assignment of very different degrees of PRI participation, reflecting in implementation responsibilities. The guidelines for part their age and the state of the devolution debate at MGNREG provide just such an example, and should act the time of framing and; (ii) there is often a mismatch as a model for future reforms of SP CSS. At the same time, between CSS-assigned institutional roles and more for CSS, there should be closer consultation with states general assignment of devolved functions to PRIs within on which unbundled functions most sensibly sit at what states.308 level of the system, and with elected or administrative one shortcoming of institutional assignments of agencies. An example of such an "unbundling" is responsibility for sp programs in the past has been a presented in Annex 1 for SGRY, looking at both de jure tendency to concentrate responsibilities for multiple and de facto assignment of the unbundled activities. program functions at specific ­ and sometimes Annex 1 also presents an indicative format for activity inappropriate ­ levels of the system. This leads to mapping developed by the Ministry of Panchayati Raj, difficulties in enforcing accountability, in part due to in response to the recommendations of the Empowered concurrent responsibilities within states. To date in most Sub-Committee of the National Development Council schemes, there has been a tendency to cluster functions (NDC) on Financial and Administrative Empowerment at specific levels of the system, rather than assigning of PRIs. In broad terms, these functions can be divided specific activities to the most appropriate level. In some into six major categories. Within these major categories cases, the clustering is de jure (e.g., SGSY), and in others of activities will be a range of sub-activities which need de facto (e.g., SGRY). For example, districts ­ through further enumeration and assignment: Collectors and BDOs - and states have historically played policydevelopmentandstandards a major role in all aspects of SP implementation, from planning prioritization of activities (e.g., works selection; priority lending areas) to oversight. This has contributed to assetcreation(and/orbenefitdelivery) 308 Seealsotheadvisorydated19.01.2009fromtheMinistryofPanchayatiRajtostatesondelineationofrolesandresponsibilitiesofPRIsinCSSs/ACAs. 309 SeeWorldBank(2006d),foradetaileddescriptionandapplicationofthe"3Fs"frameworkforIndia,andWorldBank(2004)forinternationalevidence. 1 Social Protection for a Changing India: Volume II program operation (including HR, FM and other there are different options for how to improve the supportfunctions) coherence of sp policy development and standard M&E setting, but all would imply an explicit institutional mechanism for more strategic and coherent action with respect to policy development standard setting, at both central and state levels. Different countries it is clear that the centre and states should continue have approached such coordination challenges with to play a major role. Given the patterns of financing, different institutional responses. Some have tried to and the desire for a significant degree of equity in bring programs increasingly under the purview of a programs across space within the country, it makes single Ministry (such as Labor and Social Protection/ sense for the centre and states to continue their major Welfare), others have created coordinating institutions role as policy developers. However, there are issues in or structures under the office of PM or President. how these policy and standard setting functions are These may be ambitious options for India, given the carried out within and between the centre and states institutional interests and history in the sector. A more which will require a fresh look in light of the evolving SP modest option would appear to be creation - with a policy and program mix. lead role for the CS, Planning or Finance Departments in a first issue in policy and standard setting is how states, and for Finance perhaps at the centre ­ of a Social to promote greater coordination at central and Protection Task Force or Agency. This could involve all state levels across line agencies. In the past, the ministerial players, and deal with policy issues in SP of a concentration of SP programs in the Ministry of Rural more strategic or cross-cutting nature. Development and Ministry of Consumer Affairs, Food there have been efforts at the sub-national level in and Public Distribution meant that the coordination recent years to try to promote greater institutional challenges in developing a coherent SP policy were and delivery platform harmonization across agencies more limited. However, the situation is already changing responsible for sp and other targeted programs. and can be expected to change further in coming years Uttar Pradesh approved such an agency in 2008, in several ways. Firstly, greater efforts will be need on called the Safety Nets Authority with an important development of appropriate SP policies and programs for standard setting and harmonization function across urban areas, increasing the role of both urban Ministries. departments responsible for programs targeted to Secondly, the anticipated expansion of social security to the poor. The mandate for the Authority seemed very the unorganized sector will involve Ministry of Labor and appropriate. In 2008, the Delhi Government introduced Employment more actively in SP service delivery (as well the Convergence Mission. This initiative, as the name as a range of third parties such as insurers, intermediary suggests, is a concerted attempt to streamline service organizations of unorganized workers et al). Thirdly, the delivery of multiple targeted programs. To achieve this increasing role of SP interventions to support human vision, the Government intends to bring the programs capital formation (currently midday meals, but also on a common platform, strengthen the implementation stipends for education and pilot CCTs for early child machinery at lower levels, and incorporate civil society interventions) will require increased coordination with organizations as partners in overseeing the entire other social sector Ministries such as Human Resource process (see Box 6.1). Some states, notably Kerala, have Development, Health and Family Welfare, and Women shown great progress in convergence across programs and Child Development. Finally, the increased role of at the local level through the involvement of PRIs. local institutions in delivery of SP programs will require coordinated efforts between line agencies and the a second important aspect of this issue in india is the Ministry of Panchayati Raj. Simply relying on the lead degree of policy freedom that states could be given of Rural Development and Food Ministries, and/or in policy adaptation and mix in major sp programs. expecting bilateral coordination between ministries will A common criticism of CSS has been that states have increasingly be inefficient and insufficient, and is likely insufficient policy leeway in adapting CSS program to increase the tendency to substitute broad-based design to the needs of specific states. While this policy development and oversight with "guideline concern has been somewhat less in recent years (e.g., development" within specific schemes. as states have been given more leeway in adapting local CHAPTER-6: Financing and Institutions for Social Protection 1 Box 6.1: mission convergence: institutional reform of targeted programs in Delhi Like other states/UTs, Delhi has a range of social programs - both CSS and local programs - which seek to target the poor and vulnerable. These are administered by nine different departments, including Social Welfare, Health, Education, Urban Development, Food and Civil Supplies, and Labor, among others. Many of the programs overlap in design as well as target populations, implementation and reporting arrangements. However, no single entity has a comprehensive overview of programs in terms of type of intervention, beneficiary population, delivery system, spending, etc. From the point of view of the target population, this multiplicity implies understanding various eligibility rules, making several applications and engaging with multiple authorities in accessing benefits. This is often very challenging. In 2008, the Delhi Government introduced the Convergence Mission. This initiative is a concerted attempt to rationalize and converge service delivery of multiple programs, with a view to improve access and outcomes. To achieve this vision, the Government intends to bring these programs on a common platform, strengthen the implementation machinery at lower levels, and incorporate civil society organizations as partners in overseeing the entire process. In operational terms, this has involved the following processes: (a) Defining institutional responsibilities and implementation structures. At the UT level, a registered society, the Samajik Suvidha Sangam, is responsible for implementation. The SSS is also responsible for coordination with the participating state departments. At the grassroots level, over 90 Gender Resource Centers (largely managed by NGOs and CBOs) serve as the principal agency for outreach, beneficiary identification and service delivery. (b) Rationalization of programs. The Mission, with inputs from the Commissioners of the Supreme Court, reviewed the social programs administered by the nine departments with respect to their relevance, effectiveness of implementation arrangements, and eligibility criteria. The rationalization process entailed merger of some schemes (within and between departments), closure of some, and the expansion of coverage or scope of others. This is a massive achievement. However, even further rationalization as the current of programs remains high at 35 programs. This is seen as a dynamic process. (c) Development of common beneficiary database. The Mission has adopted a new targeting method, which supersedes the previous BPL system. Vulnerable households are now classified as per three criteria: (i) Residence - those living in slums and resettlement colonies and the shelterless or precariously housed; (ii) Social deprivation ­ single unprotected children, child-headed households and households with elderly, disabled, single women, persons suffering from specific illness such as HIV/AIDS, tuberculosis or leprosy; and (iii) Occupational vulnerability ­ those households that are primarily dependent on earnings from casual work with irregular wages and poor working conditions. A subset of households is considered most vulnerable. While the move towards identifying vulnerable households on the basis of non-income dimensions is a positive step, translation of these criteria into clear, measurable and verifiable indicators remains a concern. In addition, the reliance on MC's field service units (see below) for the enumeration of households has raised problems in execution of the proposed targeting method. (d) Establishment of a common delivery platform. The next step is to develop a common delivery platform that provides access to all programs using a biometric card system. The GRCs would provide a single point of contact between beneficiaries and government. Common formats for enrolment in various programs have been developed, along with simplification and streamlining of documentation requirements. Enrolments through biometric cards are planned in order to minimize fraud and facilitate efficient and integrated delivery of benefits. Similarly, the intention is to develop common systems for information, reporting, and monitoring. Mission Convergence is an ambitious and welcome initiative. Delhi has a very complex governance structure. This presents formidable challenges to such an initiative that redefines administrative and local power structures. If successfully implemented, this could potentially increase cross-program coherence, streamline delivery and demands on administrators and beneficiaries, and increase transparency. It can also serve as a model for other states; for instance, Uttar Pradesh also announced in 2008 the formation of a Social Safety Nets Authority to perform a coordination function across state departments and programs. institutional arrangements in program delivery), the of planning, several government and donor-supported overall room for states to adapt the policy mix across initiatives have focused in recent years at district level, major SP programs remains limited. In addition, specific with the objective of achieving greater cross-program elements of the CSS policy framework remain in some convergence on priorities, in the process reducing cases more rigid than would be desirable (e.g., the risks of duplication of use of funds. The guidelines of restrictions on types of works allowed to be undertaken MGNREG again provide a useful example of efforts to under public works programs). promote greater coordination in planning between GPs, for planning, the connections between levels of the blocks and districts, though field work reveals that such system ­ and between administrative and elected coordination remains to be realized effectively in most mechanisms ­ remain the biggest challenge. In the area states. However, the challenges of achieving effective 10 Social Protection for a Changing India: Volume II district planning should not be underestimated. The guidance provided by the constitution. While at difficulties in coordinating planning efforts within one level such a proposition is obvious in light of the states include: diversity of institutional capacities across states, an at the higher levels, there is often inadequate immediate question is how far the centre can drive coordination between centre and states (and/or specific institutional assignments of roles on a scheme- DRDAs), so that communication of likely allocations specific basis. A desirable compromise between the for SP programs in the coming budget year are interests of states and the centre would seem to be a not known in advance, and at times not well into cross-scheme assessment involving both parties of the budget year. This makes planning at state and appropriate and feasible assignment of functions and lower levels challenging, as forward planning specific activities in major SP programs. The case of exercises do not have sufficient information on Kerala is the best practice to date of moving from such the resource base to make informed decisions. agreements to a generalized but specific assignment of activities for PRIs (and by default administration) in there is limited attention in allocation of program implementation. institutional responsibilities to the economies of scale/scope and the potential externalities of for urban areas, the roles of locally elected differentprogramactivities.An example is public bodies, municipalities and line agencies in sp program works planning, where GP-level decisions on water administration remain even less clear than for rural management works or construction of link roads areas in many states. While analytical work is growing may ­ in the absence of coordinated action at a on institutional roles in rural areas, the urban landscape wider geographic level ­ result in works which fail remains relatively under-examined. This is an area to maximize wider-area benefits and efficiencies. where the Ministry of Urban Employment and Poverty Often, multiple district plans are prepared without Alleviation, in collaboration with states and cities, has a an integrated vision for the district as a whole. role to play in sharing knowledge on existing institutional capacity for planning at sub-state level is arrangements in urban SP programs, and good practice typically low, so that close technical support among states. is often required to facilitate such planning on m&e, there are likely to be roles for all levels of exercises. Data from the Ministry of Panchayati the system, but in different capacities which build on Raj indicate that as of 2005-06 several states the comparative advantage of each. Some elements of had not yet constituted District Planning a desirable division of responsibilities might include: Committees (e.g., UP, Gujarat, Goa, Assam, and Andhra Pradesh), so that a basic building block forthecentre,therecouldsensiblybethreerolesin of effective district planning was absent.310 This M&E:(i)workingwithstatestodevelopcommon situation is gradually improving and by 2009, administrativemonitoringsystemsandindicators. 19 states had constituted DPCs. These could be enhanced at state level as capacity allows, but some minimum standard setting on even where district planning mechanisms are in program reporting would be a valuable central place (which is the case in the majority of large function. The collaboration between the central states), there is an inherent tension between Ministry of Rural Development and the National the tied nature of SP (and other) program funds Informatics Centre (NIC) on software development and the intention in district planning to give for MGNREG provides a good example of the panchayats, DRDAs and other local agencies a value-added of the centre in SP programs, as does less constrained voice in the planning process. the software certification function of NIC within on operational aspects of program operations, the the RSBY program; (ii) within CSS budgets, making key decisions on roles and responsibilities between explicit allowance for M&E functions in allocations, levels and between elected and administrative including capacity building within states; and machineries need to be taken by states, within the (iii) strengthening the evaluation of CSS SP 310 www.indiastat.com,citingMoPR,GoI. CHAPTER-6: Financing and Institutions for Social Protection 11 programs, in particular moving beyond the functions. Such functions would be more difficult present system of concurrent evaluations to more to perform in urban areas where community rigorous methods of impact evaluation which institutions are less developed, and options like proceed from reliable pre-program baselines. localized third party monitoring may be required; Such a function is anticipated under MGNREG (ii) the role of the block level in monitoring guidelines, though the anticipated central role should be explored more closely, as a potential in this regard has yet to be rolled-out in practice. layer which balances the desire for being close to More generally, the Planning Commission has the source of implementation with the need to argued recently for an Independent Evaluation minimize GP monitoring their own execution of Office which would have an arms' length activities. The extent to which the functions of the relationship to the implementing Ministries, and block in this respect would most usefully amount have a more analytical evaluation function than to oversight or some closer form of monitoring that of the CAG, which tends to focus more on needs further consideration. At a minimum, the compliance issues rather than impact evaluation block would continue to be an important layer in per se.311 consolidating program performance information, for states, there is a need to reconsider what and using this to inform coordinated action across rolescaneffectivelybeplayedinM&E.The need GPs where the nature of program suggests it (e.g., to "push down" certain monitoring functions public works which involve cross-GP coordination may require a narrower role for states in future. or impacts); (iii) the district level is perhaps the Nonetheless, several important tasks could most difficult to assess in terms of potential M&E sensibly be retained, including: (i) responsibility roles. It is in principle too far away from the source for financial internal and external audit of of implementation to monitor effectively, but not programs, including capacity building at lower high enough to have much capacity to conduct levels for generation of records needed to conduct more rigorous evaluations. effective audits; (ii) consolidation of lower level the biggest challenges for sp programs in terms of monitoring reports, and synthesis of trends as a institutional assignment of roles will continue to be tool for policy refinement and reorientation as two: (i) mismatch between centrally-mandated roles needed; and (iii) with financial support from the under specific programs for sub-state actors and the in centre, deepening a culture of impact evaluation practice level of devolution and capacity of PRIs; and in programs. (ii) lack of clarity on the relative roles of administrative and at the sub-state level, the challenge will be to PRI delivery channels within states, particularly where balancethestrengthsoflocallevelactorsinbeing de jure and de facto responsibilities are misaligned. "closetotheaction"withtheneedtohavesome distance between those implementing different aspectsofprogramsandthosemonitoringthem. C. PoliTiCAl eConomy oF A threshold issue in this respect is a more explicit assignment in most states of responsibility for inSTiTuTionAl ReFoRm in specific service delivery activities in SP programs between sub-state levels and between elected and SoCiAl PRoTeCTion administrative channels. However, several general the political economy of sp policy and institutional principles seems desirable: (i) communities, both reforms is complex, and will require intensive efforts through gram sabhas and through facilitated to build consensus on reforms. In particular, it will processes such as social audits, should be given be important to ensure that the interests of perceived a stronger authorizing environment for program "losers" of SP and broader economic reforms are taken monitoring of all SP programs, as allowed for under into account. While reforms that involve expanded MGNREG, and support provided to perform such coverage or new types of interventions are unlikely to 311 http://www.business-standard.com/india/news/independent-office-to-evaluate-govt-schemes-soon-montek/75523/on.October9,2009. 12 Social Protection for a Changing India: Volume II be controversial, there are strong interests in preserving easier to address through development of more the status quo in SP programs among a range of actors, outcome-based monitoring systems. The second including administrators, politicians, contractors and is more difficult in a democracy. others. Simply cutting programs or excluding certain in a number of programs, there are presently groups of beneficiaries or institutional players is significantrent-seekingopportunitiesforarange therefore unlikely to be successful unless incentives for ofactors.Such opportunities are facilitated by the institutions and households which will be affected by current complexity of the SP program mix, but reforms can be part of the reform package. The political also by the number of intermediaries who often economy of SP reform is challenging in all countries, are involved in the interactions of poor people and governments in India would benefit from more with the SP system. The generic identity of such innovation in their efforts to create a broader societal official intermediaries and unofficial middlemen understanding of the need for and benefits of reforms. is generally well-understood, but minimizing the some of the political economy challenges that the potential avenues for their continued roles has only sp system confronts if it is to become more coherent recently become a more explicit goal of SP policy and more effective as a tool for promoting poverty design. While it is too early to say, even apparently reduction and inclusive growth include: naïve blanket bans on certain actors in legislation and/or guidelines (such as the ban on contractors as in many areas of policy reform in India, under NREGA) do appear to help. However, a more consolidation and reform of the SP system comprehensive approach will require a more within a coherent strategy will run counter to thorough modernization of SP business processes. the past experience of scheme-driven initiatives Examples where such approaches already appear by a plethora of Ministries, and the observed to be making headway include greater reliance on tendency of each new government at both direct transfers to beneficiaries through banking central and state levels to want new SP programs and postal systems, and innovations in use of ICT clearly distinguishable from their predecessors. in SP program delivery. Reducing these natural bureaucratic and political amorerecentandfluiddevelopmentinthepolitical tendencies will be very challenging. A first step is economyofSP(andmuchother)servicedeliveryis obviously having an integrated SP strategy which howincreaseddecentralizationofresponsibilities is driven by the top politicians and bureaucrats to panchayats ­ in particular GPs ­ generates at central and state levels, with strong inputs different patterns of contention, cooperation from civil society in its development, including and collusion between newly elected panchayat opposition parties. However, even if such a officials and traditional loci of influence among strategy process can be developed, it will be administrators and higher level politicians such as important for it not to become a "one shot" MLAs and MPs. An essential first step in promoting exercise, but to have institutional coordination decentralization of SP service delivery as a tool for mechanisms in place which explore program contestability and hence accountability will be duplication and exploit synergies. getting a better empirical understanding of the givingstatesamoreflexiblehandinuseofcentral diversity and evolution of experience. This would SPresourceswillbeachallengingtransformation include how the gradual increase in the role of both for central administrators (whose past panchayats is proceeding (and what factors ­ tendency has been to define the parameters for such as limited control of resources and very low use of central funds quite tightly) and politicians capacity), and the extent to which panchayats (who not unexpectedly seek political attribution effectively promote accountability in SP service for centrally-financed schemes implemented by delivery or are captured by local social, political states). The first of these challenges in perhaps and administrative elites.312 312 Jha,Bhattacharyya,GaihaandShankar(2008)onMGNREGprovidesusefulempiricalinsightsintothedynamicsofprogramcaptureandthevariationsbystate forthecasesofRajasthanandAP. CHAPTER-6: Financing and Institutions for Social Protection 1 thefinalmorenascent,butpotentiallypowerful, and middle income countries. its main challenges are element in the political economy of SP reform therefore fourfold: is the role of communities and civil society in sustaining roughly the current share of public promoting more effective poverty reduction spending as overall spending rises. outcomes from spending. In this respect, the strengthening of the "authorizing environment" adjusting the composition to spending to greater for communities in SP service delivery in recent emphasis over time on ex ante risk mitigation (i.e., years is encouraging. This is both cross-cutting, social insurance programs for the unorganized through reforms such as the Right to Information sector) and promotional programs (including Act, and program specific, such as the anticipated those linked to human capital formation, role of social audits in NREGA, and new roles for discussed in Chapter 4). community groups such as SHGs in delivery of increasing the flexibility of funding from the some SP services (e.g., running Fair Price Shops). centre to states in terms of how it can be used for However, there is no guarantee of "trickle down" different SP interventions, while strengthening to citizens in terms of awareness of their emerging the emphasis on spending outcomes. entitlements. The role of NGOs, media and other deepening a range of administrative and actors in this respect cannot be under-stated, institutional reforms in SP service delivery which as various political and administrative actors can contribute to greater expenditure efficiency. at local level may not have strong incentives Many of the operational reforms are discussed to promote such citizen-based accountability in Chapter 7. mechanisms. Notable examples such as MKSS in Rajasthan demonstrate the potential impacts of the notable exception is the area of social insurance, such partnerships. where the share of public spending can be expected to increase notably as india enters middle income status, and where the imbalance between social security D. ConCluSionS AnD spending on the organized and unorganized sectors would be expected to shift in favour of the latter. ReCommenDATionS In the medium term, international experience and social expectations in India would suggest significantly It is clear that social protection is in an exciting increased spending on social insurance for the large phaseintermsofpolicyevolution,financinglevels, uncovered share of workers. It will be important that this and institutional arrangements. At the same increased expenditure priority not be at the expense of time, the situation presents an unfinished agenda social safety net spending, but rather is financed from a which in some ways lacks a coherent vision for the combination of reallocation from non-merit spending, system which is informed by the socio-economic growth in aggregate fiscal resources, and contributions developments outlined in Chapter 1, and from workers themselves. In practice, expanding social institutional capacities. Looking at the financial security to the unorganized sector is likely to require not- and institutional aspects of SP programs insignificant public subsidies to incentivize participation, covered in this chapter, some conclusions and as one sees for example with RSBY and certain pension recommendationsareasfollows: schemes.313 In addition, effective social insurance programs can in part be expected to be self-financing (i) Financing of SP programs in two ways: (i) they should help control additional demands on the safety net that might otherwise arise broadly speaking, india's share of public spending due to factors such as population ageing; and (ii) on safety nets is reasonable, and indeed quite high international evidence suggests that effective social as a share of gdp by international standards of low security systems can actually contribute to growth in a 313 Forinternationalexperienceonmatchingsubsidiestoincentivizeparticipationinpensionprogramsforinformalsectorworkers,seeHolzmannetal.(2009). 1 Social Protection for a Changing India: Volume II variety of ways, from enabling higher risk/higher return in terms of coverage and scope of benefits productive activities to cushioning the impacts of as institutional capacity and fiscal space is growth-enhancing reforms for those who lose out from developed. The core types of insurance which them in the short run.314 GoI aims to expand include life, disability, old age Thespecificproposalofthisreportisthatcentral pension, and health, and the RSBY program is SP programs over time aim for a "3 + block" already an important step forward in this regard. strategy. This would involve 3 core CSS SP Chapter 5 gives suggestions on institutional, programs or "pillars", combined with an SP block financing, sequencing and other aspects of an grant from which states could finance other SP expansion strategy. programs - or supplement benefits under the core beyond the three "pillars", states could receive an pillar programs - more tailored to the poverty and additional transfer and implement state-specific vulnerability profile of the individual state. This sp interventions. How this is programmed could would also involve significant expansion in urban vary according to state-level priorities, and include areas. This could promote both a more coherent interventions such as livelihood support of different and less duplicative SP system, but also give forms, targeted housing, interventions to incentivize states more leeway to adapt the SP policy mix to use of basic social services (such as CCTs), nutrition and/ the needs of the poor in individual states in light or early childhood care, specific urban SP programs, of available economic opportunities. The three or other options as proposed by states. A secondary core pillars proposed are: benefit of such an approach would be strengthening of a major social assistance program. The obvious complementarities between CSS and state-sponsored candidate for this is a significantly reformed PDS, schemes in order to control unnecessary duplication. merged for specific groups with existing social A common core national SP system under the three pension programs.315 Chapter 3 gives more details pilars could promote portability of basic entitlements, on the options for reform of PDS proposed by this and be increasingly useful as mobility of workers and report, with a preference for a predominantly cash households increases. transfer approach. in terms of promoting both more effective spending a public works program, for which MGNREG on sp, the "3+block" proposal would allow more would be the building block, as well as piloting cross-program flexibility to states ­ or possibly expansion in urban areas. There are several districts - in deciding their sp expenditure priorities, benefits of a reliable public works program at this while still maintaining a common national core sp stage of India's development: (i) by its demand- system. It would also allow for greater adjustment in driven nature, it can be responsive to shocks in a light of poverty levels, key vulnerabilities, etc. This could way that longer run programs typically can not. be done in a variety of ways, possibly using a menu In this way, it functions as a "quasi-insurance" approach to SP programs, and a flexible form of social program for the extended period during which protection block or matching grant which consolidates more structured insurance is expanded to the resources from existing SP CSS.316 This is an approach unorganized sector; (ii) the positive targeting which has received increased attention in India in recent outcomes of self-targeted works; and (iii) the years as policymakers seek to get greater impacts from potential for multiplier effects from asset creation SP spending. Given current financing channels from and community mobilization distinguish public the centre to states in India, a more flexible granting works from other SP programs. mechanism for SP programs to states could take abasicsocialsecuritypackageforthoseoutside different forms ­ as a more "bundled" anti-poverty CSS, the formal sector which could be expanded as Additional Central Assistance (ACA) along the lines 314 SeeWorldBank(2006c),foradiscussionofinternationalevidencethatthereisnonecessarygrowth-equitytrade-off. 315 ThiswouldbesimilartotheChineseurbanandemergingruralsocialassistancesystem,whichisbuiltaroundthe"dibao"programwhichprovidescashbenefits tothepoor,andhasadditionalprovisionforspeciallyvulnerablegroupssuchasdisabledandunsupportedelderly.SeeWorldBank(2009b). 316 SeedeNeuborg(2002)foradiscussionofthestrengthsandweaknessesofdifferentblockandmatchinggrantmechanismsinthecontextofSPprograms. CHAPTER-6: Financing and Institutions for Social Protection 1 of programs like Pradhan Mantri Gramodaya Yojana a third step could be the block grant for SP (PMGY), or through a more fungible realignment of programsoutlinedabove. Gross Budgetary Support shares between the states and an alternative approach which could be central Ministries which control CSS.317 implementedevenintheabsenceofmoreserious there are pros and cons of different possible methods financing reforms is the creation of an incentive of more flexible central funding for sp programs. fundforstatesforSPprograms.An SP incentive Firstly, there would be natural reluctance from central fund could either be stand-alone or a window of Ministries to simply transferring SP CSS resources broader proposal for a decentralization incentive to Normal or even Additional Central Assistance, as fund. Alternatively, a central innovation fund for evidenced by the failure to transfer several CSS to states social protection could be created which could as per the 1999 Planning Commission list of schemes. allow states to access central funds on a proposal- More specifically, as Normal Central Assistance is based driven basis for innovations in SP program on population and poverty under the Gadgil formula delivery, or experimentation with new initiatives. and not on performance, incentives for states to improve performance on specific SP programs would (ii) institutional roles for SP programs be relatively weak. For Additional Central Assistance, there are examples of linking transfers to state reform the most fundamental institutional challenges performance, such as JNNURM etc. However, these in sp css continue to be delineating clear lines of requirements have not been strongly enforced in accountability in service delivery, and supporting the practice. There is also no obvious mechanism for the authorized actors with adequate staff and finances. Planning Commission to monitor performance of specific This will require first and foremost greater proactivity on CSS under NCA, and for ACA only where responsibility is the part of states to approve policies and put into practice given to a central Ministry. the PRI/ULB decentralization provided for under the 73rd and 74th constitutional amendments. With notable overall, an improved transfer system for sp css exception of Kerala, and incomplete examples such as would need to balance need, capacity and fiscal and Rajasthan and Karnataka, most states have yet to define administrative effort of states. This could involve the framework for decentralized service delivery in a several elements, including: sufficiently operational manner. a first step towards such a system would be this would need to be followed by a process-intensive movingtotowardsamoreoutcomebasedfunding reconciliation of central guidelines, state-level method. Such an approach would, however, stances on service delivery decentralization, and require a clearer set of outcome indicators against capacities at sub-state levels to perform the required which state performance could be measured, and implementation functions in sp programs. This is not a stronger planning function at state and district a process which will generate a single "right" answer levels. This approach is one that is already being on the assignment of activities to different actors at used in India in some donor programs which are different levels of the system. What it could usefully results or outcome-based. achieve however is a more considered assessment by asecondstepinsuchaprocessmaybeallowingfor both centre and states of what is the chain of activities sanctioningofallocationsforstateswhichstretch from top to bottom required to deliver effective SP across two to three budget years, which could programs, and of the realistic potential of different lessen current incentives to spend SP allocations actors to deliver on their proposed responsibilities. This inefficiently (or simply parking them in accounts), could in time improve the alignment of functions, funds and also assist with programs which are subject and functionaries in SP programs. Such efforts would to cyclical demand - such as public works ­ which need to focus in particular on elements of the service is not presently synchronized with the budget delivery chain at block level and below. The process planning and release cycle. could involve several steps: 317 SeeSaxena(2006)forahistoryofcentraltransfersforAPPsincethe1960s. 1 Social Protection for a Changing India: Volume II "unbundling"eachcoreSPCSStoitsconstituent guidelines. Where such concurrent responsibilities functions or services and specific activities in are apparent, further dialogue between the states different programs, as has been done for the and centre would be needed to agree whether MGNREG and RSBY guidelines. While implicit in CSS guidelines need to provide more flexibility to guidelines of some other schemes, such a simple states on institutional arrangements for programs, exercise would be a useful building block for a or whether states need to be encouraged to move principled division of functions across levels. more actively on state-specific assignments which forallstates,producingaconsolidatedmapping are consistent with CSS programs (with Kerala of the current assignment of sectors and SP providing a good example of such a proactive program-specific functions, both de jure and role by states). Such discussions could be guided de facto. This exercise would clearly identify by both public finance criteria for assigning cases where concurrent responsibilities arise for responsibilities, and basic management principles programs where DRDAs and/or PRIs have specific for enhancing accountability.318 Both of these are responsibilities assigned to them under CSS discussed in Box 6.2. Box 6.2: guiding principles for assignment of institutional responsibilities Assigning institutional responsibility for specific services and programs, and within programs for specific activities, naturally involves a number of trade-offs. One difficulty in SP CSS to date has been that the principles for assignment, and the policy trade-offs involved in specific assignments of responsibilities, have not always been explicit. In order to move over time to a more principled assignment of institutional roles, both public finance and accountability criteria provide useful guidance. From a public finance perspective, some guiding criteria for assignment of institutional jurisdiction for service provision are: economiesofscaleorscope: Economies of scale or scope are likely to apply in SP programs in particular to issues such as record keeping, beneficiary numbering systems (e.g. unique household identifiers are most sensible when consistent across programs and space), and certain aspects of delivery such as management of targeted credit. scope of externality from the service or activity: To the extent that a service does involve externalities, the general rule would be to make the jurisdiction large enough to internalize the externality. In the case of SP programs, a number will have rather limited externalities beyond the household or village, while others such as public works potentially have a significant element of externality (e.g. small stretches of road which may if coordinated link villages). equity: This is clearly a very important welfare economics consideration for SP programs. heterogeneityofdemand:To the extent that SP-related needs of different states and sub-state areas vary (as they inevitably will in a country such as India), institutional assignments which allow for greater tailoring of priorities to local needs are attractive. From an accountability perspective, there are five key features of accountability relationships, all of which interact with each other, and are: delegation: This is simply the explicit assignment of tasks to designated agents. While obvious, it is precisely the lack of clarity in such assignments within states ­ due both to their lack of action and in some cases conflicts between CSS and state assignments ­ that is one of the major institutional challenges for SP CSS. finance: This is aligning finances with the delegated responsibilities, a task which is fraught when delegation is unclear or concurrent to start with. A second constraint in the case of SP CSS is ensuring that those in control of funds have sufficient capacity to do so in a transparent manner. performance:This is the doing of the assigned task. Given the many steps involved in delivery of most SP CSS, this requires a disaggregated breakdown of steps in the delivery chain to avoid overly lumpy assignment of many tasks to single levels of the system. informationaboutperformance: This is an area where many SP programs have been weak, with both weak linkage between information and actual performance/outcomes, and frequent reliance for monitoring on levels of the system which are too removed from the point of implementation to perform more than perfunctory monitoring. enforceability: Information on performance only matters if there are consequences for poor or good performance. How the system aligns enforceability with information on performance is this key. Sources: World Bank (2006d); World Bank (2004). 318 SeealsotherecommendationsoftheEmpoweredSub-CommitteeoftheNationalDevelopmentCouncil(NDC)onFinancialandAdministrativeEmpowermentof thePRIsJuly2008. CHAPTER-6: Financing and Institutions for Social Protection 1 havingdeterminedanappropriateassignmentof at both central and state levels, formation of functions to different levels, a fuller assessment an inter-departmental Task Force or Authority is needed of the personnel and financial for Social Protection, which would promote implications of carrying out required functions. coordination across programs targeted towards Where enhanced personnel capacity is needed, similar populations, and promote more coherent this would then need ­ as is the case with strategy development on the medium term policy MGNREG ­ to include the financing of needs mix and priorities in social protection. The cases in program costing, and the capacity building of Mission Convergence in Delhi and the Safety requirements at different levels. Nets Authority in UP provide interesting models, graduallyinstitutionalizingrolesforcommunities which could be adapted by other states. inbothselectedelementsofCSSdelivery,andin forprogramplanning,severalinitiativeswouldbe community oversight of program functioning. useful, including: (i) earlier notification to states Again, the model of MGNREG provides a and DRDAs of estimated funding envelopes for useful precedent. This is also likely to involve programs for the following fiscal year in order partnerships with NGOs to build local capacity to facilitate lower level planning and budget for effective social audit, and the more general management; and (ii) ensure that states which obligations of the state for disclosure of program have not already done so appoint District Planning information, as has proved effective for example Committees, and ensure adequate financing for in Rajasthan. The RSBY scheme also provides technical support to DPCs. useful precedents on the role of NGOs and other associalsecurityexpands,theroleoftheprivate grassroots organizations in SP service delivery sector (as insurers) and a range of member- and demand side mobilization. based organizations such as MFIs, NGOs and taking lessons from the emerging experience in workers associations is already becoming India and beyond in public-private partnerships increasingly important, and demands new in different aspects of SP service delivery. modes of engagement and partnership form Increasingly, the private sector has been the public sector. The biggest constraint on playing different roles in the Sp service delivery expanding social security to the unorganized chain. RSBY is the most advanced example, sector has been developing delivery mechanisms with involvement of private insurers, private which can deal with the transactions costs of hospitals, private sector smart card providers and reaching unorganized workers. This requires outsourced software development, as well as intensive engagement with intermediary roles for grassroots organizations noted above. partners between government/insurers and However, many other programs ­ particularly unorganized workers, as well as learning from at state level ­ have been experimenting with efficient insurance distribution channels of outsourcing of different functions.In some cases, public and private insurers. The RSBY provides this involves partnerships on technology, such as an excellent model of partnership, as do more the role of TCS in managing MGNREG databases localized initiatives such as UTI partnerships in AP, or smart card provision by commercial firms with both state governments and organizations in a number of pilots. In others, there is a more such as SEWA, and the roles of trade unions and involved role for the private sector, such as in employer organizations in some of the welfare Bihar where rural informatics service centers for funds around the country. RD programs at block level have been assigned for M&E, develop a disaggregated picture of following a tender process, in which for-profit and not-for-profit organizations operate the potential and capacities at different levels of information and database management of RD the system for monitoring and evaluation, and programs on behalf of the state Government. align program guidelines in that light. The proposal to establish a national Independent in addition to the above needs on institutional roles, Evaluation Office is a welcome step with respect specific suggestions include: to M&E. 1 Social Protection for a Changing India: Volume II CHAPTER Program Administration Issues in Social Protection 7 The "Nuts and Bolts" Chapter­7 Program Administration Issues in Social Protection The "Nuts and Bolts" the most common mantra in the social protection given india's technological and human capital field is "india has great laws and policies, but poor capacity. However, in nearly all areas, there are implementation". This section looks at some of the emerging models of good practice which suggest that "nuts and bolts" issues in SP program implementation, getting administrative systems in place which increase and attempts across programs to understand some of accountability and transparency is a realizable goal, the common administrative problems, several of which given sufficient political will and a set of institutional contribute to the coverage and targeting outcomes incentives which reward good practice. One of the key observed in Chapters 3, 4 and 5. The main operational challenges in this regard is achieving closer alignment issues covered in this chapter are: of the incentives of the centre, states and sub-state programawarenessandoutreach levels for improved program administration. This cannot be separated from allocation of institutional theapplicationsprocessandeligibility responsibilities, discussed in Chapter 6. determination publicexpendituremanagementissues paymentsystemsforbenefits A. PRogRAm AwAReneSS AnD recordkeeping ouTReACh monitoringandevaluation. although several of the main pillars of the sp overall, the basic "nuts and bolts" of program system in india have been in place for some time, most administration and procedures in most states are programs have been subject to periodic changes in far below the standards that could be possible policies, implementation arrangements, schemes CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 11 names and other features. Some of the changes media campaigns announcing the schemes and their are quite fundamental (e.g., moving to a targeted key elements. Increasingly, some program information PDS; guaranteeing the right to 100 days employment is available on the web. However, there is rarely a well- under MGNREG), while others are either a product developed outreach strategy which focuses on how of consolidation efforts (e.g., SGSY combining IRDP different audiences receive information, and what types and several related programs under one roof ), or of of information they most need either to participate in repackaging for political or administrative purposes schemes or perform their functions in program delivery. (e.g., EAS and JRY being superseded by SGRY). In addition, penetration of messages below the block level is often lacking, and in any event in formats mainly the frequent shifts in program design imply significant suitable for literate populations. demand for awareness and outreach efforts on the part of government. Despite this need, CSS SP programs evidence from four states indicates that there is have typically relied on a rather top-down approach to considerable variability in awareness of programs, dissemination of new/revised program information. and in some cases across states with respect to This has been a standard approach of central and individual programs. This can be seen in Table 7.1 state governments issuing implementing orders and and Table 7.2, which presents findings from Orissa, MP instructions, often with detailed program guidelines. and Karnataka, and additional survey information from These are distributed through line agency and general Jharkhand in the state-specific table. With respect to the administrative channels, in particular to Collectors and cross-program awareness on the three states, the main BDOs. This may be complemented by rather traditional findings are: Table 7.1: Awareness about programs, orissa, madhya Pradesh and karnataka (%) quartiles 1 2 3 4 All Cash Transfer Programs Targeted Indira Awaas Yojana 68.0 71.1 68.9 63.6 67.9 National Old Age Pension Scheme 59.6 62.8 57.1 54.7 58.6 Widow/disable pension 60.8 64.0 58.6 54.7 59.5 Universal Rural Education Scholarship 29.7 29.5 28.1 33.4 30.2 In-Kind Transfer Programs Targeted Public Distribution System 86.1 90.5 92.9 95.0 91.1 Antyodaya Anna Yojana 34.0 33.3 32.8 31.4 32.9 Universal Integrated Child Development Services 26.7 33.6 37.0 37.3 33.6 National Mid-Day Meal Scheme 67.7 77.4 78.1 68.9 73.0 Free text-book 60.2 67.6 71.6 68.6 67.0 Free hostel 18.0 31.0 31.1 33.1 28.2 Free uniform 51.2 59.8 62.7 61.5 58.8 Workfare Programs (Self Targeted) Sampoorna Grameena Rozgar Yojana 25.0 33.3 30.2 29.6 29.5 Food for work 28.5 29.8 25.7 26.0 27.5 Subsidy Based Livelihood Programs Targeted Swarnajayanti Gram Swarojgar Yojana 10.2 17.0 18.9 21.3 16.8 Source: Dev et al. (2007), based on the 2006 three state SP survey. Q1 is poorest and Q4 richest, with ranking based on wealth index. 12 Social Protection for a Changing India: Volume II Table 7.2: households aware of the safety net programs by social groups, orissa, karnataka and mP combined (%) Programs overall SC ST non SC&ST Cash Transfer Programs Targeted Indira Awaas Yojana 67.92 72.79*** 75.59*** 61.03 National Old Age Pension Scheme 58.55 56.25 62.09 57.25 Widow/disable pension 59.51 56.99 62.56 58.61 Universal Rural Education Scholarship 30.16 29.41 23.93*** 34.44 In-Kind Transfer Programs Targeted Public Distribution System 91.08 92.28 86.49*** 93.5 Antyodaya Anna Yojana 32.89 34.56 34.36 31.27 Universal Integrated Child Development Services 33.63 33.46 37.2** 31.42 National Mid-Day Meal Scheme 73.01 77.21* 73.46 71 Free text-book 66.96 72.06 62.56* 67.67 Free hostel 58.78 63.6 55.21 59.06 Workfare Programs (Self Targeted) Sampoorna Grameena Rozgar Yojana 29.5 27.21 38.63*** 24.62 Food for work 27.51 21.32 33.89*** 25.98 Subsidy Based Livelihood Programs Targeted Swarnajayanti Gram Swarojgar Yojana 16.81 13.6 21.56*** 15.11 Source: Dev et al. 2007. ***=significant a 1% level; **=at 5% level; *=at 10 percent. as would be expected, awareness is highest for betweenstates(notshowninthetable),Orissahas PDS(thoughnotAAY,forwhichawarenessisquite the highest level of program awareness overall low). This is not surprising given the program's relativetoMPandKarnataka. age and widespread coverage of FPS, and the result holds for all three states. awareness across social categories of major programs is not dramatically different, but for several key atleastthreeprogramsaresurprisingintermsof programs is somewhat higher for st, except in orissa therelativelylowlevelsofawareness­SGRY/FFW, where it is notably lower for nearly all programs. SGSY,andICDS.The low level of ICDS awareness Table 7.2 presents results across the three states broken is particularly striking given its objective of down by SC, ST and non-SC/ST. For most programs, universalism and coverage of anganwadi awareness among SC is not notably different, and the workers. difference is statistically significant only for IAY. However, overall,thereisrelativelylimitedvariationacross for ST, awareness is higher and statistically significant quartilesofthedistributioninprogramawareness, for several programs, including IAY, ICDS, SGRY/FFW but awareness is more progressive for targeted and SGSY. Unfortunately, this finding does not hold for than untargeted programs (with the notable Orissa, where ST program awareness is generally lower, exception of SGSY). While the top half of the in contrast to the overall finding of higher program distribution has somewhat higher awareness in awareness levels in Orissa. most programs, the differences are not dramatic, and awareness of targeted programs is generally at the village level, there are several factors slightly higher. associated with higher awareness, but the relative CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 importance of different factors varies notably it is necessary also to explore in more detail what between karnataka and the other two states. Table 7.3 "awareness" means, and the implications of different reports results on village level determinants of program levels of awareness for future outreach strategies. awareness, both grouped for the three states, and There is not much good survey-based evidence on this comparing the states on key indices of social capital, question, but field studies on awareness of MGNREG PRI functioning, and women's participation. Some key are instructive, and point towards the gap between results are: general knowledge of the existence of programs, and knowledge of the specific elements of entitlements that would facilitate effective participation in schemes Table 7.3: village level determinants of program by the poor. This can be seen in Table 7.4. Even in 2006, awareness, orissa, karnataka and mP, 2006 awareness of the MGNREG in different states was highly Coefficients variable, but overall reasonable in light of the young (Constant) -0.03 age of the scheme, and exceptionally good in some Female public participation (e.g. elections, GS 0.48*** cases. While Gujarat and Jharkhand were laggards on meetings) awareness, the achievements of AP, Chhattisgarh and Female empowerment index 0.23* Bihar were impressive. Subsequently, field studies that Presence of NGO in village 0.40** have revisited sample locations in Chhattisgarh and Source: Dev et al. 2007. See Annex 2 for details of construction of PRI and Bihar between 2006 and 2007 indicate that awareness, gender-based indices. ***=significant a 1% level; **=at 5% level; *=at 10 percent. though still largely low, has increased over time.319 across the three states, the factors associated Table 7.4: overall awareness of mgnReg, various withhigherawarenessincludestatusofwomenin mgnReg states, 2006 thehousehold,presenceofanNGOinthevillage, Share aware of mgnReg in and women's participation in public affairs. This State mgnReg districts (%) can be seen in Table 7.3, which reports the factors Andhra Pradesh 97.5 which were statistically significant as village level Bihar 62.5 determinants of awareness about all safety net Chhattisgarh 69.3 programs. In particular, the female participation Jharkhand 28.9 in public life (though GS meetings, elections Gujarat 38.8 etc) is highly significant in determining village- Madhya Pradesh 45.0 level awareness of SP programs. Also, female Sources: CBGA 2006 for Chhattisgarh, MP, AP and Jharkhand; IHD 2006 for empowerment index (which reflects women's Bihar; CDA 2006 for Gujarat participation in household decision making about a range of key factors, and their autonomy in despite the generally positive level of basic program social and economic behaviour) has a significant awareness, awareness of specific important elements relationship with village level awareness of SP of mgnreg entitlements and functioning is notably programs. lower in a number of cases, and even general in contrast, a range of other variables had no awareness is generally lower among lower castes. significantexplanatorypowerasdeterminantsof The results in Table 7.5 provide a more nuanced picture programawareness,including existing social and of program awareness, indicating that while MGNREG economic infrastructure, share of small farmers "brand awareness" is satisfactory or good for such a new and landless, functioning of the PRI, social scheme, awareness of some key elements of the program composition of the village, and level of trust in remain to filter down as effectively. For example, while local institutions. knowledge of the 100 day guarantee is relatively strong awareness of safety net programs is lower in (and reflective of the emphasis of Information and wealthier villages. Education Campaign (IEC)), the understanding of the 319 CBGA(2006,2007)andIHD(2006,2007). 1 Social Protection for a Changing India: Volume II Table 7.5: Awareness of specific elements of mgnReg entitlements by state, 2006 100 days minimum wages Role of Gram Basic facilities at State employment to be paid Sabha/gP work site work within 5 km Andhra Pradesh 93.8 68.8 2.5 48.8 8.8 Chhattisgarh 53.7 35.1 3.9 9.3 13.7 Madhya Pradesh 44.2 35.0 2.5 18.3 5.8 Jharkhand 10.0 8.9 1.1 1.1 2.2 Gujarat 38.8 NA 37.7 23.8 NA Bihar 77.7 44.0 0.8 4.5 26.5 Sources: CBGA (Chattisgarh, MP, AP and Jharkhand); IHD (Bihar); CDA (Gujarat), all 2006 role of GPs and gram sabhas in scheme functioning is more local posting of program information than the very low, as could be to some extent expected given past, and in principle makes information on program the innovation involved for such a scheme. The survey performance available to members of the public who of MGNREG worksites in six northern states by Drèze pay a small processing fee. This can be a useful tool in and Khera (2009) indicates that fewer than half of the awareness raising, and experience in some states (e.g., surveyed households knew about the entitlement to Rajasthan) has already demonstrated the potential of hundred days of work and minimum wages. Some the Act in circumstances where effective third parties studies note that there is little awareness that this is a such as NGO are able to navigate the system to access universal right for rural households and not limited to information. However, to date the RTI Act has been specific groups. In consequence, potential applicants more a vehicle for tracking down program information can be excluded based on certain characteristics rather than a motivation for program administrators to (e.g., widowed, single women headed, elderly etc) or be more proactive in sharing basic program indicators documentation (e.g., ration card). with the public in easily understandable formats. This will presumably be part of the agenda of central and both mgnreg and many ngo and donor programs state RTI Commissioners in the future. demonstrate that innovative methods of generating public awareness of programs are often necessary. As noted, Government IEC for programs remain typically top-down and rather traditional in their communication B. The APPliCATion methods. In contrast, many NGO schemes (or in some PRoCeSS AnD eligiBiliTy cases, public schemes partnering with NGOs) use a variety of communication methods which are more DeTeRminATion likely to be digested by target populations. These across most programs, a common complaint of include a wide range of strategies, including reliance both current beneficiaries and applicants/potential on women's and community groups with grassroots beneficiaries is that the administrative complexities penetration, and engagement with panchayat or and attitudes of officials are significant barriers to other village level notables including religious leaders, program participation. This of course is not a problem jati panchayat elders, et al. In addition, the modes of unique to SP programs, and also an area where evidence communication are much more diverse, and often better beyond the anecdotal is often lacking. The section suited to the needs of non- or low literate populations. below presents new analysis for selected programs on They include tools such as street theatre and puppetry, the reported importance of such bureaucratic factors visual communication of key messages, imaginative use in accessing programs. While much more research is of media (e.g., short human interest spots built around needed, it suggests that the widespread anecdotal the program and its benefits) etc. evidence is borne out by data. another important new factor in awareness raising sp programs in india rely heavily on the bpl card is the right to information act. This mandates much as a principle mechanism providing preliminary CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 evidence of eligibility, with both pros and cons for looking at why households do not have ration cards, program administration. A core common determinant the dominant reason nationally is bureaucratic of individual or family eligibility for benefit in most difficulties. However, there are major cross-state SP programs in India is availability of the BPL card. variations in main reasons for not having cards. Such setup helps to control the administrative cost Table 7.6 presents results across states. While bureaucratic of multiple programs by taking the burden of initial problems are the main reason nationally, there are eligibility screening away and allows the administering sharp differences across states in the importance of agencies to focus on other core tasks. This arrangement this reason. For a group of very poor states ­ including is in contrast to experiences of many other countries Bihar, Jharkhand, Orissa, and Chattisgarh, but also West where programs often operate own comprehensive Bengal ­ bureaucratic difficulties in accessing cards eligibility determination mechanisms. Given such high are much more important, though in other poor states dependence on the common external factor of eligibility such as UP, MP and Rajasthan, less specific reasons determination, ensuring efficient operation of that dominate. In contrast, for richer and southern states, external mechanism becomes crucial to the success of "other" is the dominant reason for no card. Overall, the the collective effort of poverty alleviation of various SP importance of "other" as a reason for no ration card is programs. A common limitation of the BPL mechanism, high, and requires more detailed research, particularly however, is significant element of inertia in updating the for states such as Maharashtra, TN and Karnataka where list of eligible families, as it remains part of the census it is easily the dominant reason. Also of interest from an operated only once every five years. The process of card administrative viewpoint is that around 18 percent of issue and record-keeping itself needs further analysis as household report losing their card or not being able to it exhibits serious deficiencies. get a new card after moving as the main reason for not Table 7.6: main reasons for not having ration card by state, 2004/05 (% households) moved but card States not needed Card lost Bureaucratic not transferred other Andhra Pradesh 12.8 3 29.7 10.4 44.2 Assam 1.3 42.8 31.2 11.6 13.2 Bihar 0.7 9.4 84.4 1.2 4.4 Chattisgarh 8.4 7.2 65.9 5.5 13 Gujarat 16.7 16.7 16.7 16.7 33.3 Haryana 12.6 10.2 9.8 17.3 50.2 Himachal Pradesh 25.6 3.2 17 16.7 37.5 Jammu & Kashmir 11.1 4.4 16.8 8.2 59.4 Jharkhand 8.1 7.3 67.6 3.4 13.6 Karnataka 0 8.7 6.4 17.4 67.4 Kerala 4.7 7.9 10.8 21.1 55.5 Madhya Pradesh 20.4 7.6 21.8 21.6 28.6 Maharashtra 4.7 2.8 13.3 5 74.2 Orissa 15.6 7.3 68 3.3 5.7 Punjab 7.8 21.4 28.2 16.7 25.9 Rajasthan 16.3 9.3 18.4 10.2 45.7 Tamil Nadu 11.7 0 0 0 88.3 Uttar Pradesh 13 9 30.1 4.8 43.1 Uttaranchal 27.6 4.2 32.4 8.4 27.4 West Bengal 2 7.2 65.5 5.6 19.8 All India 9.5 8.5 40.8 9.6 31.6 Source: Ajwad (2006) based on the 2005 IHDS data. 1 Social Protection for a Changing India: Volume II having a card. Finally, the share of households reporting Table 7.8: main problems in getting social pension that they did not need a card is under 10 percent, and sanctioned, Rajasthan, 2006 surprisingly, this share is higher than average in at least Primary Secondary two poor states, MP and Orissa. problem problem looking at reasons for no ration card across the Understanding the eligibility 13.9 5.3 rules distribution and social categories, the poorest and Time for getting documents 16.3 17.9 st households appear to have particularly serious problems with bureaucratic difficulties. Table 7.7 Cost of getting documents 12.7 21.8 presents results by wealth level, rural/urban, and social Proving meet criteria 3.0 19.1 category. While the difference across wealth level and Informal payments 1.3 3.6 caste in the share of households not needing a card is as Transport costs 13.0 30.2 expected, more notable is the sharp differences between Other 0.9 2.1 lower and upper income groups in the share reporting No difficulty 38.9 NA bureaucratic difficulties as the main reason for having no Source: Dutta 2008, based on Rajasthan social pensions survey, 2006. ration card. This also applies to rural households relative to urban, and notably for ST households. and indicate that just under half of applicants faced survey results on social pensions from several some form of bureaucratic difficulty, rising to around states provide further insights into the nature of two thirds when secondary reasons are included. difficulties in accessing programs. While these results The time and financial costs of documentation are of are obviously state and program-specific, they are particular note, as are the transport costs associated nonetheless instructive. By way of illustration, Table 7.8 with application. The findings on the demands of below cites the main problems encountered by those eligibility documentation are supported by previous currently receiving social pensions in Rajasthan in studies in other states, including AP, Kerala and Orissa, accessing benefits. Broadly the first five rows can be one of which refers to applicants for social pensions "lost considered different types of program related difficulty, in a bureaucratic maze which they find impenetrable".320 Table 7.7: main reasons for no ration card by wealth, location and social category, 2004/05 moved but card not needed Card was lost Bureaucratic not transferred other Poorest 4.3 9.5 53.6 7.9 24.7 Q2 5.9 8.3 48.6 8.5 28.7 Q3 6.4 8.7 41.9 8.5 34.5 Q4 14.7 7.9 29.2 9.9 38.3 Richest 21.7 6.7 23.0 8.5 40.1 Rural 6.2 9.6 46.7 6.3 31.3 Urban 15.5 6.6 30.4 15.5 32.4 OBC 7.5 8.1 44.1 8.6 31.5 SC 5.9 10.6 43.2 8.6 31.7 ST 9.3 8.4 52.7 6.9 22.6 Other 14.5 8.7 26.6 13.4 36.8 All India 9.5 8.5 40.8 9.6 31.6 Source: Ajwad 2006 based on 2005 IHDS data. Note that the other social group category excludes Brahmins. 320 See Nayak et al. (2002) for AP, and Centre for Management Development2000forKerala,thelatterfinding"non-cooperationby officials" on certification as a significantbarriertoaccess.SeealsoInternationalManagementInstitute2001forstudyinOrissa,whichfoundthecomplexityofdocumentaryevidencebeing abusedbyofficialsforrent-seekingpurposes. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 On a more encouraging note, just under 40 percent of important to stress that the term "middleman" includes successful applicants in Rajasthan reported no difficulty a range of intermediary relationships. Some ­ such as in accessing the pension. intermediation by NGOs or SHGs ­ may be quite positive processes. At the other extreme, intermediaries who insights from those whose applications failed or were exploit their relations with officials for earning "speed deterred from applying are also of interest. Among money" or contractors on public works who get people those in Rajasthan not applying the complexities of the onto worksites for a commission are less desirable. In application process were the deterrent for one third, between lie a range of social network relationships, and this was as high as 45 percent in Karnataka.321 for example with local MLAs who may exchange Among those who applied and were rejected, around intermediation for voter commitments. one quarter cited lack of personal connections to officials as the reason for rejection in Rajasthan and the prevalence of middlemen in accessing sp programs around 15 percent in Karnataka. Related to this was a is supported by evidence from the three state survey further 18 percent in Karnataka and a further 8 percent of orissa, karnataka and mp, which also provides in Rajasthan citing inability to pay a middleman as the insights into the relative importance of different types main reason for failure of their application. A further of middlemen. The results should not be interpreted factor of interest is the time taken to sanction benefits simply as malfeasance, but more as necessity (and for those who are granted them. Once more there is ability) to access intermediating officials and others. A considerable anecdotal evidence of significant delays few observations emerge: in sanctioning, or provision of work in the case of overall, reliance on panchayat officials is the workfare. The MGNREG attempts to address this issue most common form of intermediation, and for through the default option of an unemployment most programs is necessary for more than half of allowance in cases where work is not granted within beneficiaries. 15 days of a request, though in practice studies to date whiletherearenolargevariationsacrossprograms find no evidence of such allowances being paid.322 The Rajasthan and Karnataka social pension surveys inthemainformofintermediation,afewprograms also provide some insights on the question. In the are worthy of note: (i) AAY beneficiaries rely far case of both Karnataka and Rajasthan, the median more heavily on elected than administrative waiting period between application submission and officials; (ii) in child-related programs, the reliance sanctioning of the pension was 6 months, with 10 on elected or administrative official is relatively percent of current pensioners in Karnataka reporting lower and conversely, reliance on "other" forms of a wait of more than 2 years from the data of first intermediation higher; and (iii) SGSY exhibits an application.323 In both states, repeat applications by unusually high reliance on administrative officials eventual beneficiaries are also common. for accessing. in distributional terms, those in the top quartile the above discussion points to the potential importance are generally more likely than the poor to rely of middlemen in accessing benefits. One of the factors ontheintermediationofadministrativeofficials, often mentioned in the literature on SP programs is while the picture on the relative reliance on that the poor often take the help of various agents to elected officials between rich and poor is more access programs including such things as obtaining a mixed. BPL card.324 In addition, there is considerable variation across programs in the importance of middlemen, using village-level information on the role of with less reliance for programs such as NOAPS and middlemen (or other "contacts") in accessing programs, more for programs such as IAY and public works. It is it is possible to look at what factors influence the 321 SeeDutta(2008)andMurgai(2007). 322 SeereportsfromvariousstatessummarizedinChapter4. 323 Notethatinbothcases,asignificantsharereportstheneedformorethanoneapplicationbeforesanctioning,e.g.,inRajasthanamongcurrentbeneficiaries,just under30percenthadappliedmorethanonceforbenefits. 324 Nayaketal.op.cit.isagainausefulsourceontheroleofmiddlemenandhowitdiffersbetweenprograms. 1 Social Protection for a Changing India: Volume II dependence of poor households on middlemen to Table 7.9: Determinants of use of middleman to access programs. The results are reported in Table 7.9 access benefits, 2006 and Table 7.10. Overall, female literacy, women's general variables Coefficients status in the household, and households "trust" in public PRI functioning index -0.255 institutions significantly reduced the dependence on (-1.483) middlemen. The dependence on middlemen was also Economic infrastructure 0.124 higher in the relatively poorer state of Orissa than in the (0.696) other two states. Trust PRI -0.131 (-0.445) a further common claim with respect to sp programs Trust Officials -0.187 is bribery in order to access (and in some cases stay (-0.597) on) programs.325 as with many issues in program Trust Groups 0.540*** administration, this claim is widely made from field (2.796) and anecdotal evidence, but less commonly supported Total Female Participation 0.129 by data. However, closer studies of the issue have (0.596) confirmed the prevalence of direct and indirect bribery, Empowerment 0.286 but also strong variations across programs in the scale of (1.247) the problem. One can identify several forms of corrupt Control on assets -0.275 (-1.524) behavior: (i) individuals ineligible for benefit trying to Average Index 0.083 gain access to public resources; (ii) otherwise eligible (0.295) individuals trying to bribe in order to move up the priority Ratio of female and male literacy 0.769** list for benefits; (iii) eligible individuals trying to overcome (2.183) bureaucratic obstacles in accessing universal benefits. % female literacy -1.048** While (i) and (ii) are problems that are intrinsically difficult (-2.371) to address, (iii) is something that should be easier to deal Migrated -0.139 with by means of public awareness campaign, improved (-0.884) accountability, rationalized processes, and possibly Presence of NGO in village 0.050 investments in IT. (0.274) Social Composition of village (Herfindal) -0.181 a finding of previous work and supported by research (-1.077) for this study is that programs which are designed Source: Dev et al. 2007. Notes: t-statistics in parenthesis. See Annex 2 for around larger single or lumpy payments appear variable definitions. Table 7.10: Reliance on intermediaries to access benefits by program and intermediary types, orissa, karnataka and mP, 2006 officials elected leaders others q1 q2 q3 q4 All q1 q2 q3 q4 All q1 q2 q3 q4 All IAY 23.4 25.0 31.0 41.2 28.2 68.1 63.9 69.0 58.8 66.2 8.5 11.1 0.0 0.0 5.6 NOAPS 18.8 25.0 40.0 33.3 28.1 62.5 68.8 50.0 66.7 63.2 18.8 6.3 10.0 0.0 8.8 Widow/ 25.0 15.0 22.2 16.7 20.0 65.0 65.0 66.7 50.0 63.6 10.0 20.0 11.1 33.3 16.4 disabled pension Stipend 28.6 33.3 38.9 33.3 33.3 52.4 38.9 27.8 58.3 43.5 19.1 27.8 33.3 8.3 23.2 PDS 16.1 27.6 27.9 24.1 24.0 70.4 63.8 66.7 71.3 68.1 13.5 8.5 5.4 4.6 7.9 AAY 4.0 15.0 13.3 0.00 9.1 88.0 80.0 80.0 83.3 83.3 8.0 5.0 6.7 16.7 7.6 ICDS 8.0 28.6 22.7 30.0 22.5 52.0 42.9 40.9 56.7 49.0 40.0 28.6 36.36 13.3 28.6 325 Thisisdistinctfromthelargerissueofleakageofprogramfundsorgoods,anddiversionofdifferentformswithinpublicspending.ForPDS,thisisdealtwithinChapter3. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 officials elected leaders others q1 q2 q3 q4 All q1 q2 q3 q4 All q1 q2 q3 q4 All Midday 8.0 20.0 14.0 6.3 12.0 62.0 40.0 54.4 72.9 57.5 30.0 40.0 31.58 20.8 30.5 meal SGRY 18.8 25.7 32.1 40.0 27.8 59.4 60.0 60.7 45.0 57.4 21.9 14.3 7.14 15.0 14.8 FFFW 17.1 13.8 12.0 29.4 17.0 78.1 79.3 76.0 58.8 75.0 4.9 6.9 12.00 11.8 8.0 SGSY 33.3 60.0 70.0 50.0 53.7 66.7 40.0 30.0 25.0 39.0 0.0 0.0 0.00 25.0 7.3 Source: Dev et al. 2007. more subject to demands for bribes by officials and Such findings are supported in part by GoI's their associates for accessing the program. More own evaluation of the program, and by field specifically: research from AP in the early 2000s, which previousstudiesandthisfindthatIAYistypically reported a bribe of around ` 6,000 per ` 20,000 subject to a bribe running into thousands of IAY benefit.326 Box 7.1 presents examples from rupees.The results in Chapter 3 on IAY average field work in three states for this report on the benefits are very consistent with field reports stories of IAY applicants which are illustrative of from beneficiaries of a bribe of ` 4,000-5,000. the situation faced. Box 7.1: Accessing iAy ­ The view from the field in three states Field research for this report, in addition to finding evidence of bribery in quantitative data, conducted detailed personal histories of the experience of villagers in accessing schemes. Some findings with respect to accessing IAY are presented below: Govind Vishwakarma from Madhya Pradesh was asked to apply for Indira Awaas Yojana. Govind Vishwakarma contacted the Sarpanch who asked him to apply and inform him, so that he can facilitate the process. He approached some employee in the office who demanded ` 500 for writing the application and other procedural expenses. He has given ` 400 to get the work done. Later the strenuous follow up of the application began and this continued for nearly two months. Then the Sarpanch demanded ` 5000 and he had to give that amount. Within a week's time, ` 20,000 was allotted to Govind Vishwakarma. He repaid the debt of ` 5000 from that and the rest of the money was utilized for the renovation of the house. Thus, he had to give bribes at two points which amounts for more than one fourth of the total grant. Three years back, Prameshappa from Karnataka took ` 32,000 as part of Credit cum Subsidy scheme to construct a house. He thought his savings clubbed with the loan amount would suffice the needs for the small house he wanted to make. And in the villages they say, it is during the bad phase in life one would embark on making houses. It was proved as an out and out truth for Prameshappa. He exhausted all the self-earned resources in the very initial phase of house construction and the loans came through after a long point of time. Bribing officials for getting it sanctioned is a practice and Prameshappa refused to do so. It delayed the process and his house is still not finished. Ghasi Bilputia is a widow, lives along with her daily laborer 18-year old daughter in Similiguda block of Koraput district of Orissa. Her house is in dilapidated condition with walls and cracked asbestos roofs. She has tried for IAY and applied for it. The village meeting discussed it and approved their application also. But nothing proceeded further. The neighbors tell her that if she can bribe the Sarpanch and the ward members, she would get a house. For somebody who is struggling for making the two ends meet, where is the big money that needed for bribing politicians and officials to get her housing grant cleared? Shamila Yadav is an unskilled agricultural worker from Bordai village of Seoni district of Madhya Pradesh However, with their meagre incomes they are yet to build a comfortable shelter and they find it difficult to live in the broken down hut they presently live. They were told about IAY and Shamila applied for the scheme, being among the most eligible categories. She filed the application expecting that something would happen positively soon. However nothing has worked out even after her continued perusal. Later she met Sarpanch and as suggested by his accomplice, she has given ` 200 to him. He assured that it would be done positively in a month. However, even after ten months, Shamila Yadav did not get the benefits and the family continues to live in the same hut braving summers, winters and monsoons. Source: Dev et al. 2007, based on qualitative work for the three state SP study. 326 SeeCAG(2003a)IAYevaluation;Saxena(2006);Nayaketal.(2002)forAPestimate;Devetal.(2007).AlsoNair1999hasausefuldiscussionofsomeofthepolitical economyaspectsoftheprogram. 10 Social Protection for a Changing India: Volume II the other lumpy benefit is under SGSY. targeting systems, which are discussed in Chapter 8. Evidence of informal payments to officials of Some of the key issues include: intermediaries is more anecdotal in this respect. public expenditure and financial management Earlier studies by GoI of IRDP pointed to the issues prevalence of political influence in accessing paymentsystemsforbeneficiaries the program, and deduction of often significant "commissions" (e.g., in West Bengal, deduction programrecordkeeping of 10 percent of the loan amount by banks as monitoring of program performance and "charges"; in TN, "speed money" and "incidental evaluationofprogramimpacts. expenses" amounting to over 20 percent of loan amounts; in other parts of TN, request for bribes cited as the most common problem for (a) Financial management and SP CSS beneficiaries).327 An important element of this programs relates to the weak emphasis on repayment, with an informal agreement not to pursue an important element of financing of sp css relates repayment being worth more than the bribe to public financial management and accountability amount to beneficiaries.328 (pfma) of programs. As noted, implementation issues contribute to differential cross-state performance on SP incontrast,whilebriberyisconsistentlyreported CSS, and financial management is an important aspect inaccessingsocialpensions,thescaleofitrelative of this. The situation in this regard has evolved rapidly in to the cumulative flow of benefits appears to be recent years, with SGSY, SGRY, MGNREG and IAY central much lower. This can be seen from Rajasthan transfers now directly to DRDA/ZP level, bypassing and Karnataka data, where the average reported the state treasuries. This has some benefits in terms of payment to officials was only ` 100-200 in both avoiding the sometimes significant delays in approval states (variable by type of official), and just under and funds release by state treasuries. At the same time, ` 500 for the small share of beneficiaries reporting it introduces a new range of financial management and reliance on a middleman. However, also of interest accountability challenges. Some of these are due to lack is the share of rejected applicants who reported of clarity in FM/accountability procedures as they apply inability to pay a bribe/middleman as the main to DRDAs and PRIs, while others relate to limited FM reason for not being on the program (16 percent capacity as one goes down the system. of refused applicants in Karnataka, and over 8 percent in Rajasthan). on the issue of fm and accountability, previous work of goi and others has raised a number of areas for improvement in pem of css, including: 329 C. oPeRATionAl ASPeCTS oF states and districts are often not aware early enough of the quantum of funds likely to be PRogRAm ADminiSTRATion receivedfromGoIunderCSS.As a result, budget like many areas in service delivery in india, there are planning at both levels can be disconnected from a range of administrative challenges which have actual allocations. inhibited the efficient and effective implementation flow of funds problems have often been present of programs. These go beyond the broader issues of inthepast,drivenbyarangeoffactorsincluding institutional assignment of responsibilities for program GoIandstategovernmentapprovalprocesses,330 implementation, which are discussed in Chapter 6, and number of intermediaries through whom funds 327 PlanningCommission(2000);WorldBank(1998);TaylorNelsonSofresMODE(2001)regardingTN. 328 SeeNayaketal.opcitregardingAP. 329 SeevariousCAGreports,andWorldBank(2005a). 330 Biharisanexampleofstate-levelproceduresbeinganissue,withcumbersomeinternalproceduresdelayingfundreleaseby4-6months,withattendantpressure tospendquicklytowardstheendoffiscal.SeeWorldBank2005.However,thisfallswellshortofdelaysofupto24monthsinstatereleasesidentifiedbyCAGinthe SGSYschemeevaluation. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 11 flow, and failure to keep up with advances in ITC and lack of local level capacity to prepare financial for purposes of funds transfer. statements are big issues. However, several RD asCSSaretreatedas"grant-in-aid"forbudgetary programs have improved transparency and accounting purposes, release of funds by GoI availability of program performance information (and where relevant by states to lower levels) in recent years, for example putting detailed are treated as expenditure, which weakens the performance information on websites (e.g., for MGNREG in a number of states or social emphasis on accounting for fund use, financial pensions in Andhra Pradesh), and some programs reporting and audit. State-specific analysis have developed systems which report audit indicates issues in some programs with "parking" compliance. of releases in various local accounts, so that funds have not always been used for program while community level oversight of program purposes when already accounted as spent.331 expenditures has become an important feature There is no real distinction in such cases between of MGNREG through social audits in particular, expenditures and advances, an issue which is institutionalizing the social audit process is in exacerbated by incentives for states to show most states only just finding its feet even in that "expenditure" to justify further fund releases. program (though with positive exceptions like This contributes to difficulties in consolidating Rajasthan), and wider adoption of the practice year-on-year expenditures, as most programs under SP programs has not happened for the in most districts have often significant opening most part. balances. This practice does have benefits in the in programs where PRIs play a more significant face of sometimes slow releases early in budget role, such as MGNREG, there is typically very years, but some cost in terms of transparency and limited financial management capacity at the efficient planning. locallevel.This in turn is often cited as a reason in cases where funds flow directly to districts for reluctance to transfer program funds to PRIs, outside the state treasury system, the framework resulting in a classic "chicken and egg" problem for accounting, financial statements and disclosure (i.e., limited funds are handled by PRIs because requirements remains under-developed. they lack capacity, which they are unlikely to develop so long as they manage limited funds). compliance with internal control procedures is there is no provision for reassignment of SP weak.Internal audit is generally weak in practice, program funds across programs. Given the despite financial rules, and financial reporting via program-specific nature of fund release and utilization certificates tends to be used more for reporting, there is no scope for states to reallocate fiduciary and control purposes than as effective funds from programs which are not spending to financial monitoring of performance. The RD those which are. Thus, lower levels are expected Ministry has taken a number of positive steps in to be accountable for program performance, but this direction, with for example NFFW introducing have little control over either design or program conditions related to financial and physical mix at the local level. performance and monitoring to justify further fund releases. While not applicable in all schemes or in all states in the external audit and reporting are weak. In the same measure, the above factors point to significant case of programs where funds flow through state issues in accountability of funds use in CSS, including treasuries, there is a large backlog on issue of social protection programs. The results of such weak audit certificates by CAG, driven in large part by accountability mechanisms can be seen in CAG's failure of departments to provide statements of performance reports on specific SP CSS: expenditure. Where programs flow more directly theevaluationofSGSY(CAG2003b)foundthat53.5 to the sub-state level, clarity on principles of audit percentofexpenditurestestcheckedin157districts 331 Thevarietyof"parkingaccounts"isalsonotable,andnotconducivetotransparentFM,includingDRDAgeneralaccounts,personalledgeraccounts,civildeposits, currentaccountsanddeposit-at-callreceipts. 12 Social Protection for a Changing India: Volume II were diverted, misutilized, misreported, or (b) Payment systems for beneficiaries otherwise subject to spending irregularities332. The largest problems related to parking funds in apart from budgetary financing issues, there has unauthorized accounts (not necessarily indicative been some experimentation in recent years with of malfeasance, but at a minimum reflecting lack different modes of cash benefit payments to program of FM discipline), overstatement of expenditures participants. At the same time, the picture in many (typically through failure to adjust advances states reveals often outmoded payment systems, and against actual expenditure incurred), and under-developed reliance on ICT. Modes of benefit diversion of funds to other purposes. In addition, payment vary both across space within the country monitoring of repayments by borrowers was weak (and often within states), and between different benefit in a number of states (e.g., TN and Karnataka), and types within states in some cases. They range from cash default rates on borrowings ranged from 28 to payment by panchayats in the presence of the public 62 percent in Chhattisgarh, Gujarat and Kerala. (e.g., MGNREG in some districts), to cash payments from Finally, there was very poor planning in most block treasuries, to money order delivery of benefits via states for the program. the postal system (e.g., social pensions in many states), to direct deposits into bank or postal savings accounts. the IAY evaluation (CAG 2003a) found similar There are also non-regular forms of payment, including issueswithmisuseandmismanagementoffunds, transfer of MGNREG wages through field assistants or with 31.6 percent of all checked spending being in some cases contractors. Some important indications misused in some way, including diversion for non- should be considered in defining efficient disbursement program purposes, inflation of reporting, parking arrangements: accessibility, cost, transparency of in unauthorized accounts and other factors.333On record-keeping and accounting, cash in transit safety, the positive side, only a negligible portion was etc. For instance, while using banks could perhaps be consideredtobemisappropriated. the most efficient and secure method of provision, more work is needed to understand across states inadequate banking network penetration may not allow the extent to which the above factors contribute to for heavy reliance on that option. No payment method relative expenditure performance and transparency can be considered the preferred option in all cases, and of public expenditure management in sp programs.334 even apparently straightforward and low intermediary Some of the factors noted ­ such as budget procedures, options like savings account payment have been seen in prevalence of ITC in financial transactions, and field work to be subject to "deductions" by staff. commitment to social audits - are very state-specific, india's ict capacity suggests that governments could while others ­ such as generally weak FM capacity in be more proactive in experimenting with new benefit panchayats and accounting parked funds as "spent" ­ delivery models for sp programs. While some of the are more generic. To date, no clear pattern emerges to models tried internationally (and now in some parts indicate that high income states have significantly of India) will require connectivity that more remote better PFM. This is probably due to the common and areas currently lack, various E-governance initiatives uniform framework followed across the country and of both central and state governments suggest that it given the fact that the PFM reforms are still at a fairly will increasingly become realistic to experiment with nascent stage. Nonetheless, a number of States have ICT-driven payment systems (see Section D). initiated measures over the last 4-5 years to strengthen the treasury systems to obtain more timely financial while not all options will be suitable or possible information, initiated accounting reforms in the ULBs, in all areas immediately, all these solutions offer electronic transfer of salaries, setting up systems to immense opportunities, but also come with limitations monitor audit and other measures which provide good (e.g., mobile phone banking works when money has models for other states to follow. been deposited in a bank account, etc.). A combination 332 SeeCAGSGSYevaluation,SeeCAG(2003b). 333 SeeCAGIAYevaluation,SeeCAG(2003a). 334 AgoodexampleofsuchadetailedanalysisforBiharisWorldBank(2005a). CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 of various solutions would likely be required. Box 7.2 gram sabhas. While such initiatives have considerable gives some details on existing experiments in India and promise, an issue for future consideration is the extent internationally, which may provide useful lessons. to which such functions are uncompensated, which may raise questions of their sustainability and replicability. an operational complement to conventional public mechanisms that has grown in recent years is reliance on the self-help groups (shgs) and pris in certain (c) Record keeping aspects of program administration. This has included client services such as public awareness, beneficiary with the notable exception of mgnreg, record keeping identification, basic record-keeping, operation of in most programs in most states is often outmoded, FPS, and benefit disbursement. SHGs can sometimes and does not meet the needs and standards of modern organize themselves both at village and mandal levels program administration. At the same time, both and open accounts at commercial banks, which can MGNREG and the innovations of individual states (e.g., facilitate cash disbursement. PRI are also involved in Karnataka through its computerized treasury system certain programs/states in direct benefit payment, with and Andhra Pradesh with its newly developed system cash benefits handed over in community meetings/ for the social pension program), suggest that rapid and Box 7.2: innovations in payment systems While disbursing cash benefits through the organized financial sector channels is normally associated with strong operational advantage, lack of penetration of many remote and rural areas with financial services has been posing a binding constraint on this mode of delivery. While Post Office seems to operate in a greater number of locations in India, adequacy of its accounting mechanisms and capacity to assume new tasks remains to be evaluated. More generally, conventional financial intermediaries are often ill prepared to deal with mass transactions of very small amounts in remote, disbursed, and poorly educated communities. The Indian challenge is not unique. Both in India and abroad, a number of interesting and relevant innovations have been introduced to facilitate the effort of expansion of coverage of financial services. ATM machines, being a conventional means of cash disbursement, remain an expensive proposition for some communities. Research and development efforts have been focusing on bringing the advantages of conventional financial sector infrastructure to those communities. For example, lessons from Bolivia where low-cost finger print enabled ATMs were introduced to service needs of the emerging MFI sector, indicate that rural, low-income communities are both willing and capable of using high-tech services as long as they serve their needs. The cost of such device in Bolivia is around $18,000 (http://www.digitaldividend.org/case/case_prodem.htm). In India, innovations in the same field seem to be far more promising. With a prototype developed by a team lead by Dr. Ashok Jhunjhunwala of Indian Institute of Technology in Chennai, a similar device may cost less than $2,000. It can function as a stand-alone machine or be connected to a computer terminal, operating on a standard ATM platform or custom banking IT system. Accessing the account is enabled by a finger print recognition system. An ATM card or account number entry would be required to facilitate the matching process. Note that no expensive smart card is required in this case. Finally this new device is capable of working in the conditions of temporary losses of electric power supply. All of this is important for efficient service provision in otherwise infrastructurally lagging communities. The new machines have already been piloted by the ICICI Bank. Another interesting internationally emerging practice of providing access to financial services in the remote communities is mobile banking. In Kosovo, for example, when a new basic old age pension program was being introduced, bank branches were not operating in some ethnic minority enclaves. A number of mobile banking units were licensed then to service those areas lacking normal banking coverage. The mobile units visited these areas on a regular schedule, providing benefit disbursement services, until the banking sector developed to sufficiently cover the whole territory of the province through normal branch operations (Gubbels et al, forthcoming). Point of Service (POS) devices have been widely utilized, for example, in Brazil as an alternative to the conventional financial services network. Using a regular debit card, individuals can now access their accounts and perform various financial transactions, e.g., while shopping at a local drug store. Local providers have to be licensed by the host commercial bank, have a telephone line, and be equipped with a card-reading device connecting in a dial-up session to the remote bank's server. A final and more recent innovation is mobile phone-based payment systems. While still in relatively early stages in developing countries, experiences in countries such as South Africa, Kenya and Philippines indicates that there may be significant potential for exploring mobile phones as vehicles for payments of various kinds. The basic model involves topping up of mobile credits to a phone-holder in the relevant area, who can then pass on the cash value of the top-up to beneficiaries. Obviously such an approach has its limitations and would require careful piloting and evaluation, but the fact that pilots such as that supported by Vodafone in Kenya appear to be working in challenging environments suggests that it is worth considering for experimentation in India. 1 Social Protection for a Changing India: Volume II quite far-reaching improvements in record keeping can separate registers by type of social pension. The be achieved at manageable cost and despite capacity substantial variation across districts also points constraints. to lack of enforcement and standardization of record keeping. institutional transformation rather than a mere ict upgrade should be seen as an objective of operational even where registers of beneficiaries are modernization. An ICT project should be seen as an well maintained, there is typically a lack of opportunity to comprehensively assess existing business coordinationandcross-checkingacrosslevelsof processes and based on the accumulated experiences thesystem.Field work for this report found that define a new process that would help address various basic program information (including beneficiary limitations and loopholes. It is the business process, not numbers) was at times different across levels of the the ICT, that should be a driving force behind the effort system, indicating issues with data consolidation of reform. and comparability. updatingandsimplecheckingofbeneficiarylists there are a range of common challenges which sp programs in much of india face. they include: isoftenneglected.Beneficiarylistsarefrequently notupdatedforlongperiods,despiterequirements in some states (or parts thereof) and for some in several programs for annual verification programs, lack of any records at lower levels of exercises. Just as significantly, administrative data thesystem.This is for a variety of reasons, but a are not in most cases used for "reality checking" common one is the lack of capacity and training, and duplication checks. A positive example of and lack of clear guidance and enforcement the benefits of relatively straightforward (and low to the lowest levels of the system, in particular cost) exercise in this respect is found in Karnataka, GPs. An illustration is provided by the social where the Finance Department did an initial pension program in Rajasthan presented in check by taluk of social pensioners against census Table 7.11, which shows the share of villages data. This identified a range of obvious issues, by district maintaining beneficiary registers. In including high shares of very old pensioners, some districts, there were no GP registers at all taluks where the beneficiary population for of social pensioners in sampled blocks, and in NOAPS exceeded the total elderly population the 8 districts surveyed, under 30 percent of GPs (and conversely, where it was less than 5 percent), maintained a register. Of GPs that did maintain and other anomalies. This allowed for focused a register, a further quarter did not maintain program audit by the Government of Karnataka. The exercise was supplemented by running a Table 7.11: Share of gPs maintaining registers of simple algorithm for checking individual records social pensioners, Rajasthan, 2006 which seemed high risk for duplication. The latter A. Any register of B. Separate register exercise allowed an immediate reduction of social pensioners (% by social pension 6 percent of duplicate or ghost records across District of gPs) type (% of col. A) 20 taluks. This technique is described in Box 7.4. Ajmer 16.6 100 Such procedure helps to improve and make the Bundi 85.3 100 internal audit effort more directed. However it Dungarpur 4.7 0 requires at the minimum some basic electronic Ganganagar 36.0 23.1 database of records of active beneficiaries. Jaiselmer 35.2 71.0 lack of records of active members. What often Jalore 44.0 23.1 is available is only a cumulative register of all Jhalawar 0 0 beneficiaries who ever applied and/or were paid Jhunjhunu 0 0 at least once from the program. Cases of losses All sampled 29.2 76.9 of eligibility or death would remain unaccounted districts in such program statistics, therefore significantly Source: Dutta (2008) limiting monitoring and planning capacity. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 fragmentationofrecordkeepingacrossprograms, examples in CSS such as NIC's development of the resultingininefficiencies,inabilitytocross-check, basic MGNREG software, states have still to exploit and a weak information base where action is the full potential of outsourcing arrangements on needed on a cross-program basis. The lack of improvements in record keeping, including all linkages across program record-keeping deprives stages of ongoing operations like applications the system of some of the benefits of economies collection and benefit payments, and later steps of scale that might otherwise be possible. It also like archiving. Deciding what is appropriate contributes to the situation where beneficiaries to outsource is obviously a policy decision may have multiple identification numbers in (including on issues like data ownership and different parts of administrative data systems. A access protocols), but India's private sector simple but important example is the common potential in ICT suggests that there is more scope difference in the identification numbers of for exploring such options. Good computerized BPL households on the PDS ration card system records are becoming an increasingly important and in the BPL lists themselves. While a cross- element of program management in the face of program (and even broader) unique household or Right to Information Act requirements for publicly individual identifier is a Holy Grail of both Indian available program performance information. and other developing country policymakers, the capacity development for record keeping, challenge is India is not so much technical as one particular at block and GP levels. The increased of departmental coordination and political will. role of PRIs in SP program implementation has There are examples where efforts to promote data- created new demand for recordkeeping and sharing across departments are showing promise, other skills at local level. While several states are e.g., Orissa's pilots to consolidate reporting across pursuing capacity building initiatives in this area, departments in food-based programs and link this much more remains to be done. to an early warning system on "hunger blocks". there are also a range of program-specific lackofcomputerizationofrecordsonaconsistent implementation challenges. These are discussed basis, particularly at sub-district level. Where in part in Chapters 3, 4 and 5 under individual records exist at lower levels of the system, they programs. Two examples are: (i) absence of are more often than not paper-based. This is a available data under SGSY on repayment rates practical option in some areas with unreliable of beneficiaries; and (ii) perennial challenges in electricity, but nonetheless, many states lack a the veracity of muster rolls under public works consolidated plan for roll-out of computerization programs. in SP record-keeping. There are good examples of significant progress, e.g., the MGNREG databases established in several major implementing states. (d) monitoring and evaluation Beyond that, there are emerging examples of more sophisticated ICT applications such as AP's Overall, monitoring of sp programs ­ like many others ­ introduction of biometric technology in PDS remains driven by generation of input/output information ration cards which has resulted in a significant more suited to oversight than monitoring in a more reduction in duplicate and bogus ration cards. policy-oriented manner, and rigorous impact evaluation The most notable is the RSBY program, which is very under-developed and largely piecemeal. represents a path-breaking application of smart with respect to evaluation, there is no example of card technology to a public program for the a major program for which reliable pre-program poor. This is discussed in detail in Chapter 5, and baseline information has been collected. The main discussion of the experience of smart cards in evaluations conducted by GoI of major CSS are so- public programs in India is provided in Box 7.3. called "concurrent evaluations", which are useful but exploring greater reliance on outsourcing and more in the nature of assessments of key performance partnershipsinsoftwaredevelopmentandother parameters of schemes. At best ­ for example when aspects of record keeping. While there are good dedicated surveys are commissioned ­ they may give 1 Social Protection for a Changing India: Volume II Box 7.3: Smart cards in public programs: experience and issues The past few years have seen an unprecedented rise in the use of IT and associated applications in the management and monitoring of Government programs in India. Though most efforts seem to be concentrated towards increasing efficiency within Government departments, there are also efforts to increase outreach to citizens in order to facilitate access to government services. Smart Cards - plastic cards with a chip having the capability to store, maintain and manage data securely over time ­ are emerging as an important medium for enabling offline transactions with a high degree of security. The ability to enable secure transactions without the need of online connectivity (unlike debit or credit cards) makes smart cards an ideal medium for use in a country which still has large areas virtually unconnected by reliable means of communication to backend systems. Since smart cards allow only electronic transactions, an incidental benefit of using them is the ease with which authentic electronic data can be collected. The "smartness" of these cards lies not only the manner in which the embedded software is implemented, but also in the security systems that can be implemented using the encryption and authentication features available on smart cards. Deployment of these cards in association with features like biometrics (primarily fingerprints) and context dependent software further enhances their utility. Such a system can easily be used to verify usage, authenticate a person and validate the transaction without any backend connectivity to servers or external source authentication, providing a completely secure and validated environment for offline transactions. All this and more can be done using personal computers, portable computing devices and even certain mobiles, using only a minimum of external or built in smart card reading/writing devices. There have been numerous recent efforts in India using smart cards for ensuring an authenticated and validated delivery of government benefits to the beneficiary. Pilots using smart cards include ration cards in Kerala, milk contributions by women in Dairy Federations in Gujarat and Rajasthan, and providing subsidized medical aid to Bhopal Gas Victims and HIV/AIDS affected people. These pilots have used the cards for enabling offline transactions, storing the data and empowering the holders. However, not every project involving smart cards has yet realized the true potential of the medium. In fact they have even proved detrimental to the argument for use of smart cards in some cases. Projects like the Fishermen ID card in Gujarat and the much touted driver's license and registration Certificate have simply used smart cards as photo identity cards and for storing certain data that is once personalised but never read or used. Based on the lessons from other projects, the recently launched Cashless Health Insurance card of the RSBY program of GoI (see chapter 8) has proved to be a success story for smart cards in delivery of Government benefits. It has made use of the full capabilities of the card and combined it with efficient and implementable re-engineered processes to set up a system that is continuously evolving, self administrating and most importantly has made the process of utilizing free medical facilities for the Below Poverty Line very easy to use. With the exception of RSBY, India still has a considerable way to go in order to realize the potential of smart cards in public programs. In addition using smart cards as a superfluous tool to digitize existing processes, it is important to re-engineer processes to empower the beneficiary, reduce the administrative efforts and redirect them towards the actual business of delivering benefits. An important aspect for the use of smart cards in India is prevailing standards. Fortunately, the National Informatics Centre has already made a major effort in this area and Smart Card standards (SCOSTA i.e., Smart Card Operating System for Transport Applications) are already widely used and have evolved over time to keep pace with the growing demands and availability of new products. International standards already exist for smart card readers/writers. However, there is still a large gap in standards for other devices and associated systems like fingerprints, hence careful evaluation of requirements vis-a-vis available products is needed before commencing a project based on these technologies. Using SCOSTA as a distinguisher between kinds of smart cards is a common error. Commonly smart cards may be JAVA or Native based on their implementation (the JAVA card should not be confused with the open source operating system or software as this is slightly different) or they may be contact and contactless to provide a different platform for usage of the card. Source: Chopra 2009. insights into the household level benefit of programs in Government, and MGNREG provides structures in its rupee or quantity terms. However, such evaluations have governance for supporting research on program impacts yet to be designed in the public sector to explore the of different kinds. net impact of schemes, in terms of robust comparisons both to the pre-program situation and/or to control with respect to monitoring, india has a tradition of groups of non-beneficiaries with similar characteristics detailed reporting of key input/output indicators. to those participating. The skills to conduct such impact However, such data are used primarily for internal evaluations are available in India, and some evaluations reporting and justification of future funds rather than have been conducted outside government. There analytically to assess key program weaknesses by issue is increasing interest in robust impact evaluation in and state.335 Unlike many developing countries, SP 335 Whiletherearesystemsforconcurrentevaluationofprogramperformancethroughvigilanceandmonitoringcommitteesatvariouslevels,effectivenessvariesacross states.Themonitoringmechanismincludes,interalia,aperformancereviewcommittee,periodicprogressreports,auditandutilizationcertificatesandfieldmonitoring. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 programs in India do not lack for detailed data. However, include: (i) repayment rates on SGSY loans; there remain a range of issues with administrative (ii) the age composition of beneficiaries; and data used for monitoring, some related to the data (iii) actual days of employment generated under themselves and others to processes and uses of such workfare schemes prior to MGNREG. While it data for policy purposes: is important not to overburden monitoring a first shortcoming ­ noted in the FM section systems with needless reporting, some of these above­isthatexpendituredataisanimperfect indicators go to the heart of getting a good reflection of actual spending at the grassroots sense of program impacts and sustainability. This level, and conflates this with simple funds release points to a bureaucratic rather than strategic (which may remain parked in government culture in deciding which monitoring indicators accounts). are given priority. a second shortcoming in monitoring relates as noted above, the lack of data on active to consolidation of monitoring data. Given beneficiaries of the program seriously hampers any the increase reliance on CSS fund channeling effort of analysis of the program performance. through DRDAs, state budgets will frequently ontheprocessside,theavailabledataarerarely not reflect total spending in the state on specific used for any purpose other than publication programs. Getting a consolidated picture of in annual reports and to some extent audit spending within a state by program is therefore and budgeting. Both states and the centre lack not straightforward. strategic planning offices in Ministries which use a third shortcoming on some schemes is that the outputs of monitoring for assessing trends and importantindicatorsarenotcollectedasamatter patterns in program performance, and applying of course, and/or not available in digestible the insights to policy development. Monitoring form to state or central policymakers. Examples data are therefore under-exploited. Box 7.4 gives Box 7.4: Case study of cross-checking for duplicate beneficiaries in karnataka The incidence of duplicates among active ("running") records of social pensioners in Karnataka was checked by looking at the similarity across records in the administrative database. To do this, a pattern-matching program was applied to quantify the similarity between any two records in the files.a The program assigned to each record pair a "similarity score" between zero and one, found the other record most similar to each record in the file, and sorted the file according to this similarity, to put the record pairs most likely to be duplicates on top. For instance, in one of the taluks (Malur,) the two records appear near the beginning of the sorted file with an assigned similarity of 1.00 (they are almost identical, despite their different PPO codes,) meaning that they almost certainly reflect a double payment. Further below in the same file, the two records are assigned a similarity of 0.93 (they are alike enough to raise strong suspicion.) Even further below, the pair with a similarity of 0.86, may or may not be duplicates ­ a human comptroller would probably suggest an audit. Across the 20 taluks, 6 percent of the pension records have a likely duplicate in the administrative database. There was wide variation across taluks: in two (Channarayapatna and Malur), over 10 percent of the records have a close enough active duplicate (similarity score greater that 0.80) to warrant an audit. Six other taluks have over 5 percent of records with a close duplicate. The incidence of duplicates is correlated with the estimated coverage rate (the percent of elderly who benefit from the pension schemes), another indicator of hyperactivity in the application of the pension programs in the taluk. PPo iD name: Address O980744385 Venkatappa S/O Mutturayappa | Panamakanahalli-Kondashettihalli Post | Malur Tq-X-562116 O980744396 Venkatappa S/O Muthurayappa | Panamakanahalli-Kondashettihalli Post | Malur Tq-X-562116 O870725582 Muniyamma W/O Ramappa | Banuhalli Village-Tekal Hobli | Malur Taluk-X-562116 O870742919 Muniyamma W/O Ramanna | Banahalli Village-Tekal Hobli | Malur Taluk-X-562116 Source: Murgai 2006 Notes: The measure of similarity is based on Ratcliff and Obershelp's "Gestalt" algorithm. The program first transforms the "Name" and "Address" fields from both records into phonetically similar texts, applies the algorithm to each field pair separately, and finally computes the similarity score as an average. 1 Social Protection for a Changing India: Volume II a recent example from Karnataka of how simple junsunwais or public hearings on program performance. use of administrative monitoring data has been More details of that experience are provided in put to good use. Box 7.5, as well as insights from initial experience in Andhra Pradesh on social audits for MGNREG. To date, a key development on program monitoring in recent a requirement of social audit has not been included years has been involving the community itself in under any other CSS SP program, though AP is looking monitoring. With a strong push from civil society, GoI to include such a process as a standard element of all RD and states in recent years have begun to recognize programs in coming years. the potential of communities as agents for monitoring program implementation. The extent of this practice is while the community monitoring process shows highly variable across states, and in many cases is a tool potential - particularly when empowered through where the rhetoric of possibility has to date outstripped strategic use of the rti act ­ practice is still evolving practice. Nonetheless, progress is being made. MGNREG on how to make community monitoring a systematic presents the most developed example to date of a process, rather than a "one-off" exercise. To some program which has an explicit community monitoring extent, the reliance on jun sunwais is a reflection on role, through the mandate of "social audits". In the the weakness of community institutions such as the MGNREG guidelines, social audit is viewed as a continuous gramsabha. A challenge looking ahead will be trying to process of community monitoring intended to cover all incorporate some of the good practices of jun sunwais aspects of program implementation. The method builds into regular institutional mechanisms at panchayat level. on prior experience from NGO initiatives in India, in particular the work of Mazdoor Kisan Shakti Sangathan in addition, the government is increasingly partnering (MKSS), a Rajasthan NGO which pioneered a practice of with ngos and research institutes to monitor the Box 7.5: Community monitoring of social programs: experience from Rajasthan and AP MKSS (Organization for the Power of Workers and Farmers), a Rajasthani NGO, organized a series of junsunwais or public hearings on performance of public programs targeted to the poor. The meeting is an open forum which aims to include community members, social activities, lawyers, media and public officials. At the meeting, public documents on program performance are read out, and participants are encouraged to ask questions and give examples of their own experience in program operation in reaction to the official records. Initially, this approach was resisted by many officials, who objected that official documents could only be audited by official auditors. In some cases, court orders were sought to avoid sharing information through such an extra-judicial mechanism. While the junsunwais meeting has no official mandate and hence sanction authority, it has proved a useful tool for "naming and shaming" in order to improve program performance. Some officials have returned funds, others have quit or been suspended, and arrests were made in some cases. On the policy side, the process resulted in amendments to the Rajasthan Panchayati Raj Act to allow public access to all documents under anti-poverty programs, and the development of social audit guidelines. The process has also now been extended to some urban areas such as Delhi. The methodology has also been used intensively through "padyatris" on the MGNREG, and MKSS has conducted two trainings for participants from other states in the social audit methodology. Building on the Rajasthan experience and with the mandate of the MGNREG guidelines, the state of AP has taken a proactive role in promoting social audits in the districts where MGNREG is operational. The Department of RD and Panchayati Raj set up a social audit wing, which trained district resource people, who are training village level organizers. Initial social audits have now been documented in all 13 districts in the state where MGNREG is implemented. This has also involved widespread NGO participation. Training has been provided for village level workers in social audits processes. For example, in Anantpur district, 1200 workers were organized in team of 10-13 members covered 38 mandals in a 10 day period, and then conducted a mandal-level "social audit forum" to discuss main findings. The social audit process involves several steps after training: (i) teams are given a checklist of key implementation issues; (ii) key financial reports, muster rolls and other documentation are procured from local officials under the RTI Act; (iii) a household level survey is done; (iv) visits are carried out at work sites, and a village meeting is held to share information and discuss problems found; and (v) a mandal level social audit forum is called where officials and the community attend and officials are asked to explain discrepancies in implementation. While managing these processes has been challenging, the interaction between communities, PRIs and officials is generally considered useful, and has resulted in dismissal of lower level field staff in a number of cases. At the same time, the social audit has generally been better at dealing with small level grievances than systemic issues, and it remains to be seen how sustained the process and its impacts will be as implementation continues. Sources: OHCHR/UNDP workshop; Samji and Aiyar. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 performance of various programs in the field. Most developing a better picture of what are the notably, MGNREG has introduced systems for field key information sources of the poor on public monitoring of the program using national monitors as programs, including what types of information well as third party monitoring by independent agencies. theygetfromwhichchannels.The limited survey A network of institutes has been established for information indicates that typical government independent concurrent monitoring of the program and media campaigns, web-based information, and has generated a body of research on the performance of information from administrative officials are not MGNREG in the field across states.336 the channels through which the poor find out about programs providing in SP program budgets for sustained D. ConCluSionS AnD awarenessraisingbythirdpartieswherepossible. ReCommenDATionS Field work for this report indicates that program awareness dissemination is not a one-time with a myriad of programs and plans to introduce activity, but a product of repeat contacts and new schemes, expand coverage and consolidate local sources of reliable information. This tends existing schemes, it is an important time to assess not to be the approach of most IEC campaigns for relevant experiences and identify both good and bad SP programs. administrative practices. While India may not have developing a more diverse range of IEC the advantage of other countries of running a well- strategies for program awareness raising. This organized centralized machinery of administration would involve not only a more diverse range of public benefits, some strategic priorities could of actors as tools of information dissemination be defined to achieve better benefit delivery and (including NGOs, SHGs, youth groups, social transparent operation. In particular, policies seeking and religious leaders), but also a more diverse harmonization, portability, and transparency in set of information tools (including visual media benefit provisions, and otherwise ensuring minimum suitable for non-literate audiences, television operational standards, should be designed, endorsed, and radio campaigns which more creatively and promoted. Capacity to regulate and supervise engage audiences rather than simply presenting various benefit providers and their intermediaries program guidelines, use of community, etc). should be gradually built. Furthermore, a firm stand could be taken on the policy of registration and deepening the use of social audits not only as a tracking of beneficiaries of the public programs. tool for program monitoring, but as a source of From the discussion above, and from the insights program information dissemination. of many GoI and independent studies on program performance of CSS SP programs, some more specific recommendations emerge: (ii) on the applications process review program eligibility criteria on several key programs with a view to simplification of (i) on awareness and outreach eligibilityproofandunderstandingofthepublic. while program awareness and outreach is an One aspect of this would be exploring avenues inherently difficult process in a country of india's for reliable third party verification of certain facts size and diversity, experience from states and for where documentary evidence is not available or specific programs such as mgnreg indicates that much costly to gather. can be done if there is sufficient will. An improved proactive exploration of how SP programs awareness and outreach strategy would likely involve can reduce transactions costs of program several elements: application linking up with the range of ICT 336 ThisProfessionalinstitutionalNetwork(PIN)includesIndianInstitutesofManagement(IIMs),IndianInstitutesofTechnology(IITs),NationalInstituteofRural Development(NIRD),AdministrativeStaffCollegeofIndia(ASCI)andothers. 10 Social Protection for a Changing India: Volume II initiatives such as rural business kiosks and other reforms ­ such as accounting for SP CSS other village-level connectivity initiatives. The resources as "spent" when they have not been emerging examples in other contexts such as and application of accrual accounting ­ will spreading availability of land records through require coordinated efforts of the centre and rural E-kiosks (e.g., Bhoomi in Karnataka) and states on reporting processes that require availability of market information (e.g. ITC's reliable information on physical and other E-chaupal initiative) suggest that concerted outcomes of "spending". The example of effort between SP departments and their ICT MGNREG reporting in some states provides a department colleagues could pave the way for a good model which should be institutionalized. major contribution to simplification of program A first step in coordinated effort between the applications, and significant cost reductions for centre and states could be clear agreement on applicant households (not to mention lower the necessary skills needed for FM at different opportunities for undesirable middlemen). levels of the system for SP programs (and more broadly), and agreement on action plans for encourage­ifnecessarywithpublicsubsidyona ensuring adequate skills. contracting-outbasis­NGO,SHGandotherCBO intermediationtofacilitateprogramapplications, as part of strengthening the accounting and increase the use of camp approaches to and financial reporting framework, it will be registration by the administration itself. Given increasingly important in SP CSS which route that the culture of reliance on middlemen is through DRDAs to develop standard financial so prevalent, it seems sensible to encourage reporting rules for societies, which would the efforts of middlemen (or more likely subsequently be reflected in the GFRs for GoI and "middlewomen") who are more likely to have the the states. This work would most sensibly be led interests of the applicant at stake. by CAG and ICAI. in terms of program oversight, it seems sensible addressthebackloginexternalauditsofSPCSS, to target IAY and SGSY for particular scrutiny, and a more competitive process of selection of including use of social audits. third party accountants responsible for auditing societies. All Ministries implementing SP CSS share the experiences of states which are piloting should also post the findings and actions taken biometric and other ICT advances to reduce fraud on audits on websites. and human error in the beneficiary recording process. equally - and more challenging in terms of bureaucratic culture ­ there is a strong need to deepen the uses to which financial information (iii) Financial management 337 on programs is put, and hence to move away some of the needed improvements are in the from the simple input/output culture of financial domain of states, and they will need to follow reporting which prevails in most CSS. the lead of recent reformers in areas like public finally, the positive development of increased expenditure management rules and expansion citizen accountability for SP funds use seen in of ICT in release and tracking of funds. They MGNREGshouldbeinstitutionalizedovertimein will also need to facilitate sustained FM allSPprograms,with basic initiatives like posting capacity development in PRIs if the intentions of all program releases and uses, greater reliance of decentralization of SP programs are to be on social audits, and posting of beneficiary lists realized. This will also require in most states a and program rules in plain language. A number more elaborated financial accountability of these are natural corollaries of the Right to framework for PRIs and sub-state implementation Information Act, but there full potential remains agents. to be exploited in most states of India to date. 337 SeeWorldBank(2005a)formoredetailedrecommendations. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 11 (iv) Payment systems and record As one of the challenges in this work with be capacity constraints at both levels, it could be keeping useful to explore options for supplementing reforms of payment systems can take a range of this in non-costly ways, such as providing skilled forms, both technology-based, and where that is not workers from MGNREG as temporary assistants to possible, with community monitoring of payments. the gram sewak. some of the technological options include: at the state and district level, development of low cost ATMs developed at a fraction of cost of simple verification exercises which will allow the conventional machines, equipped with finger program oversight to focus on high-risk areas. print identification and long-life batteries, can Even available monitoring data often throws up facilitate cash distribution in remote areas. obvious anomalies, but is typically not used for this purpose. point-of-service (POS) devices installed in local shops and gas stations can use regular telephone at central and state levels, improve cross- lines to process on-line cash disbursement departmentalcoordinationonarangeofrecord transactions, or ­ in the case of PDS ­ facilities for keeping issues, starting with a strategy for debiting the monthly ration. convergenceofbeneficiaryidentificationnumbers. special mobile branches could make conventional Even where systems remain parallel for the short banking more accessible, enabling regular and to medium term, Collectors and BDOs should scheduled visits in the remote communities. encourage greater cross-program information exchange for programs with common elements finally, mobile phone banking is an emerging such as being food-based or child-focused. phenomenon that utilizes comprehensive as a condition of CSS receipts, the centre could penetration of the new communication networks. insist that a computerization strategy and roll- more generally, cost recovery mechanisms of different outplanforSPprogramsbedeveloped,including operational alternatives to the conventional models a training plan. This seems a bare minimum for of service provision should be further studied. movement towards a modern system of record While some models are quite promising, their financial keeping. viability as localized stand-alone solutions remains a encouragestatestobemoreproactiveinexploring concern. Clever solutions may simply need to be scaled outsourcing possibilities on development of up in order to reach the break-even point of economic program record-keeping software. India offers operation. Furthermore, strategies enabling synergies a particularly promising environment for such across multiple programs should be further explored. efforts, though there would remain an important For example, while the smart card technology may role for the central government in ensuring be a costly solution if utilized only by a single benefit consistency in basic design features, and thus program338, using it as a common platform for multiple comparability of data from different states. products (e.g., various benefit and insurance schemes) may well help to make its application a sound economic proposition. (v) monitoring and evaluation with respect to record keeping, the experiences of more ensure that any major SP program has a pre- progressive states such as karnataka offer lessons programbaselinedone, and that impact evaluations for others. Some of the issues for attention include: also include control areas where feasible. training (and subsequent monitoring and explore replicating the model of MGNREG with enforcementofcompliance)forGPsandblocksin provisionformorerobustresearchandevaluation maintainingappropriateregistersofbeneficiaries. financedfromtheprogrambudget. 338 ThemarginalcostofonesmartcardissuedinIndiaisquotedasbeinginthe$2-$3range. 12 Social Protection for a Changing India: Volume II on monitoring: (i) review the current set of building on the social audit experience in monitoring indicators for major programs to assess some states, mainstream social audits as a their utility and gaps in key program indicators; standard practice in all SP programs. While (ii) if necessary on a contracted-out basis or the experience is relatively new on any scale, under special units (e.g., as Orissa has developed social audits seem a useful tool for increasing under its Poverty Monitoring Agency), conduct community knowledge of programs and regular basic analysis of consolidated program providing a vehicle for redress where needed. monitoring information at state and central levels; However, in the absence of an authorizing and (iii) explore institutional mechanisms in states environment in program guidelines (such as and at the centre to use the analysis of such units under MGNREG), the potential of social audits for annual review of program performance and will remain limited. policy implications. CHAPTER-7: Program Administration Issues in Social Protection ­ The "Nuts and Bolts" 1 CHAPTER Targeting Mechanisms 8 BPL and Beyond Chapter­8 Targeting Mechanisms BPL and Beyond targeted programs have naturally to concern identification of poor households is done through a themselves with identifying who are the poor, and separate process by the Ministry of Rural Development how well interventions reach them. Given fiscal for rural areas; the process for urban areas seems to constraints, they also often need to ensure that the non- be more fragmented with various state departments poor do not benefit unduly from interventions which maintaining different beneficiary lists. Section C then are intended to be targeted. For nearly all its major anti- looks at comparative targeting performance across SP poverty programs, India seeks to target the poor as the programs which employ different targeting methods. exclusive or major beneficiaries of public interventions. Section D looks at alternatives to the indicator-based BPL method. The next section assesses the targeting this chapter examines the main ways in which public performance of these different methods. The chapter programs in india try to target the poor, how effective concludes with a discussion of options for improving they are in achieving their objectives, and options for targeting in SP programs, including a discussion of the improving targeting mechanisms. The structure of the proposed 2009 BPL methodology as described in the chapter is as follows: Section A outlines the different Saxena Committee draft report. While the chapter finds meanings of the term poor (in program terms, "below significant shortcomings in the BPL system, targeting poverty line" or BPL for short) in social protection (SP) of SP programs is an inherently difficult process in a policy in India. Section B then has a discussion of the country where a large share of the population clusters dominant targeting mechanism for SP programs ­ the around the poverty line, and thus the observable BPL census system ­ and its shortcomings. While poverty differences between "poor" and "non-poor" households incidence is estimated by the Planning Commission are often not easy to measure and may be small in every five years based on the "thick" NSS data, the absolute terms. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 1 The card also has many other functions, such A. whAT DoeS "PooR" meAn as an identity card, and a basis for establishing in TeRmS oF PuBliC residence and even citizenship in some cases. The determination of ration card status is done PRogRAmS? in different ways across states. In several states there are four major definitions in the sp system of (e.g., AP and UP), the Food and Civil Supplies the term "poor" or "below poverty line (bpl) people" Department conducts its own periodic survey in india: to determine the level of household welfare. In others (e.g., HP and Rajasthan), the Food and people who are below the state-specific poverty Civil Supplies Department relies on the "BPL lines determined by the Planning Commission, lists" as a basis for issuing ration cards. In others based on the NSS "thick rounds" which are (e.g., Bihar), this movement from having separate conducted every 5 years or so. These estimates bases for determining PDS ration card status are primarily for planning and fiscal allocation and BPL listing for other anti-poverty program purposes, and are important for allocation of purposes towards integration of the two lists centrally sponsored anti-poverty programs. has been fairly recent. This has been strongly peoplewhoareontheso-called"BPLlists"based encouraged for all states as the 2002 BPL lists on periodic dedicated surveys by the Ministry of were being put in place and more recently in the RuralDevelopment(forruralareas).These use a Saxena Committee draft report on the proposed standard national set of criteria for determining methodology for the 2011 BPL Census. household living standards. These lists distinguish people who are considered "destitute" or primarily between APL (Above Poverty Line) and BPL (Below Poverty Line) households, and otherwise poor or vulnerable. This includes the main purpose is ranking of households for members of specific social groups who receive determining eligibility for various anti-poverty either categorically targeted programs due programs targeted to the poor. This system has to their high average poverty rates or social been in place since 1992, with changes in the exclusion. The primary groups in this category methodology for determining eligibility since are ST and SC populations, but in specific that time. Though prepared by the Ministry programs may include other poor and/or of Rural Development, this BPL list is used for vulnerable groups such as disabled people, targeting programs (such as RSBY) administered widows and girls. by other Ministries as well. There is no comparable in terms of policy, there is therefore a divergence centrally directed process across programs for in definitions across departments, states and identifying urban poor households; instead, programs in what the term "poor people" or "below various departments identify poor households poverty line people" means. This divergence spills over using different methods and maintain separate beyond social protection programs, as other sectors beneficiary lists. "A working group at the Planning use one or other definition of "below poverty line" for Commission is currently debating the design of provision of benefits and/or exemptions (e.g., in health, an urban BPL methodology". BPL households are eligible for institutional birth peoplewhohavea"BPLrationcard".Thisrefers allowances, subsidies on participation in some public to ration cards, issued by the state Food and health insurance schemes, and exemptions on service CivilSuppliesDepartments,whichdeterminethe fees in hospitals and some other health facilities; in amount and price of grains which households education, many central and state stipends are based can access from the PDS system.339 Since the on social category). There is also frequently a divergence 1997 reforms of PDS, households may have between how policymakers and functionaries use an APL, BPL, Antodaya (AAY) or Annapurna the term "BPL" and how different segments of the ration card, depending on their poverty level. population defines it. Based on research for this report, 339 AnotableexceptionisTamilNadu,whichcontinuestohaveaPDSsystemwhichisnottargetedaccordingtoaBPL/APLdivide(seeChapter3). 1 Social Protection for a Changing India: Volume II the most commonly understood meaning of the term Table 8.1: Discrepancies between different for the general population is either possession of a estimates of "BPl" households (rural and certain type of ration card, or a more undefined sense urban combined) of destitution. hhs with BPl/AAy there has often been a significant mismatch between Poor BPl/AAy hhs as households ration cards multiples of aggregate numbers defined as poor for planning State (national)1 (national)2 poor hhs and fiscal allocation purposes and numbers who Andhra Pradesh 28.1 140.5 4.99 are poor according to other methods. The most Assam 10.3 18.9 1.84 notable example is the difference in the aggregate Bihar 61.1 71.6 1.17 numbers and state-level patterns of poverty in Planning Chhattisgarh 17.9 19.2 1.07 Commission terms and those for BPL ration cards. This can be seen in Table 8.1, which compares the state-level Gujarat 17.3 46.3 2.67 estimates of poor households according to the Planning Haryana 5.6 8.1 1.44 Commission and the number of BPL households Himachal 1.3 3.0 2.33 Pradesh identified by state governments for PDS. Two things Jharkhand 20.7 24.0 1.15 of note are that aggregate numbers of BPL ration card Karnataka 27.3 71.4 2.61 holders exceed the official estimates of the poor (by Kerala 10.5 20.2 1.93 nearly three quarters), and that the differentials differ sharply across states. Some states such as West Bengal, Madhya 45.2 52.1 1.15 Pradesh Bihar, Chattisgarh, Jharkhand, Rajasthan, Maharashtra Maharashtra 64.1 73.6 1.15 and Madhya Pradesh have coverage of ration cards Orissa 37.5 48.5 1.29 only slightly more than the official poverty rate. Only Punjab 3.9 7.1 1.85 Uttarakhand has under-coverage of BPL ration cards Rajasthan 22.3 23.1 1.04 relative to official poverty rates. In contrast, the majority of states ­ most notably Andhra Pradesh, Karnataka, Tamil Nadu 34.2 168.3 4.92 Gujarat and Orissa ­ have BPL ration card beneficiaries Uttar Pradesh 91.5 106.8 1.17 which far exceed official poverty estimates. Tamil Nadu Uttarakhand 6.8 5.0 0.73 makes no distinction between BPL and APL ration West Bengal 41.3 47.3 1.15 entitlements. In addition, there are discrepancies Major states 546. 9 955.0 1.74 between ration card holding using administrative Notes: All figures in lakhs. Tamil Nadu makes no distinction between BPL and APL ration entitlements and the number above includes both. and survey-based data. Thus, the administrative data 1. Planning Commission poverty estimates based on 2004/05 NSS data. indicates that 40 percent of all households possess a The national poverty rate is 28%. 2. As identified under PDS by state BPL/AAY ration card.340 The corresponding figure from governments in 2003/04 (dates differ somewhat across states). 2004/05 NSS household data is 25 percent.341 Source: Planning Commission; Ministry of Food, Public Distribution and Consumer Affairs. at the same time, there have been efforts ­ as yet incomplete and variable across states ­ to to promote numerical convergence between Planning promote greater convergence between the different Commission aggregate poverty estimates by state and administrative mechanisms for identifying the poor. the estimates of the poor generated from BPL lists.342 This An example in several states is the effort to develop effort was diluted in practice due to political economy common lists of BPL households for PDS and other SP challenges so that it is unclear how effective this and social programs, at least where the target groups convergence was. This, however, does not necessarily are similar. However, this initiative has not begun in a mean that all programs would need to target the same number of large states to date. Another was the attempt groups. Programs such as educational scholarships 340 Calculatedusing2005projectedpopulationandhouseholdsizeestimatesfromthe2001Census. 341 Thiscorrespondsto30percentofruralhouseholdsand11percentofurbanhouseholds. 342 SeeSaxena(2006). CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 1 that target SC/ST children could use the information in effective as a random assignment of BPL status among the common database to continue targeting on caste the bottom 40 percent of the population. Fine-tuned affiliation rather than vulnerability status. indicator-based targeting is an inherently difficult exercise in a context where the characteristics of the even more worrying is the lack of overlap across non-poor do not differ significantly from those of the administrative lists, i.e., the same households are poor.346 Finally, this section evaluates defacto targeting not necessarily identified as bpl even when the outcomes for ration card holding in the absence of underlying notion of poverty or vulnerability is data on targeting outcomes of BPL lists and because the same.For instance, a study in Bangalore revealed the most common understanding of "BPL" is linked to that there were three lists of households identified the possession of ration cards. as BPL by three different departments (including for PDS) in three notified.343 The study found discrepancies in aggregate numbers, ranging from 1841 families as (a) dejure targeting design of the BPl per the Food and Civil Supplies Department, 640 by the Directorate of Municipal Administration and 668 census families according to the Karnataka Slum Clearance in 1992, the ministry of rural development, goi, Board. What was striking was that the number of undertook the task of identifying "below poverty families that were common to all three lists was line" (bpl) households in rural india through periodic only 6 percent! (approximately five-yearly) village censuses. the first such census used self-reported household incomes to identify bpl households. Households with B. The 2002 BPl annual income below ` 11,000 were classified as poor. However, given the difficulties of measuring income, meThoDology 344 the self-reported income approach was abandoned in The BPL Census is the primary administrative the 1997 BPL census. In the modified format, a set of mechanism used for targeting in several government five questions were asked of every household in the programs, including rural housing support (IAY), village ­ whether they: (i) operated size of land was subsidized rural credit (SGSY) and, in some states as more than two hectares; (ii) owned a `pucca house' discussed above, also subsidized rations (PDS), as as defined in the Population Census; (iii) had annual well as other social programs and services targeted to household income of more than` 20,000; (iv) owned the rural poor.345 However, the 2002 BPL list has been any of the following consumer durables: television, the subject of considerable criticism. This section refrigerator, ceiling fan, motor cycle/scooter and three discusses the design and performance of the 2002 wheelers; and (v) owned farm equipment such as BPL census methodology. The analysis in this chapter tractor, power-tiller, or combined thresher/harvesters. suggests that this method is a weak mechanism for If households answered in the affirmative to any one identifying the poor. While the BPL score index is of the questions, they were declared to be "visibly non- good at distinguishing the bottom 10 percent from poor". This was done to identify "visibly poor" from the top 20 percent of the population, it is unable to "visibly non-poor" households in the village relatively differentiate between the poor and non-poor in the quickly and in an inexpensive manner. Visibly non-poor vicinity of the poverty line. Relaxing these assumptions households were excluded from the more extensive and expanding choice of indicators improves de jure BPL survey that collected information on consumption targeting outcomes somewhat, but is still not as expenditures using an abridged budget-expenditure 343 Ramanathan(2005). 344 ThissectiondrawsextensivelyonJalanandMurgai(2008),preparedasabackgroundpaperforthisreport. 345 Nocomparablecensusexistsinurbanareasandtheidentificationofpoorhouseholdsarecarriedoutseparatelybydifferentdepartments,e.g.,theFoodandCivil SuppliesDepartmentforPDS,UrbanDepartmentforSJSRY,etc. 346 AnalysisofNSSdataindicatesthatthereislittledifferenceinthestandardoflivingbetweenpoorandmiddleclass(median)households,atleastinthetypesof indicatorsthataremeasuredbytheNSS(WorldBank,2011). 10 Social Protection for a Changing India: Volume II schedule based on a mixed reference recall period of sections of states "missing out". In most cases, the cut-off 30/365 days.347 chosen was between 16 and 25. the 1997 bpl census methodology had several shortcomings, including (i) very stringent "exclusion" (b) how well does the 2002 BPl criteria, whereby e.g., households were declared visibly non-poor even if they possessed a ceiling fan; indicator perform in theory? (ii) non-availability of official poverty lines for all it is important to understand how effective the bpl states/UTs; (iii) using uniform criteria without allowing identification process is in terms of design of the for inter-state variations; and (iv) not allowing new targeting mechanism. In order to evaluate the design households to be declared poor in the interim period efficiency of the 2002 BPL method in identifying the poor before the next BPL survey is instituted.348 Prior to the as compared to the expenditure based poverty measure, next survey in 2002, an Expert Group was established this section uses information in the 1999/2000 NSS to recommend changes in the 1997 BPL identification survey to construct household scores that are as similar guidelines to overcome the criticisms. as possible to the BPL-based scores.351 The constructed BPL score indices are close, but not identical, to the a simple form of "proxy means test" was adopted administrative BPL scores because the survey instruments for the 2002 survey.349 The group recommended that are not the same and there are some differences in the rather than rely on measures like income or expenditure definitions of variables. The methodology and mapping to identify the poor, socio-economic indicators of BPL criteria and NSS household characteristics is reflecting the quality of life of households should be explained in Annex 5. Based on the distribution of BPL used to identify BPL households. Each household would scores within each state, we define a state-specific BPL be given a score of one to four for each of thirteen score cut-off such that the number of BPL poor (i.e., people "scorable" indicators and the scores would be summed with BPL scores below the cut-off) is exactly equivalent to an aggregate index ranging between zero and fifty- to the number of persons living below the poverty line in two.350 Households would be ranked based on the total that state as estimated by the Planning Commission for score that they received and categorized as poor or non- the year 1999-2000.352 Households are classified as being poor based on a cut-off score. These cutoff scores could (actually) poor using per capita monthly consumption vary locally across districts, blocks or even villages, with expenditures and the official Planning Commission one constraint that the number of BPL poor was to be state-specific rural poverty lines. NSS data for 2004/05 is the same (or not more than 10 percent) as the number also used to confirm the robustness of results. of persons living below the poverty line in that state/ UT as estimated by the Planning Commission for 1999- based on the mapping of bpl criteria to nss data, it is 2000. States were given discretion as to whether cut- clear that the bpl identification methodology is a weak off thresholds are the same across the entire state, or mechanism for identifying the poor. Table 8.2 reports determined on a more decentralized basis. In practice, the extent of under-coverage in the BPL classification they appear to have opted for uniform cut-off thresholds at the state-level. Under-coverage is defined as the rather than intra-state variation, in part due to limited percentage of the actual (i.e., expenditure-based) poor information on living standards at sub-state level, and in who are incorrectly classified as BPL non-poor. Three key part in response to political economy concerns of richer findings emerge: 347 The survey also gathered information on household demographics, housing conditions, land ownership, formal training for skill development and receipt of assistancefromvariousprogramsbutthisinformationwasnotusedforcategorizinghouseholdsasBPL. 348 Sundaram(2003)providesadetaileddiscussion. 349 SeeBox8.1inSectionD. 350 Thethirteenindicatorsincludedsizeoflandholding,typeofhouse,availabilityofclothingperperson,foodsecurity,sanitation,possessionofconsumerdurables, literacy,statusofhouseholdinlaborforce,meansoflivelihood,statusofchildrenbetween5-14years,typeofindebtedness,reasonsformigrationincaseofa migranthousehold,andpreferenceforassistancefromamongvariousschemes. 351 AdetaileddiscussionofmethodologyandresultsiscontainedinJalanandMurgai(2008),preparedasabackgroundpaperforthisreport. 352 Whilethescore-basedBPLindiceswereintroducedin2002,thedatausedarefrom1999/2000.However,thepredeterminednumberofBPLhouseholdsatstate levelsetbyGOIwasbasedon1999/2000povertyrates,andtherefore,thediscrepancy,ifany,shouldnotbesignificant. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 11 Table 8.2: Poverty rate and targeting errors in the 2002 BPl classification, by state 1999-00 2004-05 State Rural Poverty Rate (%) under-coverage (%) Rural Poverty Rate (%) under-coverage (%) Andhra Pradesh 10.5 76.9 10.0 74.4 Assam 40.3 41.6 22.0 51.8 Bihar 44.0 40.6 42.5 35.9 Gujarat 12.4 64.9 18.8 51.0 Haryana 7.4 73.8 13.3 60.1 Himachal Pradesh 7.5 74.5 10.5 56.7 Karnataka 16.8 64.2 20.6 56.1 Kerala 9.4 72.6 13.1 67.2 Madhya Pradesh 37.2 43.8 36.9 34.8 Maharashtra 23.2 54.4 29.6 45.0 Orissa 47.8 32.1 46.9 31.2 Punjab 6.0 72.4 9.0 74.4 Rajasthan 13.5 63.8 18.3 59.9 Tamil Nadu 20.0 64.5 22.8 53.5 Uttar Pradesh 31.1 51.9 33.2 47.0 West Bengal 31.7 46.3 28.4 48.2 16 Major States 27.0 49.1 27.3 45.8 Notes: Under-coverage is the percentage of the poor population wrongly classified as BPL non-poor. Rural poverty rates are estimated using the official Planning Commission state-specific rural poverty lines for 1999-00 and 2004-05. Note that the poverty rates in the two years are not strictly comparable (see Chapter 1). Sources: Jalan and Murgai (2008). across India, the BPL score misclassifies nearly Findings from 2004-05 data imply similar half (49 percent) of the poor as non-poor, and magnitudes of under-coverage and pattern conversely, 49 percent of those identified as BPL of variation across states. The BPL score poorareactuallynon-poor. misclassifies nearly 46 percent of the poor as thereisalsostrongvariationacrossstatesinthe non-poor using 2004/05 data. Under-coverage targetingerrorsimpliedbytheBPLmethodology, higher in most states, with the exception of though targeting errors are significant in all Assam, Punjab and West Bengal, in 1999/2000 states.Even in the "best" state (Orissa), 32 percent relative to 2004/05. of the poor are misclassified while in the "worst" it is also clear that the problem of exclusion of state (Andhra Pradesh), three out of every four the poor (and therefore, inclusion of non-poor) poor people are misclassified as non-poor based in the bpl method tends to be greater in the richer on the BPL indicator. states (see figure 8.1). At one level, this is a source of there are also notable variations within consolation, as SP policy would be expected to be most states (based on NSS regions) on the scale of strongly concerned with the poorest states. However, misclassification of the poor. This finding ­ not this remains a cause of concern, both with respect to shown in the table below- is simply an extension the efficiency of public programs which are based on of the above point on cross state variations, but the BPL mechanism, and of equity, given the still high is quite noticeable within several states, including exclusion of the poor by the BPL criteria even in the Assam, Bihar and Orissa.353 poorest states. 353 TheresultsbyNSSregionarepresentedinJalanandMurgai(2008). 12 Social Protection for a Changing India: Volume II Figure 8.1: BPL misclassification of the poor and rural Table 8.3: Poverty rate and targeting errors in the BPl poverty across states classification, by expenditure class (1999/00) 90.0 Poverty Rate 80.0 (Planning Targeting errors Undercoverage by state (%) NSS 55th NSS 61st 70.0 Commission) (2002 BPl census) 60.0 expenditure expendit- BPl score under- 50.0 class ure based Based coverage leakage 40.0 Poorest 10% 100.0 63.2 36.8 ­ 30.0 2nd decile 100.0 47.6 52.4 ­ 20.0 3rd decile 69.7 36.2 62.2 27.3 10.0 4th decile 0.0 31.9 ­ 100.0 0.0 3rd quintile 0.0 23.0 ­ 100.0 0 10 20 30 40 50 60 70 80 90 100 4th quintile 0.0 14.3 ­ 100.0 Rural Poverty Rate, by state (%) Richest 20% 0.0 8.0 ­ 100.0 Source: Jalan and Murgai 2008. NSS 55th round refers to 1999/200 data; NSS Total 27.0 27.0 49.1 49.1 61st to 2004/05 data. Notes: Leakage is the percentage of the BPL poor that is actually (expenditure-based) non-poor. Source: Jalan and Murgai 2008, using 1999/2000 NSS data. looking at the household level, the effectiveness of the bpl method varies significantly across the distribution in terms of how well it targets, both in including the on the positive side, targeting errors of the BPL poor and excluding the non-poor. Figure 8.2 illustrates method decrease sharply with higher per capita this; Table 8.3 also reports poverty rates (expenditure incomesabovethepovertyline.This implies that based and BPL score based) and under-coverage and the BPL indicator works relatively well in excluding leakage rates by per capita expenditure classes. Some the more rich amongst those above the poverty important observations emerge: line. Inclusion errors are largely concentrated amongst households that are only marginally Figure 8.2: Predicted targeting errors in the BPL above the poverty line. Targeting errors in the classification (1999/00) richer expenditure classes are, by comparison, marginal. however ­ and more worryingly - the BPL .8 Density of Per Capita Expenditures mechanismisproblematicbelowthepovertyline. % of households misclassi ed Even though targeting errors tend to decrease .6 with distance from the poverty line, errors of exclusion are high even amongst very poor .4 households. For example, over 20 percent of the population with expenditures which are half the .2 poverty line are misclassified as non-poor. In the poorest decile, a large share of the population 0 (around 37 percent) is incorrectly classified as being non-poor. -2 -1 0 1 2 Log reported consumption per capita, normalized by poverty line targetingerrorsarehighestintheneighborhood BPL Expenditure distribution of the poverty line. Thus, at the third decile, the Note: The height of the curve captures the extent of targeting errors made BPL indicator misclassifies 62 percent of the poor at different points in the distribution. as BPL non-poor and 33 percent of the non-poor Source: Jalan and Murgai 2008, using 1999/2000 NSS data. are classified as BPL poor.354 354 Thethirddecileofthepercapitaexpendituredistributioncorrespondstotherange(-0.1to0.3)onthex-axisinFigure7.2.Theshortrangearisesbecauseofthe largemassofpeopleconcentratedveryclosetothepovertyline.Bycontrastvaluesbelowandupto-0.26onthex-axiscorrespondtothepoorest10percentofthe population,withthemassofthepopulationtowardsthetopendofthatrange. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 1 in sum, the targeting errors of the bpl design imply omission of some socio-economic characteristics large welfare losses, both to households, and in (e.g., caste) that are known to be strong correlates terms of efficiency of public spending which is based of poverty on the bpl mechanism. The BPL method does a relatively potentially discourage households from investing good job at classifying the rich correctly. However it in schooling, housing and sanitation performs significantly worse in the lower part of the expenditure distribution, and the errors are not just Assessing the performance of indicators in identifying concentrated around the poverty line, indicative of large the poor empirically using 1999/2000 NSS data welfare losses. confirms the first of these concerns. Figure 8.3 plots the distribution of scores for each indicator across expenditure classes. There are three categories of (c) why does the BPl mechanism indicators apparent: perform poorly? some indicators are clearly ineffective in sorting Sundaram (2003) highlights two broad sets of concerns out rich from poor. For example, the scores of with the 2002 BPL methodology.355 First, the sub- indicators such as the preferred form of assistance indicators used to construct the overall BPL score and food security do not vary significantly across may themselves be a poor description of poverty and expenditure classes. second, the assumptions underlying the scoring and some indicators vary between rich and poor, aggregation method in the BPL indicator may not be but in the wrong direction. For example, while valid. This section explores both reasons empirically. the status of children 5-14 years does show some variance, the gradient is not the expected (i) Choiceofindicators direction. A larger share of the richest 20 one reason why the bpl score performs poorly may percent of the population has a score of zero be the choice of indicators. The set of indicators as compared to the poorest 10 percent of the that are included in the aggregate score are assumed population.356 to reflect a household's "quality of life". As a result, a other indicators show a shift in the distribution ranking of households based on some combination of scores in the expected direction, but the of these indicators is expected to reflect the relative differences in the poorer half of the population positioning (presumably in terms of poverty status, or arenotverystrong.That is, the types of indicators long-term economic status) of each in household in a that are included in the overall BPL score do not village. As the purpose of the census is identification of do particularly well in sorting out households poor households, there should be a clear link between into poor and less poor categories amongst the indicators and the underlying concept of poverty. the lower half of the expenditure distribution. In addition, the indicators should be clearly measurable For this fundamental reason, indexes based on and verifiable. The choice of the 13 indicators have been these scores, regardless of the weighting and criticized on the following grounds: aggregation method used are unlikely to be poor performance of some indicators in effective targeting instruments. separating the very poor from the poor or even another factor that weakens the targeting power the poor from the rich of the bpl mechanism include using the same set of imposition of the same set of indicators across indicators is used in all states. Given India's diversity, very different local contexts one question is the variability in the method's targeting 355 SeealsoHirway(2003),MehrotraandMander(2009)andJain(2004).AlkireandSeth(2008)havealsoarguedthatthecriteriafocusmainlyonresourcesrather thancapabilities.Others(e.g.,MehrotraandMander2009)pointouttheoperationaldifficultiesofmeasuringcapabilitiesinthefieldduringacensusoperation. 356 Thisrelationshiparisesbecausehouseholdswithnochildren5-14areassignedascoreofzero,asarehouseholdswithchildrenwhoareworkingandilliterate. Assigningascoreofzerotohouseholdswithnochildreninthe5-14yearagegroupartificiallypushesthesehouseholdsintotheBPLset. 1 Social Protection for a Changing India: Volume II Figure 8.3: Distribution of individual BPL indicator scores across expenditure classes Size of operational land holding Availability of normal wear clothing Food Security 100 100 100 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Ownership of consumer durables Literacy status of highert literate adult Status of household labor force 100 100 100 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Means of livelihood Status of children 5-14 years Preferred form of assistance 100 100 100 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Primary source of energy for cooking Primary source of energy for lighting 100 100 80 80 Score=4 60 Score=3 60 Score=2 40 40 Score=1 Score=0 20 20 0 0 Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Poorest 10% Next 10% Next 10% Next 10% Next 20% Next 20% Richest 10% Source: Jalan and Murgai (2008), based on 1999/2000 NSS. power across states. Table 8.4 presents results on finally, the bpl indicators do not include some under-coverage of households in the bottom two strong correlates of poverty such as caste. This deciles of the distribution in a selection of states. This information is collected in the BPL survey but is not indicates significant variation in under-coverage of built into the scoring formula. There may be sensible the poorest across states, ranging from a very high political or other reasons for these decisions, but 80 percent under-coverage in Andhra Pradesh to around it remains undeniable that a number of indicators 30 percent in Assam for the poorest decile. This confirms included in the BPL scoring formula have much less the limitations of using standard national criteria in the power as predictors of poverty than some of those that 2002 BPL methodology. were omitted. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 1 Table 8.4: under-coverage rates of BPl method for for much of the year can be treated the same as the poor, selected states non-ownership of any of the consumer durables. under- thesameaggregationprocedureisusedinevery expenditure BPl-based coverage of state, implying that the weights assigned to based poor poor BPl method indicators are the same across all states. This State/decile (% hh) (% hh) (%) implies, for example, that literacy status of the Poorest decile highest educated adult in the household has the Andhra Pradesh 100 20.0 80.0 same impact in differentiating poor versus non- Assam 100 69.9 30.1 poor in Bihar as it does in Kerala, an assumption Maharashtra 100 50.8 49.2 that will clearly not hold for every indicator. As Rajasthan 100 52.7 47.3 another example, a household with an operational Uttar Pradesh 100 63.6 36.4 holding of 0.95 hectares of unirrigated land in a Second decile high rainfall state and another household with Andhra Pradesh 100 22.6 77.4 the same size of unirrigated land in a desert state Assam 100 55.5 44.5 will be assigned the same score. Maharashtra 100 46.8 53.2 careful analysis indicates that all these assumptions Rajasthan 100 33.1 65.9 implicit in the construction of the bpl score ­ of Uttar Pradesh 100 43.8 56.2 cardinality and equal weights across indicators, Sources: Jalan and Murgai (2008). within each category of indicator, and across states - are rejected by nss data. Overly restrictive (ii) MethodologyforconstructionoftheBPL assumptions are evidently to blame, at least in part, for score high targeting errors and associated welfare losses of the BPL indicator. The question is how much? We turn to The BPL methodology entails three key assumptions the extent to which relaxing these assumptions affects with respect to the construction of the BPL score: targeting performance. the scoring method transforms the data for each indicator to a uniform cardinal scale ­ scored as 0, 1, 2, 3 or 4, with zero representing (d) Alternative specifications to the de extreme deprivation ­ such that the difference jure targeting design of the 2002 between 0 and 1 (for example, the difference BPl census between being illiterate and having some primary education) is the same as between 3 In this section, two alternative models for a proxy and 4 (for example, the difference between means test to identify the poor are presented to explore having secondary education versus having a how different types of adjustment in the 2002 BPL graduate degree). Forcing cardinality can result methodology may improve targeting power. They are: in problematic rankings, as for the indicator on an alternative index using the same indicators means of livelihood which presumes that an astheBPLmethodologybutwhichrelaxthetwo `artisan' household is always better off than one assumptions of cardinality and equal weights. engaged in `subsistence cultivation'. The first is a regression-weights based index each indicator enters the aggregate score with that uses the estimated coefficients of the most an equal weight, implicitly assuming that each flexible specification - modified to allow different indicatorhasthesameimpactonpovertystatus. coefficients across states - to predict a household's Equal weights have the appeal of simplicity and per capita expenditure. apparent objectivity, but this masks the fact that the second approach is to use an expanded list the imposition of numeric equality is completely of indicators in a regression-based index that arbitrary. It results for example in a situation captures other household characteristics that where having less than one square meal per day mayaddexplanatorypowertowardsidentifying 1 Social Protection for a Changing India: Volume II the poor. To retain maximum flexibility in the (e.g.,statusofchildren5-14years)andaddothers index and consistent with the evidence that thatarestrongcorrelatesofpoverty(e.g.,sexof equal weights for indicators across states are not household head and caste). This index results validated by the data, coefficients of all regressors in a 32 percent reduction in under-coverage are allowed to vary across states.357 compared to the BPL model. The main reason for the reduction in exclusion error stems from the Poverty and under-coverage rates for each of the fact that the model identifies 80 percent of the above methods are reported in Table 8.5, and several poor in the poorest 10 percent of the distribution, conclusions emerge: compared to only 63 percent by the BPL index. thefirstisthatrelaxingtheassumptionsimplicit Clearly, this is a significant improvement if the intheBPLmechanismimprovestargeting,butthe poorest are the group about whom policy is gains are not substantial. The regression-based most concerned. Inclusion errors are also lower, weighted index which discards the assumptions although this is mainly because again the of cardinality and equal weights across indicators augmented regression model performs better at reduces under-coverage only marginally, from correctly assigning people in the top 40 percent 49 percent to 45 percent. as non-poor. the big gains in improved targeting come from relaxingtheassumptionsandrevisingthelistof why does even the most flexible regression model, indicators to discard some that are problematic with a long list of indicators, not perform better? An examination of the poverty rates in Table 8.5 sheds some light. The augmented regression model works well Table 8.5: Poverty and under-coverage rates across in identifying the poorest among the poor. However, different targeting methods targeting errors are much higher in the vicinity of the expenditure Regression based Augmented poverty line (in the third and fourth expenditure deciles), Class weights regression model and this is a problem evident in the regression-based I. Poverty Rates by Expenditure Class model as well. With the high density of population with Poorest 10% 66.0 80.3 income close to the poverty line (see the discussion in (4.4) (27.1) Chapter 1), arriving at an effective proxy means test is an 2nd decile 51.4 62.1 inherently problematic and difficult exercise. (7.9) (30.4) 3rd decile 41. 6 45.6 as with the 2002 bpl method, there are notable (14.7) (25.8) differences across states in how well alternative 4th decile 32.2 30.1 proxy means test (pmt) specifications perform (0.9) (-5.4) relative to the bpl method. Table 8.6 presents the 3rd quintile 22.7 16.7 percentage improvement of the two specifications for (-1.2) (-27.2) a selection of states (i.e., only the figures in parenthesis 4th quintile 12.2 7.1 in the above table), focusing once more on the bottom (-14.5) (-50.6) two deciles. A few points emerge which have relevance Richest 20% 4.3 1.8 (-46.7) (-77.1) to any BPL reform: Overall 26.7 for all states except Rajasthan, the improvements II. Classification errors in targeting performance are significantly higher Under-coverage 45.4 34.5 for the poorest than the second decile for all (-11.3) (-32.5) specifications. Notes: 1. Numbers in parentheses are percentage change from the BPL score index. 2. Assumptions underlying the alternative models are given for the poorest decile, the variations are very in Annex 5. strong between states where all alternative Source: Jalan and Murgai 2008 using 1999/2000 NSS. specifications produce significant improvements 357 Theexpandedlistofindicatorsincludesdemographic,occupationalandeducationalvariablesinadditiontoselectedBPLindicatorsthatwerefoundtobestrongly correlatedwithpercapitaexpenditures. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 1 Table 8.6: improvement over BPl method in coverage associated with the implementation of the bpl census rates for lowest two deciles by state across that negatively impact data quality and eventually different targeting methods (%) targeting efficiency. International evidence suggests Regression Augmented that a large fraction of the observed differences in State/decile weights regression targeting effectiveness across systems and programs Poorest decile can be attributed to factors related to implementation Andhra Pradesh 11.5 35.7 and monitoring.358 This implies that investments aimed at correcting some or all of these limitations could go Assam 10.0 17.5 a long way in improving targeting outcomes in the Maharashtra 15.0 30.5 region. The following concerns have been highlighted Orissa 5.9 15.5 by studies and surveys in various states: Rajasthan -12.2 -3.0 Problemsindesignofthequestionnairefordata Uttar Pradesh -2.7 11.3 collection:There are concerns about the precise Second decile phrasing and coding of questions in the BPL Andhra Pradesh 8.2 21.8 census questionnaire. Some indicators, as asked Assam 5.7 10.5 in the questionnaire, are not clearly measurable Maharashtra -1.3 5.5 or verifiable (e.g., availability of clothing, form of Orissa 6.3 13.0 assistance preferred). Codes for some indicators Rajasthan 0.3 21.5 are unclear or not necessarily mutually exclusive Uttar Pradesh 6.5 16.4 with unclear guidelines for implementation, Source: Jalan and Murgai (2008) using 1999/2000 NSS data. implying subjectivity in assigning scores (e.g., status of household labor, means of livelihood; it is in coverage, and Rajasthan and UP, where they also unclear how to assign scores for households would actually worsen coverage. with no children).359 the augmented regression model dominates all Poor quality data collection processes: Effective other methods (and strongly dominates the BPL implementation of the BPL Census requires a method) in nearly all states. clear definition of institutional responsibilities and sufficient administrative capacity. While the design is the responsibility of the Center, data (e) defacto targeting ­ how well does collection in order to calculate household-level the 2002 BPl method perform in welfare scores is done at the local level. A common complaint from block-level government officials is practice? of inadequate staffing, training and time allowed The above analysis examines the BPL methodology for the enumeration. In addition, since the task in terms of its design ­ i.e., assesses how well it would is performed by teachers, health workers and perform assuming that it was implemented precisely as other village-level government employees, this intended. In this section, we turn to the problems in the responsibility comes in addition to their regular implementation of the household census and evidence work load.360 There is little standardization of on BPL outcomes in practice. processes with little or no quality control from higher levels of administration. These problems (i) ImplementationoftheBPLcensus are exacerbated given the significant expansion of the required coverage of the detailed survey in there are substantial capacity and logistic 2002 after dropping of the exclusion criteria used limitations as well as local institutional pressures in the 1997 method.361 358 Coadyetal.(2004). 359 SaxenaCommitteeReport(2009). 360 Basedonanecdotalevidencefromlocalizedsurveys. 361 Sundaram(2003). 1 Social Protection for a Changing India: Volume II Importance of local institutions and elite and significant inclusion of the non-poor.365 Since capture:Available evidence confirms widespread BPL status is the first step to access a range of anecdotal information that participation in benefits, there is considerable incentive for non- programs according to BPL criteria is affected poor households to pressure local authorities for by operation of local institutions. However, it is inclusion in the BPL list or for getting a BPL card. inconclusive in terms of systematic patterns across Grievance redressal and recertification of states in how the BPL list system is implemented households: Another important shortcoming is in practice. For instance, in three villages in UP, the the lack of attention to grievance redressal as well survey was substituted by a list of BPL households as recertification (as the eligibility of households drawn up by the Village Development Officer in may change over time) in the period between consultation with the Pradhan, and forwarded to two censuses.366 Some states have introduced district level. At the same time, it was expected an appeals process against the BPL list, a large some names from the list would be deleted at number of appeals remain unresolved. There higher levels of administration due to ceiling is no systematic process to update household on total number of poor. Similarly, studies of information or status following changes in AP, Karnataka, Kerala and TN found that: (i) the household circumstances in the five year period inclusion of landless and illiterate households in between censuses, e.g., due to deaths, migration, BPL lists was more intensive in villages that had formation of new households, changes in socio- held gramsabhas, with the probability of inclusion economic status, and so on. This lack of processes 8-10 percent higher;362 and (ii) the intensity of BPL for dynamic updating are compounded by poor coverage of SC/ST households was increased by management of the BPL database in most states. around 7 percentage points where the GP was Validation checks of the BPL administrative reserved, indicating the importance of caste database for selected states revealed two critical and other connections with the Sarpanch for problems in the BPL database:367 (i) internal purposes of being determined as BPL at the inconsistencies and inaccuracies in the database village level.363 Further evidence from West Bengal and (ii) lack of clarity on a master BPL list at the on participation in BPL-targeted programs found: state-level given dynamic updating at local levels, (i) targeting within villages appeared to be fairly with no corresponding mirror changes at the effective and not overly captured by local elites; state-level. Consequently, differences between (ii) targeting within villages did not vary greatly the state and local-level registers are not only according to village characteristics, though the significant but bound to grow over time as more employment-intensity of public works and SC/ST appeals and grievances are being processed, share of targeted credit worsened as inequalities severely undermining the system's integrity, in land ownership increased; but (iii) allocation transparency and credibility. The accuracy and of resources across villages exhibited significant reliability of BPL lists varies widely across the difficulties, with a 25 percent reduction on public country and even within states. works resources and a 70 percent reduction in all grants associated with a 5 percent increase in (ii) BPLtargetingoutcomesinpractice the share of poor SC/ST households in a village.364 Another localized study from West Bengal found it is useful to supplement these insights on design and significant manipulation by local elites of BPL data collection process with evidence on bpl outcomes lists, with severe under-coverage of the poorest in practice. Surprisingly, there is less empirical evidence 362 Besley,PandeandRao(2005). 363 Besley,Pande,RahmanandRao(2005). 364 BardhanandMookerjee(2006). 365 Mukherjee(2005). 366 Sundaram(2003),AlkireandSeth(2008). 367 InthecontextofdevelopingaconsistentbeneficiarydatabasefortheRashtriyaSwasthyaBimaYojana. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 1 than might be hoped on this question, and the existing is based on the 1997 list. As a result, it is not possible evidence is not always clear on the distinction between to conclusively assess whether field level experiences BPL ration cards, and being on BPL lists for other are different from the predictions discussed above. purposes. The 2002 BPL list was operationalized only However, it is possible to analyze targeting outcomes post-2006 due to a stay order passed by the Supreme at the household level as given by the distribution of Court on a writ petition filed by the People's Union for ration card holding. Civil Liberties which alleged that the new methodology analysis of household expenditure and ration card would reduce the number of persons identified as holding using the 2004-05 nss data indicates that BPL and a large number of the poor would lose their targeting outcomes are even worse in practice entitlements. The analysis below relies on 2004/05 NSS data when the 1997 BPL list was operational and (see Table 8.7).Several observations emerge: the 2002 BPL list had not yet become operational.368 At the national level, the percent of BPL Even in the case of two states, Himachal Pradesh and cardholders and the official poverty rate is the Rajasthan, where ration cards are issued on the basis of same. This is unsurprising given the stipulation by the BPL list, the estimates of dejure and defacto under- the central government that the number of BPL coverage are not strictly comparable. This is because the poor be the same (or not more than 10 percent) estimate of dejure under-coverage is based on the 2002 as the number of poor persons as estimated by BPL methodology while the BPL ration card holding the Planning Commission for 1999-2000. Table 8.7: defacto targeting even worse than dejure Rural Poverty Rate Rural BPl under-coverage (%) leakage (%) State (%) cardholders (%) dejure defacto defacto Adhra Pradesh 10.0 60.6 74.4 19.0 86.7 Assam 22.0 11.5 51.8 76.6 55.2 Bihar 42.5 17.1 35.9 78.2 45.8 Gujarat 18.8 38.3 51.0 43.9 72.4 Haryana 13.3 18.6 60.1 66.8 76.3 Himachal Pradesh 10.5 18.0 56.7 55.4 73.9 Karnataka 20.6 52.0 56.1 28.7 71.7 Kerala 13.1 30.0 67.2 50.3 78.3 Madhya Pradesh 36.9 35.4 34.8 47.5 56.1 Maharashtra 29.6 34.1 45.0 53.3 49.5 Orissa 46.9 45.5 31.2 40.8 39.0 Punjab 9.0 11.7 74.4 79.3 84.1 Rajasthan 18.3 17.9 59.9 66.7 65.9 Tamil Nadu 22.8 21.1 53.5 68.7 66.1 Uttar Pradesh 33.2 15.9 47.0 75.5 48.8 West Bengal 28.4 30.2 48.2 58.3 60.8 16 Major States 27.3 28.7 45.8 61.2 63.2 Notes: 1. BPL cardholders are identified in the data as those that possess a BPL or AAY card. 2. Under-coverage is the percentage of the poor population wrongly classified as BPL non-poor (dejure ­ as per 2002 BPL methodology) or not possessing a BPL card (defacto). 3. Leakage is the percentage of the BPL card holder who are actually (expenditure-based) non-poor. 4. Rural poverty rates are estimated using the official Planning Commission state-specific rural poverty lines for 2004-05. Source: Jalan and Murgai (2008) based on NSS data for 2004/05. 368 JalanandMurgai(2008).ThisNSSroundincludesadirectquestiononwhetherhouseholdspossessarationcard,andifso,whethertherationcardisaBPLcard. BPLcardholdersareidentifiedinthedataasthosethatpossessanAntodayaorBPLcard.Householdswiththe"other"typeofrationcardsareclassifiedasAPL (above-poverty-line). 200 Social Protection for a Changing India: Volume II However, this is not true across the different Table 8.8, by monthly per capita expenditure quintiles states. In states like Andhra Pradesh, Gujarat, and by social category for rural and urban households. Karnataka and Kerala, the percent of BPL On average, about 3 percent of rural households cardholders are three times more than the official possess AAY ration cards, a quarter possess BPL ration poverty rate. In these states, de facto under- cards and 52 percent possess APL ration cards, while coverage rates are lower than de jure, but the 19 percent report no ration card holding. In urban leakage rates are also very high.369 areas, ration card holding is much lower, only about At the national level, there is approximately a 11 percent reporting BPL ration cards, less than one 15 percent difference in under-coverage rates percent reporting AAY cards, and about a third of between de jure and de facto targeting. At the urban households reporting no ration card holding.370 national level and for the majority of states Several observations emerge, including: defacto targeting is even worse than de jure. possession of BPL ration cards is progressive, Nearly two thirds of the poor are under-covered though there remain significant inclusion and by BPL cards; leakage is just as high ­ 63 percent exclusion errors, and the differences in the BPL of BPL/AAY ration cards are held by the non-poor. rates of the bottom 40 percent are not as great as might be expected (at least in rural areas). findings for all-india indicate that bpl and particularly aay ration card holding is progressive, despite the general progressivity of BPL cards, though with worrying exclusion among the poorest there remain significant inclusion errors in BPL and substantial inclusion errors. Results on the cardholding, with around 35 percent of rural BPL incidence of ration card holdings are presented in cards held by households in the top 40 percent of Table 8.8: Possession of ration cards, by type and socio-economic status (% households) Rural urban quintile/Social group AAy card BPl card APl card no card AAy card BPl card APl card no card I. Expenditure quintiles Poorest 5.8 35.3 37.9 20.9 2.8 26.2 43.0 27.8 Q2 3.6 31.4 47.6 17.2 1.2 17.0 54.8 27.0 Q3 3.0 28.5 52.0 16.2 0.5 9.7 59.1 30.7 Q4 2.1 23.8 56.9 17.1 0.3 5.7 58.9 35.1 Richest 1.2 17.7 59.9 21.2 0.1 1.7 58.3 39.7 II. Social group ST 5.0 39.6 30.8 24.3 1.3 13.6 37.6 47.4 SC 4.4 34.9 43.7 16.9 1.6 17.3 49.8 31.2 OBC 2.3 24.5 54.5 18.5 0.9 14.4 51.5 33.1 General 1.9 17.3 63.0 17.6 0.4 5.1 61.8 32.5 All-India 2.9 26.5 51.8 18.6 0.8 10.5 55.6 33 Notes: 1. Households possessing "other" type of ration cards are classified as APL. 2. The expenditure quintiles are calculated for rural and urban households separately on the basis of real monthly per capita expenditure (i.e., corrected for cost of living differentials across states) expressed in 2004/05 rural and urban prices respectively. Source: Staff estimates using 2004/05 NSS data. 369 TherelativelylowpercentageofBPLcardholdersinTamilNaduisaconsequenceofthefactthatthestateprovidesuniversalaccesstosubsidizedgrainevenafter theintroductionofPDSsothatthedistinctionbetweenBPLandAPLcardholdersismeaninglessforhouseholds.SeeChapter3foradiscussion. 370 Thesefindingsaresomewhatsimilar(especiallyinruralareas)tothosereportedbyRametal.(2009)using2005/06NFHSIIIdataandbyAjwad(2006)usingthe 2004/05IHDSdata.Theformerreportsthatabout27percentofallhouseholdspossessBPLrationcards.AccordingtotheIHDSsurvey,nationally,34percentof allhouseholdspossessBPLcards,3percenthaveAAYcardsand47percenthaveAPLcardswhile15percenthavenorationcards.Thedistributionofrationcard holdingacrossassetquintilesusingthesealternativedatasetsalsodelineatesasimilarstory­i.e.,mildlyprogressiveBPLcardholding,butwithhighexclusion and inclusion errors. However, BPL card holding of the richest quintile drops more sharply when comparing across asset quintiles than across expenditure quintilesasintheNSSdata. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 201 the distribution in rural areas. The corresponding Table 8.9: Determinants of ration card holding figure for urban areas is much lower at 16 percent BPl card AAy card of urban BPL cards. variables holders holders AAY ration card holding is notably more Quintile 1 0.3536*** 0.0481*** progressive, including in the lower ends of the Quintile 2 0.3455*** 0.0472*** distribution, with almost 33 percent of all rural Quintile 3 0.2730*** 0.0298*** AAY cards held by households in the bottom Quintile 4 0.1714*** 0.0211*** quintile, and almost 55 percent by households Other caste 0.0130 0.0032 in the bottom two quintiles. In urban areas, the OBC 0.0501** 0.0045 poorest households hold up to 54 percent of all SC 0.1404*** 0.0154** urban AAY cards. ST 0.1734*** 0.0129* amongthosewithnorationcardatall,thereare Rural 0.1748** 0.0069* no sharp differences between the poor and the HH size 0.0007 - 0.0002 rich and across social categories, though tribals Female HH head 0.0081 0.0063* are somewhat less likely to have a card. Age of HH head 0.0005* 0.0001* HH head literate 0.0113 - 0.0048** probit regression models examining the role of HH head completed primary - 0.0181* 0.0005 household and village characteristics in determining HH owns agric. land - 0.0001* - 0.0001 bpl and aay ration card holding support this finding HH owns animals - 0.0071 - 0.0043 (see table 8.9).371 Thus, households in the bottom quintile have the highest probability of possessing AAY or BPL Notes: ***=significant a 1% level; **=at 5% level; *=at 10 percent. Quintile 5 and other caste are the (omitted) reference group. State and other controls cards, followed by those in the second quintile and so on. also included. However, differences between the bottom two quintiles Source: Ajwad (2006) based on the 2005 IHDS data. are small.372 With respect to inclusion errors, households in higher quintiles also have a fairly high probability of may vary across states. For example, the poor and tribals possessing cards. Caste is a significant determinant of in Orissa are more reliant on their (often weak) social ration card holding. Other household characteristics networks to access ration cards, while in Karnataka, PRIs have also the expected effects.373 In addition, Ram et play a more active role in facilitating access.375 al. (2009) find that a significant percentage of BPL card holders appear to be fairly well-off in terms of assets owned. For instance, about 18 percent of BPL card holders own more than five acres of agricultural land, C. AlTeRnATive TARgeTing 11 percent live in pucca houses with three or more meThoDS rooms, 10 percent own a motorized vehicle and so on. Also, anecdotal evidence, field experiences relayed by The BPL method examined above (and the BPL ration NGOs, and field survey based studies suggest instances cards) is the most commonly used targeting tool of both severe under-coverage of the most needy and for several anti-poverty programs, especially those coverage of the economically better-off population.374 administered by the Ministry of Rural Development. An important additional finding from the three state SP However, a number of different targeting methods are study is that the reasons underlying differential access available for directing resources to particular groups 371 For consistency with previous models of participation in various government programs (see Chapter 3 for PDS and other social assistance programs), these regressionmodelsarebasedonIHDSdata.Thesedataalsoallowsforaricherspecificationofhouseholdandvillagecharacteristics.SeeAjwad(2006)forresults. 372 However,thestrongimpactsofbeinginlowerquintilesarehighlydilutedwhenonelooksatactualusageoftheBPLrationcardsintheprevious6months (seeChapter3). 373 Inaddition,theeffectoflocationisverypronouncedforBPLrationcardholding(notreportedinthetable).Broadly,householdsinmostnorthernstatesare substantiallylesslikelytoholdBPLcards(over20percentlessinmost,andapproaching30percentinRajasthanandPunjab).Conversely,householdsineastern andcentralstatesaresubstantiallymorelikelytoholdaBPLcard,andsouthernandwesternstatesbroadlyexhibithigherpositiveparticipationeffects. 374 See,forexample,Mukherjee(2005). 375 SeeDevetal.(2007). 202 Social Protection for a Changing India: Volume II (see Box 8.1 for a brief description). All these methods, Table 8.10: Type of targeting by SP programs with the exception of the first (means testing), are in Program Type of targeting use in India. PDS grain Ration card thus, sp programs in india direct resources to Annapurna grain Ration card particular groups using a range of different Rural credit (SGSY) BPL list targeting methods. These include the BPL method Rural public works Self-targeting (proxy means test), self-targeting (in public works), and (MGNREG, SGRY, FFW) mixed methods (e.g., social pensions). See Table 8.10 for Rural housing support (IAY) BPL list a mapping of programs and targeting tools. The 2002 IGNOAPS Categorical (age) + BPL list BPL method examined above is a form of a proxy means Disability pension Categorical (disability) test, i.e., a targeting tool that relies on readily observable Widow pension Categorical (widowhood) and verifiable socio-economic proxies as substitutes Subsidized health BPL list for more expensive and difficult to collect income or insurance (RSBY) expenditure information. All programs combine their Scholarship Categorical (social group, specific targeting methodology with a geographic gender) element in terms of fiscal allocations. In addition, several Sources: Program guidelines. Note that NOAPS was earlier targeted using donor and NGO initiatives rely on community-based state-specific definitions of "destitution" in IGNOAPS; these have now been targeting methods. In some states, notably Andhra substituted by BPL status. Box 8.1: main types of targeting methods a. means tests A means test seeks to collect (nearly) complete information on households' income and/or wealth. Sometimes this information is fully/ partially verified against independent sources, other times verification is not possible and program intake workers simply record what the applicant says. Verified means tests are considered to be the gold standard of targeting but are usually difficult to implement in developing countries. b. Proxy means tests Proxy means tests generate a score for each applicant household based on fairly easy-to-observe household characteristics believed to be correlated with the household welfare status. The information provided by program applicants is usually partially verified either by program officials or by communities. Eligibility is determined by comparing the household score against a predetermined cutoff. India's 2002 BPL Census method falls in this category. c. Community-based targeting Community-based targeting uses a group of community members or leaders whose principal functions in the community are not related to the program to decide who in the community should benefit. d. geographic targeting With geographic targeting, location determines eligibility for benefits: people who live in the designated areas are eligible and those who live elsewhere are not. Few programs target only on the basis of geography, but many programs combined this criterion with other targeting methods. e. Demographic/categorical targeting The usual and simple forms of demographic targeting are based on age (e.g., elderly), or on some other individual characteristic commonly perceived to be associated with a higher likelihood of being poor, vulnerable or socially excluded (e.g., widows, disabled, certain caste groups). f. Self-targeting Self-targeted programs are technically open to everyone, but are designed in such a way that the take-up is expected to be much higher among the poor than the non-poor or the level of benefits is to be higher among the poor. A common example is the use of low wages in public work programs to induce participation primarily among the poor. Source: Grosh et al. (2008) Chapter 4. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 20 Pradesh and Kerala (to a more limited extent), this based targeting is real. At the same time, empirical approach has been adopted for the government's anti- evidence on the BPL system suggests that it is also poverty programs as well. We turn to three methods subject to capture and manipulation in such instances. that are currently in use in India, but could potentially While both methods therefore have the risk, a mediated occupy a larger place in the targeting system. community process such as described in Box 8.2 would appear to have greater channels for making such capture both explicit and subject to community account than (a) Community-based targeting the BPL system. a very different approach from the bpl targeting currently, andhra pradesh presents an interesting methodology is village-based wealth ranking of case of how community based targeting methods have households conducted by communities themselves. been rolled out, and have now been absorbed into the This is done in a growing number of states under mainstream system for identifying bpl households. donor and NGO initiatives. The obvious advantage Through project-financed activities, community wealth of the wealth ranking approach is its reliance on local ranking has been conducted in nearly all of AP in recent knowledge of factors at the household level which years. Most interesting from a systemic viewpoint is that increase or decrease poverty and vulnerability, but the credibility of the exercise has been sufficient that which cannot be well measured or taken into account by PRIs and public administration have agreed that the more aggregated administrative systems. The potential lists of poor households identified through this process disadvantages include: (i) localized assessments need to should be the basis for official identification of BPL. match with program allocation systems and financing if The experience and method of the ranking exercise is they are to be useful for allocating funds across space; outlined in Box 8.2. Interestingly, even in cases where the (ii) there may be economic and other costs (e.g., potential community-based process results in households being ostracization) incurred by the participating community removed from the official BPL lists, there has to date members; and (iii) such methods may also be subject to been fairly limited resistance, due to the village-wide local capture, just as administrative systems may be. buy-in on the process underlying the new identification international and regional evidence provides some method. A similar process is now being debated in Tamil guidance on the benefits and limitations of community- Nadu, and there is interest from PRIs in particular for based targeting.376 The basic trade-off which such exploring the Andhra Pradesh convergence experience. programs face is between the "informational advantage" Kerala follows a more hybrid model where communities that communities possess in targeting, against the are heavily involved in the validation and finalization of accountability risks of capture by local elites and loss the list of poor households. This targeting method was of spatial comparability across communities. Evidence developed in the context of Kudumbashree, but applies from the Food-for-Work program in Bangladesh, for to other government anti-poverty programs that have example, found substantial variation across village in the converged with Kudumbashree (See Box 8.3). Himachal targeting performance of community-based methods, Pradesh is also experimenting with community selection with no systematically better performance in poorer of households for Rural Development programs and villages. However, a clear result was that village with PDS from within the official BPL lists, with annual review greater within-village inequality had worse targeting of BPL status by the gram sabha. Bihar is a further outcomes.377 The empirical finding is supported by interesting case of a review and appeals process with theoretical literature that points to the importance respect to the 2002 BPL list which involved several of factors such as local asset inequality and social rounds of gramsabha meetings to revise the initial list networks.378 The risk of local elite capture of community- generated from the BPL census. 376 Ravallion(2003)providesasummary.SeealsoformoredetaileddiscussionConningandKevane(2002). 377 GalassoandRavallion(2002). 378 BardhanandMookerjee(2000)andBenabou(2000)onassetsandSpagnolo(1999)onsocialnetworks.SeealsoAlderman(2002),onAlbaniaanditsuseof community-basedtargeting,andCoudoueletal.(1998)ontheUzbekistanmahallas'scheme,whichusesacombinationofcentralguidelinesandcommunity- basedtargetingforsocialassistance. 20 Social Protection for a Changing India: Volume II Box 8.2: Participatory identification of the Poor (PiP) in Andhra Pradesh Under its Velugu program (now IKP), AP has conducted a widescale PIP process in 840 mandals in 22 districts of the state. This is based on methods developed over time in a number of NGO, donor and in some cases public programs. A key benefit is that the PIP process is able to take account of more dimensions of poverty and social exclusion than administrative systems are typically capable of. Based on this exercise, 14.2 million rural households were surveyed, of which 7.9 million. (55 percent) households were categorized as BPL, and of these 2.6 million. were rated as "the poorest of the poor". The key steps in this process are: 1. There is an intensive identification and training process for workers who act as facilitators of the PIP process. This is accompanied by a workshop with various government and civil society stakeholders to seek support for the process. 2. There is then a three day process in villages which has several elements: developing rapport with key village functionaries and service providers, including gathering information on key village and community characteristics community meeting on purpose of the exercise, including probing more on the operation and awareness of current schemes. There should be agreement on the timing and place for development of a social map of the village (see below) "transact walk" in the village as preparation for social mapping development with villagers of a social map, including social and infrastructural facilities, local institutions, use of services, social profile of the village, social clustering within the village by caste or other indicators, and other factors feedback from villagers on the social map to verify its findings, and preparing a report on these findings a "wellbeing analysis". In consultation with villagers, this looks into what are the key correlates and determinants of poverty in the village, and what are the key issues that the poor face. This should allow for a general categorization (and quantification) of households into four groups: the rich, the middle class, the poor, and the poorest of the poor identifying specific households within wellbeing categories, and listing them the household list should then be discussed in the gramsabha, and any changes incorporated. The ratification of the GramSabha is a key step in terms of community acceptance of the PIP findings. preparing a report which consolidates the findings, together with supporting documentation. This reports are tabulated at mandal and district levels, ratified by Collectors and printed. An important outcome of this process throughout AP has been agreement that the poor household identified through the PIP process should replace the administrative BPL list, subject to the aggregate "cap" on total number of BPL households allowed in any one village. Looking ahead, a key challenge will be convergence between the "macro" numbers on BPL and the micro PIP process in cases where the PIP identifies significantly greater (or lesser) number of poor households than administrative BPL numbers allow for. Box 8.3: Combining indicator-based targeting with community validation and finalization of the poor in kerala Kudumbashree was launched in 1998 by Government of Kerala as a participatory, women-oriented approach to poverty reduction. This program is implemented entirely through local self governments in both rural and urban areas. The program consists of a three-tier structure, with neighborhood groups (of women) at the bottom, federations into area development societies at the ward level and a community development society at the panchayat level. Kudumbashree presents an interesting case where indicator-based targeting is combined with a heavy reliance on community-based validation of the "BPL list" in a government program. The initial targeting tool used nine slightly different indicators for identifying the poor in rural and urban areas. Broadly, these included indicators for asset ownership, social exclusion, lack of employment or earning capability, and disability. A family was considered poor if four or more of these risk factors applied and most vulnerable if all nine risk factors applied. These indicators were chosen so as to be easily understandable by communities who had to identify poor households, to enable a ranking based on severity of poverty and yet be verifiable and easy to monitor. However, several shortcomings remained as some indicators were not directly correlated with poverty, others were amenable to manipulation, and equal weights were applied across indicators. These deliberations on the deficiencies of the previous method led to the evolution of a new set of indicators and targeting methodology. The same set of indicators are now applied in urban and rural areas. These include indicators related to capabilities (e.g., social groups, occupational groups, etc.) and entitlements (e.g., land, dwelling characteristics, basic amenities). Weights have also been assigned to these indicators but it is not clear if these have been assigned by committee or developed through a rigorous regression model. A ranking system has also been devised (along with prioritization of indicators in the case of similar scores). Certain exclusion criteria are also applied, e.g., regular employment, international migration of household members, and ownership of land, concrete house and other assets. This list of poor households ­ the equivalent of a "BPL" list ­ is prepared in four stages: a survey by officials, validation by the neighborhood groups of Kudumbashree, discussion in the gramsabha and final adoption by the panchayat. There is also an appeals process at the district level. While developed in the context of a specific program, Kudumbashree, this system of identification of the poor is used in several government programs in the state, including SJSRY (promoting wage and self-employment), VAMBAY (urban housing) and NSDP (slum development) in urban areas and for SGSY in rural areas. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 20 very little information exists which compares (iii) community ranking in this case is able to sort a small community-based wealth ranking outcomes with group of the very poor, and also to avoid large sections targeting outcomes from administrative systems. of the community being rated as poor. The fact that such The limited empirical information suggests that there an exercise in the aggregate produces a share of "poor" is overlap but also significant divergence between households which is not too far away from poverty community-based identification and administrative estimates based on NSS is also of interest. BPL identification. It also suggests that ranking by A second cut on community ranking is comparing to the community may produce somewhat different expenditure based measures of welfare, and Table 8.12 ranking outcomes among households than one based below reports findings from a smaller sample in the purely on per capita expenditures. Two examples are reported below for India, as well as one for Bangladesh in Box 8.4. Table 8.11: Comparing community-based and BPl Table 8.11 compares outcomes of a community-based targeting wealth ranking across a range of Indian states with the Community % of total % of category on status of households on BPL lists, based on a sample of wealth rank households BPl list over 3,000 households.The results indicate that: (i) BPL lists Very poor 5 71 exclude significant numbers of households considered Poor 27 57 to be poor and even very poor by communities; (ii) the Borderline 36 61 BPL lists seem to include significant shares of households Non-Poor 32 40 not considered to be poor by communities; and Sources: Sinha (2003). Box 8.4: Community targeting in Bangladesh ­ how well does it identify the poor? Given the scale of NGO poverty programs in Bangladesh, their experience with community wealth ranking is of interest. BRAC has since 2002 operated a program called "Challenging the Frontiers of Poverty Reduction: Targeting the Ultra-Poor". This relies on a several step process in identifying program beneficiaries: (i) development of program eligibility criteria based on a literature review to determine the common correlates of ultra-poverty; (ii) a community wealth ranking exercise which sought to distinguish the ultra-poor from the poor and non-poor; and (iii) based on the ranking exercise, a household survey of those identified by the community as ultra-poor, as a final tool for beneficiary selection. A study of outcomes of this process was produced in 2004 which found as follows (see table for results of survey-based analysis): the community did a good job in its ranking of distinguishing the poor and ultra-poor when cross-mapped to the "objective" indicators of poverty pre-identified from national data and studies. among the ultra-poor, the household survey was a further useful tool for distinguishing those eligible for the program and others. statistical analysis found that the differences between both the ultra-poor and others identified in the community ranking exercise, and among the ultra-poor between those eligible and not for the program were highly significant. indicator ultra-poor (% hh with non-ultra poor Beneficiaries after ranking characteristic) (% hh) plus survey (% hh) Widow 20 6 30 Divorced/abandoned 8 1 15 With no adult male 21 3 36 Physically able husband 64 88 43 Children working 12 7 18 No cultivable land 90 76 98 Average land holding 25.7 units 42.9 units NA Not own land of house 44 24 62 No non-housing assets 46 29 56 Borrowing from MFI 19 34 NA 20 Social Protection for a Changing India: Volume II Table 8.12: Comparing community-based and expenditure-based identification of poor for orissa expenditure per capita class less than 50% of 25 to 50% below 25% below poverty line Community ranking poverty line poverty line to poverty line Above poverty line Extremely poor 34.7 27.5 25.0 11.7 Highly poor 26.5 26.1 8.3 23.5 Average poor 30.6 23.2 29.1 35.2 Low poor 8.2 18.8 33.3 23.5 Non-Poor 0 4.3 4.2 5.8 Sources: Shah et al. (2005). Results based on a small sample (159 households in four villages) in Orissa. state of Orissa. While caution is necessary due to small operates in India mainly through public works programs, sample size and state-specificity, a few interesting primarily MGNREG (and implicitly to some extent in results emerge: (i) significant shares of the very poor in PDS through grain quality), and relies on some screen expenditure terms are not ranked among the poorest to allow beneficiaries to self-select into the program. by the community; (ii) conversely, around one third In the case of public works, the assumption is that the of household above the poverty line in consumption wage rate and nature of the work will lead those with terms are rated among the poorest groups by the sufficiently high opportunity cost or reluctance to community; and (iii) as one approaches the poverty line, undertake work to avoid the program. The targeting distinguishing among the poor becomes less effective. outcomes presented in the next section indicate that self-targeted public works have often done better than the only firm conclusion that one can draw from BPL-based targeting. the above studies is that community rankings differ to both bpl and expenditure-based measures. What while self-targeting can be an important element this implies depends on whether community ranking of targeting strategies, it is also important to is accepted a priori as more "correct" in identifying the look closely at which situations to which it is poor. In the case of the BPL comparison, the analysis of best suited. Self-targeting will be more suitable in this chapter suggests that that is a valid assumption. conditions where administrative capacity is weak (and However, for the expenditure-based comparison, the hence more intensive targeting methods unsuitable), conclusion is less clear. To some extent, this imperfect in crisis situations where the time lag in conducting overlap with consumption expenditure is not entirely more demanding forms of targeting are not desirable, surprising. This is because local definitions of poverty and where incomes are irregular. In addition, there and/or vulnerability typically encompass indicators that needs to be a suitable screen for the self-targeting. The may or may be tightly correlated with consumption- most common ones worldwide are the nature of work based poverty.379 and/or a low wage rate in public works, and subsidies on certain food items which are not consumed by (b) Self-targeting better-off people. On the other hand, it is important to acknowledge the limitations of self-targeting. A an important method of household-level targeting classic example is the exclusion of certain groups like which appears to have value both internationally and elderly and disabled (and in some countries, women) in india is self-targeting. The approaches examined from self-targeted public works due to the nature of previously assume an active process of identification of the work demanded. Another is that ­ by the nature poor households, whether through the BPL method or of the targeting tool ­ it would typically be difficult community wealth ranking. Self-targeting in contrast to provide high benefits even to the poorest, as this 379 Communitiesoftendefinepovertyintermsof(i)economicindicatorsassociatedwithlivelihoods,assetsandincome,(ii)abilitytomeetbasicneedsoffood, shelter,andclothing,(iii)healthandeducation,and(iv)indicatorsofinsecurity,exclusion,andlackofparticipation.See,forexample,WorldBank(2008b)and Krishna(2004,2006). CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 20 would blunt the targeting mechanism.380 Self-targeting but many programs combined this criterion with other programs by their nature impose participation costs on targeting methods. The advantage of this approach beneficiaries. While these are higher for the non-poor, is that it is administratively simple, requiring none of they are often also significant for the poor, e.g., evidence the machinery for individual assessment. However, on various workfare programs find opportunity costs geographic targeting will perform poorly when poverty of participation ranging from around one quarter is not spatially concentrated. It also depends on the (Maharashtra EGS) to around one half the gross earnings accuracy and level of disaggregation of the poverty in Argentina's Trabajar workfare program.381 estimates. Finally, political compromises may be required, as politicians from each jurisdiction will lobby to have their districts included and this could mean that a few (c) geographic targeting districts in each area, rather than the poorest districts, most anti-poverty programs in india include, in will benefit. Indeed, the list of the 200 backward districts combination with the methods described above, an includes at least one district in each state. element of geographic targeting in terms of fiscal allocations for programs. In addition, some programs explicitly rely on geographic targeting at least in initial D. how Do TARgeTing phases. For instance, the Mahatma Gandhi National Rural Employment Guarantee Scheme was introduced in meThoDS STACk uP? a phased manner, starting with the 200 most backward while robust comparisons of all targeting methods districts, then rolling out to the next 330 districts and are not available, it is possible from national data finally to the most advanced districts.382 The indicators to compare targeting outcomes from selected css sp used to select these "backward" districts included programs which rely on different methods. Targeting indicators such as value of output per agricultural outcomes by program are presented in Table 8.13 and worker, agricultural wage rate and percentage of SC/ Table 8.14, looking first at coverage rates across the ST population as well as some consideration special distribution, and then benefit incidence, the latter also category status and extremism. using the "targeting differential" method. Looking across With geographic targeting, location determines programs, some observations on coverage rates across eligibility for benefits: people who live in the designated the distribution are: areas are eligible and those who live elsewhere are not. theuseofself-targeting(inSGRY)andcombined Few programs target only on the basis of geography, categorical"destitutiontargeting"throughsocial Table 8.13: Cumulative share of beneficiaries of major programs by wealth quintile, 2004/05 (%) Poorest q2 q3 q4 Richest PDS grain 27.0 47.8 72.3 89.8 100 Annapurna grain 51.9 74.8 85.5 98.1 100 Rural housing support (IAY) 28.6 48.4 73.7 91.9 100 IGNOAPS 32.8 49.1 66.8 82.1 100 Disability pension 30.7 49.7 73.4 86.3 100 Widow pension 43.2 60.8 77.9 91.7 100 Rural credit (SGSY) 32.9 53.6 74.2 86.3 100 Rural public works (SGRY) 43.4 65.6 85.3 96.1 100 Scholarship 22.6 52.2 73.0 88.4 100 Source: Ajwad (2006) based on 2005 IHDS data. 380 SeeCoadyetal.(2004)foramoredetaileddiscussion. 381 Ravallionetal.(1993). 382 The Backward Regions Grant Fund (BRGF), introduced in 2006/07, an untied transfer to panchayats in order to strengthen local governance and village infrastructure,isalsotargetedtobackwarddistrictsofIndia. 20 Social Protection for a Changing India: Volume II Table 8.14: Cumulative share of total benefits captured by quintile by program, 2004/05 (%) Targeting Poorest q2 q3 q4 Richest differential1 PDS grain 29.7 54.8 77.4 93.6 100 4.8 Annapurna grain 37.9 64.8 76.0 98.6 100 20.4 Rural housing support (IAY) 24.6 43.1 68.0 90.7 100 5.6 IGNOAPS 31.1 50.6 66.0 81.6 100 8.0 Disability pension 27.6 44.1 64.0 75.2 100 4.1 Widow pension 32.8 51.0 70.4 87.6 100 14.5 Rural credit (SGSY) 7.7 23.1 33.5 50.2 100 -21.7 Rural public works (SGRY) 34.2 51.7 67.9 83.2 100 16.4 Scholarship 11.7 33.3 45.8 65.9 100 -10.4 Notes: Targeting differential of benefits between lower and upper quintiles. Higher numbers indicate better targeting performance. Source: Ajwad (2006) based on 2005 IHDS data. pensionsandAnnapurnaarethemostpowerfulin indicating better aggregate targeting performance. termsofpromotingcoverageamongthepoorest While rankings are broadly similar, the very regressive quintile.SGRY, Annapurna and widows pensions performance of SGSY shows up as even more of a stand out for sharply higher inclusion of the concern (see Table 8.14). poorest as a share of all beneficiaries.383 however, targeting effectiveness is a function BPL-basedtargetingunderPDS,andruralhousing not just of the design and implementation of the programs performs less well among targeted targeting tool, but also the design, implementation programs,although they are mildly progressive. and monitoring of the targeted program. As a result, benefit incidence for the bottom quintile is worse programs targeted using the same tool can exhibit across all programs except pds than simple coverage radically different targeting outcomes. For instance, rates would indicate. At the same time, the relative both PDS and old age pensions (NOAPS) are targeted performance of programs in terms of targeting using the BPL indicator (combined with categorical performance does not change, with the notable targeting in the case of NOAPS). However, the programs exception of SGSY and school scholarships, which present varying degrees of targeting effectiveness which perform particularly poorly in benefit incidence terms. can be attributed to differences in program design and The more useful insight in policy terms, however, is how implementation. robust the ranking of targeting performance remains, A few common lessons for the link between program with self-targeting and specific categorical targeting design and targeting outcomes emerge from the continuing to outperform other methods. analysis of program performance in previous chapters. The targeting differential provides an alternative single Clear program design with well-defined program measure which takes account of the share of program rules and institutional roles and responsibilities is benefits captured by the poor (for this purpose, the a critical prerequisite. The nature and frequency of bottom two quintiles of the distribution) and the benefits also has implications for minimizing fraud and non-poor (the top two quintiles), which takes account leakage ­ (i) cash transfers appear to be preferable than of both inclusion and exclusion errors.384 As such, it in-kind transfers; and (ii) regular (smaller) payments provides useful shorthand for targeting performance in appear to be preferable than one-time large payments. both dimensions. A negative number denotes regressive While the PDS deals in providing monthly rations of food benefit incidence, and vice versa, with higher numbers grains and other items, the NOAPS offers regular small 383 ThoughinformationontargetinginMGNREGisyettoberigorouslyconfirmedfromhouseholdsurveys,administrativedatasuggestlargelypositiveasthereis over-representationofSC/SThouseholdsinemploymentprovided(seeChapter4). 384 SeeRavallion(1999)andRavallionandGalasso(2005)foranexplanationoftheestimationmethod. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 20 and regular cash benefits. The regularity of payments, which must be taken into account in reforming targeting relatively small amount of each pension payment and in SP programs: monetization of benefits under NOAPS make it a more the distribution of households exhibits effectively delivered and better targeted program as considerable clustering around the poverty line compared to the PDS. nationally, making fine-tuned targeting design With respect to program implementation, the following inherently challenging contribute to maximize access and minimize de facto there is considerable diversity across states in exclusion errors, as well as to reduce program leakage: the factors that are correlated with poverty, so (i) adequate outreach and information campaigns; that reliance on standard national indicators in (ii) streamlined enrolment process to as to minimize targeting systems has inherent weaknesses. transaction costs; (iii) creative use of IT for payment community level beneficiary identification has delivery; and (iv) lean administrative structures (i.e., attractions in such an environment. However, minimum number of program intermediaries). In its strength is within-community targeting at a general, greater awareness among the poor, both of very localized level in relative terms rather than programs and eligibility criteria, are a critical first step the kind of absolute measure of poor people in their ability to access benefits.385 In recent years, necessary for comparisons across space. awareness of the BPL process is fairly high in rural areas, partly due to re-verification drives undertaken in several a key challenge therefore is how might the system states after the large number of appeals against the marry the "top down" targeting necessary for spatial 2002 BPL list. With respect to administrative structures distribution of social protection resources, with the for delivery, there appears to be a weak link between "bottom-up" information on the characteristics lean structures with shorter delivery chains and better of the poor which might do a better job of ranking targeting outcomes. For instance, the PDS system is households at the grassroots level. There appear to operated by heavy bureaucratic machinery involving be several elements of a potential strategy. These are many stages and levels of intermediaries, including discussed below, first presenting more modest reform a network of fair price shops. Some other relatively proposals, and then presenting an option which would poorly targeted programs also rely on intermediaries, imply more substantial reform. e.g., SGSY relies on banks, SHGs and supporting NGOs international evidence on targeting outcomes or other mobilizers. In contrast, NOAPS is fairly simple provides useful insights in thinking about an in design, making direct payments to pensioners. This appropriate targeting mix. Three observations emerge distinction ­ and its mapping to targeting outcomes ­ is from a review of targeting methods and outcomes in not airtight (e.g., why does Annapurna perform so much 122 targeted programs in 47 developing and transition better than PDS itself? SGRY saves on the targeting countries: transactions and intermediaries through self-targeting, but needs coordination on organization of works and some methods have better targeting outcomes payment determination), but is worth keeping in mind than others when taken across a wide range of in reflecting on relative targeting outcomes. countries. This can be seen in Box 8.5. combining targeting methods generally results in improved outcomes. For example, a combination e. oPTionS FoR imPRoving of three methods improves targeting outcomes by around 30 percent. This potential improvements TARgeTing in SP PRogRAmS needs to be traded off however with additional the above analysis suggests there is major scope to administrative costs and demands, and the risks improve targeting systems for public programs in of too much complexity introducing possibilities india. At the same time, there are several generic issues for manipulation. 385 Forinstance,asurveyafterthelaunchofRSBYinaparticularstaterevealedthat,alargeproportion(82percent)ofhouseholdswhowereawareoftheschemehadenrolled fortheprogram.Researchforthisreportalsopointstothelackofawarenessabouteligibilitycriteria(e.g.,forsocialpensions)andentitlementsunderMGNREG. 210 Social Protection for a Changing India: Volume II Box 8.5: international findings on targeting methods It is useful to look at targeting methods in light of their targeting performance in other developing countries. This can be seen in the table below. Targeting effectiveness of different targeting methods Targeting method median targeting performance All methods 1.25 Any form of individual assessment 1.50 Means testing 1.55 Proxy means testing 1.50 Community assessment 1.40 Any categorical method 1.32 Geographic 1.33 Age: elderly 1.16 Age: young 1.53 Other categorical 1.35 Other selection methods 1.10 Work 1.89 Consumption 1.00 Community bidding 1.10 Source: Coady et al. (2004). Higher scores indicate better targeting performance. An important additional issue is the reliance on multiple targeting criteria. This raises the question of the benefits of using multiple targeting methods, which implies a relatively high degree of administrative effort to implement. International evidence gives some support for the use of multiple targeting methods. In an econometric analysis of 122 targeted programs from 45 developing countries, it has been found that each additional targeting method used improved targeting outcomes by around 15 percent, controlling for factors such as country income level, governance, inequality and other factors. Of course, such improvements in targeting performance come at the cost of additional costs of administration and additional costs for households in compliance. However, it is unusually difficult to make a reliable estimate of the costs of administering UDB for several reasons, so that it is not possible to have an informed discussion of the trade-off between improved targeting outcomes and increased targeting costs. Source: Coady et al. (2004). at the same time, there is also strong variation and that the policy question is whether a more across countries in targeting outcomes within effective mix of methods can be employed for sp (and each method, pointing to the importance of other programs).386 This would include improvements country specificities and implementation capacity in methods for each type of targeting, and potentially a in considering options. Across the sample of different mix of methods. The discussion below focuses programs, around 80 percent of the variation in initially on improvements in geographic targeting and targeting outcomes is accounted for by within the "backbone" BPL method, before a discussion of more method variation, and around 20 percent by substantial reform options. between method variations. One important dimension of this is community "voice" in the society, which internationally improves targeting (a) improving geographic targeting outcomes by around 23 percent. with respect to geographic targeting of central sp international evidence therefore clearly suggests allocations, it would be useful to generate sub-state that no single targeting method will ever be sufficient, estimates of poverty using poverty mapping techniques, 386 SeeCoadyetal.(2004)forausefuldiscussionofdifferenttargetingmethodsandcombinationsindevelopingcountries. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 211 and use these as the basis for future spatial resource improvement which may contribute to a more effective allocation.387 Poverty mapping allows for reliable mix of targeting methods. small area estimates of poverty below the state level by combining information from the NSS household survey data with unit record data from the Census. Since the (c) Reforming the BPl method: the late 1990s, detailed "poverty maps" have been prepared proposed 2011 BPl methodology for a growing number of developing countries. These given the concerns over widespread exclusion and maps provide estimates of poverty and inequality at the inclusion errors in the 2002 bpl method, an expert local level ­ such as the district, sub-district, and even committee has suggested significant changes in village level. Such information is not commonly available design of the proposed 2011 bpl census (as described because household surveys are typically too small in in the saxena committee report).388 One, the method sample size to permit sufficiently fine disaggregation. Yet, with ongoing efforts to apply detailed spatial targeting proposes the automatic exclusion of visibly non-poor of public interventions, or to realize the gains from households and the automatic inclusion of the most decentralization and community-centered development, vulnerable households, with a survey and scoring only there is a pressing need for information on distributional of the remaining households. Two, the new method outcomes at the local level. overlays geographic targeting criteria by specifying district, block and GP-level quotas with respect to the in india, this could reliably generate poverty estimates proportion of BPL households. In districts that have a very at block level. While progress on this front would not high proportion of BPL households (over 80 percent), address the issue of household level targeting, it would all households (except the visibly non-poor) would be significantly reduce issues of targeting across space considered to be poor and automatically included in the within states. In India, a pilot effort to produce poverty BPL list. This design reduces administrative demands and maps in three states - West Bengal, Orissa and Andhra cost to some extent by reducing the scope of the survey. Pradesh ­ is ongoing. Once that is complete, it would be The underlying principle of the proposed methodology a more precise basis for allocations of CSS SP resources is to minimize errors of exclusion, without worrying across space. In particular, these poverty maps can over-much about errors of inclusion. The elements of potentially be an important part of a strategy to marry the proposed method are described below. "top down" targeting (essential for fiscal allocations of resources across space) with "bottom-up" targeting of setting quotas for the distribution of bpl households households or individuals (e.g., through community- across space: district-level thresholds for the based methods that rely on local knowledge but are not proportion of bpl households could be set using comparable across space). the ratio of sc/st population in the district to that in the state, the inverse of agricultural production per rural person and agricultural wage rate of (b) improving household-level the district.389 These could then be used to estimate targeting block-level proportions of the poor using any one or a combination of indicators on soil and irrigation In addition, there is a major need to improve India's quality, road connectivity, female literacy and share household level targeting systems, though the range of non-agricultural workers. Finally, these block-level of options varies between urban and rural areas, and estimates would be used to generate GP-level quotas by program type. This section looks at three possible for BPL households on the basis of population estimates. reforms in the BPL methodology as well as the potential Estimating these disaggregated sub-district thresholds role of other methods, including community-based and as well as choosing alternative rules to generate GP- self-targeting methods. None of them provide a "magic level quotas could be left to the discretion of the state bullet" for all programs, but are suggestive of options for government or district authorities. 387 SeeElbersetal.(2003)fordiscussionofthepovertymappingmethodologyandWorldBank(2011)foradiscussiononthepilotsinIndia. 388 SaxenaCommitteereport(2009). 389 ThisisthestandardformulausedforfiscalallocationsforRuralDevelopmentprograms. 212 Social Protection for a Changing India: Volume II identifying poor households: The basic principle (ii) automatic inclusion of certain categories of behind the methodology being proposed is to ensure households that are considered to be among the the inclusion of groups that are regarded as particularly poorest and most vulnerable: These include the vulnerable (e.g., workers in low income or highly following categories: (a) designated Primitive Tribal vulnerable occupational categories, households where Groups; (b) designated most discriminated against earning capacity is severely curtailed by external SC groups (called Maha Dalit Groups), if so identified constraints, etc.) and/or groups that are historically by the state; (c) single women headed households; subjected to social exclusion. The proposed method (d) households with disabled person as bread-earner; of identifying households has the following three (e) households headed by a minor; (f ) destitute elements: households that are dependent predominantly on i. Automatic exclusion of visibly non-poor house- alms for survival; (g) homeless households; and (h) holds households where any member is a bonded laborer. However, there is some confusion in the treatment ii. Automatic inclusion of the poorest and most of households and individuals while operationalizing vulnerable households these criteria. While the report focuses for the most iii. Scoring of the remaining households (except part on households (i.e., a joint family that resides in districts with poverty above the specified under a common roof and eat from a common threshold ­ 80 percent - where all households, kitchen) and nuclear families, some of the specific except those automatically excluded, are categories noted above relate to individuals. In this identified as poor without a survey). case, the report suggests treating these groups as separate households (e.g., single women with (i) automatic exclusion of households that are meet no major son, elderly couples or individuals, etc.). certain criteria and are identified as non-poor: The However, it is not clear how to score the remaining intention is to exclude households that are visibly household members. Also, not all of these criteria non-poor, but whose position of economic, social or are easily measurable or verifiable in the field ­ e.g., political privilege makes it possible for them to get a functional assessment of disability requires several (wrongly) included in BPL lists.The exclusion criteria detailed questions and highly trained investigators. include the following: (a) households with double the land of the district average of the agricultural (iii) grading of the remaining households (identified land per agricultural household if partially or wholly as poor) in order to prioritize programs and irrigated (three times if completely unirrigated); services to the poorest among these households: (b) households who have two-wheeled (or more than The proposed scoring system ranks households on two wheeled) motorized vehicles; (c) households a scale from one to ten using the following criteria: who have at least one mechanized farm equipment; (a) caste and religious affiliation ­ SC/ST (3 points), (d) households who have any person who is denotified tribes and designated `Most Backward drawing a salary of over ` 10,000 per month in non- Castes' (2 points), Muslim/OBC (1 point); (b) any government/private organizations or is employed one household member engaged in the following in government (including parastatals) on a regular occupations ­ landless agricultural worker (4 points), basis with pension or equivalent benefits; (e) income agricultural laborer with some land (3 points), casual tax payers.390 These criteria are an improvement over workers (2 points), self-employed artisans or fisher those used in the 1997 BPL method as (i) they allow folk (2 points); (c) no adult above the age of 30 with for some local variation to a limited extent (e.g., schooling up to grade five (1 point); (d) any household land holding is assessed relative to the local average member with TB, leprosy, mental illness or HIV/AIDS instead of the all-India threshold of two hectares or disability (1 point); and (e) households headed by previously) and (ii) focus only on high-value assets an old person of age 60 and above (1 point). Since the that may be more successful in separating the rich bunching of a large number of households is possible from the poor. at certain scores, the method proposes a ranking of 390 Thesecriteriaapplytotheentirejointfamily­i.e.,thefirstcriteriaofhigherthanaveragelandownershipwouldapplyeventothesonofalargefarmereven thoughhemaynotactuallyhavelandregisteredinhisname. CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 21 households with the same score, with those in the test using NSS data as done with the 2002 BPL method special category group such as SC/ST at the top, would yield interesting insights into how well the new followed by landless agricultural laborers and so on. de jure targeting design would perform. this proposal is a modest approach to reform in however, several drawbacks of the previous bpl that it proposes continuing with the bpl system with method remain, particularly with respect to the some improvements. The proposed methodology is an imposition of central criteria across diverse local improvement over the 2002 BPL method in some ways: contexts, cardinality of indicators, and the same (i) The indicators chosen are likely to be highly correlated process of aggregation. In addition, there are likely with poverty for the most part (though not necessarily to be reporting issues with some of the indicators (e.g., for the elderly).391 Some of these indicators (gender of chronic health problems, disability). The adoption of this household head, education, occupation) were among a methodology will also have significant fiscal implications. wider set of indicators used by Jalan and Murgai (2008) The inclusion criteria are defined on the basis of nuclear in their augmented regression model that performed families and individuals who are considered as separate much better than the 2002 BPL method. However, while households for the purpose of this exercise. This would the reduction in the number of indicators increases lead to the number of BPL households increasing simplicity in field processes, it may make it more difficult significantly, implying possible rationing within these to separate the rich from the poor and the poor from households if program budgets do not rise adequately. the poorest; (ii) Weights have been assigned to the indicators as noted above. However, weights have as a result, there is still scope to move to a more not been applied across states; and (iii) The use of the developed form of proxy means-test, using methods geographic, automatic inclusion and exclusion criteria which are good practice in developing countries, for reduce the scope of the survey and hence administrative both rural and urban areas. Box 8.6 provides some demands and costs. Subjecting this methodology to the insights from international experience with PMT. Box 8.6: international experiences with proxy means testing Targeting on the basis of proxy means test like the BPL is becoming more common in developing countries. Generally such methods are used for large benefits and/or for multiple programs. Current examples include use of PMT for cash transfers (Armenia, Colombia, Mexico), targeting food subsidies and rations (Indonesia, Turkey), rationing entry for subsidized health insurance schemes (Colombia) etc. PMTs are also currently being designed and piloted in Bangladesh and Pakistan. In general, PMTs are relatively insensitive to quick changes in welfare and require time to design, pilot and set up systems for implementation, monitoring and redressal. As a result, PMTs are best suited for targeting the chronic poor in stable situations and not as a response to crises. From reviews of international experience with PMT, several common features emerge which can be considered good practice. India in its BPL system relies on some of these but not others: Design:Typically a PMT is designed using regression models of total household income or consumption on the selected variables, often separately for different regions of the country. This process, usually iterative, should yield: l Indicators that are the best possible proxies to predict poverty in rural and/or urban settings. l Indicators that are easily observable and not easily manipulated by households. l A weighting across the indicators, with the weights derived from the regression model. l Not too many indicators to be used in the PMT. Around two dozen variables are used typically when one looks at programs internationally. l Piloting before finalizing the PMT allows validation of the tool against other indicators of household welfare using current data. Implementation: The use of PMT requires institutional capacity for collection of household-level information and for subsequent database management. l Governments or project teams may decide to collect information through a door-to-door census type exercise or through targeting centers (combined with outreach and communication strategies to minimize exclusion). l Some countries combine the use of PMT with home visits or other methods for ensuring both that those identified through the PMT are actually poor, and to check on others. While desirable, this needs to be trade off with additional administrative costs and capacities. l In some countries, the scoring system is made public. l There is often an appeals process to mitigate the negative impacts on some poor households of the mechanical nature of the PMT formula. 391 SeePalandPalacios(2008)foracomparisonofpovertyratesamongtheelderly. 21 Social Protection for a Changing India: Volume II there are several possible elements in such as householdwelfare,measuredusingotheraccepted reform: measuressuchaspercapitaconsumptionand/or assetindices. evenifstandardnationalBPLcriteriaareretained, these should be determined using rigorous BPLlistsshouldbesubjecttolocalizedverification regression models for greater targeting power. by communities, using gram sabhas or other While the proposed 2011 BPL methodology does mechanisms. In principle, this should happen, include criteria such as caste affiliation that have and does in some states (e.g., HP) and/or for been found to be strong correlates of poverty, specific programs. However, the process is not the preferred method for such an exercise is to systematic and made more complicated by the use the NSS or other representative national fact that gram sabhas are held infrequently and data to generate a set of indicators that provide participation is often low in most states.392 If it more targeting power, and which could be easily were to become more standard, it may be useful verifiable at reasonable cost. The Ministry of Rural in some areas to have the verification process Development is currently analysing data from a facilitated by NGOs or other third parties where sample survey to inform the debate on the 2011 there are concerns of elite capture. Verification BPL design. and finalization of the BPL list by the gramsabha is one of the suggestions made by the Saxena whatever indicators are used, they should have Committee report. some weighting ­ preferably on a state-specific basis. The basis for such weights and a more build in systems for dynamic updating of BPL sensible ranking between levels of each indicator listsinbetweencensuses.The Saxena Committee should be the NSS data, or other representative report suggests holding a census once every national data where relevant. In contrast, ten years, combined with a system for updating the proposed BPL methodology and Kerala's the lists every two years by registering changes Kudumbashree use a set of weighted indicators in household circumstances. However, the for the identification of poor households, but the report does not specify the implementation weights do not appear to be based on regression arrangements for doing this. analysis. These are likely to be assigned by requirestatestohavecommonBPLlistsandBPL committee or, in the case of Kudumbashree, rationcardaggregates,preferablyconsistentalso developed by communities in a participatory atthehouseholdlevel.Several states already do manner. this, and it has the benefit of saving administrative a preferred approach beyond the above would demands and avoiding multiple targeting be to let the indicators in the BPL methodology criteria. The Saxena Committee report makes this vary by state, or perhaps groups of states. As recommendation. However, in states where the the purpose of BPL is increasingly ranking of two do not presently coincide, political economy households rather than determining aggregate may make convergence challenging, particularly numbers of poor people (which remain controlled where BPL ration cards significantly exceed the through the linkage to Planning Commission numbers on BPL lists. aggregates for states), there seems no inherent no comparable central guidelines for a census of need to insist on standard national indicators. As poor households exists for urban areas. Typically, part of a process to generate indicators, it would the identification of poor households is carried out be worthwhile to simulate state or region-specific separately by different departments, such as the indicator sets and test their power in ranking of Food and Civil Supplies Department for PDS ration households. cards. However, methods differ across states and piloting the proposed method would allow an departments and there is little systematic information assessment of how well these indicators capture on this process. Recently, a working group was 392 Eveninthefoursouthernstates,only20percentofhouseholdsreportedattendedagramsabha(Besleyetal.2005). CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 21 formed in the Planning Commission to deliberate on database for multiple programs, thereby reducing the design of an urban BPL. Notable exceptions are administrative costs and increasing transparency. This, Delhi and Kerala (see Box 8.3 above). More recently, however, does not necessarily mean that all programs Delhi has defined a common process of identification would need to target the same groups. Programs such of the poor across programs. This defines a set of as educational scholarships that target SC/ST children thematic indicators capturing residential location, could use the information in the common database social deprivation and occupational vulnerability to continue targeting on caste affiliation rather than to identify vulnerable households. The underlying vulnerability status. principle of promoting inclusion and several of the specific indicators are similar to those proposed in the while the above reforms could significantly improve rural 2011 BPL methodology (see Box 8.7 for details). the bpl system, international and indian evidence A survey is currently underway to collect information suggests that indicator-based targeting ­ even on households in poor localities, with the intention of where generated robustly from reliable survey covering other areas over time. The list of vulnerable data ­ will struggle to explain a significant portion households so generated would be used by the nine of variations in living standards across households. state departments that are engaged in delivering This is true in a static sense, where such methods rarely programs or services to the urban poor. This is a very explain more than half the variance in household positive step towards generating a common beneficiary consumption.393 Box 8.7: Targeting under mission convergence for urban Delhi The emphasis of Mission Convergence (MC), initiated in March 2008, has been on improving targeting and outreach of social protection programmes in the National Capital Territory of Delhi. The Mission targets through a revised methodology, premised on the enumeration and identification of prospective beneficiaries employing a survey of localities and sites which house `vulnerable' populations in Delhi. Three thematic indicators ­ type of residential locality, social deprivation and occupation are used to define `vulnerability' and classify target vulnerable households as distinct from other households canvassed for survey data. vulnerable households are defined as per the following parameters: Residence ­ those living in notified and non-notified slums, resettlement colonies (F,G and H), and are shelterless and precariously housed; Social deprivation ­ households with elderly either living alone or as dependents or others, households with persons with disabilities, households with any member suffering from a specified chronic illness, single women living in households as dependents or alone as heads of households, single unprotected children and households headed by children; Occupationalcategory ­ Households that are primarily dependent on earnings from occupations and forms of employment or self employment which are casual, irregular and hazardous with insecure wages and poor working conditions. An indicative list of such occupations include rag picking, unskilled construction labor, casual daily wage labor, street hawking, casual domestic work, cycle rickshaw, unskilled workers in household enterprises. In addition, a sub-category of the most vulnerable households includes those that are (a) homeless or precariously housed households; (b) households residing in notified slums, non-notified slums and resettlement colonies (F,G,H) and who report at least one family member as being socially vulnerable as per the criteria noted above. MC relies on the institutional capacity and expertise of Gender Resource Centres (GRCs) ­ managed and staffed by non-governmental organisations, civil society groups and community-based organisations for the data collection process. Such an institutional framework, whilst providing unique opportunities in relation to the delivery of benefits and services, has also been accompanied by problems in the pursuit of the planned targeting method. As noted in the Saxena Committee draft report, the actualization of such a non-income based assessment require clear, measurable and standardized definitions of concepts and terms used, which can facilitate strict instructions for data collection. Thus, a concern with the revised vulnerability targeting criterion set by MC is the difficulty in translating the specified criteria into practicable inputs for implementation by agencies with limited survey experience and capacity. In addition, some of the groups noted above are broadly defined (e.g., a large number of households would have at least one member engaged in some form of casual work). This combined with no weighting across criteria, indicated the need for rules for prioritization or ranking within these indicators if faced with budget constraints for programs. 393 Forexample,Egypt'sPMTforitsfoodrationingsystemcapturesonly43percentofthevariationinconsumptionsacrosshouseholds,andArmenia'sonlyaround onequarterofthevariation. 21 Social Protection for a Changing India: Volume II (ii) Otherreformoptions pronounced), and would need facilitation by competent NGOs. However, an evaluation of the AP experience, and while reform of the bpl system is highly desirable, well-evaluated pilots in other states seem warranted to there may be additional options in rural areas of see whether some of the shortcomings of even a well- some states for community-based targeting to play designed PMT can be mitigated through a stronger a stronger role in sp programs. As the centre enforces community role in identifying beneficiaries. See Box 8.8 convergence on aggregate numbers of poor people for a discussion of the operational challenges in scaling between Planning Commission estimates and the BPL up community-based targeting methods. system, possibilities for more significant reforms of targeting systems could be considered in rural areas as happens presently with community wealth ranking, (though perhaps not in urban, where an enhanced PMT the process would benefit from being framed within would seem more feasible). general criteria for ranking households. From an implementation viewpoint, there is a tension between once a poverty map is available at block level, more localized guidelines and minimizing intensity reliance on community-level wealth ranking becomes of process and costs in developing the guidelines. a feasible option as a tool within the formal system However, experience from India (e.g., Kerala's for identifying sp beneficiaries. While not an ideal Kudumbshree) and from other countries which have match, the availability of aggregate numbers to block formalized community identification into their formal level and community ranking at GP/village level comes SP systems (e.g., Uzbekistan), suggests that this is a close to the "top-down" and the "bottom-up" meeting. surmountable challenge. To the extent there remains a gap, solutions such as assuming even poverty rates within the block, or some whether the modest or more fundamental reform simple indicators such as population share (as is being options are pursued, self-targeting (as happens in proposed in the Saxena Committee draft report) or SC/ST public works) would continue to be a useful tool for share or landless share in different villages might be used programs where it is suitable. This approach is most to allocate SP resources within a block. The precedent effective when used to target goods or services that are of Andhra Pradesh is instructive, where a list of poor more heavily consumed by the poor than the non-poor, household generated entirely by community wealth and for programs where monitoring costs associated ranking has replaced the administratively generated BPL with verification of eligibility are high. Programs such as list across the state for the main anti-poverty programs public works are amenable to self-targeting approaches administered by the Rural Development Department. through the wage rate and types of work offered. Others This may not be feasible in all states (e.g., where caste such as subsidized credit lend themselves less readily fragmentation is high and risks of local elite capture more to self-targeting. For cash benefits, the possible role Box 8.8: operational considerations in scaling up community-based targeting methods Several operational challenges need to be addressed with respect to scaling up community-based methods if they are to be used in government programs: Definingcommunity in a manner that balances the sociological necessity of cohesion among households with the operational compulsions of scale. The unit of the exercise should be small and socially cohesive, but this adds to the cost of the exercise and leads to an additional problem of aggregation of these locally collected lists to an operational/administrative unit level. Mobilizingparticipation is typically earlier in a project context where there is expectation of immediate benefits but may require more outreach when part of a targeting exercise that is delinked from a specific project and with more diffuse expectations of future benefits. Monitoringthequalityofparticipation with respect to standardized and high quality facilitation to minimize dominance. Definingpovertyandcategorizingthepoor requires balancing the maxim of allowing communities to define characteristics of the poor with the operational need to ensure an acceptable degree of comparability across communities (even in a project-context). Typically `guidelines' are adopted for the characterization of various poverty strata. Institutionalizingtheprocess by setting up systems for verification of poverty lists and for regular updating of poverty status. Source: Conning and Kevane (2002). CHAPTER-8: Targeting Mechanisms ­ BPL and Beyond 21 of self-targeting will depend on benefit levels, ease of requirement or of benefits offered permits. This suggests accessing programs and other factors. Despite these that self-targeting should continue to play a role in obvious limitations, the empirical results in Chapter 4 programs where it is suitable. An important additional highlighting the superior targeting performance of factor supporting this conclusion is that self-targeting public works suggests that "cutting out the middleman" may minimize the political costs of clear designation in targeting of SP programs remains an important tool of target groups, by allowing policymakers to describe for programs where the nature of the self-targeting such programs as "open to all". 21 Social Protection for a Changing India: Volume II CHAPTER Getting More from India's Social Protection System 9 Directions for the Future Chapter­9 Getting More from India's Social Protection System Directions for the Future india will undoubtedly continue its strong Building consensus around the reforms to be commitment to a social protection system which seeks undertaken. It will be necessary for political to serve the poor. The question is how best to do that constituents, including a significant share of those more effectively in the short and medium term. There who may lose from reform, to support the stages are likely to be three broad elements necessary for an of policy and implementation evolution if they effective SP reform strategy for India: are to be successful. Reorientingthepolicymixandspecificprogram if such a reform agenda can be delivered, the benefits policies across SP programs in order to meet for the poor of india could be substantial, and make the diverse needs of India's poor ­ diversity growth significantly more inclusive. In addition, there which has several dimensions: spatial across and within states; across socio-economic are likely to be positive impacts on growth itself from an categories; and between households trying SP system which more effectively addresses a range of to manage different types of risks and shocks. market failures which result in poor and unproductive This would include exploring new SP tools for citizens. The traditional view of social protection leveraging improved human capital outcomes systems and the redistributional objectives underlying and household productivity, and innovating them was that there was a clear growth versus equity with policies and programs to support the trade-off. However, empirical evidence increasingly urban poor; highlights that a well-designed and implemented Getting better poverty reduction outcomes from SP system provides dynamic efficiency gains to the existing programs through improvements in economy through positive impacts on productivity, and financing, targeting, institutional arrangements as an important tool for governments in managing the andadministrationofexistingschemes;and impacts of reforms in the wider economy.394 394 SeeWorldBank(2004). CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 221 as is currently being proposed under NRLM; A. PoliCy ReFoRm in SoCiAl and secondly considering the options for use of PRoTeCTion safety net transfers (see next bullet) to leverage participation in core education, health and there is a need to deepen the ongoing policy possibly nutrition services, in order to promote reorientation of the indian social protection system long term movement out of poverty. to meet the changing and increasingly diverse needs of its population. Marginal changes alone will not moving to more consolidated and more cash- deliver the kind of safety net which a changing India based social assistance programs for the needs for its poor and for its economy. This would chronicallypoor.The "big elephant in the room" involve several elements: (i) a rebalancing of the in this respect is fundamental reform of PDS. policy mix across different types of public SP priorities; The very poor and long run performance of (ii) consolidation of the large number of central and the program in many states suggests that the state schemes to a core set of flagship programs; medium term vision of a reformed PDS for most (iii) in the context of consolidation, introducing an groups should be cash-based, though this would element of choice and flexibility for states in the specific face substantial resistance in light of the ongoing program mix of centrally-supported schemes that they debate around the Right to Food legislation. A operate; and (iv) in some areas and for some programs, reformed PDS could still provide food-based actively exploring the possibilities for leveraging the support for specially vulnerable groups (in line role of private players (both non-governmental and for- with Supreme Court orders), and in specific areas profit) in delivery of interventions. facing chronic or acute food shortages, but for most areas and most people, a cash-based social in terms of reorienting the policy mix, the report assistance system seems a more efficient and suggests several directions: transparent means of providing an income floor. increasingtheemphasisonpreventiveprograms An intermediate solution currently being mooted which help the poor and those vulnerable to in the 2010 Economic Survey is to transfer the poverty to manage risks and shocks better. This subsidy directly to households (rather than the implies a significant expansion in coverage of PDS store owner) through food coupons with a different social insurance instruments, though lumpsum entitlement that can be exchanged at in a phased manner consistent with institutional any PDS store. and fiscal capacity. Experience to date suggests in the face of demographic change and slower that phasing would benefit from: (i) starting with rates of poverty reduction in urban areas, starting simpler-to-administer insurance products such as toaddresstheneglectofurbansocialprotection life and permanent disability, while continuing to policy. While some of the needs of the urban pilot and evaluate experience with more complex poor are common to their rural counterparts, products such as health insurance; and (ii) for the possibilities (and constraints) of the urban reasons of ease of worker mobilization and to environment suggest that simple mimicking of control transactions costs, focusing initially on the rural models of SP programs and service delivery "low hanging fruit" of unorganized workers who mechanisms is unlikely to be an adequate are members of groups (e.g., MFIs, cooperatives, response. For example, the options for "voice trade union and worker associations, SHG accountability" of service providers which can federations) that could play an intermediary be mobilized in rural areas through collective function between workers and the state/insurers. community action are likely to be less possible rethinking programs which seek to promote in urban areas, while the possibilities for "choice movement out of poverty in two ways: firstly, accountability" (through income enhancement moving from administratively driven subsidized and offering options in service providers where credit to public financing of a more diverse range possible) are likely to be greater. The JNNURM of livelihood promotion approaches better suited program had appeared to offer a solid base for to the labor market conditions of individual states such an urban SP reform, but has largely failed to 222 Social Protection for a Changing India: Volume II deliver in this regard and closer consideration is GoI aims to expand include life, disability, old age needed of how to affect urban SP policy change. pension, and health, and the RSBY program is already an important step forward in this regard. the specific proposal of this report is that central Chapter 5 gives suggestions on institutional, sp programs over time aim for a "3+block" strategy. financing, sequencing and other aspects of an This would involve 3 core CSS SP programs or "pillars", expansion strategy. combined with an SP block grant from which states could finance other SP programs - or supplement beyond the three "pillars," states could receive an benefits under the core pillar programs - more tailored additional transfer and implement state-specific to the poverty and vulnerability profile of the individual sp interventions. How this is programmed could state. This would also involve significant expansion in vary according to state-level priorities, and include urban areas. This could promote both a more coherent interventions such as livelihood support of different and less duplicative SP system, but also give states more forms, targeted housing, interventions to incentivize use leeway to adapt the SP policy mix to the needs of the of basic social services, nutrition and/or early childhood poor in individual states in light of available economic care (e.g., through conditional cash transfers as being opportunities. The three core pillars proposed are: piloted in some states), specific urban SP programs, a major social assistance program. The obvious or other options as proposed by states. A secondary candidate for this is a significantly reformed PDS, benefit of such an approach would be strengthening merged for specific groups with existing social of complementaries between CSS and state-sponsored pension programs.395 Chapter 3 give more details schemes in order to control unnecessary duplication. on the options for reform of PDS proposed by A common core national SP system under the three this report, with a preference for a predominantly pillars could promote portability of basic entitlements, cash transfer approach. and be increasingly useful as mobility of workers and households increases. a public works program, for which MGNREG would be the building block, as well as piloting in terms of promoting both more effective spending expansion in urban areas. There are several on sp, the "3+block" proposal would allow more benefits of a reliable public works program at this cross-program flexibility to states ­ or possibly stage of India's development: (i) by its demand- districts - in deciding their sp expenditure priorities, driven nature, it can be responsive to shocks in a while still maintaining a common national core sp way that longer run programs typically cannot. system. It would also allow for greater adjustment in In this way, it functions as a "quasi-insurance" light of poverty levels, key vulnerabilities, etc. This could program for the extended period during which be done in a variety of ways, possibly using a menu more structured insurance is expanded to the approach to SP programs, and a flexible form of social unorganized sector; (ii) the positive targeting protection block or matching grant which consolidates outcomes of self-targeted works; and (iii) the resources from existing SP CSS.396 This is an approach potential for multiplier effects from asset creation which has received increased attention in India in recent and community mobilization distinguish public years as policymakers seek to get greater impacts from works from other SP programs. See Chapter 4 for SP spending. Given current financing channels from a discussion. the centre to states in India, a more flexible granting abasicsocialsecuritypackageforthoseoutside mechanism for SP programs to states could take the formal sector which could be expanded different forms ­ as a more "bundled" anti-poverty CSS, in terms of coverage and scope of benefits as Additional Central Assistance (ACA) along the lines as institutional capacity and fiscal space is of programs like Pradhan Mantri Gramodaya Yojana developed. The core types of insurance which (PMGY), or through a more fungible realignment of 395 ThiswouldbesimilartotheChineseurbanandemergingruralsocialassistancesystem,whichisbuiltaroundthe"dibao"programwhichprovidescashbenefits tothepoor,andhasadditionalprovisionforspeciallyvulnerablegroupssuchasdisabledandunsupportedelderly.SeeWorldBank(2009b). 396 SeedeNeuborg(2002)foradiscussionofthestrengthsandweaknessesofdifferentblockandmatchinggrantmechanismsinthecontextofSPprograms. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 22 Gross Budgetary Support shares between the states and term policy mix and priorities. This could in turn assist central Ministries which control CSS.397 the transition from the current program-driven approach the above approach would require close consideration to SP to thinking in terms of a social protection system of the policy design of existing programs. The findings which is animated more by poverty outcomes and of this and other reports suggest that the PDS require less in terms of scheme-based target fulfillment. Such fundamental consideration of its current design, SGSY institutional reforms have been important elements is currently undergoing significant change under NRLM of successful SP reforms in a range of developing and and attention to implementation than in the past is developed countries. warranted, while two others warrant experimentation a second overarching need in such policy reorientation to see whether innovative approaches can yield better will be greater willingness to experiment in program outcomes than seen to date ­ IAY and school stipends. design and base reforms on results of evaluations. The other major programs ­ public works and social This will require changes in two tendencies of Indian pensions - would also benefit from policy improvements SP policy since the 1970s: firstly, what CAG has called but these are more in the nature of incremental policy "rechristening and revamping" of programs at the reform which can be expected in the normal course of expense of genuine experimentation and innovation, program evolution. Urban programs remain small, but and secondly a limited willingness on the part of the pre-conditions suggest that merely transferring rural central government to give states (and in some cases, the SP models to urban settings will limit potential impacts, sub-state level) a freer hand in adapting their policy mix and equally that the SP system needs to explore stronger among programs, by allowing flexibility in adjustment linkages with the livelihood opportunities available to of specific programs to suit their diverse circumstances. the poor in urban areas. The experience of a number of developing countries a number of cross-cutting issues in policy evolution including Bangladesh in recent decades provides a of the sp system will also be important. These positive example of the social benefits of experimentation include cross-program convergence and consolidation, in SP policy. Closer to home, there is growing innovation willingness to experiment with new types of programs at the state level in India which demonstrates the value and modes of delivery, and more structured efforts to of such an approach, and the increased buy-in among build understanding of the benefits of reforms and politicians and administrators for innovations which are manage the political economy of the reform process. "home grown." An additional challenge will be ensuring increased a third overarching theme of policy reform is that attention in SP policies to the needs to the urban poor. These are discussed in turn below. "government cannot do it alone", and programs would benefit from appropriate partnerships with firstly, central and state-level policymakers will the non-government sector. This partnership could need to accelerate cross-program convergence and be in both policy formulation and the specifics of policy consolidation in policy and administration to mAkE design. The "non-government sector" in this respect thE systEm morE unDErstAnDABlE to thE populAtion, could range from communities themselves (in the form rEAlizE EConomiEs of sCAlE in poliCy DEvElopmEnt AnD of SHGs and other forms of CBOs), to the NGO sector, ADministrAtion, AnD simplify ongoing plAnning AnD to the for-profit private sector in specific programs ExECution of sp progrAms. This will require enhanced and functions. The design of MGNREG is a promising efforts of institutional coordination within and across example of such a reorientation of policy formulation levels of government and administration. Both at central and program design, with its clear roles for community and state levels, there seems a need for formation of and NGO actors, and willingness to bring in private an inter-departmental Task Force for Social Protection sector expertise and research institutions on areas which would promote coordination (and possibly reduce such as M&E. But there is room for much more active duplication) across targeted programs, and promote engagement with the commercial private sector also, more coherent strategy development on the medium including in areas such as public grain distribution, 397 SeeSaxena(2006)forahistoryofcentraltransfersforanti-povertyprogramssincethe1960s. 22 Social Protection for a Changing India: Volume II targeted credit and livelihood interventions for the poor, rapidandsubstantialimprovementsinthebasic and low-income urban housing. "nuts and bolts" of program administration and procedures. The detailed suggestions in this regard are outlined in Chapter 7. Broadly, B. imPRoving they would involve overhauling a range of bureaucratic procedures which impede funds imPlemenTATion oF SoCiAl flow, strengthening processes for administrative PRoTeCTion PRogRAmS and social accountability of service providers, a through modernization of program record even if the necessary reorientation of the sp policy keeping and reporting arrangements (including and program mix can be achieved, it will not improve computerizing systems and taking advantage of outcomes for the poor unless accompanied by a India's ICT prowess to look for "technology leap- thorough overhaul of sp program administration, frogging" opportunities such as introduction of including institutional arrangements. Whatever smart cards and other innovations), building on the evolving mix of SP policies, there will be several improved rural banking infrastructure to overhaul key elements of administration and institutional payment systems, and building a strong culture of arrangements which will need to be confronted if India M&E. Recent reforms in RSBY and to a lesser extent is to achieve the poverty reduction outcomes that its MGNREG, together with a number of state-specific significant spending on SP warrants, including: program pilots, provide many lessons in this area, delineating clear lines of accountability and it is hoped that they can be systematically accompanied by adequate staff and finances. incorporated in other SP programs over time. Delineating appropriate institutional overhauling existing targeting mechanisms, responsibilities for all links of the SP service both at the household level and geographically. delivery chain, and aligning the division Any social protection system needs to be able of functions with assignment of personnel to identify who are the poor with a reasonable and allocation of resources for program degree of accuracy. Innovations already operating implementation will be critically important for in India and good practice from other developing improved implementation of SP programs. This countries offer a range of options for significant will require first and foremost greater proactivity improvements in targeting mechanisms. These on the part of states to approve policies and include: (i) development of "poverty maps" at put into practice the PRI/ULB decentralization a sub-state (probably block) level which would provided for under the 73rd and 74th constitutional allow more precise geographical allocation of SP amendments. This would need to be followed funds to poor areas; (ii) overhaul of the BPL 2002 by a process-intensive reconciliation of central methodology in line with good practice in design guidelines, state-level stances on service delivery of proxy means-tests (PMT) in other developing decentralization, and capacities at sub-state countries, including allowing for cross-state levels to perform the required implementation and urban/rural variations in the PMT formula; functions in SP programs. Achieving this goal will and (iii) in rural areas, continued piloting and require coordination and gradual convergence strengthened evaluation of community-based across the many departments of government beneficiary identification for SP programs with an responsible presently for different programs, and eye to convergence with a reformed BPL system, commitment to strengthening implementation and possibly ­ as has already happened in AP ­ its capacities at the lower levels of the system, in replacement in appropriate settings. While the particular at the block and GP levels. It will also new BPL methodology proposed by the Saxena mean building on innovations in institutional Committee improves upon the 2002 system in roles to promote greater accountability among SP several ways, several drawbacks of the previous service providers, as exemplified by social audits method remain. Piloting the proposed methods in MGNREG. and subjecting this methodology to the test CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 22 using NSS data as done with the 2002 BPL method range of reasons which are presented in this and other would yield interesting insights into how well the reports. Firstly, it remains easily the largest safety net new de jure targeting design would perform. program. Secondly, it has poor targeting outcomes, in the area of social security for unorganized and in many of the poorest states appalling levels of workers, past experience suggests that direct leakage which have persisted over many years. Thirdly, public provision, financing and administration the procurement system on which it relies is becoming is neither feasible nor desirable. It seems more increasingly strained in the face of gradual liberalization feasible to partner with existing non-governmental of agricultural markets in India, both domestically and in entities (for-profit, NGOs, and membership - terms of openness to trade. based organizations) and restrict the role of looking at results from the poor states in particular, government to: (i) providing targeted subsidies; it is hard to argue that pds comes anywhere near and (ii) regulating these entities and setting basic achieving its food security and poverty alleviation standards. This model already exists in India in objectives. However, reform options for PDS have several forms, such as the JBY scheme operated by to be assessed with a clear acknowledgement of the LIC for life insurance, and more importantly the RSBY context: technical, political economy, and possibly program. In addition, many other schemes falling legal. This context is of course strongly affected by also into the community based or micro-insurance the commitment of GoI to the Food Security Bill which category could be incorporated under an umbrella would convert PDS from a scheme to a legislated right. program that provided matching contributions or These include: premia but set certain standards in terms of benefit reformofPDSisinextricablylinkedtothepublic targets, eligibility conditions, investment policy and recordkeeping, among others. procurement system. While the benefits of PDS to households are spread across India, the main sp programs can be classified into three main pillars. beneficiaries of public procurement of grains to "Protective" measures or programs, which provide feed the PDS are concentrated among farmers relief against deprivation and shocks once they have in a few states: Punjab, Haryana, some parts of occurred. "Promotional" measures attempt to improve Uttar Pradesh, and Andhra Pradesh to a lesser incomes in the short and longer term through livelihood extent. Without some reform of the public and human capital interventions. "Preventive" measures procurement system ­ in terms of pricing policy seek to avert deprivation by supporting the ability of with respect to the Minimum Support Price and households to manage different risks and shocks they the control of government in grain procurement may face ex ante. The main sets of programs falling into more broadly ­ there will continue to be large each of these categories are discussed below, together grain stocks purchased each year which need to with a number of cross-cutting areas critical for program be drawn down. success, including financing and institutions, program there is a large internal bureaucracy running administration, and targeting. PDSwhichislikelytoresistfundamentalreform options which would undermine their role. The (a) Protective programs FCI alone employs around 450,000 people in India, and SFCs a further number. This in itself is a The programs intended primarily to protect individuals strong lobby which is likely to resist any changes and families once risks have been realized, include PDS, in the PDS which would imply either a potential social pensions and targeted housing programs. The role for the private sector in grain provision, or section reviews the performance of each and identifies more fundamentally a shift in use of the PDS some program reform options. subsidy from food purchase and distribution to cash for poor households. (i) PublicDistributionSystem(PDS) despitetheshiftattheaggregatelevelfromfood clearly, the dominant program as far as safety deficittofoodsurplus,thereare­andwillcontinue net reform in india is concerned is tpds. This is for a tobe­areasofthecountrywhichareperiodically 22 Social Protection for a Changing India: Volume II foodinsecure.This reality will continue to support aggregate level of public spending devoted to social arguments for a direct food security role for the assistance for the poor: state. At the same time, penetration of private food anincrementalapproachtoreformwhichwould markets has increased sharply in recent years, so retain the current PDS model but with a host of that the actual availability of food is there with the improvementsinthepolicyandimplementation exceptions of some tribal and dry land areas, and systemstoincreaseefficiencyfromitsoftenvery market infrastructure of well-developed and sub- poorstate. markets are increasingly well-integrated. Thus, anintermediatereformoption,whichwouldretain even with the food crisis of recent years, in most areas the challenge is not so much availability per afood-basedentitlementprogrambutintroduce se as price and affordability. privatesectorparticipationingrainprocurement and delivery and a more fundamental overhaul the maturity of the PDS has naturally created of the PDS administration through use of smart socialexpectationsthatitispartofthefabricof cards. socialpolicy.The results on awareness of PDS (see fundamental reform which allows for cash Chapter 7) are instructive in this respect, and the transfers instead of food-based transfers, either social constraints of any enforced fundamental whenthestateprovesitselfunabletofulfillitsfood change in PDS should not be under-estimated. transferobligationsorbyofferinghouseholdsthe theremayalsobelegalconstraintsontheability choiceofgrainorthecashequivalentofthegrain ofthestatetowithdrawfromdirectprovisionof subsidy.400 grainsthroughPDS.The Supreme Court of India, in response to Public Interest Litigation take up the overall position of this report is that fundamental by the People's Union for Civil Liberties in 2001 reform options for pds should be considered and with Union of India, FCI and eventually all state that offering households the option of a cash governments as respondents, has clarified that transfer - while retaining the core food security Article 21 of the Constitution on the Right to and buffer stock functions of fci, and ensuring pds Life also implies a right to live with dignity. The grains in areas where access is a genuine issue ­ is an SC agreed that one aspect of this was a right to attractive option. While there is certainly potential to food, with the PDS a central plank of the right. improve PDS performance (as is currently being done The Court has made various orders subsequently through the Ministry's nine point action plan), there in efforts to improve the functioning of food- remain a host of structural issues with a SP program so based schemes.398 A point of note is that the initial dependent on many intermediaries operating within order from 2001 indicated that "what is of utmost such weak governance systems. In that light, the importance is to see that food is provided to the incremental approach represents a triumph of hope aged, infirm, disabled, destitute women, destitute over experience which cannot be expected to resolve men who are in danger of starvation, pregnant the situation of PDS in much of the country. The long and lactating women and destitute children, term problems of PDS cannot be solved quickly or with especially in cases where they or members of their a single prescription for all states and all time. However, family do not have sufficient funds to provide food offering options which allow households choices in how for them".399 The implications are taken up below. they benefit from the massive public subsidy of PDS could allow for flexibility, greater efficiency, and accountability in light of this context, three reform options are for the system. While the proposed Food Security Bill presented for pds, which could themselves have will potentially have a major impact on the legally variants borrowing from the approaches discussed. possible range of options, it is hoped that it is not too Each approach is outlined below. It is stressed that none prescriptive, but leaves open approaches which would of the options assumes any necessary decrease in the appear to have the possibility to improve the welfare of 398 Ausefulsummaryandbackgroundmaterialscanbefoundatwww.righttofoodindia.org. 399 SeeSupremeCourtOrder,of23rdJuly,2001. 400 ThisoptionhasbeentakenbyarangeofIndiancommentatorsovertheyears,includingmostrecentlyKapuretal.(2008)andPanagariya(2008). CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 22 poor households. In this light, the default provided for likely to operate the FPS on a regular basis. Results in the consultation draft of the Bill for provision of cash from the SPS surveys on safety nets are instructive where the PDS system is unable to provide adequate and in this respect. Despite kerosene distribution in decent quality grains seems sensible. The system has so PDS being notorious for being high leakage, the clearly demonstrated its inability to do so for so long in incidence of kerosene from PDS in Orissa is far so many places that ruling out such an option seems more progressive than that of grains. A simple likely to leave many poor households with a stronger explanation appears to be that kerosene has been legal right but no better a real world situation. the first commodity transferred to operation by introducing a range of measures to make women's SHGs in the state.401 By early 2009, about honestly operated FPS financially viable, so that 83,000 FPS (out of the 4.98 lakhs shops across leakageisnotafinancialinevitabilityofthePDS the country) had been allotted to cooperatives, atthatlevel.The measures proposed ­ many of women's and other SHGs, PRIs, etc. which are in practice in some states - include: decentralization of grain purchases. This would (i) doorstep delivery of grains to FPS at the cost of have several possible advantages, and a number the state (currently being done in 17 states/UTs); of states are increasingly exploring this option in (ii) provision of rent-free premises by panchayats their grain purchasing. Firstly, locally purchased for FPS operations; (iii) allowing FPS to sell non- grains are more likely to accord with local dietary PDS commodities which will both increase their preferences, and hence be in greater demand. viability and increase the likelihood of being open Secondly, local procurement could spread the more days in a month. States such as Himachal economic benefits of the procurement subsidy Pradesh and Gujarat are already moving in this from the current concentration in a few states direction, which is supported by the Planning (two of which are very well-off ) to farmers in Commission in its PDS report; (iv) ensuring all states. Thirdly, the costs of transportation that FPS have a minimum catchment area and and storage of grains are likely to be reduced turnover of grain which allows for sufficient scale considerably. A number of states have started to to ensure viability. The Planning Commission move in this direction, including lagging states estimates that substantial share of FPS fail to meet such as Chhattisgarh and Orissa, though how far this benchmark. There is an obvious challenge this can advance before hitting resistance on the to enforce such a requirement in remote and supply side in traditional supply states such as inaccessible areas, which are likely to be those Punjab and Haryana remains to be seen. most reliant on FPS. One option for "squaring makingoptionsforhousehold-levelofftakefrom the circle" in this respect could be the use of FPS more flexible. The most obvious reform in mobile FPS vans such as Himachal Pradesh is this respect is obviously allowing, as a number already using in a number of remote districts; and of states already do, for weekly drawing of the (v) in line with Planning Commission suggestions, household's allocation. The only constraint on increase the margin on grains for FPS to 2 percent this is the FPS being open on a more regular of the economic cost. basis, but that should be addressed through the deepening ongoing reforms in management of management reforms outlined above. A second FPS.States ­ both rich and poor - are increasingly measure which may be more challenging from a moving from FPS operated by private operators bureaucratic perspective ­ but seems desirable ­ to operation by community-based institutions, would be allowing BPL and/or AAY households to including PRIs, SHGs, and cooperatives (e.g., in carry over their monthly ration from one month Tamil Nadu in 2009, almost 90 percent of FPS to another in months when they had less need are run by cooperatives). This seems eminently for their full ration. Given seasonal fluctuations sensible in terms of having operators who are in PDS to market prices, this could imply some more accountable to their communities and more lumpiness. A third option would be allowing 401 SeeDevetal.(2007). 22 Social Protection for a Changing India: Volume II households a more flexible mix of grains and transporting PDS commodities. Funds for piloting other basic commodities up to the value of their this technology in Chhattisgarh, Delhi and Tamil ration subsidy, an approach which GoI plans to Nadu were sanctioned in 2007-08. In addition, pilot in the coming year (see below). computerization of TPDS operations have been strengthening monitoring of PDS operations. A initiated in several states. number of measures to increase transparency conductinganindependentreviewofFCI'spricing in the operations of PDS are feasible within structure, with an eye to reducing the economic the current framework and are indeed being costs of grain procurement and handling. It implementing in several states. For instance, one appears that FCI is inefficient in its handling of element of management reform may be more grains, and the costs of this are naturally borne by direct community oversight in PDS operations, e.g., the budget and the end beneficiary. Despite this, through community verification of PDS delivery previous reviews have not been very probing in quantities where they are not already operating their efforts to look at possibilities for improved the FPS and/or through the involvement of PRIs in efficiency in FCI. A truly independent review of vigilance committees to monitor FPS. Some states their cost structure with a prior commitment by GoI have made some progress in implementing a to implement its findings would seem desirable system of monthly certification by PRIs/vigilance (though the technical demands of "efficient" cost committees for delivery of food grains to FPS estimation in the face of such a dominant market and allocation to ration card holders on time. In position should not be under-estimated). addition, public display of BPL lists and of district anotheroptionisintroductionoffoodstampsor and FPS-wise allocations of food grains for public coupons,anapproachthatisprevalentinAndhra scrutiny as proposed under the nine point action Pradesh and Rajasthan, and was introduced in plan would improve transparency. The review of Biharin2007.The current food coupon approach BPL/AAY lists in 14 states as part of this process has involves coupons which are redeemable only in already led to the elimination of 100.51 lakh bogus/ the FPS network, though there is no reason not to ineligible ration cards. Concurrent evaluations of allow for a coupon which would be redeemable PDS performance by NCAER and IIPA have been also in approved private food retailers (see below). commissioned in recent years by the Department Coupons may have attractions as a tracking and of Food and Consumer Affairs. leakage-reduction tool if FPS owners are able to strengthening the use of Information and replenish their grain stock based strictly on the Communication Technology (ICT) in the PDS at basis of redeemed coupons. In Andhra Pradesh, it the beneficiary end in order to promote more is estimated that the introduction of coupons has robust identification and in reducing leakage. reduced leakage in the PDS by up to 25 percent. This can take a variety of forms, some ­ such as The system has some obvious risks, including in Andhra Pradesh already ­ simple biometric forgery of coupons. Another issue may be the identification but not fully integrated into a "smart" development of a secondary market for coupons system, others involving bar coding and other (though how much of a concern this is if poor applications. The range of options for greater ICT households place more value on the cash than use in the PDS system have been exhaustively grains is a question). The Bihar experience with reviewed in a report for the GoI Ministry of Finance food coupons has been closely assessed through financed by the World Bank, which also conducted a two round survey-based assessment.403 While detailed feasibility assessments in Anand district only around 60 percent of eligible households in Gujarat and Thane district of Maharashtra in had been provided with food coupons in the 2007.402 In addition, the use of new technologies first year of implementation, access among them such as GPS for tracking movement of vehicles dramatically increased as a result of the reform, 402 SeeCal2Cal(2007)foraPDSsmartcardfeasibilitystudyproducedatrequestofMinistryofFinance. 403 SeeVashishtetal.(2009)whichanalyzesurveysfromapre-reformbaselineandafollow-upsurveyjustoverayearaftertheintroductionoffoodcoupons. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 22 rising from only 2 percent to around half of BPL intervals at which he/she could be supplied rations, etc. households. Bar coded coupons/ration cards have Point of Service (POS) terminals are simple machines that been introduced under TPDS in six states. read the cards and have the capability of authorizing the transactions via phone lines or the internet or recording the above options can be considered a minimalist the transactions on the smart card itself. approach to pds reform. the option of allowing grain purchases with coupons from approved private food smart card systems are not new to india or to south retailers would represent a more bold form of pds asia. Several experiments are already in operation, reform which feasibility studies indicate is viable. and the Rashtriya Swastya Bima Yojana (RSBY) health Some of the potential attractions of such an approach insurance program introduced by GoI in 2008 clearly include competition between PDS and private traders demonstrates that smart card applications are very (which could act as an accountability check and perhaps feasible for wide scale use even in poor and low incentivize efficiency improvements in FCI and SFCs), capacity settings, provided the right mix of institutional greater frequency of availability in areas where FPS players is involved and there is rigor in standard setting open irregularly, and possibly reduction in the transport and implementation (see Chapter 5). Indeed, there is and holding costs of PDS for the portion of grains discussion presently of "piggy-backing" other programs purchases from private traders. The United Progressive on the RSBY cards, and this is technically possible. There Alliance (UPA) Government in 2007 announced plans to has already been progress with respect to MGNREG in pilot such a reform in 20-50 districts, using coupons for this regard. Another notable small scale private initiative beneficiaries which can be redeemed in neighborhood is that of Swayam Krishi Sangam (SKS) in operation in stores other than FPS ("kirana" stores), though there one of the poorest districts of Andhra Pradesh (in appears to have been limited progress. In addition, Medak). The card is being used essentially as electronic beneficiaries would be allowed to have a more flexible passbook to record all the transactions for micro-finance form of ration, allowing for purchase of any mix of loans. Recently, GoI has provided in the 2008 and 2009 grains, pulses or other household basics up to the value budgets funds for piloting of a smart card-based reform of the coupon. This reform option was also mentioned in Haryana and Chandigarh which would also allow in the 2010 Economic Survey. While details remain to access to authorized private dealers. be worked out, the initiative seems worth pursuing and the worrying condition of pds as reported by goi, evaluating.404 national researchers and this report, suggests a more technically advanced form of food coupons however that a more fundamental reform of pds is could be introduction of "smart cards" in the pds, warranted which would introduce the option of which could be redeemable also at approved private cash transfers.406 The following paragraphs outline traders and/or fps.405 A smart card system can two variants of a fundamental reform option for PDS: facilitate two key functions: individual identification (i) switching to a cash transfer for the poor, or offering and remote transaction processing/storage. Smart options for poor households between food and cash; cards have the ability to store and record a large amount and (ii) switching to a conditional cash transfer, in order of program and authorized biometric information to leverage improved human capital outcomes from (signature or fingerprint image) that can be matched to the huge spending on PDS.407 Neither option would the actual fingerprint or signature of individual involved eliminate the need for food buffer stocks, nor for the in a transaction. In the context of the food ration system, continued use of food-based transfers for specific the card can store information on the identity of the situations (e.g., where relief aid is needed due to climactic individual, eligibility for rations, quantity, price and time or other disasters) or specific areas (e.g., remote or dry 404 TimesofIndia,July3,2007,"FoodcouponstoendPDSmess?",seehttp://timesofindia.indiatimes.com/news/india/Food-coupons-to-end-PDS-mess/articleshow/2165884. cms,lastaccessedonSeptember22,2009. 405 GoIhaslauncheditspilotofthisapproachinHaryanaandChandigarh.AndhraPradeshhasalreadygonesomewayinthisdirectionbyrecordingbiometric(iris) identificationofallBPLrationcardholdersinacentraldatabase.Thisallowedforsignificantreductionsinduplicateandghostrationcardholders.SeeCal2Cal (2007)foradetaileddiscussionofthetechnologicaloptionsandfeasibilityofsmartcardsinPDS. 406 See,forexample,RadhakrishanandSubbarao(1997). 407 Inessence,thisisthestrategythatMexicofollowedinitstransitionfromtheso-called"tortillasubsidy"totheCCTprogramProgresa(nowcalledOpportunidades). 20 Social Protection for a Changing India: Volume II land areas), and possibly specific population groups poor. There is also no solid or systematic evidence (e.g., those noted in the Supreme Court order above, from field work of the concern that cash grants which largely conform to the AAY category presently). are drunk or smoked by the men in recipient Despite the ongoing debate around the Food Security households, though of course that cannot be bill, Indian policymakers appear to be willing to explore discounted as a possibility. such options, as evidenced by the recent announcement there are several obvious challenges and issues in by the Government of Delhi of a "cash for ration" pilot in such a proposal. they include: one district which would provide ` 1,100 per month to women in poor households (an amount which would likely strong resistance from the stakeholders purchase around 30 kg of rice at market price as of mid- outlined above in the discussion of political 2009).408 The 2010/11 Economic Survey suggests that economy constraints. This is natural. The main the GoI is considering the introduction of food coupons way of managing this risk is to have a strategy for for households with a lumpsum entitlement (rather than "reinvention" of the FCI along the lines outlined, specific amounts for rice, wheat, sugar, etc.) that can be so that it becomes an active player in the new encashed at a PDS store of their choice.409 system, with important residual functions like buffer stock management, rather than simply a the elements of a transition from a food to cash "loser" in the reform process. based pds could be the following: resistance from FPS owner groups. Given the reform would have to start at the procurement existence of over 400,000 FPS owners, thought endofthesystemifitistobefeasible,andinvolve would need to be given to how to manage this areorientationofFCI'sfunctions.This essentially risk. The first element of a strategy would be involves a reduction in government controls continuing transfer of FPS functions to PRI/CBOs over grain markets and procurement operations. who are not "sole purpose" organizations in the Even in such a reformed system, FCI could still way that FPS owners would be, and hence are less be expected to play important roles. Firstly, it likely to resist changes that may be supported could compete with private players in the market, by their communities. Secondly, a sub-set of FPS relying on the economies of scale in operations, would continue to be needed for the continued existing infrastructure, distribution networks and PDS functions which would continue even in a other advantages to be competitive. Secondly, reformed system. Thirdly, the government could and very importantly, FCI could retain a role as as necessary support transitional livelihoods manager of India's buffer stocks of grain and support for FPS owners who lost their businesses perhaps their distribution in situations or areas as a result of reform (though this has not proven where they were needed. Finally, to the extent necessary to date in states which have transferred that specific groups or areas continued to receive FPS ownership from private individuals to PDS grains, FCI could continue to perform that community groups). distribution function in coordination with SFCs. legalimplicationswithrespecttoexistingSupreme if procurement reform could be achieved, Court decisions and the proposed Food Security substantialresourceswouldbefreedupforusein Bill.This is an important issue, as any reform must acashtransferprogramforthepoor.In essence, be consistent with the law of the land, more so public funds that are now spent on the large food when it flows from constitutional provisions. One subsidy bill of GoI would be freed up to be used for reading of the SC order is that Government may direct cash grants to the poor. The analysis of this not be constrained if it provided sufficient cash report shows that regular cash grants are subject transfers to the identified groups to allow them to less leakage than food in nearly all cases, are far to purchase food. A more demanding reading easier to administer, and are highly valued by the suggests that a legal obligation directly to provide 408 See Hindustan Times, August 27, 2009: Pro-poor schemes get the thumbs-up from plan panel, see http://www.hindustantimes.com/Pro-poor-schemes-get-a- thumbs-up-from-plan-panel/H1-Article1-447365.aspx,lastaccessedonSeptember22,2009. 409 GoI(2010). CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 21 food may apply to the groups noted. At the to those for a simple cash grant, though conditionalizing same time, the expansion of midday meals and transfers may face more legal complexities. The types Integrated Child Development Scheme (ICDS), of conditions that could be considered would vary together with the existence of other programs according to household characteristics, but could for the destitute such as Annapurna (and of include registration of girl births and school attendance course social pensions) goes some way towards for households with children. For other household meeting such an obligation if supplemented by types, conditionality may not be appropriate, especially a cash transfer. Clearly, however, this would need those in the categories outlined above. The obvious close consideration of the legal issues and their attraction of a CCT approach relative to a simple cash implications for a reform strategy. Another way of grant would be that PDS spending could be used to approaching this may be to offer households the leverage improved human capital outcomes, or other option of grains or cash and letting them decide. socially desirable goals such as better treatment of girl This would presumably meet the state's legal children. The additional challenges would include the obligations while offering choice to households administrative demands of operating a CCT, but the in exercising their rights. Janani Suraksha Yojana (JSY) institutional birth grant the existence of areas where food insecurity is and other state-specific experience in India suggests chronic. Despite the aggregate move to food that such demands can be dealt with. surplus, there is a natural concern about ensuring in light of various constraints and given the food security in these areas. It may be that fundamental nature of such a reform, it would be PDS continues in those areas until there is an advisable to experiment with a cash-based pds, perhaps assessment of the impact of transition to cash in focusing on more food secure states/districts first. It other areas. would also be sensible to ensure that the groups covered aggregate evidence indicates that caloric intakes under the Supreme Court order continue to receive are falling in india, even among the very poor.410 In grains, possibly in addition to a cash transfer. Such a light of this, moving to cash for PDS may seem counter- pilot would need careful monitoring and evaluation to intuitive. However, the counter-balances to that in the assess impacts and ensure that food security was not above proposal would include: (i) buffer stocks remain compromised, but for poorer states in particular seems an important part of the system, possibly with a "buffer a more promising way of ensuring (albeit in an indirect stock plus" to allow a more generous provision for manner) the right to food. grains for emergency and special situations; (ii) the recent expansion of midday meals may for children (ii) Socialpensions provide a caloric floor (see below); and (iii) it is proposed the evidence suggests that social pensions are a that AAY continue to provide food for the poorest, as reasonably effective part of the indian sp system, and would Annapurna for a share of the elderly destitute; their enhancement in coverage and central financing and (iv) ongoing efforts to improve the performance in recent years is well directed. In light of this, the reform of the nutritional program under ICDS could ­ if options for social pensions are more of an incremental effective ­ provide improved coverage of poor infants nature, both in terms of relative spending/coverage and pregnant/lactating women. Cumulatively, these by priority, and in terms of improvements in current policies no means suggest a wholesale withdrawal from direct and implementation. The main recommendations are: provision of food for the poor. there is a need in a number of states for better the second variant of a cash-based reform of pds educationofofficialsandotherinvolvedinsocial would be introducing a conditional cash transfer pension administration on eligibility policies. (cct) rather than a simple cash grant as outlined Field work in several states finds either partial above. The basic arguments for such an approach ­ and knowledge or problematic interpretations the political economy and legal constraints ­ are similar of previous eligibility criteria on "destitution" 410 SeeDeatonandDrèze(2009). 22 Social Protection for a Changing India: Volume II (now modified to BPL) and even the reforms under checks such as comparison of census and NSAP are likely to face some continuing problems beneficiary information by district is typically of identifying the poor. Following a review of not done, and has proven very informative where policy, a more thorough awareness raising has been done in identifying outliers on the up and perhaps development of implementation and downsides (e.g., in Karnataka). This would guidelines could help improve the situation. allow a more targeted process of performance morebroadly,asprogressismadeonexpansion audit. In addition to state departments, there of social security to the unorganized sector, it may be a potential role in such an enhanced willbecriticalforthosemanagingsocialpension process for PRIs. policy development to be involved in thinking areviewofpaymentsystemsforsocialpensions through options for pension expansion, and the wouldbedesirableinanumberofstates,aswould complementarities with social pensions as a athroughefforttoreconcilesometimesdivergent possible "zero pillar" of any contributory system locallevelandstatelevelreportingonbeneficiary fortheunorganizedsector. numbersandotherinformation. centralmonitoringofNSAPperformanceshouldbe furtherstrengthened,despitetherelianceonACA (iii) Targetedhousingprograms ratherthanCSSfinancingmode.Social pensions while there are clearly some significant challenges are one of the few major national programs for in making the targeting and performance of targeted which simple consolidated reporting on various housing programs more equitable and efficient, performance indicators is not available in annual specific recommendations on improvement are more reports of the Ministry of Rural Development. At difficult to suggest. Nonetheless, some suggestions a minimum, this would be desirable. A welcome include: recent initiative is the computerization of attempt to develop and implement mechanisms NSAP beneficiaries that significantly increases in states for minimizing the significant rent- transparency and enables the central Ministry seeking that appears to be happening in IAY. to monitor outcomes. States are currently in the This is inherently challenging so long as benefits process of moving to a full-fledged MIS for NSAP. involve large lump sums. However, the nature of As importantly, the central Ministry could act as a rural housing needs and behavior among the poor more effective conduit for sharing of good state- suggest that the program can be better adapted level experience in different aspects of policy and to needs. implementation. ensure that safeguards in current guidelines experience suggests that the previous fund flow are actually enforced. The recent introduction modelofsendingsocialpensiontransfersdirectly of "permanent waiting lists" based on the BPL to DRDA level is preferable to routing through list and the requirement of displaying these state treasuries, and that its reintroduction waiting lists in a public area (e.g., the wall of the should be considered. The strongest argument panchayat building) is the first step in increasing for this is evidence of even rich states such as transparency in the program. In this respect, Gujarat diverting social pension transfers for options such as social audits seen in the Mahatma other purposes, something that is less likely with Gandhi National Rural Employment Guarantee DRDA routing. In addition, delays in on-payment Scheme may offer an additional mechanism. by state treasuries could be avoided with such A second obvious improvement would be to a model. However, this potentially entails a provide for direct transfers to households via the policy decision to make the program a centrally banking system or post offices. Such a reform in sponsored scheme. IAY in Bihar has been introduced recently, and field theverificationprocessforcurrentbeneficiaries visits indicate significant reductions in leakage of bystatesneedstobemaderegular,andaprocess funds due to the reduction in intermediation by forremedialactiondeveloped.Presently, simple officials in the benefit payment process. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 conductadetailedstudyonpracticesamongthe view these programs in the context of the larger housing rural poor in housing improvement, to explore and habitat policy for the country. options for better matching scheme design with needsandpractice. (b) Public works and promotional a longer term option for targeted housing reform may be exploring possibilities for more market-led social protection programs provision of housing, particularly in urban areas. both public works and promotional social protection Experience in some metros may provide guidance here, programs have received increased political and and exploring the role of the private sector as financier budgetary priority in recent years. This seems a sensible and constructor but with subsidies to households effort to get greater leverage on poverty reduction and seems warranted. There are a range of challenges in human capital formation from the safety net. Given that market-based solutions which are particularly acute significant spending on these programs, particularly for the informal and self-employed sectors of the MGNREG and mid-day meals, can be expected to workforce due to difficulties in credit assessment, the continue, it will be increasingly important to ensure that need to mitigate risks for lenders, and transactions costs. the desired equity and public expenditure efficiency Nonetheless, small scale experiences of Micro Finance objectives are achieved. The following sections provide Institutions (MFIs) such as Swadhaar, Ujjivan and BASIX, some recommendations in this respect. and interest from small and medium housing finance companies indicate both interest in non-traditional (i) Publicworks market players in serving the urban poor for housing upgradation needs. Initial assessments indicate that many of the appropriate reforms of public works policy market-based solutions may be possible for households are already reflected in the guidelines of mgnreg, with monthly incomes in the range of ` 4,500-8,000.411 which in several ways represents the most serious In addition, there is a need to integrate housing delivery effort to date to address many of the institutional under IAY into a larger process of facilitating access to and implementation problems encountered in previous safe, sustainable and adequate housing. This could works (and indeed several other) programs. In many include access to credit in order to meet the funding gap states there has been greater political and institutional between the IAY subsidy and the cost of housing as well commitment to trying to "make the program work". as access to environmentally sustainable technologies Examining the implementation experience of MGNREG and building materials and services to assist people in in the last four years suggests some lessons for improving construction. the delivery of the program in states that are not doing as well as others. It also identifies some areas that warrant the more fundamental question on housing programs increasing attention as the program matures. for the poor is the extent to which they should remain as standalone css, or might more usefully be the variable implementation experience across states allowed for within an sp (or generalized) block grant offers several valuable cross-state learning. One system.412 The JNNURM approach of having an option of the biggest challenges in implementing MGNREG is within a larger program but requiring more pro-active to match the expression of demand with the supply of proposal development and management on the part worksites and employment opportunities. It is vital to of local authorities in order to use central funds for low- address the constraints ­ formal and informal ­ on this income housing seems worth exploring in the context process. At the broadest level, this goes to the heart of of IAY also. This would be more consistent with giving what a rights-based, demand-driven approach means in states/cities greater local determination in use of CSS practice. Some states have done better at establishing resources. Whatever the approach, it would certainly systems to improve the responsiveness of supply to the seem that innovation is needed in the area of public demand for work. This has typically involved attention subsidies for housing for the poor. It is also important to to or innovations in the following areas: 411 SeeMonitorGroup(2007)forausefulmarketassessmentofbothdemandandsupplysidesoflowcosturbanhousingoptions. 412 AproposalalongtheselineshasbeenmadebyKapuretal.(2008)usingIAYandSGSYfundsaspartofanenhancedPRIblockgrant. 2 Social Protection for a Changing India: Volume II Establishing the implementation structure by mandating that the GP MGNREG account and early in the game and ensuring adequate staff accounts of MGNREG households to be in the with the appropriate orientation and skills. This same branch). includes serious and sustained efforts at building Revisingtheruralscheduleofrates(SoRs)through capacity at all levels of the delivery process, detailed time and motion studies for different often in partnership with the State Institute for localesandgroupstoenablea"normal"workerto Rural Development. Under the Act, a portion of earntheminimumwageatMGNREGworksites. funds are available for capacity building of those Partneringwithcivilsocietyorganizationstowork involved in MGNREG implementation, including PRIs. Using these funds effectively will be a critical assupportagencieson a variety of areas, including element of program success over time. orientation and capacity building of MGNREG staff, awareness generation and mobilization among Generating awareness of the rights and workers, promoting participatory planning of entitlements under MGNREG as the first step works, and enhancing accountability. towards establishing a right to work. Over time, Emphasizing the commitment to transparency there has been some evolution of IEC campaigns in terms of moving from the dissemination of rights and accountability. While the design of to an emphasis of the need to and the means by MGNREG contains many safeguards in terms which individuals can demand that right. It is also of transparency and accountability, actual necessary to make communities aware of the implementation on this front has been highly unemployment allowance provision in the Act for variable. This includes institutionalizing social the guarantee function to be credible. It is vital accountability mechanisms, making the MIS up- to deepen awareness raising efforts on MGNREG to-date, and a continued emphasis on monitoring entitlements, in close collaboration with civil and community mobilization. society and using strategies that are tailored to a in addition, increased attention by policy-makers largely illiterate audience. and implementing agencies on improving the quality Leapfrogging technical manpower constraints and relevance of assets created under mgnreg is (e.g.,theshortageofengineersattheblocklevel) warranted as the program matures. This is critical if by developing detailed technical specifications MGNREG is to have any long-term impact on the rural of MGNREG works for different geo-climatic economy and future livelihoods. In addition, the creation conditionsasapreparatorystage.This minimizes of appropriate, durable and productive assets could the technical input required at the block and GP potentially help garner the support of non-MGNREG level at the planning stage and while starting participants for the program and a stake in improving a worksite. This can be done without the aid of program outcomes. There are many elements to this technology (e.g., as in Madhya Pradesh) or with increased focus on the second objective of the Act: technology as an integral part of a transactions- Explore options for a wider range of works based Management Information System (e.g., as authorized under MGNREG to reflect variable in Andhra Pradesh). needsandtodovetailwithotherprogramssoasto Streamliningtheflowoffundsinvariouswaysso contributetowardsacoherentvillagedevelopment astopreventfundingdelaystoconstrainopening plan.The list of eligible works needs to be flexible of worksites or payment of wages. For instance, enough to incorporate seasonality, differences in some states (e.g., MP) make available advance geo-climatic conditions and the needs of specific funds (linked to the volume of MGNREG work) groups, particularly those not capable of hard with GPs that makes it easier to open worksites manual labor. As experience grows, this would be in response to demand. In addition, some states worth looking at from several perspectives, some have reduced delays in payment of wages from of which are currently under discussion. The first the GP to worker post office accounts (e.g., by is the range of works that could be implemented placing a "float" with post offices to make wage in flood-prone, water-logged, heavily forested payments while waiting for funds transfer; and or mountainous areas. The second is the range CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 of possible works which could be offered to all rates of return on assets. In addition, the use of MGNREG workers, and whether some "softer" technology such as GIS in both planning and work options ­ e.g., related to provision of social monitoring would be. welfare or community services for vulnerable an additional issue for consideration is whether populations ­ could be introduced which have any element of direct human capital formation can positive social externalities. The third is looking at be factored into mgnreg as it matures. Presently the specific needs of sub-groups for whom special efforts may be needed to provide appropriate there is no provision under MGNREG for skill formation work, e.g., disabled people who may not be among workers. This may be something that could be able to carry out hard physical labor. The fourth considered in due course with the view of enhancing is exploring options for MGNREG to finance the not just current but also future livelihoods. One option labor component while the community (or other to consider is the South African public works program line departments) co-financing works that may which provides for two days training per month of not be on the approved list of MGNREG works but work for those undertaking public works. While such are part of the larger village development plan. an approach obviously requires a supply side agency ­ probably on a contracted-out basis ­ which can provide Revitalize Gram Sabhas and institutionalize the useful training, it seems a useful option to consider in directinvolvementofcommunitiesinidentification future development of MGNREG. ofworksundertakenunderMGNREGinawaythat isintegratedwiththelargervillagedevelopment a final issue is that public works for the poor remain plan.This has to date been the weakest element in restricted to rural areas. Recently, policy-makers the chain, largely because gramsabhas are often have started to discuss the possibility of designing a not held. If the Act's objective of empowering self-targeted public works programs for the urban poor. communities and strengthening grassroots Such programs already exist in the works schemes of democracy is to be achieved, it will be important a number of developing countries such as Ethiopia, to ensure that the role of gramsabhasanticipated Liberia, Colombia, and some other African and Latin under the Act is made a reality. This includes American countries. Drawing on the experience ensuring gramsabhas are held regularly, building of these countries, for example the role of urban the capacity of gram sabhas in participatory communities in identification of beneficiaries, planning planning, community oversight and other and execution of works, and oversight, would be useful relevant areas and the mobilization of groups in designing an urban public works program for Indian such as SHGs and CBOs to have their priorities cities and towns. In fact, one of the north eastern states reflected through the gramsabha process. is planning to introduce an employment guarantee for Establishsystemsforprovidingin-timetechnical urban areas as a state-funded scheme. Combining such inputsforassetplanningandevaluation,beyond a program with vocational or technical training would the currently mandated technical supervision enable young participants to upgrade their skills and during asset creation. This includes providing also compete in the labor market. An example of such technical inputs (e.g., through village-level resource an approach is being developed in Kenya for youths mapping to ascertain the technical feasibility living in urban slums.413 of different types of works) to the gram sabha during the planning process and development (ii) Programstopromotemovementoutof of shelf of works. Similarly, monitoring of the povertyintheshortandlongrun quality and durability of assets created will (a) Targeted rural credit increasingly become important. In this respect, MGNREG could draw on international experience the government has recently undertaken a significant (e.g., Bangladesh's Food For Work program) on overhaul of the sgsy in the form of the national developing cost effective methods for estimating rural livelihood mission. The main reasons for SGSY's 413 SeedelNinnoetal.(2009). 2 Social Protection for a Changing India: Volume II and predecessor programs' weak performance strongly MFI financial, reporting and other procedures and suggested the need for a fundamental reform that went those of government. beyond marginal improvements. In many ways, the where local labor markets are stagnant and/or restructuring of the existing program into NRLM goes a migrationisalreadysignificant,financingsupport long way in doing this. services for poor migrant workers which would the nrlm moves away from the precisely defined credit reducesomeoftheeconomicandsocialcostsof program for the poor as exemplified by sgsy and irdp migration, and increase its benefits. There are and instead includes a range of livelihood support already interesting small-scale examples of such options for the poor. This is consistent with the options initiatives, e.g., in southern Rajasthan with workers in the SP block grant proposed within the "3+ block" migrating to Gujarat. Some of the services that outlined in the Executive Summary. The rationale behind could be provided include: (i) reliable remittance such a restructuring and reorientation is necessary for a mechanisms; (ii) support for children of migrating variety of reasons: (i) the nature of labor markets across workers either in situ or in destination sites to and within states varies sufficiently that a "one size fits ensure that they do not drop out of school; and all" credit program seems increasingly inappropriate; (iii) provision of reliable information on labor (ii) the livelihood support needs of individual groups market conditions in destination areas, probably and poor households that go beyond credit are also through a contracted-out service provider diverse, and uniform CSS have failed to respond to this; contract. (iii) the growing penetration of both commercial banks and other non-bank players such as MFIs make the (b) School stipends challenge for the poor making them "bankable" rather while the evidence on school stipends suggests that they than being the sole source of formal credit. are not as well targeted as many safety net schemes, their objective is distinguishable. The position of this such an approach would require greater effort by report is that they will continue to be an intervention states to develop state specific poverty reduction worth expanding, though with serious thought about strategies for the effective use of central subsidies some design elements. The more pertinent concerns for livelihoods promotion. The NRLM design gives with school stipends seem to be two-fold: (i) in their states a fair degree of flexibility in trying different current form, where stipends are based on enrollment approaches to livelihood support; some further options and not attendance and thus with no assurance that the could be explored in this regard: intended outcome is being promoted, stipends are less using reputable MFIs as a channel for credit likely to achieve the goal of improved human capital delivery where they have a presence. MFIs have acquisition among the poor; and (ii) stipend schemes a strong interest in developing a sustained do not provide choice for households in schooling, relationship with BPL clientele, and more diverse being linked to enrollment only in government and and community-based sources of information aided schools. This may reduce their potential welfare on borrowers, allowing for a more informed impacts not only for the household, but also for the assessment of risk in the lending transaction. They education system, by increasing the marginal costs of are also more likely to enforce repayment, reducing private schooling and reducing the accountability of the rent-seeking opportunities seen presently, and public system that might otherwise be encouraged by provide appropriate support services to their households "voting with their feet". members. MFIs often have savings as an integral element of their relationship with clients, so that there seem solid arguments for exploring ways in which credit provision is reinforced by savings that can to make stipend receipt conditional on a specified level act as cushions against household shocks. At the of actual attendance rather than simply enrollment, same time, a significant issue would be the relative which in itself may not contribute to human capital terms of SGSY lending and those of most MFIs to investments. The obvious challenge that such a transition their existing clientele, and such an option would would present is operating an effective system to record demand some degree of harmonization between and verify school attendance. However, evidence from CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 a growing number of countries which have introduced at least on enrolments. As a result, the MDM program conditional cash transfers based on a specified level of provides an interesting example of a major demand school attendance indicates a range of positive effects side SP intervention which has great potential. In this on attendance, and other household welfare indicators. context, a stronger focus on monitoring and evaluation Though several central and state-specific CCT programs is warranted. This requires establish systems to monitor exist in India, these operate more as cash transfers the performance in the field with respect to inputs rather than conditional cash transfers. Key challenges (such as already being done for example for school in the effective implementation of these programs as infrastructure for providing meals), outputs (besides CCTs include the monitoring and enforcement of the administrative data on children availing of MDM) and conditionality and a weak institutional framework for outcomes. The latter requires conducting more widely such cross-sectoral programs. In addition, supply-side representative studies on the various impacts of midday constraints in the provision of services, particularly in meals, in particular nutritional impacts. In addition, rural areas, could also play a role. while several concerns on coverage and implementation have been addressed through the recent expansion and the arguments on introducing a demand side element to revised guidelines, more can be done to reduce the stipends through either expansion to private unaided variability in performance across states. schools or use of education vouchers redeemable in private schools are more complex. It is probably not possible to think about any wholesale shift to systems (c) Preventative measures: Closing like vouchers which would be relatively new in the Indian the coverage gap in social security context, and one where international experience ­ particularly in developing countries ­ is far more limited. in addition to social assistance programs aimed at However, this should not preclude piloting in states alleviating existing poverty, many social protection where the conditions are appropriate, or perhaps in systems insure against a range of shocks that can areas in selected states where the participation in private often lead to poverty.415 These social security schemes schooling by poor households is more pronounced (e.g., generally include provisions for old age, disability and in urban areas where the experience of a pilot in urban death of the breadwinner under the umbrella term, Delhi was fairly positive).414 pensions.416 Many countries have also tied health insurance coverage to membership in social security (c) Midday meals schemes which require contributions or premia while others have opted for general revenue financing.417 This like mgnrega, the mid-day meals program is also section is about the schemes that operate in India today conceived as a universal right rather than a targeted for both organized and unorganised sector workers. In program. Mid-day meals are a universal entitlement for addition, based on recent experience, it outlines thinking all children enrolled in government or government aided about how to expand pension and health insurance schools. With the recent expansion, midday meals have coverage to the vast majority of India's unorganized labor become an increasingly important part of the Indian force. Closing the "coverage gap" is an important policy safety net. While the national level impacts of midday objective in India and many developing countries.418 meals in educational, nutritional and other dimensions remains to be understood in depth, the evidence india's unorganized labor force is vast and available suggests that MDM have had positive effects heterogeneous and this reality must be taken into 414 CMSSocial(2009).SeealsoShahandBraun-Munzinger(2006)foracriticalreviewoftheexperienceofelevencountrieswitheducationvouchers,withadiscussion oflessonsforIndia. 415 Otherimportantsourcesofincomeprotectionsuchascropandlivestockinsurancearenottypicallycoveredbysocialinsuranceprograms,butarecriticalinterms ofprotectionagainstcovariateshocksintheruralsectorincountrieslikeIndia.Itisalsoworthnotingthatlumpyexpendituresforweddingsarealsoimportantin theIndiancontext. 416 Unemploymentandcashbenefitsformaternity(asopposedtomedicalinsurance)areoftencoveredbygovernmentsponsoredsocialinsuranceschemes.There aremoralhazardproblemsinbothcases,especiallywithregardtounemployment.Moreover,thedefinitionandmonitoringofunemploymentforunorganised sectorworkersisdifficulttoadminister.ESISoffersanunemploymentbenefit. 417 SeeWagstaff(2007)forasummaryoftheongoingdebatebetweenthesemodels. 418 Foradetaileddiscussionofextensionofsocialinsurancecoverageinthecontextofpensions,seeHolzmann,RobalinoandTakayama(2009). 2 Social Protection for a Changing India: Volume II account in any attempt to extend coverage of advantages over the alternative, the creation of a new social security programs. One important distinction layer of government bureaucracy with potentially is between households that are either too poor to lakhs of employees. In particular, the PA model has contribute or have already experienced the shocks become increasingly relevant in India in recent years that could otherwise be insured through contributory as financial sector reforms have resulted in competitive schemes. In these cases, social assistance i.e., cash markets for insurance and asset management and have transfers on a means-tested basis are the appropriate led to the creation of specialized regulatory institutions. intervention. These programs are dealt with in Chapter 3. This favorable situation should be exploited in order to At the same time, a large proportion of households would reduce costs and preempt the problems observed with benefit from ex ante interventions that allowed for monopoly, quasi-state entities that cover formal sector risk pooling and were made more affordable through workers. The key question is whether this `bottom up' direct subsidies. Among these households, a subset approach can be successfully scaled up with government is already participating in a number of uncoordinated support. Answering this question could begin with a attempts ­ mostly through group arrangements ­ to series of meetings that could bring together potential provide protection against a range of life-cycle risks. group participants (SHGs, MFIs, coops etc.), insurance Some are sponsored by government, such as welfare companies, asset managers, regulators, government funds while others are NGO-based initiatives that often ministries and researchers. involve microinsurance. initially targeting existing groups such as shgs and the lessons from some of the schemes that already exist mfis is recommended for several reasons including in india should be distilled and applied in the process for reducing transaction costs, achieving effective of adopting a new national policy for extending risk pooling420 and ensuring a minimum critical mass social protection coverage. These experiences could of voluntary take-up in the early stages of the provide policymakers with information ranging from program. However, there are tradeoffs involved in the factors affecting voluntary participation to data such a strategy. First, while the number of such groups that could help with actuarial calculations of cost.419 has been increasing rapidly in recent years, coverage Implementation lessons might include estimates of the is still relatively low and regionally concentrated costs of specific transactions, such as processing a claim (predominantly in the Southern states).421 Second, any or issuing a statement and the potential savings from subsidies involved in such a scheme should not be the scale economies usually found in administration expected to mainly reach the very poorest households. and recordkeeping operations. Comparing notes across Various studies have shown that membership in existing schemes might yield useful information about groups tends to be concentrated in the second and how to efficiently use technology or how to set up third quintiles rather than the bottom quintile.422 As processes to certify eligibility and verify claims. In some mentioned above, for many of these households, cases, there may be evidence that supports government the condition that would have been insured already action through better regulation or direct provision of exists (e.g., life insurance for widows). In these cases, public goods. The case of unique identifiers, already ex-post transfers in the form of social pensions seem mentioned, is likely to be one such case. the appropriate policy response. many of the existing schemes use a form of the at least three measures are required if these initiatives partner-agent model and this approach has several are to be successfully scaled up. 419 Unorganised sector workers and subsets therein will exhibit mortality and morbidity patterns that may differ greatly from the typical clientele of insurance companies.Forexample,thelifetablesusedbyLICarebasedonannuitantsmostofwhomarehigherincomeindividualswithmuchlowermortalityratesthan thegeneralpopulation.Adatabasethatsystematicallycollectedthisinformationcouldbeusedtoproduceactuarialtablesthatwouldserveasabenchmarkfor assessingcostswhendesigningtheschemeaswellassupervisingprovidersoncetheschemewasoperational. 420 Thesegroupsare,forexample,inamuchbetterpositiontoensurethatallmembersparticipateinahealthinsurancescheme,reducingadverseselectionproblems. 421 AsGhate(2006)reportshowever,thegrowthinthenumberofSHGslinkedwithbankshasincreaseddramaticallyandregionaldisparitiesareshrinkingsomewhat. 422 Seeforexample,BasuandSrivastava(2005)fortherelationshipbetweenincomelevelandmembershipinSHGsinAndhraPradesh.Inthesamevein,Ghate(2006) concludesthat"Microfinanceisbestsuitedtoreachtheeconomicallyactivepoor,whichmayexcludesomeinthelowestdecileortwoofthepopulationthatsuffer fromoldage,illhealthordisability." CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 first, in order to take advantage of economies depend on them for income.424 Nevertheless, extending of scale, to ensure portability and to facilitate coverage for other risks in a shorter time frame could supervision, some common standards would prevent millions of households from falling into poverty have to be developed, particularly in the area in the meantime. ofrecordkeeping.A good example is a universal achieving widespread coverage in an effective manner standard for identification of covered workers is not possible without the development of information that, in itself, would be a public good with many systems that allow the goi to track members of these other uses. Another is a ratings system for SHGs programs and the financial flows (contributions, and other groups based on objective criteria. premia and benefits) efficiently. Innovative use of second, appropriate contribution and insurance technology may be part of the answer, but in order to premia would be calculated based on rigorous keep costs down, existing infrastructure such as post actuarial calculations that were adjusted over offices and banks will have to be harnessed. In addition, timetoreflectexperience.This would help ensure and especially during the first phase of implementation, that unfunded liabilities (for example, those that the recordkeeping that already exists for groups such as arise when guarantees are offered) did not arise SHGs and MFIs should be utilized and, where necessary, and compromise financial sustainability. It would upgraded so that it is possible to `plug in' large numbers also make any subsidy involved explicit and of participants in a cost effective manner. Both transparent. governmental and external assistance could be made direct government subsidy is the third element available to groups willing to participate and meeting of successful coverage expansion. In order to recordkeeping standards (including compliance with encourage voluntary take up among low income the unique ID system) should be a condition for receipt segments of the unorganised labor force, the of subsidies. A centrally managed recordkeeping required premia and contributions would have to system is one option, but common standards that be subsidized. This subsidy would have to be set allow supervision entities to monitor effectively are at a realistic level given budget constraints.423 a minimum prerequisite. The time and resources required to design and construct a national system of fiscal constraints, along with the need to keep this kind should not be underestimated but getting the transaction costs low and to establish the credibility implementation machinery right at the beginning will of the new program, all suggest that the new avoid bigger problems down the line. At the same time, programs be phased in carefully and gradually. This India is a world leader in MIS applications so that state of kind of sequencing will inevitably imply tradeoffs. For the art solutions are possible if policymakers accord the example, as experience in India already shows, it is easier issue the importance it merits. to implement cash-based benefits covering risks such as death and old age than it is to provide health insurance finally, although not mentioned above, there are many due to supply side constraints and the complexities disparate and disconnected initiatives in progress by of monitoring providers. Add to this the problems of central and state governments in india that should adverse selection and moral hazard and it becomes be brought under a consolidated national policy clear that health insurance is much more difficult to framework. The passage of the Social Security Act at design and implement than pensions. Yet, medical care the end of 2008 is a positive step towards establishing is likely to be a greater priority for most unorganised such a framework. It now requires implementation. The sector workers and the members of the households that Social Security Board envisioned under this act can serve 423 TheGOIalreadysubsidizesmembersoftheEPFOwithanannualcontributionof1.16percentofthecoveredwagebill.Inaddition,favorabletaxtreatmentof superannuationproductsandthePersonalProvidentFund(PPF),amediumtermsavingsinstrument,resultsintaxexpendituresofanunknownamountthat mostlyaccruetohigherincomeworkerswhopayincometax.Itcouldbearguedthatamatchingsubsidyofthekinddescribedherewouldbebothmoreprogressive thanandhaveamorepositiveimpactfromapublicpolicyperspectivethanexistingsubsidies. 424 Interestingly, some surveys, (e.g., Madheswaran et al. 2005a show that unorganised sector workers rate old age pensions higher than insurance for sickness orhealth.However,thedataonactualshocksexperiencedsuggestthathealthshocksaremoreimportanttomosthouseholds,atleastonashorttermbasis. Furthermore, there is evidence that many workers do not value the benefit of health insurance if they do not make a claim and consider the premium `lost'. Pensions,ontheotherhand,aredeferredincomethat,atleastinprinciple,willeventuallybereceived(atleastbyasurvivorincaseofdeath). 20 Social Protection for a Changing India: Volume II as a coordinating body bringing various efforts under adjusting the composition to spending to greater one umbrella, spell out a national policy complete with emphasis over time on ex ante risk mitigation (i.e., target benefit levels, costing, subsidy and strategy for social insurance programs for the unorganized phased implementation. In light of the importance of sector) and promotional programs (including implementation issues and, especially recordkeeping, those linked to human capital formation, a special advisory group to make recommendations discussed in Chapter 4). on these aspects with the best available technical increasing the flexibility of funding from the support from public - private sector partners in India centre to states in terms of how it can be used for could be formed. different SP interventions, while strengthening the emphasis on spending outcomes. (d) Financing and institutions for deepening a range of administrative and institutional reforms in SP service delivery which social protection can contribute to greater expenditure efficiency. like other public interventions, effective social Many of the operational reforms are discussed in protection systems rely on a mutually supportive Chapter 7. web of appropriate policies, capable and empowered the notable exception is the area of social insurance, institutions for implementation, and transparent where the share of public spending can be expected and adequate financing. This section provides an to increase notably as india enters middle income overview of financing and institutional aspects of SP status, and where the imbalance between social programs, key determinants of SP system effectiveness security spending on the organized and unorganized in India. It first explores financing sources, mechanisms, sectors would be expected to shift in favour and patterns in recent years for the major centrally of the latter. In the medium term, international sponsored schemes of social protection, before experience and social expectations in India would examining the rapidly evolving institutional framework suggest significantly increased spending on social for SP program delivery. insurance for the large uncovered share of workers. it is clear that social protection is in an exciting It will be important that this increased expenditure phase in terms of policy evolution, financing levels, priority not be at the expense of social safety net and institutional arrangements. At the same time, spending, but rather is financed from a combination the situation presents an unfinished agenda which in of reallocation from non-merit spending, growth in some ways lacks a coherent vision for the system which aggregate fiscal resources, and contributions from is informed by the socio-economic developments workers themselves. In practice, expanding social outlined in Chapter 1, and institutional capacities. security to the unorganized sector is likely to require Looking at the financial and institutional aspects of SP not-insignificant public subsidies to incentivize programs covered in this chapter, some conclusions and participation, as one sees for example with RSBY and recommendations are as follows: certain pension schemes.425 In addition, effective social insurance programs can in part be expected to be (i) FinancingofSPprograms self-financing in two ways: (i) they should help control additional demands on the safety net that might broadly speaking, india's share of public spending otherwise arise due to factors such as population on safety nets is reasonable, and indeed quite high ageing; and (ii) international evidence suggests that as a share of gdp by international standards of low effective social security systems can actually contribute and middle income countries. Its main challenges are to growth in a variety of ways, from enabling higher therefore fourfold: risk/higher return productive activities to cushioning sustaining roughly the current share of public the impacts of growth-enhancing reforms for those spending as overall spending rises. who lose out from them in the short run.426 425 Forinternationalexperienceonmatchingsubsidiestoincentivizeparticipationinpensionprogramsforinformalsectorworkers,seeHolzmannetal.(2009). 426 SeeWDR,(2006),foradiscussionofinternationalevidencethatthereisnonecessarygrowth-equitytrade-off. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 21 there are pros and cons of different possible methods fundforstatesforSPprograms.An SP incentive of more flexible central funding for sp programs. fund could either be stand-alone or a window of Firstly, there would be natural reluctance from central broader proposal for a decentralization incentive Ministries to simply transferring SP CSS resources fund. Alternatively, a central innovation fund for to Normal or even Additional Central Assistance, as social protection could be created which could evidenced by the failure to transfer several CSS to allow states to access central funds on a proposal- states as per the 1999 Planning Commission list driven basis for innovations in SP program of schemes. More specifically, as Normal Central delivery, or experimentation with new initiatives. Assistance is based on population and poverty under the Gadgil formula and not on performance, (ii) InstitutionalrolesforSPprograms incentives for states to improve performance on the most fundamental institutional challenges specific SP programs would be relatively weak. For in sp css continue to be delineating clear lines of Additional Central Assistance, there are examples of accountability in service delivery, and supporting the linking transfers to state reform performance, such authorized actors with adequate staff and finances. as JNNURM. However, these requirements have not This will require first and foremost greater proactivity on been strongly enforced in practice. There is also no the part of states to approve policies and put into practice obvious mechanism for the Planning Commission to the PRI/ULB decentralization provided for under the monitor performance of specific CSS under NCA, and 73rd and 74th constitutional amendments. With notable for ACA only where responsibility is given to a central exception of Kerala, and incomplete examples such as Ministry. Rajasthan and Karnataka, most states have yet to define overall, an improved transfer system for sp css the framework for decentralized service delivery in a would need to balance need, capacity and fiscal and sufficiently operational manner. administrative effort of states. This could involve this would need to be followed by a process-intensive several elements, including: reconciliation of central guidelines, state-level a first step towards such a system would be stances on service delivery decentralization, and movingtotowardsamoreoutcomebasedfunding capacities at sub-state levels to perform the required method. Such an approach would, however, implementation functions in sp programs. This is not require a clearer set of outcome indicators against a process which will generate a single "right" answer which state performance could be measured, and on the assignment of activities to different actors at a stronger planning function at state and district different levels of the system. What it could usefully levels. This approach is one that is already being achieve however is a more considered assessment by used in India in some donor programs which are both centre and states of what is the chain of activities results or outcome-based. from top to bottom required to deliver effective SP asecondstepinsuchaprocessmaybeallowing programs, and of the realistic potential of different for sanctioning of allocations for states which actors to deliver on their proposed responsibilities. This stretch across two to three budget years, which could in time improve the alignment of functions, funds could lessen current incentives to spend SP and functionaries in SP programs. Such efforts would allocations inefficiently (or simply parking them need to focus in particular on elements of the service in accounts), and also assist with programs which delivery chain at block level and below. The process are subject to cyclical demand ­ such as public could involve several steps: works ­ which is not presently synchronized with "unbundling"eachcoreSPCSStoitsconstituent the budget planning and release cycle. functions or services and specific activities in a third step could be the block grant for SP different programs, as has been done for the programsoutlinedabove. MGNREG and RSBY guidelines. While implicit in an alternative approach which could be guidelines of some other schemes, such a simple implementedevenintheabsenceofmoreserious exercise would be a useful building block for a financing reforms is the creation of an incentive principled division of functions across levels. 22 Social Protection for a Changing India: Volume II forallstates,producingaconsolidatedmapping in different aspects of SP service delivery. of the current assignment of sectors and SP Increasingly, the private sector has been program-specific functions, both de jure and playing different roles in the Sp service delivery de facto. This exercise would clearly identify chain. RSBY is the most advanced example, cases where concurrent responsibilities arise for with involvement of private insurers, private programs where DRDAs and/or PRIs have specific hospitals, private sector smart card providers and responsibilities assigned to them under CSS outsourced software development, as well as guidelines. Where such concurrent responsibilities roles for grassroots organizations noted above. are apparent, further dialogue between the states However, many other programs ­ particularly and centre would be needed to agree whether at state level ­ have been experimenting with CSS guidelines need to provide more flexibility to outsourcing of different functions. In some cases, states on institutional arrangements for programs, this involves partnerships on technology, such as or whether states need to be encouraged to move the role of TCS in managing MGNREG databases more actively on state-specific assignments which in AP, or smart card provision by commercial firms are consistent with CSS programs (with Kerala in a number of pilots. In others, there is a more providing a good example of such a proactive involved role for the private sector, such as in role by states). Such discussions could be guided Bihar where rural informatics service centres for by both public finance criteria for assigning RD programs at block level have been assigned responsibilities, and basic management principles following a tender process, in which for-profit for enhancing accountability.427 Both of these are and not-for-profit organizations operate the discussed in Chapters 6 and 7. information and database management of RD havingdeterminedanappropriateassignmentof programs on behalf of the state Government. functions to different levels, a fuller assessment in addition to the above needs on institutional roles, is needed of the personnel and financial specific suggestions include: implications of carrying out required functions. at both central and state levels, formation of Where enhanced personnel capacity is needed, an inter-departmental Task Force or Authority this would then need ­ as is the case with for Social Protection, which would promote MGNREG ­ to include the financing of needs coordination across programs targeted towards in program costing, and the capacity building similar populations, and promote more coherent requirements at different levels. strategy development on the medium term graduallyinstitutionalizingrolesforcommunities policy mix and priorities in social protection. inbothselectedelementsofCSSdelivery,andin The cases of Mission Convergence in Delhi community oversight of program functioning. and the Safety Nets Authority in UP provide Again, the model of MGNREG provides a interesting models, which could be adapted by useful precedent. This is also likely to involve other states. partnerships with NGOs to build local capacity forprogramplanning,severalinitiativeswould for effective social audit, and the more general be useful, including: (i) earlier notification obligations of the state for disclosure of program to states and DRDAs of estimated funding information, as has proved effective for example envelopes for programs for the following fiscal in Rajasthan. The RSBY scheme also provides year in order to facilitate lower level planning useful precedents on the role of NGOs and other and budget management; and (ii) ensure that grassroots organizations in SP service delivery states which have not already done so appoint and demand side mobilization. District Planning Committees, and ensure taking lessons from the emerging experience in adequate financing for technical support India and beyond in public-private partnerships to DPCs. 427 SeealsotherecommendationsoftheEmpoweredSub-CommitteeoftheNationalDevelopmentCouncil(NDC)onFinancialandAdministrativeEmpowermentof thePRIsJuly2008. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 associalsecurityexpands,theroleoftheprivate should be gradually built. Furthermore, a firm stand sector (as insurers) and a range of member- could be taken on the policy of registration and based organizations such as MFIs, NGOs and tracking of beneficiaries of the public programs. workers associations is already becoming From the discussion above, and from the insights increasingly important, and demands new of many GoI and independent studies on program modes of engagement and partnership form performance of CSS SP programs, some more specific the public sector. The biggest constraint on recommendations emerge: expanding social security to the unorganized sector has been developing delivery mechanisms (i) Onawarenessandoutreach which can deal with the transactions costs of while program awareness and outreach is an reaching unorganized workers. This requires inherently difficult process in a country of india's intensive engagement with intermediary size and diversity, experience from states and for partners between government/insurers and specific programs such as mgnreg indicates that much unorganized workers, as well as learning from can be done if there is sufficient will. An improved efficient insurance distribution channels of awareness and outreach strategy would likely involve public and private insurers. The RSBY provides several elements: an excellent model of partnership, as do more developing a better picture of what are the localized initiatives such as UTI partnerships key information sources of the poor on public with both state governments and organizations programs, including what types of information such as SEWA, and the roles of trade unions and theygetfromwhichchannels.The limited survey employer organizations in some of the welfare information indicates that typical government funds around the country. media campaigns, web-based information, and for M&E, develop a disaggregated picture of information from administrative officials are not potentialandcapacitiesatdifferentlevelsofthe the channels through which the poor find out systemformonitoringandevaluation,and align about programs. program guidelines in that light. The proposal providing in SP program budgets for sustained to establish a national Independent Evaluation awarenessraisingbythirdpartieswherepossible. Office is a welcome step with respect to M&E. Field work for this report indicates that program awareness dissemination is not a one-time (e) Program administration in social activity, but a product of repeat contacts and local sources of reliable information. This tends protection not to be the approach of most IEC campaigns for with a myriad of programs and plans to introduce SP programs. new schemes, expand coverage and consolidate developingamorediverserangeofIECstrategies existing schemes, it is an important time to assess forprogramawarenessraising.This would involve relevant experiences and identify both good and bad not only a more diverse range of actors as tools administrative practices. While India may not have of information dissemination (including NGOs, the advantage of other countries of running a well- SHGs, youth groups, social and religious leaders), organized centralized machinery of administration but also a more diverse set of information tools of public benefits, some strategic priorities could (including visual media suitable for non-literate be defined to achieve better benefit delivery and audiences, television and radio campaigns which transparent operation. In particular, policies seeking more creatively engage audiences rather than harmonization, portability, and transparency in simply presenting program guidelines, use of benefit provisions, and otherwise ensuring minimum community, etc.). operational standards, should be designed, endorsed, deepening the use of social audits not only as a and promoted. Capacity to regulate and supervise tool for program monitoring, but as a source of various benefit providers and their intermediaries programinformationdissemination. 2 Social Protection for a Changing India: Volume II (ii) Ontheapplicationsprocess ofICTinreleaseandtrackingoffunds.They will review program eligibility criteria on several also need to facilitate sustained FM capacity key programs with a view to simplification of development in PRIs if the intentions of eligibilityproofandunderstandingofthepublic. decentralization of SP programs are to be realized. One aspect of this would be exploring avenues This will also require in most states a more for reliable third party verification of certain facts elaborated financial accountability framework for where documentary evidence is not available or PRIs and sub-state implementation agents. costly to gather. other reforms ­ such as accounting for SP CSS proactive exploration of how SP programs can resources as "spent" when they have not been reducetransactionscostsofprogramapplication and application of accrual accounting ­ will throughlinkingtotherangeofICTinitiativessuch require coordinated efforts of the centre and as rural business kiosks and other village-level states on reporting processes that require connectivity initiatives. The emerging examples reliable information on physical and other in other contexts such as spreading availability of outcomes of "spending". The example of land records through rural E-kiosks (e.g., Bhoomi in MGNREG reporting in some states provides a Karnataka) and availability of market information good model which should be institutionalized. (e.g., ITC's E-chaupal initiative) suggest that A first step in coordinated effort between the concerted effort between SP departments and centre and states could be clear agreement on their ICT department colleagues could pave the the necessary skills needed for FM at different way for a major contribution to simplification levels of the system for SP programs (and more of program applications, and significant cost broadly), and agreement on action plans for reductions for applicant households (not to ensuring adequate skills. mention lower opportunities for undesirable as part of strengthening the accounting middlemen). and financial reporting framework, it will be encourage­ifnecessarywithpublicsubsidyona increasingly important in SP CSS which route contracting-outbasis­NGO,SHGandotherCBO through DRDAs to develop standard financial intermediationtofacilitateprogramapplications, reporting rules for societies, which would and increase the use of camp approaches to subsequentlybereflectedintheGFRsforGoIand registration by the administration itself. Given thestates.This work would most sensibly be led that the culture of reliance on middlemen is by CAG and ICAI. so prevalent, it seems sensible to encourage addressthebackloginexternalauditsofSPCSS, the efforts of middlemen (or more likely and a more competitive process of selection of "middlewomen") who are more likely to have the third party accountants responsible for auditing interests of the applicant at stake. societies. All Ministries implementing SP CSS in terms of program oversight, it seems sensible should also post the findings and actions taken to target IAY and SGSY for particular scrutiny, on audits on websites. includinguseofsocialaudits. equally ­ and more challenging in terms of sharetheexperiencesofstateswhicharepiloting bureaucratic culture ­ there is a strong need to biometricandotherICTadvancestoreducefraud deepen the uses to which financial information and human error in the beneficiary recording on programs is put, and hence to move away process. from the simple input/output culture of financial reporting which prevails in most CSS. (iii) Financialmanagement finally, the positive development of increased some of the needed improvements are in the citizen accountability for SP funds use seen in domain of states, and they will need to follow MGNREG should be institutionalized over time the lead of recent reformers in areas like public in all SP programs, with basic initiatives like expenditure management rules and expansion posting of all program releases and uses, greater CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 reliance on social audits, and posting of beneficiary with respect to record keeping, the experiences of more lists and program rules in plain language. A progressive states such as karnataka offer lessons number of these are natural corollaries of the for others. Some of the issues for attention include: Right to Information Act, but their full potential training (and subsequent monitoring and remains to be exploited in most states of India to enforcement of compliance) for GPs and blocks in date. maintaining appropriate registers of beneficiaries. As one of the challenges in this work with be (iv) Paymentsystemsandrecordkeeping capacity constraints at both levels, it could be reforms of payment systems can take a range of useful to explore options for supplementing forms, both technology-based, and where that is not this in non-costly ways, such as providing skilled possible, with community monitoring of payments. workers from MGNREG as temporary assistants to Some of technological the options include: the gram sewak. low cost ATMs developed at a fraction of cost at the state and district level, development of of the conventional machines, equipped simple verification exercises which will allow with finger print identification and long-life program oversight to focus on high-risk areas. batteries, can facilitate cash distribution in remote Even available monitoring data often throws up areas. obvious anomalies, but is typically not used for this purpose. point-of-service (POS) devices installed in local shops and gas stations can use regular telephone at central and state levels, improve cross- lines to process on-line cash disbursement departmental coordination on a range of record transactions, or ­ in the case of PDS ­ facilities for keeping issues, starting with a strategy for debiting the monthly ration. convergence of beneficiary identification numbers. Even where systems remain parallel for the short special mobile branches could make conventional to medium term, Collectors and BDOs should banking more accessible, enabling regular and encourage greater cross-program information schedules visits in the remote communities. exchange for programs with common elements finally, mobile phone banking is an emerging such as being food-based or child-focused. phenomenon that utilizes comprehensive as a condition of CSS receipts, the centre could penetration of the new communication insist that a computerization strategy and roll-out networks. plan for SP programs be developed, including a more generally, cost recovery mechanisms of different training plan. This seems a bare minimum for operational alternatives to the conventional models movement towards a modern system of record of service provision should be further studied. keeping. While some models are quite promising, their financial encourage states to be more proactive in exploring viability as localized stand-alone solutions remains a outsourcing possibilities on development of concern. Clever solutions may simply need to be scaled program record-keeping software. India offers up in order to reach the break-even point of economic a particularly promising environment for such operation. Furthermore, strategies enabling synergies efforts, though there would remain an important across multiple programs should be further explored. role for the central government in ensuring For example, while the smart card technology may consistency in basic design features, and thus be a costly solution if utilized only by a single benefit comparability of data from different states. program428, using it as a common platform for multiple products (e.g., various benefit and insurance schemes) (v) Monitoringandevaluation may well help to make its application a sound economic ensure that any major SP program has a pre- proposition. program baseline done, and that impact 428 ThemarginalcostofonesmartcardissuedinIndiaisquotedasbeinginthe$2-$3range. 2 Social Protection for a Changing India: Volume II evaluations also include control areas where very localized level in relative terms rather than feasible. the kind of absolute measure of poor people explore replicating the model of MGNREG with necessary for comparisons across space. provision for more robust research and evaluation a key challenge therefore is how might the system financed from the program budget. marry the "top down" targeting necessary for spatial on monitoring: (i) review the current set of distribution of social protection resources, with the monitoring indicators for major programs to assess "bottom-up" information on the characteristics their utility and gaps in key program indicators; of the poor which might do a better job of ranking (ii) if necessary on a contracted-out basis or households at the grassroots level. There appear to under special units (e.g., as Orissa has developed be several elements of a potential strategy. These are under its Poverty Monitoring Agency), conduct discussed below, first presenting more modest reform regular basic analysis of consolidated program proposals, and then presenting an option which would monitoring information at state and central levels; imply more substantial reform. and (iii) explores institutional mechanisms in international evidence on targeting outcomes states and at the centre to use the analysis of such provides useful insights in thinking about an units for annual review of program performance appropriate targeting mix. Three observations emerge and policy implications. from a review of targeting methods and outcomes in building on the social audit experience in some 122 targeted programs in 47 developing and transition states, mainstream social audits as a standard countries: practice in all SP programs. While the experience some methods have better targeting outcomes is relatively new on any scale, social audits seem a than others when taken across a wide range of useful tool for increasing community knowledge countries. of programs and providing a vehicle for redress where needed. However, in the absence of an combining targeting methods generally results in authorizing environment in program guidelines improved outcomes. For example, a combination (such as under MGNREG), the potential of social of three methods improves targeting outcomes by around 30 percent. This potential improvements audits will remain limited. needs to be traded off however with additional administrative costs and demands, and the risks (f) Targeting mechanisms of too much complexity introducing possibilities for manipulation. the analysis in the report suggests there is major scope at the same time, there is also strong variation to improve targeting systems for public programs in across countries in targeting outcomes within india. At the same time, there are several generic issues each method, pointing to the importance of which must be taken into account in reforming targeting country specificities and implementation capacity in SP programs: in considering options. Across the sample of the distribution of households exhibits programs, around 80 percent of the variation in considerable clustering around the poverty line targeting outcomes is accounted for by within nationally, making fine-tuned targeting design method variation, and around 20 percent by inherently challenging. between method variations. One important there is considerable diversity across states in dimension of this is community "voice" in the the factors that are correlated with poverty, so society, which internationally improves targeting that reliance on standard national indicators in outcomes by around 23 percent. targeting systems has inherent weaknesses. international evidence therefore clearly suggests community level beneficiary identification has that no single targeting method will ever be sufficient, attractions in such an environment. However, and that the policy question is whether a more its strength is within-community targeting at a effective mix of methods can be employed for sp (and CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 other programs).429 This would include improvements (ii) Improvinghousehold-leveltargeting in methods for each type of targeting, and potentially a in addition, there is a major need to improve india's different mix of methods. The discussion below focuses household level targeting systems, though the range initially on improvements in geographic targeting and of options varies between urban and rural areas, and the "backbone" BPL method, before a discussion of more by program type. This section looks at three possible substantial reform options. reforms in the BPL methodology as well as the potential (i) Improvinggeographictargeting role of other methods, including community-based and self-targeting methods. None of them provide a "magic with respect to geographic targeting of central sp bullet" for all programs, but are suggestive of options for allocations, it would be useful to generate sub-state improvement which may contribute to a more effective estimates of poverty using poverty mapping techniques, mix of targeting methods. and use these as the basis for future spatial resource allocation.430 Poverty mapping allows for reliable reforming the bpl method. given the concerns over small area estimates of poverty below the state level by widespread exclusion and inclusion errors in the combining information from the NSS household survey 2002 bpl method, an expert committee has suggested data with unit record data from the Census. Since the significant changes in design of the proposed 2009 late 1990s, detailed "poverty maps" have been prepared bpl census (as described in the Saxena Committee for a growing number of developing countries. These report).431 One, the method proposes the automatic maps provide estimates of poverty and inequality at the exclusion of visibly non-poor households and the local level ­ such as the district, sub-district, and even automatic inclusion of the most vulnerable households, village level. Such information is not commonly available with a survey and scoring only of the remaining because household surveys are typically too small in households. Two, the new method overlays geographic sample size to permit sufficiently fine disaggregation. targeting criteria by specifying district, block and GP- Yet, with ongoing efforts to apply detailed spatial level quotas with respect to the proportion of BPL targeting of public interventions, or to realize the households. In districts that have a very high proportion gains from decentralization and community-centered of BPL households (over 80 percent), all households development, there is a pressing need for information (except the visibly non-poor) would be considered to on distributional outcomes at the local level. be poor and automatically included in the BPL list. This design reduces administrative demands and cost to in india, this could reliably generate poverty estimates some extent by reducing the scope of the survey. The at block level. While progress on this front would not underlying principle of the proposed methodology is address the issue of household level targeting, it would to minimize errors of exclusion, without worrying over- significantly reduce issues of targeting across space much about errors of inclusion. The elements of the within states. In India, a pilot effort to produce poverty proposed method are described below. maps in three states - West Bengal, Orissa and Andhra Pradesh ­ is ongoing. Once that is complete, it would be setting quotas for the distribution of bpl households a more precise basis for allocations of CSS SP resources across space: District-level thresholds for the proportion across space. In particular, these poverty maps can of BPL households could be set using the ratio of SC/ potentially be an important part of a strategy to marry ST population in the district to that in the state, the "top down" targeting (essential for fiscal allocations of inverse of agricultural production per rural person and resources across space) with "bottom-up" targeting of agricultural wage rate of the district.432 These could then households or individuals (e.g., through community- be used to estimate block-level proportions of the poor based methods that rely on local knowledge but are using any one or a combination of indicators on soil and not comparable across space). irrigation quality, road connectivity, female literacy and 429 SeeCoadyetal.(2004)forausefuldiscussionofdifferenttargetingmethodsandcombinationsindevelopingcountries. 430 SeeElbersetal.(2003)fordiscussionofthepovertymappingmethodologyandWorldBank(forthcoming-a)foradiscussiononthepilotsinIndia. 431 SaxenaCommitteereport(2009). 432 ThisisthestandardformulausedforfiscalallocationsforRuralDevelopmentprograms. 2 Social Protection for a Changing India: Volume II share of non-agricultural workers. Finally, these block- benefits; (e) income tax payers.433 These criteria level estimates would be used to generate GP-level are an improvement over those used in the quotas for BPL households on the basis of population 1997 BPL method as (i) they allow for some local estimates. Estimating these disaggregated sub-district variation to a limited extent (e.g., land holding is thresholds as well as choosing alternative rules to assessed relative to the local average instead of generate GP-level quotas could be left to the discretion the all-India threshold of two hectares previously) of the state government or district authorities. and (ii) focus only on high-value assets that may be more successful in separating the rich identifying poor households: The basic principle behind from the poor. the methodology being proposed is to ensure the inclusion of groups that are regarded as particularly vulnerable (e.g., Automatic inclusion of certain categories of workers in low income or highly vulnerable occupational households that are considered to be among categories, households where earning capacity is the poorest and most vulnerable: These include severely curtailed by external constraints, etc.) and/or the following categories: (a) designated Primitive groups that are historically subjected to social exclusion. Tribal Groups; (b) designated most discriminated The proposed method of identifying households has the against SC groups (called Maha Dalit Groups), following three elements: if so identified by the state; (c) single women headed households; (d) households with disabled Automatic exclusion of visibly non-poor house- person as bread-earner; (e) households headed holds by a minor; (f ) destitute households that are Automatic inclusion of the poorest and most dependent predominantly on alms for survival; (g) vulnerable households homeless households; and (h) households where Scoring of the remaining households (except any member is a bonded laborer. However, there in districts with poverty above the specified is some confusion in the treatment of households threshold ­ 80 percent - where all households, and individuals while operationalizing these except those automatically excluded, are criteria. While the report focuses for the most identified as poor without a survey). part on households (i.e., a joint family that resides Automatic exclusion of households that are under a common roof and eat from a common meet certain criteria and are identified as non- kitchen) and nuclear families, some of the specific poor: The intention is to exclude households categories noted above relate to individuals. In this that are visibly non-poor, but whose position of case, the report suggests treating these groups economic, social or political privilege makes it as separate households (e.g., single women with possible for them to get (wrongly) included in BPL no major son, elderly couples or individuals, etc.). lists. The exclusion criteria include the following: However, it is not clear how to score the remaining (a) households with double the land of the district household members. Also, not all of these criteria average of the agricultural land per agricultural are easily measurable or verifiable in the field ­ household if partially or wholly irrigated (three e.g., a functional assessment of disability requires times if completely unirrigated); (b) households several detailed questions and highly trained who have two-wheeled (or more than two investigators. wheeled) motorized vehicles; (c) households who Grading of the remaining households (identified have at least one mechanized farm equipment; as poor) in order to prioritize programs and (d) households who have any person who is services to the poorest among these households: drawing a salary of over ` 10,000 per month in The proposed scoring system ranks households non-government/private organizations or is on a scale from one to ten using the following employed in government (including parastatals) criteria: (a) caste and religious affiliation ­ SC/ST on a regular basis with pension or equivalent (3 points), denotified tribes and designated `Most 433 Thesecriteriaapplytotheentirejointfamily­i.e.,thefirstcriteriaofhigherthanaveragelandownershipwouldapplyeventothesonofalargefarmereven thoughhemaynotactuallyhavelandregisteredinhisname. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 Backward Castes' (2 points), Muslim/OBC (1 point); to be reporting issues with some of the indicators (e.g., (b) any one household member engaged in the chronic health problems, disability). The adoption of this following occupations ­ landless agricultural methodology will also have significant fiscal implications. worker (4 points), agricultural laborer with some The inclusion criteria are defined on the basis of nuclear land (3 points), casual workers (2 points), self- families and individuals who are considered as separate employed artisans or fisher folk (2 points); (c) no households for the purpose of this exercise. This would adult above the age of 30 with schooling up to lead to the number of BPL households increasing grade five (1 point); (d) any household member significantly, implying possible rationing within these with TB, leprosy, mental illness or HIV/AIDS or households if program budgets do not rise adequately. disability (1 point); and (e) households headed by as a result, there is still scope to move to a more an old person of age 60 and above (1 point). Since developed form of proxy means-test, using methods the bunching of a large number of households is which are good practice in developing countries, for possible at certain scores, the method proposes a both rural and urban areas. There are several possible ranking of households with the same score, with elements in such as reform: those in the special category group such as SC/ ST at the top, followed by landless agricultural evenifstandardnationalBPLcriteriaareretained, laborers and so on. these should be determined using rigorous regression models for greater targeting power. this proposal is a modest approach to reform in While the proposed 2009 BPL methodology does that it proposes continuing with the bpl system with include criteria such as caste affiliation that have some improvements. The proposed methodology is an been found to be strong correlates of poverty, improvement over the 2002 BPL method in some ways: the preferred method for such an exercise is to (i) The indicators chosen are likely to be highly correlated use the NSS or other representative national with poverty for the most part (though not necessarily data to generate a set of indicators that provide for the elderly).434 Some of these indicators (gender of more targeting power, and which could be easily household head, education, occupation) were among a verifiable at reasonable cost. wider set of indicators used by Jalan and Murgai (2007) whatever indicators are used, they should have in their augmented regression model that performed some weighting ­ preferably on a state-specific much better than the 2002 BPL method. However, while basis. The basis for such weights and a more the reduction in the number of indicators increases sensible ranking between levels of each indicator simplicity in field processes, it may make it more difficult should be the NSS data, or other representative to separate the rich from the poor and the poor from national data where relevant. In contrast, the poorest; (ii) Weights have been assigned to the the proposed BPL methodology and Kerala's indicators as noted above. However, weights have Kudumbashree use a set of weighted indicators not been applied across states; and (iii) The use of the for the identification of poor households, but the geographic, automatic inclusion and exclusion criteria weights do not appear to be based on regression reduce the scope of the survey and hence administrative analysis. These are likely to be assigned by demands and costs. Subjecting this methodology to the committee or, in the case of Kudumbashree, test using NSS data as done with the 2002 BPL method developed by communities in a participatory would yield interesting insights into how well the new manner. de jure targeting design would perform. a preferred approach beyond the above would however, several drawbacks of the previous bpl be to let the indicators in the BPL methodology method remain, particularly with respect to the vary by state, or perhaps groups of states. As imposition of central criteria across diverse local the purpose of BPL is increasingly ranking of contexts, cardinality of indicators, and the same households rather than determining aggregate process of aggregation. In addition, there are likely numbers of poor people (which remain controlled 434 SeePalandPalacios(2008)foracomparisonofpovertyratesamongtheelderly. 20 Social Protection for a Changing India: Volume II through the linkage to Planning Commission political economy may make convergence aggregates for states), there seems no inherent challenging, particularly where BPL ration cards need to insist on standard national indicators. As significantly exceed the numbers on BPL lists. part of a process to generate indicators, it would no comparable central guidelines for a census be worthwhile to simulate state or region-specific of poor households exists for urban areas. indicator sets and test their power in ranking of Typically, the identification of poor households is households. carried out separately by different departments, such piloting the proposed method would allow as the Food and Civil Supplies Department for PDS an assessment of how well these indicators ration cards. However, methods differ across states capture household welfare, measured using and departments and there is little systematic other accepted measures such as per capita information on this process. Notable exceptions are consumptionand/orassetindices. Delhi and Kerala. More recently, Delhi has defined a BPLlistsshouldbesubjecttolocalizedverification common process of identification of the poor across by communities, using gram sabhas or other programs. This defines a set of thematic indicators mechanisms. In principle, this should happen, capturing residential location, social deprivation and does in some states (e.g., HP) and/or for and occupational vulnerability to identify vulnerable specific programs. However, the process is not households. The underlying principle of promoting inclusion and several of the specific indicators are systematic and made more complicated by the similar to those proposed in the rural 2009 BPL fact that gram sabhas are held infrequently and methodology (see Chapter 8 for details). A survey participation is often low in most states.435 If it is currently underway to collect information on were to become more standard, it may be useful households in poor localities, with the intention of in some areas to have the verification process covering other areas over time. The list of vulnerable facilitated by NGOs or other third parties where households so generated would be used by the nine there are concerns of elite capture. Verification state departments that are engaged in delivering and finalization of the BPL list by the gramsabha programs or services to the urban poor. This is a very is one of the suggestions made by the Saxena positive step towards generating a common beneficiary Committee draft report. database for multiple programs, thereby reducing build in systems for dynamic updating of BPL administrative costs and increasing transparency. This, listsinbetweencensuses.The Saxena Committee however, does not necessarily mean that all programs draft report suggests holding a census once would need to target the same groups. Programs such every ten years, combined with a system for as educational scholarships that target SC/ST children updating the lists every two years by registering could use the information in the common database changes in household circumstances. However, to continue targeting on caste affiliation rather than the report does not specify the implementation vulnerability status. arrangements for doing this. while the above reforms could significantly improve requirestatestohavecommonBPLlistsandBPL the bpl system, international and indian evidence ration card aggregates, preferably consistent suggests that indicator-based targeting ­ even also at the household level. Several states where generated robustly from reliable survey already do this, and it has the benefit of saving data ­ will struggle to explain a significant portion administrative demands and avoiding multiple of variations in living standards across households. targeting criteria. The Saxena Committee draft This is true in a static sense, where such methods rarely report makes this recommendation. However, in explain more than half the variance in household states where the two do not presently coincide, consumption.436 435 Eveninthefoursouthernstates,only20percentofhouseholdsreportedattendedagramsabha(Besleyetal.2005). 436 Forexample,Egypt'sPMTforitsfoodrationingsystemcapturesonly43percentofthevariationinconsumptionsacrosshouseholds,andArmenia'sonlyaround onequarterofthevariation. CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 21 (iii) Otherreformoptions community identification into their formal SP systems (e.g., Uzbekistan), suggests that this is a surmountable while reform of the bpl system is highly desirable, challenge. there may be additional options in rural areas of some states for community-based targeting to play whether the modest or more fundamental reform a stronger role in sp programs. As the centre enforces options are pursued, self-targeting (as happens in convergence on aggregate numbers of poor people public works) would continue to be a useful tool between Planning Commission estimates and the BPL for programs where it is suitable. This approach is system, possibilities for more significant reforms of most effective when used to target goods or services targeting systems could be considered in rural areas that are more heavily consumed by the poor than the (though perhaps not in urban, where an enhanced PMT non-poor, and for programs where monitoring costs would seem more feasible). associated with verification of eligibility are high. Programs such as public works are amenable to self- once a poverty map is available at block level, targeting approaches through the wage rate and reliance on community-level wealth ranking becomes types of work offered. Others such as subsidized credit a feasible option as a tool within the formal system lend themselves less readily to self-targeting. For cash for identifying sp beneficiaries. While not an ideal benefits, the possible role of self-targeting will depend match, the availability of aggregate numbers to block on benefit levels, ease of accessing programs and other level and community ranking at GP/village level comes factors. Despite these obvious limitations, the empirical close to the "top-down" and the "bottom-up" meeting. results in Chapter 4 highlighting the superior targeting To the extent there remains a gap, solutions such as performance of public works suggests that "cutting assuming even poverty rates within the block, or some out the middleman" in targeting of SP programs simple indicators such as population share (as is being remains an important tool for programs where the proposed in the Saxena Committee draft report) or SC/ nature of the self-targeting requirement or of benefits ST share or landless share in different villages might offered permits. This suggests that self-targeting be used to allocate SP resources within a block. The should continue to play a role in programs where it is precedent of Andhra Pradesh is instructive, where a list suitable. An important additional factor supporting of poor household generated entirely by community this conclusion is that self-targeting may minimize the wealth ranking has replaced the administratively political costs of clear designation of target groups, by generated BPL list across the state for the main allowing policymakers to describe such programs as anti-poverty programs administered by the Rural "open to all". Development Department. This may not be feasible in all states (e.g., where caste fragmentation is high and risks of local elite capture more pronounced), and would need facilitation by competent NGOs. However, C. PoliTiCAl eConomy oF an evaluation of the AP experience, and well-evaluated SoCiAl PRoTeCTion ReFoRm pilots in other states seem warranted to see whether the political economy of sp policy reorientation is some of the shortcomings of even a well-designed PMT complex, and will require intensive efforts to build can be mitigated through a stronger community role in consensus on reforms. In particular, it will be important identifying beneficiaries. to ensure that the interests of perceived "losers" of SP as happens presently with community wealth ranking, and broader economic reforms are taken into account. the process would benefit from being framed within While reforms that involve expanded coverage or new general criteria for ranking households. From an types of interventions are unlikely to be controversial, implementation viewpoint, there is a tension between there are strong interests in preserving the status quo more localized guidelines and minimizing intensity of in SP programs among a range of actors, including process and costs in developing the guidelines. However, administrators, politicians, contractors and others. experience from India (e.g., Kerala's Kudumbshree) Simply cutting programs or excluding certain groups of and from other countries which have formalized beneficiaries or institutional players is therefore unlikely 22 Social Protection for a Changing India: Volume II to be successful unless incentives for institutions and with the SP system. The generic identity of such households which will be affected by reforms can be official intermediaries and unofficial middlemen part of the reform package. The political economy of SP is generally well-understood, but minimizing the reform is challenging in all countries, and governments potential avenues for their continued roles has only in India would benefit from more innovation in their recently become a more explicit goal of SP policy efforts to create a broader societal understanding of the design. While it is too early to say, even apparently need for and benefits of reforms. naïve blanket bans on certain actors in legislation and/or guidelines (such as the ban on contractors some of the political economy challenges that the under MGNREGA) do appear to help. However, a sp system confronts if it is to become more coherent more comprehensive approach will require a more and more effective as a tool for promoting poverty thorough modernization of SP business processes. reduction and inclusive growth include: Examples where such approaches already appear as in many areas of policy reform in India, to be making headway include greater reliance on consolidationandreformoftheSPsystemwithin direct transfers to beneficiaries through banking a coherent strategy will run counter to the past and postal systems, and innovations in use of ICT experience of scheme-driven initiatives by a in SP program delivery. plethoraofMinistries,andtheobservedtendency amorerecentandfluiddevelopmentinthepolitical of each new government at both central and economyofSP(andmuchother)servicedeliveryis state levels to want new SP programs clearly howincreaseddecentralizationofresponsibilities distinguishablefromtheirpredecessors.Reducing to panchayats ­ in particular GPs ­ generates these natural bureaucratic and political tendencies different patterns of contention, cooperation will be very challenging. A first step is obviously and collusion between newly elected panchayat having an integrated SP strategy which is driven by officials and traditional loci of influence among the top politicians and bureaucrats at central and administrators and higher level politicians such as state levels, with strong inputs from civil society MLAs and MPs. An essential first step in promoting in its development, including opposition parties. decentralization of SP service delivery as a tool for However, even if such a strategy process can be contestability and hence accountability will be developed, it will be important for it not to become getting a better empirical understanding of the a "one shot" exercise, but to have institutional diversity and evolution of experience. This would coordination mechanisms in place which explore include how the gradual increase in the role of program duplication and exploit synergies. panchayats is proceeding (and what factors ­ givingstatesamoreflexiblehandinuseofcentral such as limited control of resources and very low SPresourceswillbeachallengingtransformation capacity), and the extent to which panchayats both for central administrators (whose past effectively promote accountability in SP service tendency has been to define the parameters for delivery or are captured by local social, political use of central funds quite tightly) and politicians and administrative elites. (who not unexpectedly seek political attribution a more nascent, but powerful, element in the for centrally-financed schemes implemented by politicaleconomyofSPreformisthepolicyshift states). The first of these challenges in perhaps towards a rights based approach. Government easier to address through development of more of India is increasingly operationalizing such an outcome-based monitoring systems. The second approach in a number of areas through legislation is more difficult in a democracy. and specific policies and programs. For instance, in a number of programs, there are presently the Right to Information Act was passed in 2005 significantrent-seekingopportunitiesforarange and mandates the government to release timely ofactors.Such opportunities are facilitated by the information demanded by citizens. It has been current complexity of the SP program mix, but widely hailed as one of the most important also by the number of intermediaries who often drivers of governance reform and transparency are involved in the interactions of poor people in India. The Right to Food and Right to CHAPTER-9: Getting More from India's Social Protection System ­ Directions for the Future 2 Livelihood movements are led by civil society, for communities in SP service delivery in recent but have managed to influence government so years is encouraging. This is both cross-cutting, that the National Rural Employment Guarantee through reforms such as the Right to Information Act, (now renamed the Mahatma Gandhi Rural Act, and program specific, such as the anticipated Employment Guarantee Act) was passed in role of social audits in MGNREGA, and new roles 2005. Other similar Acts include the Right to for community groups such as SHGs in delivery of Education Act (2009), the Scheduled Tribes and some SP services (e.g., running Fair Price Shops). Other Traditional Forest Dwellers (Recognition of However, there is no guarantee of "trickle down" Forest Rights) Act (often called simply the "Tribal to citizens in terms of awareness of their emerging Rights Act", 2006) while legislations such as for entitlements. The role of NGOs, media and other food security are on the anvil. actors in this respect cannot be under-stated, as various political and administrative actors a consequence of the above is the increasingly at local level may not have strong incentives prominent role of communities and civil society to promote such citizen-based accountability in promoting more effective poverty reduction mechanisms. Notable examples such as MKSS in outcomes from spending. In this respect, the Rajasthan demonstrate the potential impacts of strengthening of the "authorizing environment" such partnerships. 2 Social Protection for a Changing India: Volume II Annexes annex 1 India Human Development Survey (IHDS) ­ II (2005) The India Human Development Survey (IHDS) ­ II (2004-05) is a household survey collected by the National welFARe meASuRe Council of Applied Economic Research (NCAER), as a Income and consumption data were not available from collaborative exercise with the University of Maryland, the IHDS at the time of the analysis, but the asset index USA. The data canvassed around 216,000 individuals allows for analysis across the wealth distribution. We can living in 41,500 households in India. Questions on safety compute a measure of wealth based on the ownership net programs were included in the survey with financial of durable goods and housing characteristics. The wealth support from the World Bank, and analysis of the data index is generated using a principle components model carried out by Bank staff after entry and cleaning by (see Filmer and Pritchett, 1998). Separate wealth indices NCAER. The data are representative at the state level are created for urban areas and for rural areas, with for 33 states and Union Territories and for urban and data from all the states pooled together. Household rural areas separately. Table A1.1. presents the sample wealth quintiles too are computed for urban and rural size (number of households) canvassed in the survey for areas separately. Filmer and Pritchett demonstrate that urban and rural areas separately and by state. measuring long-run wealth (asset index) using a principle Table A1.1: number of households surveyed for urban and rural areas separately and by state States urban Rural Total Andhra Pradesh 900 1,535 2,435 Arunachal Pradesh 45 119 164 Assam 315 701 1,016 Bihar 464 966 1,430 Chandigarh (UT) 80 80 Chattisgarh 269 905 1,174 Dadra & Nagar Haveli (UT) 60 60 Daman & Diu (UT) 60 60 Delhi (UT) 840 120 960 Goa 44 121 165 Gujarat 900 1,178 2,078 Haryana 268 1,351 1,619 Himachal Pradesh 315 1,058 1,373 Jammu & Kashmir 315 400 715 Jharkhand 405 519 924 Annexes 2 States urban Rural Total Karnataka 1,168 2,852 4,020 Kerala 627 1,104 1,731 Madhya Pradesh 645 2,168 2,813 Maharashtra 1,125 2,078 3,203 Manipur 45 60 105 Meghalaya 45 116 161 Mizoram 45 60 105 Nagaland 30 100 130 Orissa 600 1,466 2,066 Pondicherry (UT) 45 60 105 Punjab 540 1,052 1,592 Rajasthan 892 1,595 2,487 Sikkim 45 60 105 Tamil Nadu 1,108 991 2,099 Tripura 45 184 229 Uttar Pradesh 1,125 2,387 3,512 Uttaranchal 135 323 458 West Bengal 1,124 1,256 2,380 All India 14,549 27,005 41,554 Source: Ajwad 2006. components model can perform as well as other measures whether or not a household owns/uses each of these of welfare, such as household per capita consumption. items and not the quantity or quality of the items439. The asset index employed in this note is computed The asset index satisfies internal coherence (Table A1.2). using 24 durable goods. The durable goods used are: Sharp differences are seen in ownership across households bicycles, sewing machines, generator sets, mixer/ in the poorer versus richer households. For example, 1.3, grinder, motor cycle, black and while television, color 0.1, 0.2 percent of households in the poorest quintile television, air cooler, clock or watch, electric fan, chair own a sewing machine, motor bike or scooter or color or table, cot, telephone, cell phone, fridge, and pressure television, while in the richest quintile 59, 59, and 65 cooker. Among households that had at least 6 of the percent respectively own the same assets. As expected, above durable goods, households are asked about the all households in the richest quintile own at least two following: car, AC, washing machine, computer, credit pairs of clothes and two pairs of shoes for all members of card437. If a household has less than 4 items from the main the family, while only 85 and 64 percent of households list of assets, then they are also asked about owning: two in the poorest quintile own those assets respectively. pairs of clothes, shoes or chappals for all the members of the household438. These last two categories of assets Asset deprivation is broadly consistent with state are particularly good at separating out the poor from rankings using other sources of information such as per the very poor. Note that we only have information on capita gross state domestic product or the HDI. Consider 437 Sinceanswersareonlyrecordedforhouseholdswithatleast6durablegoods,weassumethathouseholdswithfewerthan6durablegoodsdonotpossessanyof these5items.Sincethese6itemsarelikelytobeownedbyverywelloffhouseholds,theassumptionislikelytobemet. 438 Sinceanswersarerecordedonlyforhouseholdswithmorethan4durablegoods,weassumethathouseholdswithmorethan4durablegoodsalsohavetwopairs ofclothesforeachofthehouseholdmembersandthateveryoneinthehouseholdalsohasshoesorchappals. 439 Notincludedinthedataaresomefarmassetslike:tubewells,electricpumps,dieselpumps,bullockcarts,tractors,threshers,biogasplant.Thesevariablesmaybeuseful tocreateawelfareaggregate,butthefactthatnon-farmhouseholds(whetherrichorpoor)arenotlikelytohaveanyoftheassetsmakesinclusionlesscompelling. Alsonotincludedarehousingcharacteristicssuchas:themainsourceofdrinkingwater(tapinsidetheresidence,sharedorpublictap,hand-pump,etc.),typeof housingmaterial(kachcha,semi-pucca,pucca),typeoftoiletfacility(ownflushtoilet,ownpittoilet,sharedtoilet,etc.),mainsourceoflightingforthehouse(electricity, kerosene,other),andmainsourceoffuelforcooking(LPG/electricity,kerosene,wood,other).ThesedataarenotavailableintheIHDS.Itshouldalsobenotedthat abouthalftheassetsinthemodelrequireanelectricityconnectionintothehouseholdoragenerator.Inourmodel,onlyhalftheassetsareelectricitybased. 2 Social Protection for a Changing India: Volume II Table A1.2: Asset ownership ownership for all ownership for households ownership for households households in quintile 1 in quintile 5 Own (1); rent home (2); neither (3) 1.11 1.11 1.07 Bicycle 58.1% 39.5% 71.1% Sewing machine 20.2% 1.3% 58.5% Generator 1.1% 0.0% 5.4% Mixer 22.5% 0.4% 54.7% Motor bike/Scooter 16.0% 0.1% 58.7% TV ­ Black & White 24.7% 12.3% 32.9% TV - Color 24.1% 0.2% 65.4% Air cooler 9.8% 0.2% 34.2% Clock or watch 83.2% 38.1% 99.3% Fan 58.5% 14.2% 96.3% Chair or table 64.9% 14.0% 97.7% Cot 85.1% 64.9% 97.9% Phone 13.7% 0.0% 48.6% Cell phone 7.0% 0.0% 26.0% Fridge 13.4% 0.0% 44.4% Pressure cooker 38.2% 4.3% 83.1% Car 1.6% 0.0% 8.0% AC 0.5% 0.0% 2.2% Washing machine 3.2% 0.0% 14.4% Computer 1.0% 0.0% 4.2% Credit card 1.4% 0.0% 7.7% Clothes 97.1% 84.5% 100.0% Shoes 93.0% 64.3% 100.0% Source: Ajwad 2006. a measure such as the proportion of households in the Bihar (60 percent), Madhya Pradesh (55 percent), and state falling into the all-India bottom 40 percent of the West Bengal (53 percent) respectively rank among welfare distribution. With that measure, Maharashtra states with the largest number of poor people. By this (around 67 percent of its residents fall into the measure, the smallest contributors to overall poverty bottom 40 percent of the all-India welfare ranking), are Mizoram, Goa, and Sikkim. Table A1.3: Proportion of state's population belonging to the all-india welfare quintiles States Poorest q2 q3 q4 Richest Andhra Pradesh 12.80% 14.10% 32.00% 27.30% 13.80% Arunachal Pradesh 0.20% 4.00% 12.70% 40.50% 42.60% Assam 6.50% 30.80% 34.30% 17.20% 11.10% Bihar 33.00% 27.30% 21.20% 11.20% 7.10% Chandigarh (UT) 4.80% 8.60% 19.50% 14.70% 52.40% Chattisgarh 25.40% 21.20% 25.00% 15.50% 12.90% Dadra & Nagar Haveli (UT) 10.40% 18.60% 12.70% 10.40% 47.90% Daman & Diu (UT) 0.70% 0.40% 4.30% 29.20% 65.50% Delhi (UT) 5.20% 11.70% 17.30% 25.60% 40.10% Annexes 2 States Poorest q2 q3 q4 Richest Goa 1.10% 1.00% 3.30% 21.20% 73.40% Gujarat 5.10% 14.50% 24.20% 26.90% 29.30% Haryana 3.60% 5.10% 11.40% 21.90% 58.00% Himachal Pradesh 2.00% 5.60% 15.50% 25.30% 51.50% Jammu & Kashmir 9.90% 15.10% 20.20% 22.70% 32.10% Jharkhand 20.20% 22.50% 25.50% 16.50% 15.30% Karnataka 28.00% 17.50% 18.00% 16.60% 19.90% Kerala 1.70% 11.20% 15.40% 18.70% 53.00% Madhya Pradesh 32.60% 22.70% 16.80% 13.00% 14.90% Maharashtra 37.90% 28.90% 21.60% 10.90% 0.60% Manipur 6.20% 2.00% 15.80% 23.20% 52.80% Meghalaya 29.60% 18.60% 21.10% 17.20% 13.50% Mizoram 0.00% 0.90% 24.60% 26.40% 48.10% Nagaland 2.50% 3.10% 17.10% 58.90% 18.40% Orissa 38.80% 12.50% 22.60% 13.30% 12.70% Pondicherry (UT) 11.40% 9.80% 20.90% 17.50% 40.30% Punjab 1.30% 4.50% 9.40% 20.60% 64.20% Rajasthan 14.90% 24.70% 19.00% 18.40% 23.00% Sikkim 0.00% 2.60% 11.90% 31.80% 53.60% Tamil Nadu 18.30% 15.60% 21.20% 21.70% 23.30% Tripura 12.90% 12.60% 26.60% 31.20% 16.70% Uttar Pradesh 19.10% 29.50% 17.60% 17.30% 18.70% Uttaranchal 9.20% 12.30% 18.30% 31.30% 28.80% West Bengal 25.70% 27.40% 19.50% 18.00% 9.50% All India 20.30% 21.10% 20.60% 18.30% 20.00% Source: Ajwad 2006. 20 Social Protection for a Changing India: Volume II annex 2 Social Protection Survey in Three States This study financed a special household survey represented the most developed area of the district and conducted in three Indian states, Orissa, Madhya Pradesh the other the least developed area. For this purpose, a and Karnataka. A multi-stage stratified sample scheme comprehensive development index was constructed for (see Table A2.1) was followed for selecting the sample all the blocks in the sample district on the basis of Census households, as described below. data like sex ratio, literate population and literacy rate, worker and non-worker population, etc. A total of 5 villages were selected from each district SAmPle SeleCTion (two blocks). From the most developed block, 2 villages In each state, three districts were selected. While sampling were selected randomly using PPS method. In the least the districts within each state, it was ensured that they developed block three villages were selected with are located in different agro-climatic zones with differing PPS, one from the list of all villages having more than socio-economic characteristics. For this purpose each 60 percent SC and ST population and two from the rest. state was divided into different socio-cultural regions In each village a sample frame of households was (SCRs)440 based on the agro-climatic zones and socio- prepared by listing all the households using a separate economic characteristics (urban, remote, tribal, etc). From listing schedule. The listing schedule also contained each state three SCRs were selected with probability some ancillary information like the caste, land proportional to size (PPS). Further one district was selected randomly using PPS method from each sample possessed, education, participation in four major safety SCR. The list of selected districts is given in Table A2.2. net programs, etc. From the listing of households, a sample of 30 households was selected for administering From each sample district, two blocks were selected. the main household questionnaire and the women's While making the selection, it was ensured that one block questionnaire. In order to gain insight into the benefits Table A2.1: Sample coverage of SPS survey level Details Sample in the Study States Madhya Pradesh, Orissa & Karnataka 3 Districts 3 per State 9 Blocks 2 per District 18 Villages 5 per District 45 Households Survey 30 per Village 1356 440 Khan(1992). Annexes 21 Table A2.2: Districts selected for the survey State SCR District Madhya Pradesh Bhagelkhand Rewa Bundelkhand Datia Chhatisgarh & Gondwana Seoni Karnataka Kannada (Bombay) Bagalkot Karnatak (Deccan) Bidar Old Mysore & Maland Shimoga Orissa Chhatish Garh & Gondwana Sundargarh Coastal Orissa (Kalinga) Jajapur Dandekaranya Koraput availed from government-funded safety net programs, absent, it was administered to any knowledgeable adult it was decided to over-sample households which have member of the household. The section for the females was availed benefits under Food for Work, Indira Awaas administered to the spouse of the head of the household. Yojana, SGRY, SGSY.441 For this purpose all the listed In case she was not available, it was administered to any households were divided into five strata. The participants adult married female available in the household. The in the programs Food for Work, Indira Awaas Yojana, coverage for the study is summarized in the tables. SGRY, SGSY were treated as four separate strata442. The rest of the households were treated as the fifth stratum. Finally, we selected 3 households from each of the first meThoDology AnD ToolS four strata. The balance of 18 sample households were a mix of quantitative and qualitative approaches was selected randomly from fifth stratum443. The survey was adopted in the survey. The instruments used to collect conducted in 45 villages covering 1356 households. the data are shown in Table A2.3. The main questionnaire was administered to the head The survey also used a wealth index to rank households of the household. In case the head of the household was for purposes of incidence analysis. This was constructed Table A2.3: instruments used instrument Respondent Sample information Household questionnaire Household member 1,350 Information about the households ­ ­ Socio-economic characteristics, (450 per ­ Programme participation, State) ­ Profile of benefits, shocks/risk profile etc. Village schedule Local key informants - 45 Information about the village ­ Sarpanch, Patwari ­ Demographic characteristics and Infrastructure facilities, ­ Program coverage, Utilization of funds, ­ Functioning of Panchayat, NGOs etc., Focus Group Discussions in Stakeholders (separately 90 ­ Social mapping, the village for men and women) ­ Political economy and institutional incentives. Case studies 30 ­ Households who successfully used and benefited from programs and those who did not, ­ Major shocks faced and coping mechanisms, ­ Seasonal migration and household/community impacts, ­ Detailed situation of specific vulnerable groups, ­ Dynamics of poverty and poverty traps. Interview with BDOs 18 Seek information on implementation of the SSN schemes, its Interview with NGOs 9 success/failure and reasons. 441 Suitablemultiplierswereused;themultiplierswereconstructedbasedonthepopulationsizesusedfordrawingsamplesateachstage. 442 Incaseofhouseholdsaccessingmorethanoneprogram,thehouseholdisselectedbasedonitsparticipationinoneprogram. 443 Incaseofanyshort-fallofsamplesundertheprogramstrata,theremainingsamplesweredrawnfromthegeneralstrata. 22 Social Protection for a Changing India: Volume II from the household questionnaire and is outlined in variable and condition for each index are described the Table A2.4 below. For constructing the durable index, weights are assigned A. Index of household structural social capital to each durable. The normalized weights are derived (score 1 to 16, if anyone in the household is a from the reciprocals of the frequency of possession. member): Thus the most commonly owned durable gets a lower Village Development Committee weight while one with less frequently owned gets NGO- SHG higher weight. Vana Samrakshana Samithi ­SHG The analysis also relied upon indices of several social Finance, credit group (Other than DWCRA/ and institutional factors, including household social NGO SHG) capital, women's trust in institutions, women's civic participation, women's autonomy and decision making Religious group power (all from the household survey). An infra Political parties structure index was also created for the sample Caste Association villages from the village schedule. The nine household- level indices were constructed by aggregating Mahila Mandal the score on each of the component variables Development of Women and Children in Rural (1 if the condition is met, 0 otherwise). The component Areas (DWACRA/NGO SHG) Table A2.4: wealth index: (constructed from household questionnaire) Sl no indicator Score I Durables Cycle 1, if household possess, 0 otherwise Radio/Transistor -do- Fan -do- Cot/Charpai -do- Almirah -do- TV -do- Refrigerator -do- Car -do- Scooter/motorcycle -do- Watch/Clock -do- Furniture (Chair, stool, Table etc.) -do- Telephone (Landline or Mobile) -do- Durable Index Weighted average of above* II Utensils used for cooking Earthen pots 0 Aluminum 0.50 Stainless steel 1.00 Index for Utensils Average III Clothing One set 0 Two sets 0.33 Three sets 0.67 More than three 1.00 Index for Clothing Average Welath Index Average (I,II,III) Annexes 2 Vana Samrakshana Samithi Management Zila Parishad committee State Legislature (MLA) Rythu Mitra/Farmers' association Lok Sabha (MP) Mothers' Committee G. Women's role in household decision making Watershed Committee (score 1 to 14; 1 if women involved in the final Village Education Committee decision): Village Tribal Development Agency Daughter's Marriage Water Users Association Son's Marriage B. Women's Trust in PRIs (score 1 to 4; if women Food related trust): Debt related Ward-members Children's education ­ Boy Village vice-sarpanch Children's education ­ Girl Sarpanch Care of elderly people Village Assistant (Patwari) Health Care No. of children C. Women's Trust in officials (score 1 to 8; if Family Planning women trust): Spacing between births Teachers Application for ration card Extension officers Participation of HH members in public works ANM Participation in SHG Banks Post office H. Women's control over assets (score 1 to 4; 1 if Aaganwadi teacher women owns): Aaganwadi helper Agricultural Land Police House/plots Gold & Silver D. Women's Trust in community groups (score 1 Durables to 5; if women trust): Mother Committee I. Women's autonomy (score 1 to 9; 1 if women Village Education Committee can go to the first 8 events/places without permission; 1 if allowed to set money aside Parents Teachers Committee for self use; 1 if purchased the last variable): Any Other important village committee Can go alone without permission E. Women's participation in meetings (score 1 to Marketing 6; score of 1 if women participated in the first Visiting friends three meetings; 2 if women participated and Visiting relatives spoke or raised any issue): Cinema, local entertainment, etc. Gram Panchayat Local health center/Doctor Water users association Outside village for Work Village Education Committee Community Center/Park in the village F. Women's participation in elections (score 1 to Community functions 5; 1 if women participated): Allowed set aside some money for self use Gram Panchayat Did you purchase clothes for yourself on your own Block elections during the last 12 months? 2 Social Protection for a Changing India: Volume II annex 3 Cumulative Coverage Across Major Programs it is useful to look at the cumulative coverage at looking at the distributional incidence of multi- household level of major sp programs by examining program coverage, over one fifth of the poorest (and evidence on households accessing more than one st) households access no sp program at all. at the program. This section looks briefly at access to major same time, the situation is for the most part remarkable SP programs in a cumulative sense.444 The results are mainly for its uniformity across wealth levels and presented by wealth, location and social category in social category, though as would be expected rural Figure A3.1. households are significantly more likely to access more than one program. This can be seen in Figure A3.1. the proportion of households benefiting from The most notable feature is how high up the distribution more than one significant sp program is fairly low similar patterns of coverage persist, and how the poorest nationally, though significant in some states. Around are not specially better off in coverage terms than those 23 percent of all households accessed no program at in the middle and even upper reaches of the distribution. all, while the large majority of households (around On a social category basis, the story is somewhat more 58 percent) accessed only one program. Above that, positive, with SC and ST having notably higher coverage in only Orissa, UP and AP of the major states had more more than one program. However, particularly when one than 20 percent of households accessing more than considers some of the results on average benefit levels one program. Over all states, the share of households among SC/ST, the overall picture is one of considerable accessing more than two programs (generally PDS plus public spending spread rather thinly across a large share one other) is insignificant. of the population. Figure A3.1: Number of SP programs per household, by wealth and caste Number of SP programs per HH by wealth Number of SP programs per HH by caste 70 70 60 60 % of HH in quintile 50 50 % of HH in quintile Poorest 40 Q2 40 OBC 30 Q3 30 SC Q4 ST 20 20 Richest Other 10 10 0 0 0 1 2 3 4 0 1 2 3 4 progs progs progs progs Source: Ajwad 2006 based on 2005 IHDS data. 444 TheprogramsincludedinthisanalysisarebenefitsfromPDS,scholarships,SGSY,NOAPSandothersocialpensions,maternitybenefits,Annapurna,andhousingprograms. Annexes 2 annex 4 Unbundling SP Service Delivery Activities 445 Chapter 7 pointed to the frequent bundling of SP example of an unbundling for SGRY. It presents the"delivery service delivery activities at specific ­ and at time chain" for SGRY, and then seeks to map both the de jure inappropriate ­ levels of the system, and recommended and de facto assignment of activity responsibilities across that all programs undertake an "unbundling exercise" levels of the system. Just such an unbundling exercise which could allow for more considered assignment of was done for MGNREG in its operational guidelines, activities in the service delivery chain. The following is an which offer a good model of both process and outcome. Table A4.1: De-jure Functions and Activities matrix Function Activity Responsibility village Panchayat District Central Sabha gram gram Block State Policy Design/Standards Rules of implementation Targeting Budgeting Standards Planning Activity Prioritization/action plan Activity selection Asset Creation Human Capital Skill development Social Capital Information dissemination Physical Capital Public Works Operation Beneficiary Selection Identification of beneficiaries Awareness Raising Recurring Provision of wages/food grains Supervision & quality control 445 ThisannexisdrawnentirelyfromAiyarandSamji(2006). 2 Social Protection for a Changing India: Volume II Function Activity Responsibility village Panchayat District Central Sabha gram gram Block State Personnel Hiring/firing Maintenance Accounting & financial management Repairs Monitoring And Evaluation Assets Record of assets Physical verification of assets created Audits Financial audit Social audit De-facto Functions and Activities Matrix Policy Design/Standards Rules of implementation Targeting Budgeting Standards Planning Activity Prioritization/action plan Activity selection Asset Creation Human Capital Skill development Social Capital Information dissemination Physical Capital Public Works Operation Beneficiary Selection Identification of beneficiaries Awareness Raising Recurring Provision of wages/food grains Supervision & quality control Personnel Hiring/firing Maintenance Accounting & financial management Repairs Monitoring and Evaluation Assets Record of assets Physical verification of assets created Audits Financial audit Social audit Annexes 2 Table A4.2: indicative format for activity mapping (ministry of Panchayati Raj) Panchayati Raj gS/gP Stg. Com. institutions Central govt. inermediate CBos, etc. State got. District village DPCs ulBs Broad function Specific activity (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (A) Functions Framing scheme Objective, design, standards, etc. Planning Building database Preparation of macro plans Preparation of micro plans Approval of micro plans Consolidation of plans Asset creation and Identification of project/land/site operation Construction Operation & maintenance User charges Identification of beneficiaries Procurement/distribution of asses Monitoring And Reporting evaluation Display of information Social audit Independent evaluation IEC (B) Functionaries Oversight over Selection/appointment each category of Training/capacity building functionaries Payment of salary etc Disciplinary control Attendance monitoring Performance evaluation (C) Funds Financial approval, Untied/flexible funds accounting & audig Financial approval Authorization to release Reporting on expenditure Expenditure review Maintenance of accounts Quick audit Source: Ministry of Panchayati Raj (Advisory to states delineating the role and responsibilities of PRIs in CSSs/ACAs, Annexure III, 19th January, 2009). 2 Social Protection for a Changing India: Volume II annex 5 Methodology for Comparing BPL "Poor" and NSS "Poor" at the Household Level This technical annex describes the methodology In order to examine the nature of the targeting errors, adopted by Jalan and Murgai (2008) for the analysis i.e., where in the consumption distribution these described in Chapter 8. Based on the distribution of errors manifest themselves, we follow the approach BPL scores within each state, we defined a state-specific developed by Skoufias and Coady (2002) to display the BPL score cut-off such that the number of BPL poor (i.e., targeting errors graphically. We construct a variable people with BPL scores below the cut-off ) is exactly that takes a value one when households that are equivalent to the number of persons living below the classified as "poor" and "nonpoor" according to poverty line in that state as estimated by the Planning consumption are classified incorrectly as "nonpoor" Commission for the year 1999-2000 and 2004-2005. and "poor", respectively, according to the BPL indicator. Undercoverage ­ or misclassification - was defined as Otherwise this variable take the value zero. Using the percentage of the actual or expenditure-based poor non-parametric methods described by Skoufias and who are incorrectly classified as BPL non-poor. In this Coady, we plot the mean of this variable against the context, at the state-level, the extent of under-coverage log of reported per capita expenditures normalized is identical to leakage ­ the percentage of BPL poor who by the poverty line. The value on the y-axis is the are actually (expenditure-based) non-poor ­ because "predicted error probability" (PEP). The height of the of the restriction that the total number of BPL poor is curve captures the extent of targeting errors made at the same as the number of expenditure-based poor in a different points in the distribution. The shape captures state. Therefore, whenever we refer to state-level under- where in the distribution these errors are being made. coverage estimates as a measure of targeting accuracy, For example, a bell-shaped curved concentrated it is useful to keep in mind that it implies the same extent around 0 (where expenditures equal the poverty of aggregate state-level leakage, by assumption. line) indicates that most of the misclassifications involve households that are just below and above the A shortcoming of undercoverage and leakage measures poverty line. is that they do not differentiate between exclusion (inclusion) of households who are just below (above) The impact of targeting losses can be summarized the poverty line and households that are far below in a welfare index that places a greater "welfare (above) the poverty line. If most of the targeting errors weight" on income transfers to households that (exclusion and inclusion) were concentrated around are lower income households (Skoufias and the poverty line, the welfare losses suggested by these Coady, 2002). Equation (1) derived from standard measures would be considerably lower. welfare theory defines an index p for program p. Annexes 2 households.446 h is each household's share in the total dy h h budget. h is the weight assigned to household h and is p h = h h dy h h the social valuation of income transfer to household h. yz h is the state-specific poverty line, yh refers to consumption dy h of household h, and is the inequality aversion parameter. where = ( y ) h , = h z h y (1) For example, = 0 implies no aversion to inequality and dy h h all welfare weights take a value of unity, i.e., transfers to all households are viewed equally. When = 1, if household h refers to households who receive transfers and dyh h has half (twice) the consumption of the poverty line, is the level of transfers for household h, so that the then its welfare weight is 2.0 (0.25), and so on. denominator is the total budget to be allocated across Source: Jalan and Murgai (2008). 446 Strictlyspeaking,pisabenefit-costratiobutitisequivalenttoawelfareindexsinceweassumethatthebudgetisfixedacrosssimulations. 20 Social Protection for a Changing India: Volume II References Aakella, K. V. & S. Kidambi (2007a). "Challenging Afridi, F. (2005). "Mid-day Meals in Two States". Economic Corruption with Social Audits". Economic and Political andPoliticalWeekly. Vol. 40(15): 1528-1534. Weekly. Vol. 42(5): 345-347. Afridi, F. (2007). "Child Welfare Programs and Child Aakella, K. V. & S. Kidambi (2007b). "Social Audits in Nutrition: Evidence from a Mandated School Meal Andhra Pradesh: A Process in Evolution". Economic and Program in India". Department of Economics, Syracuse Political Weekly. Vol. 42(47): 18-19. University. Acharya Akash and Kent Ranson (2005) `Health Care Agarwal, B. (1990). "Social Security and the Family: Financing for the Poor: community based Health Insurance Schemes in Gujarat', Economic and Political Coping with Seasonality and Calamity in Rural India". Weekly. Vol. 40 (38) pp. 4141- 4150. JournalofPeasantStudies, Vol. 17(3): 341-412. Action for Food Production (2009). "Infrastructure Ahmed. A. and A. Narain (2010). "India's employment Development and Beyond: exploring the scope for challenge: Creating jobs, helping workers". Oxford sustainable livelihood support under NREGA in Chainpur University Press. Block, Gumla District, Jharkhand". A study commissioned United Nations Development Programme (UNDP) Ahuja, R. (2005). "Health insurance for the poor in with the support of Ministry of Rural Development, India: selected issues". Journal of Insurance and Risk Government of India. management. Vol. 6(7): 89 -114. Adhikari, A. and K. Bhatia (2010). "NREGA Wage Payments: Aiyar and Samji (2006). "Improving the effectiveness of Can We Bank on the Banks?". Economic and Political National Rural Employment Guarantee Act". Economic Weekly. Vol. 45(1): 30-37. andPoliticalWeekly.Vol. 41(4): 320-326. References 21 Ajwad, I. (2006). "Coverage, Incidence and Adequacy of Ban, R., M. Das Gupta and V. Rao (2008). "The Political Safety Net Programs in India". Prepared as a background Economy of Village Sanitation in South India: Capture paper for Social Protection for a Changing India, or Poor Information?". Policy Research Working Paper World Bank. Number 4802, World Bank. Alam, M. (2004). "Ageing, Old Age Income Security Bardhan, P. and D. Mookherjee (2000). "Capture and and Reforms: An exploration of Indian Situation". governance at local and national levels". American Economic and Political Weekly. Vol. 39(33): 3731-3740. Economic Review. Vol. 90(2): 135-139. Papers and August 2004. Proceedings of the One Hundred Twelfth Annual Meeting of the American Economic Association, May Alderman, H. (2002). "Do local officials know something 2000. we don't? Decentralization of targeted transfers in Albania," JournalofPublicEconomics, Vol. 83(3): 375-404, Bardhan, P. and D. Mookherjee (2006). "Pro-poor targeting March, 2002. and accountability of local governments in West Bengal". JournalofDevelopmentEconomics. Vol. 79(2): 303-327. Alkire, S. and S. Seth (2008). "Multidimensional Poverty and BPL Measures in India: A Comparison of Basu, P. and P. Srivastava (2005) `Exploring Possibilities: Methods". Working Paper 15, Oxford Poverty & Human Microfinance and Rural Credit Access for the Poor in Development Initiative. India', EconomicandPoliticalWeekly, 40(17). Ambasta, P., P. S. Vijay Shankar and Mihir Shah (2008). Benabou, R. (2000). "Unequal Societies: Income "Two years of NREGA: The Road Ahead". Economic and Distribution and the Social Contract". AmericanEconomic Political Weekly. Vol. 43(8): 41-50. Review, Vol. 90(1): 96-129. Amis, P. (1997). "Indian Urban Poverty: Where are Benjamin, S. (2000). "Governance, economic settings and the levers for its Effective Alleviation". IDS Bulletin, poverty in Bangalore", Environment and Urbanization, Vol. 28(2): 94-105. Vol. 12(1): 35-56. Angelini, J. and K. Hirose (2004). "Extension of Social Berman, P. and R. Ahuja (forthcoming). "Changing Security Coverage for the Informal Economy in utilization and expenditure patterns of health care in Indonesia". ILO Subregional Office for South-East Asia India", manuscript. and the Pacific Working Paper. December 2004. Besley T., R. Pande, L. Rahman and V. Rao (2005). "The ARC (2006). "National Rural Employment Guarantee Act, Politics of Public Good Provision: Evidence From Indian Government of India, New Delhi". Local Governments". Journal of European Economic Association, Vol. 2(2-3): 416: 426. Badiani, R., S. Dercon, P. Kishnan and K. P. C. Rao (2006). Besley, T., R. Pande and V. Rao (2005). "Participatory "Changes in Living Standards in Villages in India democracy in action: survey evidence from South 1975-2004". CPRC Working Paper 85. Prepared as a India". Journal of European Economic Association. background paper for Social Protection for a Changing Vol. 3(2-5): 648-657. India, World Bank. Bhandari, L. and S. Shresth (2003). "Health of the Poor Badiani, R. S. Dercon and P. Krishnan (2006b). and their Subgroups in Urban Areas". Report prepared "Development Programmes, Caste and Capture in for World Bank by Indicus Analytics. Villages in India 1975-2004", published paper. See http://www.isid.ac.in/~planning/PramilaKrishnan.pdf, Bhatia, B. and J. Drèze (2006). "Employment Guarantee last accessed on February 17, 2009. in Jharkhand: Ground realities". Economic and Political Weekly. Vol. 41(29): 3198-3203. Balachander, J., P. V. Dutta, P. O'Keefe (2009). "Jharkhand: Social Protection". Prepared as a background paper Bhatty, K. (2006). "Employment Guarantee and for Social Protection for a Changing India, South Asia Child Rights". Economic and Political Weekly. region, World Bank. Vol. 41(20): 1965-1967. 22 Social Protection for a Changing India: Volume II Bhide, S. and A. K. Mehta (2004). "Chronic Poverty in (Gujarat)". Ministry of Finance, Government of India and Rural India: issues and findings from panel data". Journal World Bank. December 2007. ofHumanDevelopment, Vol. 5(2): 195-209. Calvo C. and S. Dercon (2005). "Measuring Individual Blue, J. (2005): The Government Primary School Mid- Vulnerability." Economics Series Working Papers 229. day Meals Scheme: An Assessment of Programme University of Oxford. Implementation and Impact in Udaipur District, Sewa Mandir, Udaipur. CBGA (2006). "Draft Report on Implementation of NREGA in Andhra Pradesh, Chattisgarh, Jharkhand and Madhya BRAC (2004)."Stories ofTargeting: Process Documentation Pradesh". New Delhi. of Selecting the Ultra poor for CFPR/TUP Programme". CFPR Working paper Series No. 1. BRAC, Dhaka. CDA (2006). Concurrent Monitoring of National Rural Employment Guarantee Act in Guajarat, Centre for Breman, J. (1985). Of Peasants, Migrants, and Paupers. Development Alternatives, Ahmedabad. Delhi: Oxford University Press. Center for Budget and Governance Accountability Breman, J. (1996). FootlooseLabour:WorkingintheIndian (2006). "Report on implementation of NREGA in Andhra Informal Economy. Cambridge: Cambridge University Pradesh, Chhattisgarh, Jharkhand and Madhya Pradesh". Press. CBGA: New Delhi. Brugere, C. and L. John (2003). "Irrigation deficits and Centre for Institute of Fisheries Education (2009). "A Study farmers' vulnerability in Southern India", Agricultural report on appraisal of NREGA programme in Thane and Systems, Vol. 77 (1): 65­88. Akola districts of Maharashtra". Buckley, R. M., J. Kalarickal, M. Singh (2005). "Strategizing Centre for Management Development (2000). "Quick Slum Improvement in India: A Method to Monitor and Evaluation Study of National Social Assistance Refocus Slum Development Programs". Draft World Programme". Study submitted to the Ministry of Rural Bank note. See http://www.worldbank.org/urban/ Development, Government of India. November 2000. symposium2005/papers/buckley.pdf, last accessed on February 18, 2010. CGAP. 2009. "Banking the Poor via G2P Payments". Focus note no. 58. December 2009. CGAP, Department for CAG (2003a). "Report of the CAG on the Union International Development. December 2009. Government". Chapter 3: Rural Housing. Comptroller and Auditor General of India. Chakraborty, P. (2007). "Implementation of Employment Guarantee: A preliminary Appraisal". Economic and CAG (2003b). "Report of the CAG on the Union Political Weekly. Vol. 42(7): 548-551. Government". Chapter 2: Swarnjayanti Gram Swarozgar Yojana. Comptroller and Auditor General of India. Chathukulum, J. and K. Gireesan. (2006). "NREGA: Absence of Ownership". Economic and Political Weekly. CAG (2008). "Performance Audit of implementation Vol. 41(31): 3338, 3340. of NREGA". Office of the Principal Director of Audit, Economic and Service Ministries, New Delhi. Comptroller Chaudhuri, S. (2000). "Empirical Methods for Assessing and Auditor General. Comptroller and Auditor general. Household Vulnerability to Poverty", mimeo, Department of Economics, Columbia University: New York. CAG (2008). "Report No. 13- Performance Audit of National Programme for Nutritional Support to Chaudhuri, S., J. Jalan, and A. Suryahadi (2002). "Assessing Primary Education (Midday meal scheme)". Comptroller household vulnerability to poverty: A methodology and Auditor General of India, Government of India, and estimates for Indonesia". Department of Economics New Delhi. Discussion Paper No. 0102-52. New York: Columbia University. Cal2Cal (2007). "PDS Technical and Feasibility reports. For the pilot of smart card public distribution system Chen, S. and M. Ravallion (2008). "The developing world at Thane district (Maharashtra) and Anand district is poorer than we thought, but no less successful in the References 2 fight against poverty". Policy Research Working Paper Datt, G. and M. Ravallion (1994). "Transfer Benefits from 4703, World Bank. Public-Works Employment: Evidence for Rural India". EconomicJournal. Vol. 104(427): 1346-69. Chopra, K. (2001). "Wastelands and Common Property Land Resources". Seminar, Vol. 499: 24-31. De, A., C. Noronha and M. Samson (2005). "Towards more benefits from Delhi's midday meal scheme". CORD ­ Christiaensen, L. J., and K. Subbarao (2005). "Towards Collaborative Research and Dissemination, New Delhi. an Understanding of Household Vulnerability in Rural October 2005. Kenya". Journal of African Economies, Vol. 14(4): 520-58. de Haan, A. (1999). "Livelihoods and Poverty: The Role of CIESEN (2009)."In Search of Shelter: Mapping the Effects of Migration ­ A Critical Review of the Migration Literature". Climate Change on Displacement and Migration". CIESIN, JournalofDevelopmentStudies, Vol. 36(2): 1-47. United Nations University, and CARE International. de Neuborg (2002). "Incentives and the Role of CMS Social (2009). Delhi voucher project first assessment Institutions in the Provision of Social Safety Nets". Social report, Centre for Media Studies. Protection Discussion Paper Series No. 0226. Social Coady, D. and R. Harris (2004). "Evaluating Transfer Protection Unit, Human Development Network, World Programs within a General Equilibrium Framework". Bank. September 2002. EconomicJournal, Vol. 114(498): 778­99. Deaton, A. and V. Kozel (2005). TheGreatIndianPoverty Coady, D., G. Margaret and H. John (2004). "Targeting of Debate. Delhi: Macmillan India Ltd. Transfers in Developing Countries: Review of Lessons Deaton, A. and J. Drèze. (2009). "Nutrition in India: and Experience". World Bank, Washington DC. facts and interpretation". EconomicandPoliticalWeekly, Conning, J. and M. Kevane (2002). 'Community Based Vol. 44(7): 42-65. February 2009. Targeting for Safety Nets: A Critical Review", World del Ninno C., G. Vecchi and N. Hussain (2006). "Poverty, Development, Vol. 30(3): 375-394. Risk and Vulnerability in Pakistan". Prepared as a Coudouel, A., S. Marnie, and J. Micklewright (1998). background paper for "Social Protection in Pakistan: "Targeting Social Assistance in a Transition Economy: The Managing Household Risks and Vulnerability". World Mahallas in Uzbekistan". Innocenti Occasional Papers, Bank and Centre for Research in Poverty and Income Economic and Social Policy Series No. 63, UNICEF. Distribution (CRPRID). CSE (2008a). "An Ecological Act: A backgrounder to the del Ninno, C., Subbarao, K. and Milazzo, A. (2009) How to National Rural Employment Guarantee Act (NREGA)". Make Public Works Work: A Review of the Experiences. Natural Resource Management and Livelihood Unit. SP Discussion Paper No. 0905. Social Protection and Centre for Science and Environment. Centre of Science Labour, Washington, D.C.: World Bank. and Environment. del Rosso J. (1999). "School Feeding Programmes: CSE (2008b). "NREGA: Opportunities and challenges". Improving Effectiveness and Increasing the Benefit to Natural Resource Management and Livelihood Unit, Education. A Guide for Programme Managers". Centre for Science and Environment, New Delhi. 2008. Dercon (2002). "Income Risk, Coping Strategies, CSE (2008c). "An Assessment of the Performance of The and Safety Nets". The World Bank Research Observer, National Rural Employment Guarantee Programme in Vol. 17(2):141-166. Terms of its Potential for Creation of Natural Wealth in Dercon, S. (2005). "Vulnerability: a micro perspective", India's Villages". Centre for Science and Environment. unpublished manuscript. Das, M. (2009). "South Asia: Jobs in the time of the crisis". Desai, S., A. Dubey, B. L. Joshi, M. Sen, A. Shariff Unpublished manuscript, World Bank. and R. Vanneman (2010). Human Development in Das, M. (2010). "The Below Poverty Line List: A note on India: Challenges for a society in transition, Oxford Database Administration". Unpublished manuscript. University Press. 2 Social Protection for a Changing India: Volume II Deshingkar, P., and D. Start (2003). "Seasonal Migration Dror (2006). "Health Insurance for the Poor: Myths for Livelihoods in India: Coping, Accumulation and and Realities". Economic and Political Weekly. Exclusion". Working Paper 220, Overseas Development Vol. 41(43 & 44): 4541-4544. Institute, London. Duflo, A. (2005). "Health Shocks and economic Deshpande, R. S. (2002)."Suicide by Farmers in Karnataka". vulnerability in rural India: break the vicious EconomicandPoliticalWeekly, Vol. 37(26): 2601-2610. circle". Centre for Micro Finance Research Working Paper, IFMR. Dev. M, C. Ravi, B. Viswanathan, A. Gulati, and S. Ramachander (2004). Economic Liberalisation, Dutta, B. and Ramaswami, B. (2001), "Targeting and Targeted programmes and Household food security: Efficiency in the Public Distribution System: Case of A case study of India. Markets Trade Institutions Andhra Pradesh and Maharashtra", EconomicandPolitical Division discussion paper 68. International Food Policy Weekly, Vol. 36(18), Pages 1524-1532, May 05 - May 11, Research Institute, Washington, D.C. 2001. Dev, S. M. (2002). "Pro-poor Growth in India: What do Dutta, K. (2008). Study of Labor Market Interventions we know about the Employment Effects of Growth for Disadvantaged Youth in India, report prepared for 1980­2000?". Working Paper 161, Overseas Development the World Bank. Institute, London. Dutta, P. V. (2005). "Accounting for Wage Inequality Dev, S. M., K. Subbarao, S. Galab and C. Ravi (2007). in India". Indian Journal of Labour Economics, "Safety net programmes: Outreach and effectiveness". Vol. 48(2): 273-295, 2005. Prepared as a background paper for Social Dutta, P. V. (2006). "Poverty, inequality and vulnerability Protection for a Changing India, South Asia region, in India". Prepared as a background paper for Social World Bank. Protection for a Changing India, World Bank. Drèze J. and A. K. Sen (1989). HungerandPublicAction. Dutta, P. V. (2008). "The Performance of Social Pensions Oxford University Press. in India: The Case of Rajasthan". Social Protection Drèze, J., P. Lanjouw and N. Stern (1992). "Economic Discussion paper no. 0834, World Bank. Mobility and Agricultural Labour in Rural India: A Case Elbers, C., T. Fujii, P. Lanjouw, B. Ozler and W. Yin (2007). Study" IndianEconomicReview, Vol. 27 (special number "Poverty alleviation through geographic targeting: in memory of Sukhamoy Chakravarty): 25-54. How much does disaggregation help?". Journal of Drèze, J. and P. V. Srinivasan (1997). "Widowhood DevelopmentEconomics, Vol. 83(1):198-213. and Poverty in Rural India: Some Inferences from G. B. Pant Social Science Institute (2007). "NREGA in Household Survey data", JournalofDevelopmentStudies, Orissa: Ten Loopholes and the Silver Lining". Interim Vol.54(2): 217-234. Survey Report, 21 October. Drèze, J. and A. Sen (2002). India Development and Gaiha, R. (1988)."Income Mobility in Rural India". Economic Participation.Oxford University Press. Development and Cultural Change, Vol. 36(2): 279-302. Drèze, J and C. Oldiges (2007). "Commendable act". January 1988. Frontline. Vol. 24(14). Jul. 14-27, 2007. Gaiha, R. and A. B. Deolalikar (1993). "Persistent, Drèze, J and C. Oldiges (2009). "Work in progress". Expected and Innate Poverty: Estimates for Semi-Arid Frontline. Vol. 26(4). Feb. 14-27, 2009. Rural South India". Cambridge Journal of Economics, Vol. 17(4): 409-421. Drèze, J and R. Khera (2008). "Banned but still there". Frontline. Vol 26(1). 3-16 January 2008. Gaiha, R. and Imai, K. (2004). "Vulnerability, Shocks and Persistence of Poverty: Estimates for Semi- Drèze, J. and R. Khera (2008), The battle for employment Arid Rural South India". Oxford Development Studies, guarantee, Frontline, Vol. 26 (1). Jan. 03-16, 2009. Vol. 32(2): 261-281. References 2 Gaiha, R., S. Shankar, R. Jha (2010). "Targeting Accuracy Alleviation, Government of India. Oxford University of the NREG: Evidence from Rajasthan, Andhra Pradesh Press: New Delhi. and Maharashtra." Australia South Asia Research Centre Government of India (2010). "Economic Survey". Working Paper 2010/03, revised January 2010. Goyal, A. (2005)."Making the Politician and the Bureaucrat Galasso, E. and M. Ravallion (2003). "Social protection in Deliver: Employment Guarantee in India". Indira Gandhi a crisis - Argentina's Plan Jefes y Jefas," Policy Research Institute of Development Research. Working Paper Series 3165, World Bank. Grosh, M., C. Del Ninno, E. Tesliuc, and A. Ouerghi Galasso, E. and M. Ravallion (2005). "Decentralized (2008). "For Protection and Promotion: The Design and targeting of an antipoverty program". JournalofPublic Implementation of Effective Safety Nets". Washington Economics,Vol. 89(4):705-727. DC: World Bank. Gangopadhyay, S., B. Ramaswami, and W. Wadhwa. Gubbels, J., D. Snelbecker and L. Zezulin (2007). "The (2005). "Reducing Subsidies on Household Fuels Kosovo Pension Reform: Achievements and Lessons". in India: How will it affect the Poor?". Energy Policy Social Protection Discussion Paper No. 0707, April 2007. Vol.33 (18). pp 2326­36. World Bank. Garand, D. (2005). Vimo SEWA, India (Case study No. 16). Guhan, S. (1994). "Social Security Options for Developing CGap Working Group on Microinsurance, World Bank, Countries". InternationalLabourReview,Vol. 133(1): 36-53. Washington, DC. Gupta, I., and M. Trivedi (2005). "Social Health Insurance Garg, C. and A. Karan (2008). "Reducing out-of-pocket Redefined: Health for All through Coverage for All". expenditures to reduce poverty: a disaggregated Economic and Political Weekly. Vol. 40(38): 4132-4140 analysis at rural-urban and state level in India". Health PolicyandPlanning. Vol. 24(2):116-128. Haberfeld, Y., R.K. Menaria, B.B. Sahoo and R.N. Vyas (1999). "Seasonal Migration of Rural Labour in India". Gentilini, U. (2005). "Mainstreaming Safety Nets in the PopulationResearchandPolicyReview, Vol. 18(4): 73­89. Social Protection Policy Agenda: A New Vision or the Same Old Perspective?". Manuscript, United Nations Haddad, L., M. T. Ruel, and J. L. Garret. (1999). "Are urban World Food Programme. Mimeographed document. poverty and undernutrition growing? Some newly assembled evidence". Food Consumption and Nutrition Ghate, Prabhu 2006, "Microfinance in India: A State of the Division Discussion Paper No. 63. International Food Sector Report, 2006", Microfinance India, New Delhi, and Policy Research Institute, Washington, DC. "Indian Microfinance: The Challenges of Rapid Growth, Sage, New Delhi. Hammer, J. S., S. Salimah and A. Yamini (2006). "Bottom's Up: To the Role of Panchayati Raj Institutions Gopal, K. S. (2009)."NREGA Social Audit: Myths and Reality". in Health and Health Services". Social Development Economic and Political Weekly. Vol. 44(43): 70-71. Papers, South Asia Series, Paper No. 98, World Bank, Washington, DC. Government of Himachal Pradesh (2005). "Evaluation Study of Beneficiaries under Old Age, National Old Harriss, J., K. P. Kannan, G. Rodgers (1990). UrbanLabour Age and Widow Pension Scheme". Project Report, Market Structure and Job Access in India: A Study of Department of Economics and Statistics, Government Coimbatore. International Institute for Labour Studies, of Himachal Pradesh. ILO, Geneva. Government of India (2005). "Financing and Delivery Heltberg, R. (2007). "Helping South Asia Cope Better of Health Care Services in India". National Commission with Natural Disasters: The Role of Social Protection". on Macroeconomics and Health, Ministry of Health and DevelopmentPolicyReview, Vol. 25(6): 681-698. Family Welfare, New Delhi. Himanshu (2007). " Recent Trends in Poverty and Government of India (2009). "India Urban Poverty Inequality: Some Preliminary Results". Economic & Report 2009". Ministry of Housing and Urban Poverty PoliticalWeekly, Vol.42(06): 497-508. 2 Social Protection for a Changing India: Volume II Himanshu (2009). Cheaper grain's only one part of a Food Indian Institute of Management (2009). "Appraisal of Security Act, Live mint, Fri. Jul 24 2009, see http://www. NREGA in the states of Meghalaya and Sikkim". Rajiv livemint.com/Articles/2009/06/23195210/Cheaper- Gandhi Indian Institute of Management. grain8217s-only-one.html, last accessed on February Indian Institute of Management Bangalore (2008). 18, 2010. "NREGA Surveys in Anantapur, Adilabad, Raichur and Hirway, I. (2003): "Identification of BPL Households for Gulbarga (2007-08)". Study Commissioned by Ministry Poverty Alleviation Programmes". Economic & Political of Rural Development and UNDP. Weekly, Vol.38(45): 4803-38. Indian Institute of Management Calcutta (2009). Hoddinot, J. and A. Quisumbing (2003). "Methods for "National Rural Employment Guarantee Act". Draft Microeconometric Risk and Vulnerability Assessments". report. Social Protection Discussion Paper No. 324, World Bank. Indian Institute of Management Lucknow (2009). Holzmann, R. and S. Jorgenson (2000). "Social Risk "Quick Appraisal of 5 districts under NREGS in Management: A New Conceptual Framework for Social Uttar Pradesh". Centre for Food and Agribusiness Protection and Beyond." Social Protection Discussion Management, Indian Institute of Management Paper Series No. 0006, World Bank. Lucknow. Holzmann, R., D. Robalino and N. Takayama (2009). Indian Institute of Sciences Bangalore (2009). ClosingtheCoverageGap:RoleofSocialPensionsandOther "Environment Services, Vulnerability Reduction and Retirement Income Transfers. World Bank, Washington Natural Resource conservation from NREGA activities". D.C., June 2009. Institute of Development Studies (2008). "Evaluation of Hou, X. and R. Palacios (2010). "Hospitalization of NREGA in Rajasthan". BPL Households in RSBY: Preliminary Evidence", draft Institute of Human Development (2006). "Evaluation paper. and Impact Assessment of National Rural Employment Howes. S. and S. Jha (1992). "Urban bias in Indian public Guarantee Scheme in Bihar". Institute for Human distribution system". Economic and Political Weekly, Development. New Delhi. Vol. 27(19), pp 1022-30. International Institute of Population Sciences (IIPS) and IAMR (2007). All India Report on evaluation of NREGA: A Macro International (2007). "National Family Health survey of twenty districts". Institute of Applied Manpower Survey (NFHS-3) 2005-06". Mumbai: IIPS. Research, Planning Commission of India. New Delhi. International Labor Organization (2004). "India: The IHD (2006): "Evaluation and Impact Assessment of Welfare funds' experience". Technical paper series National Rural Employment Guarantee Scheme in Bihar". 1, June 2004. Institute for Human Development, New Delhi. International Labor Organization (2005a). "Insurance IIT Madras (2009). "Evaluation of National Rural products provided by insurance companies to the Employment Guarantee Act". Rural Technology and disadvantaged groups in India". Working paper, series: Business Incubator, Indian Institute of Technology Special studies, International Labor Organization Madras. 2005. IIT Roorkee (2009). "System and process review and International Labor Organization (2005b). CaseStudyof impact assessment of NREGS in the state of Uttrakhand". YeshashviniHealthInsuranceScheme, New Delhi: ILO. Indian Institute of Technology Roorkee. IPCC (2007). "Climate Change 2007: Impacts, Adaptation India Centre for Research in Rural and Industrial and Vulnerability". Contribution of Working Group II to Development (2009). "Appraisal of impact assessment of the Fourth Assessment Report of the Intergovernmental NREGS in selected districts of Himachal Pradesh, Punjab, Panel on Climate Change Cambridge University Press, Haryana". Cambridge, UK. References 2 Irudaya Rajan, S. (2001). "Social Assistance for Poor for Social Protection for a Changing India, South Asia Elderly: How Effective?". Economic and Political weekly, region, World Bank. Vol. 36(8): 613-617. February 2001. Jayaraman, R. and P. Lanjouw (1999). "The Evolution Irudaya Rajan, S. (2004). "Welfare funds in India: of Poverty in Indian Villages". The World Bank Research an overview". Centre for Development Studies, Observer,Vol. 14(1): 1-30. Thiruvananthapuram, Kerala, India. Jenkins, R. and A.M. Goetz (1999). "Accounts and ISWSD (2006). "Monitoring and Evaluation of National accountability: theoretical implications of the right-to- Rural Employment Guarantee Scheme with Special information movement in India". ThirdWorldQuarterly, Focus on Gender Issues", Indian School of Women's Vol. 20(3): 603­622. Studies Development. Jha, R. R. Gaiha, S. Shankar (2008). "Reviewing the ISWSD (2008). "Impact of National Rural Employment National Rural Employment Guarantee Programme". Guarantee Scheme on the living and working conditions Economic and political weekly. Vol. 43(11): 44-48. of women in rural India", Indian School of Women's Studies Development. Jha, R., S. Bhattacharya, R. Gaiha (2010). "Social Safety Nets and Nutrient Deprivation: An Analysis of the Iversen, V., K. Sen, A. Verschoor and A. Dubey (2009). National Rural Employment Guarantee Program and the "Job recruitment networks and migration to cities Public Distribution System in India." Australia South Asia in India". Journal of Development Studies, Vol. 45(4): Research Centre draft, February 2010. 522-543. Jha, R., S. Bhattacharya, R. Gaiha, S. Shankar (2008). Jacob, A. and R. Varghese (2006). "NREGA Implementation- "Capture of Anti-Poverty Programs: An Analysis of the I: Reasonable Beginning in Palakkad, Kerala". Economic National Rural Employment Guarantee Program in and Political Weekly. Vol. 41 (48): 4943-4945. India". Australia South Asia Research Centre working paper 2008/07. August 2008. Jacob, A. and R. Varghese. 2006."NREGA implementation-I: Reasonable Beginning in Palakkad, Kerala". Economicand Jha, S. and P. V. Srinivasan (2001). "Taking the PDS to the PoliticalWeekly, Vol. 41(51): 4943-45. December, 2006. poor: Directions for further reform",EconomicandPolitical Weekly, Vol. 36(39): 3779-3786. September 2001. Jacoby, H. and Skoufias, E. (1998). "Testing Theories of Consumption Behavior Using Information on Jhabvala, R., R. M. Sudarshan, J. Unni (2003). Informal Aggregate Shocks: Income Seasonality and Rainfall in economy centre stage: new structures of employment. Rural India". AmericanJournalofAgriculturalEconomics, Thousand Oaks, Calif.: Sage Publications. vol. 80(1): 1-14. Jodha, N. S. (1986). "Common Property Resources Jain, J. and M. Shah (2005). "Antyodaya Anna Yojana and and Rural Poor". Economic and Political Weekly, Mid-Day Meals in MP". Economic and Political Weekly. Vol. 21(26): 1169-81. Vol. 40(48): 5076-5088. Jodha, N. S. (1995). "Common Property Resources Jain, S. (2006). "Poverty, Migration & National Rural and Dynamics of Rural Poverty in India's Dry Regions". Employment Guarantee Scheme". See http://www. Unasylva, No. 180, 46(1):23-30. Online at http://www. righttofoodindia.org/data/ega_sahariya_sachin.doc; fao.org/docrep/v3960e/v3960e00.htm last accessed on March 11, 2010. Kannan, K. P (2002). "The welfare fund model of Jain, S. K. (2004). "Identification of the Poor: Flaws in social security for informal sector workers: The Kerala Government Surveys". Economic & Political Weekly, Experience". Working paper 332, Center for Development Vol.39(46): 4981-84. Studies, April 2002. Jalan, J. and R. Murgai (2008). "An Effective `Targeting Kapur, D., P. Mukhopadhyay, A. Subramanian (2008). Shortcut'? An Assessment of the 2002 Below-Poverty "More on Direct Cash Transfers". EconomicandPolitical Line Census Method". Prepared as a background paper Weekly,Vol. 43(47): 85- 87. November 2008. 2 Social Protection for a Changing India: Volume II Kartika, B. and A. Ranjan (2008). "Alternative to Migration". Krishnamurthy, J. (2006). "Employment Guarantee Frontline. Vol. 26(1), 3-16 January 2008. and Crisis Response". Economic and Political Weekly. Vol. 41(9): 789-790. Khan, R. (1992). "Federal India - A design for change". Vikas Publishing House Pvt. Ltd. Kuruvilla, S., Liu, M. and Jacob, P. (2005). "The Yeshaswini health insurance scheme for rural farmers and peasants: Khera (2006)."Employment Guarantee act and migration". Towards health care for all?". Mimeo, School of Industrial The Hindu. July 13, 2006. Relation Cornell University. Khera, R. (2002): `Mid-Day Meals in Rajasthan', TheHindu, Lakdawala Committee Report and NIUA (1993). "Expert November 13. Group on Estimation of Proportion and Number of Khera, R. (2006). "Mid-Day Meals in Primary Schools: Poor". Planning Commission, Government of India, Achievements and Challenges". Economic and Political 1989-1993. Weekly. Vol. 41(46): 4742-4750. Laxmaiah A., K. V. Rameshwar Sarma, D. Hanumantha Khera, R. (2008). "Pati Experience". Frontline. Vol. 26(1). Rao, Ch. Gal Reddy, M. Ravindranath, M. Vishnuvardhan 3-16 January 2008 Rao and K. Vijayaraghavan (1999). "Impact of mid day meal program on educational and nutritional status Khera, R. (2009). "Empowerment Guarantee Act". of school children in Karnataka". Indian Pediatrics. Economic and Political Weekly. Vol. 43(35): 8-10. Vol. 36(12): 1221-1228. Khera, R. (2009). "Right to Food Act: Beyond LIC (2006). "Booklet of Social Security group insurance Cheap Promises". Economic and Political Weekly, schemes". Life Insurance Corporation of India, Pension Vol. 44(29): 40­44. July 2009. and Group Schemes Department, Central office, Mumbai. Khera, R. and A. Vanaik (2008). "Rajasthan way ahead". Frontline. Vol. 26(1). 3-16 January 2008. Ligon, E., and L. Schechter (2003). "Measuring Vulnerability". Economic Journal, Vol. 113(486): Khera, R. and N. Nayak (2008). "What Works Against C95-C102. Women". Frontline. Vol. 26(1). 3-16 January 2008. Loughhead, S., O. Mittal, G. Wood (2001). "Urban Poverty Khera, R. and N. Nayak (2009). "Women Workers and Vulnerability in India: DFID's Experiences from a and Perceptions of the National Rural Employment Social Policy Perspective". Department For International Guarantee Act". Economic and Political Weekly. Vol. Development, London. August, 2001. 44(43): 49-57. Louis, P. (2006). "NREGA Implementation II- Birth Pangs Kochar, A. (1995). "Explaining Household Vulnerability in Bihar". Economic and Political Weekly. Vol. 41(48): to Idiosyncratic Income Shocks". American Economic 4946-4947. Review, Vol. 85(2): 159-64. Madheswaran, S., D. Rajasekhar, G. K. Karanth Kozel, V. and B. Parker (2005). "A Profile and Diagnostic (2005a). "Design and management of Social Security of Poverty in Uttar Pradesh". Chapter in Deaton, A. and benefits for unorganized workers in Karnataka". V. Kozel (eds), The Great Indian Poverty Debate. Institute for Social and Economic Change, Bangalore. New Delhi: Macmillan India Ltd. 2005. Submitted to GTZ/Micro-Insurance Project, New Delhi. Krishna, A. (2004). "Escaping Poverty and Becoming Poor: May 2005. Who Gains, Who Loses, and Why?". WorldDevelopment, Vol. 32(1): 121-136. Madheswaran S., D. Rajasekhar and K. G. Gayathri Devi (2005b). " A comprehensive study on the status of beedi Krishna, A. (2006). "Pathways Out of and Into Poverty in industry in Karnataka". Institute for Social and Economic 36 Villages of Andhra Pradesh, India". WorldDevelopment, Change, Bangalore. Submitted to the Department of Vol. 34(2): 271-288. Labor, Government of Karnataka. References 2 Mander, H. (2009). "Living Rough Surviving City Streets: Mukherjee, N. (2005)."Political Corruption in India's Below A Study of the Homeless Populations in Delhi, Chennai, the Poverty Line (BPL) Exercise: Grassroots' Perspectives Patna and Madurai". Chapter 15 in India Urban Poverty on BPL in Perpetuating Poverty and Social Exclusion & Report 2009. Good Practice in People's Participation and BPL Exercise from Bhalki Village". Unpublished manuscript. Mathur, B., D. Hariprasad and P. Joshi (2005). "Situational Analysis of Mid Day Meal Programme in Rajasthan". Munshi, K. and M. Rosenzweig (2006). "Traditional Paper presented at the National Seminar on Girls' Institutions Meet the Modern World: Caste, Gender and education ­Towards Equality organized by the Department Schooling Choice in a Globalizing Economy". American of Education and UNICEF on 14-15 November 2005. EconomicReview, Vol. 96(4): 1225-1252. Mazdoor Kisan Samiti & Nari Shakti Sangathan (2006). Murgai, R. (2006). "Do Public Funds Reach the Destitute? "Fact finding report on the status of implementation Assessment of Leakages and Targeting of Social Pension of scheme under NREGA in Bengabad block of Giridih Programs in Karnataka". Prepared as a background paper district Jharkhand". September 2006. See http://www. for Social Protection for a Changing India, World Bank. righttofoodindia.org/data/mksnss06giridihnrega.doc Muthiah (2008). "A scene from the south". Frontline. Mehrotra, S. (2008). "NREG two years on: where do we Volume 26 - Issue 1, 3-16 January 2008. go from here? Economic and Political Weekly August 2, 2008. Naik, R. (2005): `Report on Akshara Dasoha Scheme of Karnataka', University of Dharwad. Mehrotra, S. and H. Mander (2009). "How to identify the poor? A proposal". Economic & Political Weekly, Nair, T. S. (1999)."Housing: the Missing Concerns".Economic Vol. 44(19): 37 ­ 44. May 2009. andPoliticalWeekly, Vol. 34(28): 10-16. July 1999. Mehta, A. K and A. Shah (2003). "Chronic Poverty in India: Narayanan, S. (2008). Employment guarantee, Women's Incidence, Causes and Policies". World Development, work and Childcare. Economic and Political Weekly. Vol. 31(3): 491-511, March 2003. March 1, 2008. Menon, S. V. (2008). "Right to Information Act and National Federation of Indian Women (2008). "A study NREGA: Reflections on Rajasthan". Retrieved March 11, on socio - economic empowerment of women under 2010 from http://mpra.ub.uni-muenchen.de/7351. National Rural Employment Guarantee Act (NREGA)". Ministry of Housing and Urban Poverty Alleviation, Nayak, R., N.C. Saxena and J. Farrington (2002). "Reaching Government of India. Oxford University Press: New Delhi. the Poor: The Influence of Policy and Administrative Processes on the Implementation of Government Mor, N. (2005). `Expanding access to financial Poverty Schemes in India". Working Paper 175, Overseas services ­ where do we go from here?'. Center for Development Institute, London. Microfinance Research Working Paper series, Institute for Financial Management and Research, Chennai. NCEUS (2006). "Report on social security for unorganized workers". Report, National commission for enterprises in Morduch, J. (1999). "Between the State and the Market: the unorganized sector. New Delhi. May 2006. Can Informal Insurance Patch the Safety Net?". World BankResearchObserver,Vol.14(92): 187-207. Noponen, H. (1991). "The Dynamics of Work and Survival for the Urban Poor: a Gender Analysis of Panel Data from Morduch, J. (2004). "Consumption Smoothing Across Madras". DevelopmentandChange. Vol. 22(2): 233-260. Space: Testing Theories of Risk-Sharing in the CIRISAT Study Region of South India". in S. Dercon (ed.) Insurance NSSO (2001). "Migration in India, 1999-2000". National againstPoverty,Oxford University Press. Sample Survey Organization Report No. 470, Government of India. MSSRF and WFP (2004). "Atlas of the Sustainability of Food Security in India". M.S. Swaminathan Research NSSO (2005a). "Situational Assessment Survey of Foundation and World Food Programme. Farmers: Indebtedness of farmer households". National 20 Social Protection for a Changing India: Volume II Sample Survey Organization Report 498. Government Poorest Area Campaign Society (2007). "Status of NREGA of India. implementation 2006-07". Second monitoring report of PACS. PACS series. NSSO (2005b). "Household Indebtedness in India as on 30.06.2002". National Sample Survey Organization PRAXIS (1999). "Consultations with The Poor: India Report 501. Government of India. 1999". Prepared as a background paper for the World Development Report 2000-01, World Bank, O'Keefe, P. and R. Palacios (2006). "Evaluating the NCEUS social security proposals in light of international Washington DC. experience". Economic and Political Weekly, Vol. 41(32): PRIA (2006). "National Study: Phase II: Role of Panchayati 3483-3487, August 2006. Raj Institutions in Implementation of NREGA". ORG-MARG (1998). "Evaluation of National Social Participatory research in Asia. Assistance Programme in Selected States". Report PRIA (2007). "National Study: Phase II: Role of Panchayati prepared by the Operations Research Group for Ministry Raj Institutions in Implementation of NREGA". of Rural Areas and Employment, New Delhi. Participatory research in Asia. Pais, J. (2002). "Casualisation of Urban Labour Force: PRIA (2008). "National Study: Phase III: Role of Panchayati Analysis of Recent Trends in Manufacturing". Economic Raj Institutions in Implementation of NREGA". andPoliticalWeekly,Vol. 37(7): 631-652, February 2002. Participatory research in Asia. Pal, S. and R. Palacios (2008). "Understanding Poverty Pritchett (2005). "A Lecture on the Political Economy of among the Elderly in India: Implications for Social Targeted Safety Nets". Social Protection Discussion Paper Pension Policy," IZA Discussion Papers 3431, Institute for Series 0501. Social Protection Unit, Human Development the Study of Labor (IZA). Network, World Bank, Washington DC, January 2005. Pal, S. (2004). "Do Children Act as Old Age Security in Pritchett, L., S. Sumarto, A. Suryahadi (2002). "Targeted Rural India? Evidence from an Analysis of Elderly Living Arrangements", unpublished manuscript. Programs in an Economic Crisis: Empirical Findings from the Experience of Indonesia." SMERU working paper, Pal, S. and Kynch, J. (2000)."Determinants of Occupational Jakarta, October 2002. Change and Mobility in Rural India". AppliedEconomics, Vol. 32(12):1559 ­ 1573, October 2000. Radermacher, R. N. Wig, O. Van Putten-Rademaker, V. Muller, D. M. Dror (2005). "Yeshasvini Trust, Karnataka Panagariya, A. (2008). India:TheEmergingGiant. Oxford India". CGAP Working Group on Microinsurance Good university press, 2008. and Bad Practices, International Labour Organisation, Pankaj, A. (2008)."Processes, institutions and mechanisms Geneva. of Implementation of NREGA: Impact assessment of Bihar Radhakrishna Committee (2009). Report of the and Jharkhand". Institute of Human Development. Committee on Credit Related Issues under SGSY, Planning Commission (2000). "Performance Evaluation Department of Rural Development Ministry of Rural of Sampoorna Grameen Rozgar Yojana." Program Development Government of India. Evaluation Organization, Planning Commission, Radhakrishna, R. and K. Subbarao (1997), with Government of India. S. Indrakant and C. Ravi. `India's Public Distribution Planning Commission (2005)."Performance evaluation of System: A National and International Perspective', World targeted Public Distribution System". Program Evaluation Bank Discussion Paper No. 380, Washington DC: World Organization, Planning Commission, Government of Bank. India. Ram, F., S. K. Mohanty and U. Ram (2009). "Understanding Planning Commission (2007). "Towards faster and more the Distribution of BPL Cards: All-India and Selected inclusive growth: An approach to the Eleventh Five Year States". Economic and Political Weekly. Vol. 44(7): Plan 2007-2012", Government of India. 66-71. References 21 Ramanathan, R. (2005). Urban Poverty Issues in with Vikas Samvad. April 2007. See http://www. India ­ Challenges and Responses. Vikalp Alternatives, righttofoodindia.org/data/mdms_mp.pdf, last accessed Issue: "Urban Poverty in India", December 2005. on March 10, 2010. Ramaswami, B., S. Ravi and S.D. Chopra (2003). "Risk Rogaly, B., J. Biswas, D. Coppard, A. Rafique, K. Rana Management in Agriculture", Discussion Paper 03-08, and A. Sengupta (2001). "Seasonal Migration, Social Indian Statistical Institute, Delhi, June 2003. Change and Migrants Rights, Lessons from West Bengal". Economic and Political Weekly, Vol. 36(49): 4547­59, Ramji, M. (2009). "Financial Inclusion in Gulbarga: Finding December 2001. Usage in Access." Working Paper Series No. 26. Chennai: IFMR Centre for Microfinance. Roth, J., C. Churchill, G. Ramm, and Namerta. (2005). "Microinsurance and microfinance institutions: Evidence Rani Si, A. and N. K. Sharma (2008). "An Empirical Study from India". CGAP Working Group on Micro insurance, of the Mid-Day Meal Programme in Khurda, Orissa". Good and Bad Practices Case Study No. 15 (Geneva, ILO EconomicandPoliticalWeekly. Vol. 43(25): 46-55. Social Finance Programme). Rao, C. H., S. K. Ray and K. Subbarao (1988). Unstable Saxena (2006)."Rural Poverty Reduction through Centrally Agriculture and Drought. New Delhi: Vikas Publishing Sponsored Schemes". Prepared as a background paper House. for Social Protection for a Changing India, World Bank. Rao, U. (1994). "Palamoor Labour: A Study of Migrant Saxena Committee Report (2009). "Identification of BPL Labour in Mahabubnagar District". C. D. Deshmukh Impact households in rural India". Draft report from the Ministry Centre, Council for Social Development, Hyderabad. of Rural development. Available online at http://rural. Ravallion (1999). "Are poorer states worse at targeting nic.in/latest/rpt_bpl_census2009.pdf, last accessed on their poor?". EconomicsLetters. Vol. 65 (3): 373-377. This January 14, 2010. is not as puja said. Scandizzo, P., R. Gaiha, and K. Imai, (2007). "Does the Ravallion, M. (2002). "On the urbanization of poverty". Employment Guarantee Scheme stabilize the incomes JournalofDevelopmentEconomics, Vol. 68(2), pages 435- of households in rural India?", School of Economics 442, August 2002. Discussion Paper Series, Number 0706. Ravallion, M. (2003). "Targeted Transfers in Poor Sen, S. (2008). "Bihar: Social Protection". Manuscript, Countries: Revisiting the Tradeoffs and Policy Options". South Asia region, World Bank. World Bank Policy Research Working paper 3048, World Sen (2010) for a summary of the field studies on Bank, Washington D.C. educational related outcomes of MDM. These are largely Ravallion, M. and S. Chaudhuri (1997). "Risk and small-scale studies with the estimates of impact typically Insurance in Village India: Comment". Econometrica, based on the perceptions (typically of parents and/or Vol. 65(1): 171-184. teachers) or on school-level administrative information (such as registers, etc.) rather than survey-based data. Ravallion, M., S. Chen and P. Sangraula. 2007. "New As a result, generalizations are difficult and point to the Evidence on the Urbanization of Global Poverty". importance of rigorous assessments of performance Population and Development Review, Vol. 33(4): and impact. 667-702. Shah, A. and A. K. Mehta (2008) ."Experience of the Ravi, S. and M. Engler (2009). "Workfare in Low Income Maharashtra Employment Guarantee Scheme: are there Countries: An Effective Way to Fight Poverty? The Case lessons for NREGS?". Chronic Poverty research center. of India's NREGS", manuscript. Working paper 118. September 2008. Robinson, N. (2007). "Madhya Pradesh: A Report on the Shah, A., S. Kumar Nayak and B. Das with Joshi, H. (2005). Implementation of the Mid-Day Meal Scheme in Four "Remoteness and Chronic Poverty in a Forest Region in Districts". Jawaharlal Nehru University in association Southern Orissa: A Tale of Entitlement Failure and State's 22 Social Protection for a Changing India: Volume II Apathy". Paper presented at CPRC-IIPA seminar on Swaminathan, M. (1995). "Aspects of Urban `Chronic Poverty: Emerging Policy Options and Issues'. Poverty in Bombay". Environment and Urbanization, Vol. 7(1): 133-144. Shah, M. (2007). "Employment Guarantee, Civil Society and Indian Democracy". Economic and Political Weekly. Swaminathan, M. and N. Misra (2001): Errors in Vol. 42(45): 43-51. Targeting­Public Distribution of Food in a Maharashtra Village (1995-2000)". Economic & Political Weekly, Shah, P. J. and C. Braun-Munzinger (2006). Education Vol. 36(26), June 30 - July 06 2001. vouchers: Global Experience and India's Promise, Policy review No. 1, Centre for Civil Society, New Delhi. Swarup (2009). "Marketing of the Rashtriya Swasthya Bima Yojana". Unpublished manuscript. Sharma et al. (2009), Evaluating performance of National Rural Employment Guarantee Act, NCAER-PIF Tesliuc, E. and K. Lindert (2002). "Vulnerability: A study, National Council of Applied Economic Research, quantitative and qualitative assessment". Guatemala New Delhi. Poverty Assessment Program, World Bank, Washington, DC. Siddhartha and A. Vanik (2008). "CAG Report on NREGA: Fact and Fiction". Economic and Political Weekly. Townsend, R. M. (1994). "Risk and Insurance in Village Vol. 43(25): 39-45. India," Econometrica, Vol.62(3): 539-591. Sinha, F. (2003). "Understanding and assessing Umali-Deininger, D.L. and K. Deininger (2006). "Targeted poverty: multi-dimensional assessment versus Public Distribution System: Review of Performance". `standard' poverty lines". Paper presented at the Prepared as a background paper for Social Protection EDIAIS conference, University of Manchester, UK, 24- for a Changing India, World Bank. 25 November 2003. Unni, J. and G. Raveendran (2007)."Growth of Employment Sivakumar, S. K. (2006). "Walking with a purpose". (1993-94 to 2004-05): Illusion of Inclusiveness?". Economic Frontline. Vol. 23(9). May 2006. andPoliticalWeekly, Vol. 42(3): 196-199. January 2007. Sjoblom, D. and J. Farrington (2008). "The Indian National Vanaik, A. (2008). "Accounts of Corruption". Frontline. Rural Employment Guarantee Act: Will it reduce poverty Vol. 26(1). 3-16 January 2008. and boost the economy". Project briefing. Overseas Vanaik, A. (2008). "NREGA and the Death of Tapas Soren". Development Institute. January 2008. Economic and Political Weekly. Vol. 43(30): 8-10. Spagnolo, G. (1999). "Social relations and cooperation Vanik, A. and Siddhartha (2008). "Bank Payments: End of in organizations". Journal of Economic Behavior & Corruption in NREGA?". Economic and Political Weekly. Organization, Vol. 38(1): 1-25 January 1999. Vol. 43(17): 33-39. Sun, C. (2010). Patterns of enrolment in RSBY: Preliminary Vashisht, P., C. Ravi and K. Subbarao (2009). "Evaluation Evidence from the MIS", manuscript. of food coupon program in rural Bihar: An analysis of baseline and end line surveys", manuscript. Sundaram, K. (2003). "On Identification of Households Below Poverty Line in BPL Census 2002: Some Comments Vijay Shankar, P. S., R. Rao, N. Banerji, M. Shah on the Proposed Methodology", EconomicandPolitical (2006). "Government's Schedule of Rates working Weekly, Vol. 38(9): 896-901. March 2003. against Rural labor". Economic and political weekly. Vol. 41(17): 1616-1619. Supriti, S., M. Barnhardt and R. Ramanathan (2002). "Urban Poverty Alleviation in India: A General Assessment Wagstaff, A. (2007). "Social health insurance reexamined," and A Particular Perspective". Ramanathan Foundation, Policy Research Working Paper Series 4111, World Bank. Bangalore, India. Walker, T. S. and G. R. James (1990). VillageandHousehold Swami, K. (2008). "All is not well with the Wells". Frontline. Economies in India's Semi-air Tropics. Baltimore: Johns Vol. 26(1). 3-16 January 2008 Hopkins University Press. References 2 Weigand and Grosh (2008). "Levels and Patterns of Safety World Bank (2006c). "World Development Report 2006: Net Spending in Developing and Transition Countries". Equity and Development". World Bank, Washington DC. Social protection discussion paper 0817, World Bank, World Bank (2006d). "Inclusive Growth and Service June 2008. Delivery: Building on India's Success". Development World Bank (1999). "India Food grain Marketing Policies: Policy Review, South Asia Region. Reforming to Meet Food Security Needs". Volume 1, World Bank (2007a). "India 2025: Inputs for an Urban Main report. Report No. 18329-IN, Rural Development Strategy". Presentation to the Planning Commission. Sector unit. South Asia region, World Bank, August 17, World Bank. 1999. World Bank (2007b). "Rajasthan, India: An Assessment World Bank (2001), "India: The Challenge of Old Age of the Mid-Day Meal Scheme in Chittorgarh District". Income Security", World Bank, Washington DC. Social accountability series, South Asia Sustainable World Bank (2004). "World Development Report 2004: Development Department. Making Services Work for Poor People". World Bank, World Bank (2008a). "India: Accelerating Growth and Washington DC. Development in the Lagging Regions of India". Report World Bank (2005a). Bihar, Towards a Development No. 41101-IN, South Asia Region. Strategy. A World Bank Report, World Bank, World Bank (2008b). "Moving out of Poverty: The Promise Washington DC. of Empowerment and Democracy in India". World Bank: World Bank (2005b). "World Development Report 2005: Palgrave Macmillan. A Better Investment Climate for Everyone". World Bank, World Bank (2009a). "People with Disabilities in India: Washington DC. From Commitments to Outcomes". South Asia Region. World Bank (2005c). "India: Unlocking Opportunities for World Bank (2009b). "China: From Poor Areas to Poor Forest-Dependent People in India". Report No. 34481-IN, People, China's Evolving Poverty Reduction Agenda". South Asia region. East Asia and Pacific Region. World Bank (2006a). "Overcoming Drought: Adaptation World Bank (2011). "Perspectives on Poverty in India: Strategies for Andhra Pradesh". World Bank, Stylized Facts from Survey Data". South Asia Region. Washington DC. Yamini Aiyar and Salimah Samji (2006). "Towards World Bank (2006b). IndiaRuralGovernmentsandService Ensuring the Success of the National Rural Employment Delivery. Oxford University Press. Guarantee Act". Economic and Political Weekly. Vol. 41(4): 320-326. 2 Social Protection for a Changing India: Volume II India's economic growth and the expansion in public spending on the social sector have created new possibilities for its social protection system. While India spends over 2 percent of GDP on core safety net programs, the reduction in poverty and improvement in livelihoods of the vulnerable have not reached full potential. This report represents a first comprehensive review of the performance of India's key anti poverty and social protection programs, including the Public Distribution System, Mahatma Gandhi National Rural Employment Guarantee scheme (MGNREG), Indira Awaas Yojana (IAY), Swarnajayanti Gram Swarozgar Yojana (SGSY) and others, as well as programs to expand social security coverage to the unorganized sector such as the Indira Gandhi National Old Age Pension Scheme and the growing Rashtriya Swasthya Bima Yojana (RSBY) health insurance program. The review draws on new empirical analysis of primary and secondary data sources, including several rounds of the National Sample Survey data, a special Human Development Profile of India survey, dedicated state-level household surveys on associated themes and a rich body of analysis on program performance and impact by national researchers and Government agencies. The report concludes with suggestions for future directions in reform to help India get the most from its social protection system as it enters the second decade of the 21st century. Volume I highlights the main conclusions and recommendations, while Volume II contains the full report with all analyses and findings in detail. human Development unit South Asia