Poverty and Social Assistance in the TCc POVERTY AND SOCIAL ASSISTANCE IN THE TCc February 2019 This study is part of the Economic Analysis Programme for Growth and Sustainable Development which is funded through the European Union’s aid regulation “Council Regulation (EC) No. 389/2006” of 27 February 2006. The opinions expressed in this study do not reflect any official opinion by the European Commission and the World Bank’s Board of Executive Directors, nor do they in any way constitute recognition of boundaries or territories. l i Poverty and Social Assistance in the TCc © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 +1 202-473-1000 www.worldbank.org Disclaimer The contents of this document are the sole responsibility of its author and should in no way be taken to reflect the official views of the World Bank and its Board of Directors, the governments they represent or the European Commission. The Republic of Cyprus covers the entire territory of the island of Cyprus. The term ‘Turkish Cypriot community’ refers, solely for the purposes of this study, to the areas in which the Administration of the Republic of Cyprus does not exercise effective control. If reference is made in the report to any ‘ministries’, ‘departments’, ‘services’, ‘bodies’, ‘organizations’, ‘institutions’, and ‘autho- rities’ in the areas not under the effective control of the Administration of the Republic of Cyprus, or if respective acronyms or abbreviations are used, this is done to allow a clear factual understanding of the administrative structures in the Turkish Cypriot community without intention to support any claims based on international law. Similarly, comparisons between the areas where the Administration of the Republic of Cyprus exercises effective control, and those areas where it does not, are factual only. ii l Poverty and Social Assistance in the TCc ACKNOWLEDGEMENTS This Report was prepared by a team led by Michele Zini, Senior Economist. Principle authors of the report are Metin Nebiler, Research Analyst, and Michele Zini. Mertkan Hamit, Consultant, provided invaluable support and comments. The team would like to thank the following colleagues for their advice, guidance, and support: Facundo Cuevas, Senior Economist (peer-review) Aylin Isik-Dikmelik, Senior Economist (peer-review), Marcus Heinz, Enrique Aldaz-Carroll, Vincent Belinga, Reena Badiani-Magnusson, and Cem Mete. The team would like to thank the Turkish Cypriot administration (and colleagues from the ’State Planning Organization’, the ‘Ministry of Labor and Social Insurance’ , the ‘Ministry of Education and Culture’, the 'EU Coordination Centre', and the EU Program Support Office in particular) for their availability, comments, and for providing the team with the necessary data. l iii Poverty and Social Assistance in the TCc iv l Poverty and Social Assistance in the TCc TABLE OF CONTENTS Table of Figures........................................................................................................................................... vi ............................................................................................................................... vii EXECUTIVE SUMMARY. Policy Recommendations.................................................................................................................... viii INTRODUCTION............................................................................................................................................ 1 1. DATA AND METHODOLOGY................................................................................................................... 3 1.1 Caveats and Limitations.................................................................................................................6 2. A PROFILE OF POVERTY IN THE TCC....................................................................................................... 7 2.1 Poverty Indicators in the TCc........................................................................................................... 7 2.2 Inequality in the TCc........................................................................................................................ 9 2.3 Non-Monetary Poverty Indicators................................................................................................. 11 2.4 Profiles of poor households in the TCc.......................................................................................... 12 3. AN ASSESSMENT OF SOCIAL ASSISTANCE IN THE TCc.......................................................................... 15 3.1 The Social Assistance System in the TCc........................................................................................ 15 3.2 Overview of Social Assistance Programs in the TCc....................................................................... 16 3.3 Social Assistance Spending............................................................................................................ 19 3.4 The Performance of the Social Assistance System......................................................................... 19 3.5 Profiles of Social Assistance Beneficiaries..................................................................................... 23 .................................................................................... 27 CONCLUSIONS AND POLICY RECOMMENDATIONS. Main findings....................................................................................................................................... 27 Policy Recommendations.................................................................................................................... 28 ANNEX ..................................................................................................................................................... 30 l v Poverty and Social Assistance in the TCc TABLE OF FIGURES .......................................................................................... 7 Figure 1- Poverty rates in the TCc and EU, 2015. Figure 2- Poverty Gap in the TCc and EU, 2015............................................................................................ 8 Figure 3- Poverty rate by province (left) and geographical distribution of the poor (right).........................8 Figure 4- Gini Coefficient – TCc vs. EU28, 2015.......................................................................................... 10 Figure 5- Share of total income by quintile in the TCc, 2015...................................................................... 10 Figure 6- Human Development Indicators: TCc vs. Selected Countries, 2015............................................ 11 ....................................................... 12 Figure 7- Household Size (Left) and Dependency Rates (Right), 2015. Figure 8- Gender (left) and age (right) of the household head, 2015........................................................ 12 Figure 9- Education level of the household head, 2015............................................................................. 13 .......................................................................................... 13 Figure 10- School Enrollment in the TCc, 2015. ............................................................... 14 Figure 11- Sector of employment of the household head, 2015. Figure 12- Labor Force status (left) and employment status (right) of the household head, 2015............14 Figure 13- Consumption expenditure of households, 2015....................................................................... 15 Figure 14- Housing conditions (left) and ownership of assets (right) in the TCc, 2015.............................. 15 Figure 15- The Benefit Structure of the “Poor in Need” Program.............................................................. 17 Figure 16- Social Assistance Expenditure in real and nominal terms, 2014 – 2017 (Million TL)................ 19 Figure 17- Social Assistance Expenditure (% of GDP), 2017....................................................................... 19 Figure 18- Coverage: TCc vs Selected Countries (%) - All Social Assistance, by quintile............................. 20 Figure 19- Coverage of Social Assistance Programs in the TCc, by quintile, 2015...................................... 20 ................. 21 Figure 20- Beneficiary Incidence in Selected Countries (%) - All Social Assistance, by quintile. Figure 21- Benefit Incidence in Selected Countries (%) - All Social Assistance, by quintile........................ 21 Figure 22- Targeting of Social Assistance Programs (%), by quintile, 2015, Beneficiaries (left) and Program Resources (Right)................................................................................................. 22 Figure 23- Adequacy in Selected Countries (%) - All Social Assistance, by quintile....................................23 Figure 24- Adequacy rate of SA programs (%) - by quintile, 2015.............................................................. 23 Figure 25- Beneficiaries of Social Assistance Programs, Household size (left) and dependency ratio (right)..................................................................................................... 24 Figure 26- Beneficiaries of Social Assistance Programs, Gender (left) and Age (right) of Household Head. ........................................................................................... 25 Figure 27- Beneficiaries of Social Assistance Programs - Employment Status of Household Head............26 vi l Poverty and Social Assistance in the TCc EXECUTIVE SUMMRARY member or a student in tertiary education. In I addition, the TCc administration provides free n the TCc, relative poverty - defined as healthcare services for households already per EUROSTAT’s methodology as the receiving SA benefits. proportion of people with an equivalized The total expenditure on social assistance disposable income below 60 percent of the has been increasing both in nominal and real median equivalized disposable income - was terms since 2014. However, as a percentage of 22.2 percent in 2015. This figure is higher than the EU average of 17 percent. While the GDP, social assistance expenditure remains most of the poor lived in the two biggest low and below international standards, having provinces in the TCc (Gazimağusa∕Famagusta dropped from 1.24 percent of GDP in 2014 to and Lefkoşa∕Nicosia), the poverty rate was 1.05 in 2017. significantly higher in the Iskele/Trikomo The TCc’s SA system had both some strengths region. Income inequality, at 34 percent, was and some weaknesses. also high in 2015 and above the EU28 average of 31 percent. • The targeting performance fared well even by international standards, with Poverty displayed some key characteristics. around 42 percent of all SA beneficiaries Poor households were larger and had a belonging to the bottom quintile (only 11.3 higher dependency ratio relative to non-poor percent were from the richest quintile), households. The poor were more likely to work in low-skilled private jobs and less likely and around 56 percent of benefits to work in the ‘public’ sector and in education. reaching the poorest quintile. However, They were also more likely to be unemployed social assistance programs covered a and out of the labor force. Among the working small share of the poor (27.1 percent of population, the poor were more likely to work households in the first decile). in temporary or casual jobs, and to be self- • Adequacy levels were high enough to lift employed. Poor households’ heads had lower households out of poverty; in particular, education compared to the non-poor. While cash transfers constituted a large share school enrolment rates were high throughout of the total household budget of the the TCc, there was a significant difference in beneficiaries. Expectedly, however, social tertiary education enrolment between the assistance transfers have a limited impact poor and the non-poor. Only half of the poor on poverty and inequality. owned the houses in which they live in, versus 81 percent of non-poor households. Car and Social assistance beneficiary households computer ownership were much lower for tended to be smaller and have less poor families. dependents than poor households. They Social assistance (SA) programs aim to support were also more likely to be headed by women individuals who are not able to work and or elderly relative to poor households, or to is less focused on reducing poverty. The SA be out of the workforce compared to poor system delivers transfers to those households households’ heads. This suggests that many that include either an elderly or disabled poor households with more members and l vii Poverty and Social Assistance in the TCc higher dependency ratios, or with working resources devoted to social programs may need members as heads (especially men), were to be increased. To allow for greater spending being left out of the social assistance system. on social assistance programs, a rebalancing This is unusual, as in most countries, larger of ‘public’ budgets would be needed to create families (especially families with children) additional fiscal space. are typically favored or targeted by the social The “Poor in Need” could serve as a strong assistance system through minimum income starting point to design a true poverty- schemes or separate ad hoc programs. targeting program in the TCc, to better target The most common types of cash-based poor households, and to better protect them support are the “Poor in Need” program in the fact of shocks and crisis. The program first and foremost, and disability transfers, could be tweaked to include an explicit poverty followed by education scholarships. The only reduction objective. Policymakers may wish sizeable in-kind noncontributory program is to consider simplifying its eligibility criteria, the provision of free health services for the introducing clear equivalence scales, revising poor. There is no conditional cash transfer benefit amounts, and reducing the implicit program in place to incentivize specific disincentives to work that the current set up outcomes (i.e. education, health, employment provides beneficiaries. The program could etc.), or a program with the explicit objective be modified to ensure that poor households of reducing poverty and protecting households with members of working age, which are from shocks. currently excluded from accessing benefits, can apply and receive support if they meet The “Poor in Need” program is an certain income criteria. The program should unconditional cash transfer program aimed also be redesigned to ensure that poor larger at certain categories of people, with some families, especially families with children, income testing elements. The program is could qualify for the benefit, as the absence, designed mainly for elderly people and is in the TCc, of a program with the objective very complex in terms of eligibility criteria. supporting families with children is unusual Whether someone is or isn’t eligible varies relative to most EU countries. Working should in fact according to a long list of categories not automatically exclude individuals from and income criteria. The benefit structure is applying to the program. In other words, the equally complicated and unclear and provides program could be gradually transformed into beneficiaries with a clear disincentive to work. a poverty-targeting tool, moving away from Policy Recommendations categorical eligibility criteria and towards To increase the efficiency of the resources income and assets criteria. devoted to social assistance, and to increase To complement the currently used relative the impact of programs on poverty indicators, poverty measures, an absolute poverty line existing programs would need to be reformed. should be constructed to monitor poverty The design parameters of some programs and identify poor households. An absolute should be reconsidered, moving away from poverty measure would define the minimum categorical targeting and towards an explicit monetary amount that households would poverty reduction objective. This would allow need to meet their basic needs and is essential to achieve greater poverty reduction impacts to designing an efficient social assistance within the current budgetary allocation. system. An absolute poverty measure would However, given the overall low coverage of in fact inform policymakers on the number of the TCc’s social assistance system, if the TC people failing to meet the most basic needs, administration aims to achieve the levels and to quantify the transfers necessary for of developing and developed countries, the them to do so. An absolute poverty line would viii l Poverty and Social Assistance in the TCc also allow to assess whether poor households’ policies for the poor households. However, the incomes have increased through time, and HBS survey is conducted only every 6-7 years, hence whether their situation has improved and this limits the frequency with which deep in absolute terms. It would also introduce a poverty and social assistance analysis can clear and transparent metric by which poverty be carried out. Furthermore, more detailed would be determined. To define and measure household survey is required to assess the internationally comparable absolute poverty, performance of the SA system more accurately. a purchasing power parity (PPP) index is The changes suggested above become even required, which has not been calculated for more important in the face of the recent the TCc until today. sharp depreciation of the Turkish Lira. As Finally, household budget survey should be ‘public’ revenues shrink further, making conducted more frequently, and in a more spending on social programs more efficient detailed manner to allow for evidence- will be paramount. At the same time, based discussions on poverty monitoring designing program which can effectively target and reduction. Poverty measurement in the poor households and support them in the face TCc is an important challenge when designing of crises will be crucial towards weathering l ix Poverty and Social Assistance in the TCc the macroeconomic shocks which the TCc is currently undergoing. x l Poverty and Social Assistance in the TCc INTRODUCTION This report profiles poverty and assesses the performance of the social assistance system in the Turkish Cypriot community (TCc). The aim of this report is to provide information and evidence-based analysis to policy-makers on the determinants of poverty in the TCc and on the effectiveness of the community’s social assistance programs, with the ultimate objective of putting forward recommendations on how to improve policy- alleviating measures. To the best of the World Bank’s team knowledge, the analysis included in this note (and in particular the one related to social assistance programs) has not been carried out in over 13 years. l 1 Poverty and Social Assistance in the TCc The poverty profile developed in this report can provide policymakers with vital information on the characteristics and living conditions of poor households in the TCc. The report presents information on the poverty and inequality rates in the TCc, highlights the individual and household characteristics of the poor, as well as the regional aspects of poverty (Haughton & Khandker, 2009). Notably, neither the poverty trends nor the characteristics of the poor are currently well documented (mainly due to data constraints); this report aims to fill this gap. This report also presents the first assessment of the TCc’s SA system since 2006. The analysis of household survey data, as well as of administrative data, allows to understand the coverage, targeting performance, strengths and weaknesses of the SA system in the TCc, and whether or not the existing range of programs matches the profiles of the poor. Said analysis can enable policymakers to improve the effectiveness of social assistance programs in the TCc. This report consists of three chapters. The first chapter describes the data used and the methodology adopted in the report. The second chapter presents poverty and inequality indicators for the TCc and provides a profile of poor households. The third chapter focuses on the social assistance system in the TCc. After briefly describing such system, the chapter assesses its performance. Some concluding remarks and policy recommendations then bring the note to a close. To the best of the World Bank’s team knowledge, the analysis included in this note (and, in particular, the one related to social assistance programs) has not been carried out in over 10 years. The analysis is based on the most recent data available (2015 for HBS data, 2017 for administrative data); therefore, some of the results presented may differ from the situation in early 2019. However, especially for the social assistance related findings, the analysis presented should broadly apply to the current scenario, as not much has changed in terms of programs over the last 4 years. 2 l Poverty and Social Assistance in the TCc 1. DATA AND METHODOLOGY This report uses a relative poverty line to sketch a profile of poor households in the This report relies on multiple sources of TCc. As per EUROSTAT’s methodology, and data to profile a picture of poverty in the as done throughout the EU, for the purposes TCc and assess the performance of social of this report, 60 percent of the median assistance transfers. The main data source equivalized household income is used as the is the 2015 Household Budget Survey (HBS) poverty threshold (Bradshaw & Mayhew, that is conducted by the ‘State Planning 2010). To determine households’ poverty Organization’ ('SPO') in the TCc. The HBS is status, household income per adult equivalent a TCc representative household survey that is calculated by using the modified OECD scale, includes information on households and as per EUROSTAT methodology1. Poverty is individuals’ characteristics, demographics, defined at the household level, and if a given education, employment, income, assets and households is deemed as poor, every individual consumption. The most recent survey was in it is deemed as poor. conducted by interviewing 2,913 households and 8,679 individuals in a period of twelve The decision to use a relative poverty mea- months from September 2014 to September sure is linked to data limitations and for 2015. The HBS is conducted every seven to comparability purposes. First, using a rela- eight years, mainly due to the prohibitive costs tive poverty line allows to compare the TCc’s of data collection. poverty indicators with EU countries. Second, absolute poverty (defined by the World Bank The report also uses data from the European Upper Middle-Income Countries as living with Statistics Office (Eurostat) to benchmark less than US$ 5.5 2011 PPP per capita per day) the TCc’s poverty indicators with countries cannot be calculated in the TCc since there is across the EU. Eurostat provides statistics at no official Purchasing Power Parity (PPP) in- the country level in the European Union (EU) dex defined for the Turkish Cypriot economy. and enables comparisons between countries Similarly, the “cost of basic needs approach” through standardized indicators. With respect (which defines an absolute poverty threshold to social assistance, data from the World in monetary terms) is not followed by the TCc Bank’s ASPIRE database is utilized to compare authorities. the performance indicators of the TCc’s SA system with selected countries. 1 The EU uses the modified OECD scale (in which the first adult receives a weight of 1.0, every other adult 0.5 and each child below 14 years of age 0.3) l 3 Poverty and Social Assistance in the TCc Box 1 - Measurement of Poverty in the TCc Poverty in the TCc is measured and monitored by SPO using HBS data. The HBS database includes information on individual and household level annual income and household consumption expenditure. SPO prefers income as a welfare indicator rather than consumption expenditure, in line with EUROSTAT’s methodology. Income information is collected at the individual level within households and the derived income aggregate includes all incomes each household member received in twelve months leading to the survey. Household income per adult equivalent is calculated by using the “modified OECD scale”. SPO defines poverty in relative terms (i.e. uses a relative poverty line as per EU methodology). However, SPO defines a household as poor if its equivalized adult income falls below 50 percent of the median equivalized disposable income of the TCc, a more stringent definition that than adopted by EUROSTAT. In 2015, SPO’s relative poverty line corresponded to 1001 TL per month per adult equivalent. Poverty is calculated for urban and rural areas differently. According to this measure, in 2015, the poverty rate in the TCc was 15.4 percent, and the rate was higher in rural areas than in urban areas, measuring 17.2 and 14.0 percent respectively ('State Planning Organization’, 2016). Poverty in the TCc cannot be estimated every year due to lack of data. The HBS is in fact carried out every five to seven years due to budget constraints. This does not allow close monitoring of poverty in the TCc and hampers the evaluation of implemented projects and policies. Moreover, it prevents policymakers from taking informed decisions and implementing evidence-based policies to reduce poverty and inequality. Source: SPO HBS 2015 Note: The official poverty line in the TCc is 50 percent of the median equivalized disposable income. Throughout this report, social assistance analyzed include the following: the “Poor benefits are defined as non-contributory in Need” program, scholarships, disability public transfers that aim to help poor benefit, and free health care provision for the households (The World Bank, 2018). HBS has poor. Pensions, and survivor benefits programs information on all existing social assistance and are classified as social insurance because of social insurance programs, but unfortunately their contributory nature. The “Families of the does not classify them in a consistent manner. Fallen” program, a noncontributory benefit The main shortcoming of the current data extended to the families of those who died collection methodology (and of the current during the 1960-70s conflict, is not considered survey questionnaire) is that information on as a social assistance program as it does not contributory and non-contributory transfers have a poverty mitigating objective and is is merged under the same question, thereby therefore analyzed separately. making it very difficult to separate different The performance of social assistance programs and to discern contributory and programs is assessed by measuring coverage noncontributory measures. Changes to of each programs (and of the system as a HBS’s survey questionnaire and methodology whole), beneficiary and benefit incidence are therefore needed to better assess the (i.e. targeting), as well as adequacy of the performance of the SA system and to measure benefits (Demery, 2000). These indicators the performance of individual programs; provide insights into how effective social the World Bank stands ready to assist in this assistance programs are in aiding the poorest endeavor. segments of the population, and help identify For the purposes of this report, the social gaps and challenges. The definition of each assistance (noncontributory) programs indicator is provided in Box 2. In most of the 4 l Poverty and Social Assistance in the TCc analysis, households are divided into five compared to EU countries as per EUROSTAT’s groups (i.e. “income quintiles”) based on their database. However, for the performance of “pre-transfer” disposable income per adult the social assistance system, the choice of equivalent. comparators is driven by data availability, as the World Bank’s ASPIRE database only covers This report benchmarks the TCc’s poverty a limited number of countries, and data for EU measures and social assistance system countries is not available.2 against the EU, where possible, or against selected countries. Poverty indicators are Box 2 - A Glossary of the Main Performance Indicators of Social Assistance Programs Coverage: Percentage of the population in a quintile/decile benefitting from social assistance pro- grams (includes direct and indirect beneficiaries). Specifically, coverage is calculated as: (number of individuals in the quintile/decile who live in a household where at least one member receives the transfer) / (number of individuals in that quintile/decile). Adequacy: Total transfer amount received by all beneficiaries in a quintile/decile as a share of the to- tal “pre-transfer” welfare of beneficiaries in that quintile/decile. If the generosity is high, the transfer is a large share of the beneficiary household’s consumption. Beneficiary Incidence (Targeting): Percentage of program beneficiaries in a quintile/decile relative to the total number of beneficiaries in the population. Benefit Incidence: Percentage of benefits going to each quintile/decile of the post-transfer welfare distribution relative to total benefits going to the population. Targeting Error: Exclusion error refers to the share of the poor who are not beneficiaries of the pro- gram. Inclusion error refers to the share of non-poor receiving program benefits. Poverty impact: poverty head-count reduction is computed as (poverty headcount pre-transfer - pov- erty headcount post transfer) / (poverty headcount pre-transfer) and the poverty gap reduction is computed as (poverty gap pre-transfer - poverty gap post-transfer) / (poverty gap pre-transfer). Note: Definitions are taken from the World Bank Database Aspire: The Atlas of Social Protection - Indicators of Resilience and Equity. 2 Where applicable, Kazakhstan, Malaysia, Turkey, Slovak Republic, Albania, Croatia, Russian Federation, Chile, Belarus, Kosovo, Poland, Lithuania, Latvia, Romania, Bulgaria, Montenegro, Mexico, Serbia, Bosnia and Herzegovina, Hungary, Fiji, Mauritius and Maldives have been used as compar- ators. For these countries, latest available year is taken into consideration. l 5 Poverty and Social Assistance in the TCc 1.1 Caveats and Limitations allow researchers, and hence policymakers, to more precisely assess the performance of the Perhaps the most important limitation is SA system and increase the effectiveness of that the absolute poverty rate cannot be programs. measured in the TCc. First, and as discussed, the TCc does not have an absolute poverty line When compared to administrative data, the which does not allow researchers to calculate HBS captures seventy percent of the social absolute poverty. Second, the international assistance beneficiaries and only fifty five poverty lines that the World Bank uses to percent of the total benefits. Household monitor poverty also cannot be measured surveys always capture a portion of social since there is no purchasing power parity index assistance transfers and of the total budget calculated for the TCc. The report, therefore, of transfers. This happens for many reasons uses relative poverty measures. While this is (households failing to report benefits a useful metric, and one which is used widely recipiency, sampling errors, etc.). The results within the EU, relative poverty can be seen of any social assistance assessment can more of a measure of inequality as opposed to therefore be biased if there is a systematic an index of deprivation. exclusion of social assistance beneficiaries Although there are only four social assistance from the survey. For instance, if all social programs in the TCc, the HBS does not assistance beneficiaries who are not captured include detailed questions for each different in a survey are from the bottom of the income social assistance transfer. As mentioned distribution, performance indicators would be above, social assistance transfers are reported biased downwards (i.e. the results look worse in the same question with social insurance than what happens in reality). In contrast, it transfers. For instance, noncontributory would be upwardly biased if they are all from disability benefits are reported together with the top of the income distribution. This is an contributory disability benefit, contributory inherent limitation of this type of analysis, sickness benefit, and with the “Poor in Need” and an important caveat to bear in mind: the program. Since it is not possible identify each findings presented in this report are limited to social assistance program separately, the what is reported in the household survey. report refers to noncontributory disability benefits and to the “Poor in Need” transfer In the next section we present information on jointly as “cash transfers” and present poverty and inequality in the TCc and highlight results accordingly. Asking about the four- the individual and household characteristics social main assistance programs in the TCc of the poor, as well as the regional aspects of through separate question in the HBS would poverty. 6 l Poverty and Social Assistance in the TCc 2. A PROFILE OF POVERTY IN THE TCc the average poor household is relative to the reference poverty line – indicates, at almost 2.1 Poverty Indicators in the TCc 27 percent, that the median poor household In the TCc, relative poverty - defined in line in the TCc was significantly below the poverty with EUROSTAT methodology as the propor- threshold. Like the poverty rate, the poverty tion of households with an equivalized dis- gap was higher in rural areas compared to posable income below 60 percent of the me- urban areas. The poverty gap was closer to dian equivalized disposable income, was 22.2 the EU average compared to poverty rate. This percent in 20153. implies that the poor in the TCc were relatively close to the EU28 average (of 25 percent) but The relative poverty rate in the TCc was high higher relative to Malta (17.3 percent), Iceland compared to EU countries. While EU averages (18.5 percent) and the Republic of Cyprus were around 17 percent, poverty rate varied (19.8 percent). from 9.2 percent in Iceland to 25.4 percent in Poverty in the TCc showed differences among Serbia. The TCc ranked among the countries provinces. Based on HBS 2015 data, more with the highest relative poverty incidence, than half of the TCc’s poor households lived after Serbia, Romania, Turkey and Latvia. in two provinces, Famagusta (29.1 percent) When compared to island economies, the and Nicosia (25.4 percent). Although only 13.6 TCc’s relative poverty rate was significantly percent of the poor lived in the Trikomo/Iskele higher than in Iceland (9.2 percent), and Malta province, given the relatively small size of said (16.3 percent)4. Relative poverty in the TCc is province, poverty incidence was high, reach- also higher relative to the Republic of Cyprus ing 37.2 percent6. The Nicosia/Lefkosa prov- (16.2 percent). ince displayed the lowest poverty incidence The poverty gap in the TCc was 26.7 percent (17.2 percent) and accounted for the smallest in 20155, close to the EU28 average. The share of poor households out of total poor poverty gap – a measure of how “distant” households in the TCc. Figure 1 - Poverty rates in the TCc vs. EU and Non-EU countries, 2015 Source: SPO HBS 2015 and EUROSTAT 3 This measure is also referred to as the “at-risk-of-poverty” rate, as per EUROSTAT methodology and definitions, and is a measure of vulnerability. 4 The relative poverty line was 1201TL per month per adult equivalent in 2015. 5 Poverty gap is measured by the difference between median equivalized disposable income of people below the at risk of poverty threshold and at the risk of poverty threshold, expressed as a percentage of the at-risk-of-poverty threshold (cut-off point: 60% of median equivalized income). It is calculated with the following formula: Poverty gap = ((At risk of poverty line – Median equivalized income of the poor)/At risk of poverty line) *100 6 Total population in TCc was 286,257 according to the results of the 2011 population census. Total number of residents was 94,824 in Nicosia, 69,741 in Famagusta, 69,163 in Kyrenia/Girne, 30,037 in Morphou and 22,492 in Trikomo/Iskele province (State Planning Organization, Statistical Yearbook 2016, 2017). l 7 Poverty and Social Assistance in the TCc Figure 2 - Poverty Gap in the TCc vs. EU and Non-EU countries, 2015 Source: SPO HBS 2015 and EUROSTAT Figure 3 – Poverty rate by province (left) and geographical distribution of the poor (right) Source: SPO HBS 2015 and EUROSTAT 8 l Poverty and Social Assistance in the TCc Box 3 – Absolute Poverty in the TCc The TCc does not have an absolute poverty line to measure the fixed standard of living (or monetary amount) that is needed by households to meet their basic needs. As discussed, ‘SPO’ measures poverty using a relative poverty line (defined as having an equivalized disposable income below 50 percent of the median equivalized disposable income). To complement the currently used relative poverty measures, an absolute poverty line should be constructed to monitor poverty and identify poor households. An absolute poverty measure would define the minimum monetary amounts that households would need to meet their basic needs. An absolute poverty measure would in fact inform policymakers on the number of people failing to meet the most basic needs, and to quantify the transfers necessary for them to do so. An absolute poverty line would also allow to assess whether poor households’ incomes have increased through time, and hence whether their situation has improved in absolute terms. It would also introduce a clear and transparent metric by which poverty would be determined. Even though an absolute measure does not link households’ conditions to changes in inequality or takes into account the overall unfair distribution of resources, absolute poverty rates are less sensitive to overall economic changes (as the threshold is not directly linked to overall shifts in the income distribution). When constructing an absolute poverty line, the most common approach is to use cost of basic needs methodology (Ravallion, 2016). The methodology calculates the cost of acquiring enough food for an adequate nutrition (often 2,100 calories per person per day), and the cost of other expenditures such as clothing and shelter. This exercise could be carried out by ‘SPO’ since it monitors monthly prices to calculate inflation. If a basket of basic goods and needs could be constructed, monitoring the cost of the basket can be implemented by the same department. To measure the international absolute poverty rate, a purchasing power parity (PPP) index is required, which has not been calculated for the TCc until today. First, a predicted PPP index can be estimated by using the regression-based formula used for estimating PPPs for countries who did not participate in the PPPs. Second, a detailed price level comparison between Turkey and the TCc can be conducted. Both economies have detailed information on unit prices of consumption items in their household budget surveys. Comparing unit prices of similar items in both surveys can inform us about the general price level between Turkey and in the TCc. This way, PPP for the TCc can be estimated by adjusting Turkey’s PPP with the general price level difference. The World Bank stands ready to assist in defining an absolute poverty line. 2.2 Inequality in the TCc the EU varied significantly across the EU and other countries, from 24 percent in Slovakia, Income inequality in the TCc was high in 2015 to above 40 percent in Turkey. Inequality in and above the EU28 average. According to HBS the TCc was significantly higher than in other 2015 data, the Gini coefficient was 34 percent, small islands such as Malta (28.1 percent) and 3 percentage points above the EU average Iceland (24.7 percent), and comparable to the of 31 percent. It should be noted however, Republic of Cyprus (33.6 percent). as Figure 4 shows, that income inequality in l 9 Poverty and Social Assistance in the TCc Figure 4 – Gini Coefficient – TCc vs. EU and Non-EU countries, 2015 Source: SPO HBS 2015 and EUROSTAT The poorest households in the TCc held only 5, inequality in 2015 was high in the TCc. less than 7 percent of the total TCc income. Households in the bottom quintile received An alternative way to display inequality is only 6.8 percent of the total income, six times by looking at the shares of total income less than the income share received by the top generated by an economy received by different quintile (41.1 percent). households. By this metric, shown in Figure Figure 5 - Share of total income by quintile in the TCc, 2015 Source: SPO HBS 2015 10 l Poverty and Social Assistance in the TCc 2.3 Non-Monetary Poverty around 78 percent in the EU. However, and worryingly, at 12 every 1,000 births, infant Indicators mortality was well above the EU average of Human development indicators in the TCc 3.7. were comparable to those across the EU, Non-monetary poverty indicators, which can except for infant mortality rate. Literacy and help quantify and document the degree of school enrollment rates hovered around 97 poverty in an economy, acquire all the more percent in the TCc, close to the EU-28 figures. importance in the absence of an absolute Life expectancy was the highest among poverty line. However, the available HBS data all comparator countries and even slightly does not allow for a deeper analysis in this higher than the EU average. Basic sanitation respect, and limits it to the above-mentioned and electricity were provided almost to indicators. For instance, constructing a multi- every household in the TCc, much like in dimensional poverty index, or the human any EU country, with the coverage rates for capital index is not currently possible. This these services being essentially 100 percent. shortcoming points once again to the need However, at 56 percent, the TCc was lagging to urgently improve the HBS data collection when it came to access to internet, which was process. Figure 6 - Human Development Indicators: TCc vs. Selected Countries, 2015 Source: SPO HBS 2015 WDI and EUROSTAT l 11 Poverty and Social Assistance in the TCc 2.4 Profiles of poor households in poor households the split was 22.8/77.2. the TCc However, this small difference (0.8 percentage points) was not statistically significant, Expectedly, poor households were on pointing to the fact that the gender of the average larger and had a higher dependency household head was not a significant correlate ratio7. Poor households included on average of poverty. However, poor households seemed almost one more person compared to non- to be headed, on average, by a younger person poor households (3.6 vs. 2.9, respectively). relative to non-poor households. Households classified as poor had also more dependent children compared to non-poor Somewhat unsurprisingly, the education level households. These findings are consistent with of the household head was an important the features of poverty around the world. determinant of poverty. In line with findings Figure 7 – Household Size (Left) and Dependency Rates (Right), 2015 Source: SPO HBS 2015 In terms of gender of the household head, around the world, in 2015, heads of poor poverty seems to comparable between households had on average much lower households headed by men and headed educational attainments compared to non- by women. While 22 percent of non-poor poor household heads. Around 50 percent of households were female headed, and 78 heads of poor households had only primary percent were male-headed households, among education or less (including 7.1 percent non- Figure 8 – Gender (left) and age (right) of the household head, 2015 Source: SPO HBS 2015 7 The child dependency ratio is the ratio of the population aged 0-14 to the population aged 15-64, presented as the number of dependents per 100 persons of working age (15-64). 12 l Poverty and Social Assistance in the TCc Figure 9– Education level of the household head, 2015 Heads of poor households tended to work in low-skilled private sector jobs. Among the poor, employed household heads were mainly working in the construction and services sectors. A striking difference between poor and non-poor household heads concerned the share of employment in ‘public’ sector, measuring at 2.1 percent and 17.6 percent respectively. Given the generosity of the wages and benefits of the TCc’s ‘public’ sector, it is Source: SPO HBS 2015 no surprise then that the sector of occupation literate, 5.4 percent literate without a diploma) was (and likely continues to be) an important while the same share among the non-poor was determinant of poverty in the TCc. only half, 26.2 percent. Similarly, 23.1 percent Heads of poor household were more likely to of non-poor household heads finished tertiary be unemployed or out of the labor force. Out education, but only 4.8 percent of the heads of the whole population, only 0.5 percent of of poor households had done so in 2015. non-poor household heads were unemployed in the TCc, while the ratio was 5.7 percent While net school enrolment rates in primary among poor household heads. Similarly, the and secondary education were high in probability of being out-of-the-labor-force was the TCc, the poor displayed lower school more than double among the heads of poor enrolment rates at each level of education, households relative to non-poor households and particularly at the tertiary level. The (18.2 and 8.5 percent, respectively). Another difference between the enrolment rates of the important difference was the share of retirees poor and non-poor children was very small; between the two groups, 22.4 percent for the around 95 percent of poor children attended non-poor and 7.9 percent for the poor, pointing primary education and 93 percent attended to the fact that some of the poor may have secondary education, versus 98 and 96, been (and probably still are as the situation in respectively, among non-poor children. terms of social programs is unchanged relative to 2015) excluded from the social protection Figure 10 – School Enrollment in the TCc, system, and that not receiving a pension was 2015 an important correlate of poverty. If we restrict our analysis to employed household heads only, although the share of employees was very similar between poor and non-poor households, the heads of poor households tended to have casual jobs and be self-employed. Having a permanent job decreases the chances of falling into poverty. Not surprisingly then, in the TCc, the heads of poor households were less likely to Source: SPO HBS 2015 have permanent jobs compared to non-poor household heads. While 16.5 percent of poor household heads had casual jobs, this share By contrast, the school enrolment rate of the was 3.2 percent for the non-poor household poor in tertiary education was lower than heads. that of the non-poor (62 percent versus 77, The consumption patterns of poor house- respectively). holds, unsurprisingly, differed significant- l 13 Poverty and Social Assistance in the TCc Figure 11 – Sector of employment of the household head, 2015 Source: SPO HBS 2015 ly from that of non-poor households. Poor most striking difference between the poor households spent a little over half of their con- and the non-poor was in asset ownership and sumption expenditure on food and housing access to services, the poor were significantly while this share was only 39.1 percent among less likely to hold assets and access to services non-poor households. Non-poor households compared to non-poor families. Only half of were more likely to spend more on education the poor owned the houses in which they live in, while this share was 81 percent among and leisure activities while the poor had very the non-poor households. Similarly, car and little income left for those activities. computer ownership were much lower for Not surprisingly, poor households held less poor families, 66 percent of the poor owned assets compared to non-poor households. a car and 50 percent of the poor owned a Both poor and non-poor families had decent computer. Access to internet was also much sanitation and shower conditions in the TCc. lower among poor households. Only 35 The most important difference concerns access percent of the poor had access to internet in to hot water; the poor were significantly less their homes while this share was 62 percent likely to have hot water in their houses. The among the non-poor. Figure 12 - Labor Force status (left) and employment status (right) of the household head, 2015 Source: SPO HBS 2015 14 l Poverty and Social Assistance in the TCc Figure 13 - Consumption expenditure of households, 2015 Source: SPO HBS 2015 Figure 14 - Housing conditions (left) and ownership of assets (right) in the TCc, 2015 Source: SPO HBS 2015 3. AN ASSESSMENT OF SOCIAL 3.1 The Social Assistance System in ASSISTANCE IN THE TCc the TCc While the previous section presented poverty Social assistance programs in the TCc mostly indicators and described the profile of poor aim to support individuals who are not able households in the TCc, this section focuses on to work. The SA system delivers transfers to the social assistance system in the TCc. First, a those households that contain either an elder- brief introduction to the social assistance sys- ly or disabled member, or a student in tertia- tem in the TCc is provided. The section then ry education. In addition to that, the Turkish investigates the effectiveness of the social as- Cypriot administration provides free health- sistance system based on selected performan- care services for those households already ce indicators. receiving SA benefits. Individuals who are for- mally employed can also benefit from trans- fers such as retirement pensions, unemploy- ment benefits, contributory disability benefits, sickness and health insurance, etc., thanks to the contributions paid by their employer and by themselves to the social insurance system. l 15 Poverty and Social Assistance in the TCc The SA system was first established in 1989 the case of single parents; however, the age with the adoption of Regulation No 72/1989. requirement seems not to be binding) ii) the This regulation allowed the ’Social Services applicant’s “demonstrated” lack of income Department’ (‘SSD’) of the ‘Ministry of Labor from his/her own resources or through and Social Insurance’ to deliver transfers financial support from his/her children. In to individuals who meet certain eligibility other words, the “Poor in Need” includes both criteria. The SA system aims to ensure that categorical eligibility criteria (i.e. to benefit selected households have a level of income from the program the applicant must belong to sufficient to cover basic food needs. However, a predetermined list of categories of people), eliminating poverty per se, or supporting and income criteria (i.e. to benefit from the those able to work, was and is not a main program the applicant has to have an income objective of the SA system ('Ministry of Labor below certain levels). This is not uncommon and Social Protection’, 1989). across the EU, with many countries using similar criteria for noncontributory program 3.2 Overview of Social Assistance for the poor (Italy, Malta, the UK being all but Programs in the TCc some examples). For the purposes of the analysis that follows, The benefit structure of program is equally the SA system in the TCc consists of four pro- complicated and unclear and provides ben- grams and aims to support beneficiaries cov- eficiaries with a clear disincentive to work. er their basic needs, and access tertiary edu- The benefit varies with the income of the ap- cation and health services. The most common plicant. Single applicants with no income who types of cash-based support are the “Poor in are deemed eligible for support receive 896 TL Need” program first and foremost, and dis- per month. However, applicants with income ability transfers, followed by education schol- greater than zero but less than 498 TL (or half arships. The only sizeable in-kind noncontrib- the benefit for a single person) only receive utory program is the provision of free health partial assistance (500 TL, or 550 TL or 600 TL services for the poor. Other in-kind programs per month, whichever amount brings them are very rare. In the TCc, there is no conditional closer to TL 896). For families with more than cash transfer program in place to incentivize one member, the benefit is increased but it is specific outcomes (i.e. education, health, em- not clear by how much and according to which ployment etc.), or a program targeting poverty equivalence scale, making the benefit diffi- regardless of categorical elements related to cult to communicate and monitor. Not only households. this structure is administratively complicated The “Poor in Need” program but provides a clear disincentive to work for single beneficiaries with no dependents as, as The “Poor in Need” program is an shown in Figure 15, for incomes between 1 TL unconditional cash transfer program aimed and 896 TL one is better off not working and at particular categories of people, with some receiving the full benefit amount. income testing elements. The program, which is implemented by the 'SSD', is designed mainly The “Poor in Need” program represents, for elderly people and is very complex in despite the shortcomings listed above, a good terms of eligibility criteria. Whether someone basis on which one can build upon to endow is or isn’t eligible varies in fact according to a the TCc with a poverty targeted benefit. The long list of categories and income criteria. To “Poor in Need” transfer, in its current structure simplify matters, however, the main eligibility and design, somewhat resembles a Guaranteed criteria to apply for the program can be seen Minimum Income (GMI) program which can as two8; i) a minimum age requirement of 55 be found in many EU countries. GMI programs years for women and 60 years for men (in typically extends selected households with 8 The list of categories of people that can however qualify under very specific conditions is very long and detailed, as well as somewhat unclear. 16 l Poverty and Social Assistance in the TCc Figure 13 - Consumption expenditure of households, 2015 Source: World Bank based on interviews with TCc officials a monetary transfer to ensure that a certain income and assets criteria. The program could standard of living is met by beneficiaries (i.e. be better monitored and communicated more they are normally supplemental transfers the clearly and simply to the population to ensure amount of which varies depending on the maximum take up by the neediest. recipient’s needs and income). For instance, Scholarships a person with no income will receive the full amount of the transfer, whereas a person with Scholarship programs in the TCc aim to in- a higher income will receive a lower amount crease the participation and completion of to reach a pre-determined threshold. In this tertiary education. There are three types of sense, the “Poor in Need” could serve as a scholarship transfers; i) achievement-based9 ii) strong starting point to design a true poverty- quota-based10 iii) need-based. All scholarships targeting program in the TCc, to better target are implemented by the ‘Ministry of Educa- poor households, and to better protect them tion and Culture’ (MoEC) but only the need- in the face of shocks and crisis. The program based scholarship is considered as social assis- could be tweaked to simplify its eligibility tance. Tertiary education students are entitled criteria, revise and/or clarify equivalence to need-based scholarships if they can prove scales, revise benefit amounts, and reduce that they come from a low-income family. Full the implicit disincentives to work that the tuition fee of orphans and fifty percent of the current set up provides beneficiaries, so that tuition fee of needy students whose parents its impact on poverty could be increase. The earn less than the minimum wage is paid by program could be modified to ensure that poor the administration. For monthly regular schol- households with members of working age, arship stipends, income eligibility criterion is which are currently excluded from accessing much higher. Students are entitled to receive benefits, can apply and receive support if they monthly stipend if their household income is meet certain income criteria. In other words, less than 3.2 times of the minimum wage. ‘SSD’ the program could be gradually transformed evaluates applicants income eligibility through into a poverty-targeting tool, moving away household visits and reports the information from categorical eligibility criteria and towards to the ‘MoEC’. It is however unclear how the 9 Achievement based scholarships are delivered to students who have high grade point average (GPA) in tertiary education. 10 Students who do well in Turkey’s university entry exam are entitled to receive quota-based scholarships in case they register in TC universities that are decided together with the ‘Ministry of Education and Culture’ and Turkey’s High Education Institution. l 17 Poverty and Social Assistance in the TCc income eligibility criterion is calculated. Since the amount is 60 percent of the minimum the HBS does not differentiate between schol- wage iii) for beneficiaries with one child, the arships, the report treats all scholarships as amount is 70 percent of the minimum wage iv) social assistance. for beneficiaries with 2 or more children, the amount is 80 percent of the minimum wage. Disability Benefit The objective of disability transfer is to sup- Free Healthcare Service port disabled individuals to meet their basic The provision of free healthcare in the TCc needs. Individuals who have a disability rate aims to eliminate the financial barriers for of 40 percent or more are eligible to apply for poor families when accessing to health ser- disability benefits. The disability transfer is vices. Families who receive support from the implemented by the ‘SSD’ under the ‘Minis- “Poor in Need” program and/or disability so- try of Labor and Social Insurance’. If the indi- cial assistance transfers are also eligible for vidual has a disability rate below 40 percent, free healthcare services in ‘public’ hospitals they might be directed to apply for sickness in the TCc. No other eligibility requirement benefits. The disability transfer used to be a seems to exist, and access to free health care means-tested transfer, but the MoLSI removed seems largely at the discretion of SSD. This “in- the means-testing in 2016 to support all dis- kind” provision of healthcare does not deliver abled people in the TCc. To be eligible appli- any cash to beneficiaries, but instead elim- cants need to undergo a medical examination inates the costs of health services, indirectly in a ‘public’ hospital. Currently, 4,635 people contributing to households’ budget. It is there- are receiving disability benefits. The amount is fore difficult to measure the true budgetary adjusted as follows; i) for beneficiaries below impact of such program (from an administra- 18 years of age, the amount is 50 percent of tive point of view). the minimum wage ii) for adult beneficiaries, Box 4 – The “Families of the Fallen” Program The “Families of the Fallen” program is a categorical program which delivers benefits to families of individuals who died during the 1960s-70s conflict. It is implemented by the ‘Social Services Depart- ment’ in TCc and is considered, strictly speaking, a public assistance program as it is a non-contribu- tory program. However, it does not have a poverty alleviating objective, and has no income criteria or means testing elements. The report therefore examines its performance separately from more traditional social assistance program. The beneficiaries of the family of the fallen program are either the spouse or children of the fallen of past conflict. Beneficiaries are entitled to receive the benefit regardless of their income level. The spouses of the fallen receives the benefit for life. The sons of the fallen person receive the benefit until they finish their military service. The daughters of the fallen person receive the benefit depending on their marital status (i.e. until they get married, but if they divorce they can access the benefit once more). The amount of the benefit is high, but the budgetary impact of the program is modest due to its lim- ited size. Currently, beneficiaries receive 3,300 TL per month (1.27 times the minimum wage in 2018). By January 2018 there were about 800 beneficiaries, while by June this number had dropped to 785 people. The total cost of this program to TCc is 2,5 million TL per month, around 30 million TL per year. If the family of the fallen program were to be considered a social assistance transfer in the figures that follow, the total social assistance budget would increase from 1.05 percent to 1.24 percent of GDP. The performance of the family of the fallen program differs significantly from that of social assistance transfers. The family of the fallen program covered only 1.2 percent of the poorest quintile. Only about 22 percent of beneficiaries were poor, and about 34 percent of the program’s budget went to poor households. 18 l Poverty and Social Assistance in the TCc 3.3 Social Assistance Spending Although the GDP of the TCc grew 26.5 percent in real terms between 2014 and 2017, The total expenditure on social assistance SA expenditure did not grow at the same rate has been increasing both in nominal and real and its share in GDP dropped in the past years. terms since 2014. The TCc spent around 110 The TCc, (which can be classified as an upper million TL on social assistance in 2014 and middle-income economy using the World 152 million TL in 2017, a 38 percent increase Bank’s standards) spent on social assistance in nominal terms. In real terms (in 2014 TL), much less than the average of this group. The the increase was however modest, only 6.5 TCc’s overall SA spending was low even when percent. The “Poor in Need” (which increased compared to low-income countries, where from 36 million TL in 2014 to 48.2 million TL the average social assistance expenditure was in 2017) and especially disability benefits 1.47 percent. The gap was much larger when (which rose from 48.9 million TL in 2014 to compared to high-income and OECD countries 70.3 million TL in 2017) were the main drivers where the SA expenditures were 1.91 and 2.83 of this rise (7.6 percent in real terms), while percent of GDP, respectively. scholarship increased only by 2.6 percent in real terms, from 25.1 million TL in 2014 to 33.5 million TL in 2017. Figure 16 - Social Assistance Expenditure in real and nominal terms, 2014 – 2017 (Million TL) Source: TCc Administrative Data The TCc’s expenditure on social assistance was however low as a share of GDP and 3.4 The Performance of the Social below international standards. Social Assistance System assistance expenditure absorbed 1.24 percent Coverage of GDP in 2014 but dropped to 1.05 in 2017. In 2015, social assistance reached a small share of the poorest twenty percent of the Figure 17 – Social Assistance Expenditure population by income (i.e. the bottom quin- (% of GDP), 2017 tile), well below the figures observed in se- lected countries. The coverage rate of social assistance programs in the poorest quintile was 17.5 percent (27.1 percent in the poorest decile and 7.5 percent in the second poorest decile). When free healthcare provision is in- cluded, coverage increased to 22.8 percent for the poorest quintile (34.4 for the poorest decile and 11 for the second poorest decile). These figures point to the fact that, while the Source: SPO HBS 2015 and ASPIRE social assistance system was successful at l 19 Poverty and Social Assistance in the TCc Figure 18 - Coverage: TCc vs Selected Countries (%) - All Social Assistance, by quintile Source: SPO HBS 2015 for TCc data and WB ASPIRE database for other countries (most recent year available) reaching first and foremost the poorest, cov- Targeting erage remains low overall. As shown in Figure Very encouragingly, a high share of SA 18, coverage of the SA system in the TCc was beneficiaries came from the bottom quintile in fact significantly low compared to selected in the TCc, faring well by international countries. standards. Around 42 percent of all SA Cash transfers (defined as the “Poor in Need” beneficiaries belonged the bottom quintile, program and disability transfers) performed while only 10.8 percent were from the richest the best in terms of coverage. Cash transfers quintile. Further breaking down the income reached about 15 percent of households in the distribution, within the first quintile, 32.2 of bottom quintile, and free healthcare services all SA beneficiaries were in the poorest decile, covered about 11 percent. Scholarships, on and only 10.2 percent in the second decile, the other hand, reached very few individuals showing that the social assistance system across all quintiles. Generally speaking, as was successful is prioritizing households and Figure 19 confirms, coverage rates of each supporting the neediest. To put this in an individual SA programs were very low. international context, beneficiary incidence in Figure 19 - Coverage of Social Assistance Programs in the TCc, by quintile, 2015 Source: SPO HBS 2015 20 l Poverty and Social Assistance in the TCc Figure 20 - Beneficiary Incidence in Selected Countries (%) - All Social Assistance, by quintile Source: SPO HBS 2015 for TCc data and WB ASPIRE database for other countries (most recent year available) the TCc compared well against a set of selected The targeting performance of cash transfers countries for which data was available. Even (the “Poor in Need” program and disability though the TCc’s social assistance system can transfers) was the highest among SA programs. improve in terms of targeting (countries like Around 54 percent of the beneficiaries of cash Montenegro, Turkey, and Kosovo had more transfers came from the bottom quintile, a than 50 percent of SA beneficiaries in the first remarkable result. Free healthcare services for quintile), these are very encouraging results on the poor came next, with around 42 percent which policymakers should build upon. of beneficiaries from the bottom quintile. The Figure 21 - Benefit Incidence – TCc vs. Selected Countries (%) - All Social Assistance, by quintile Source: SPO HBS 2015 for TCc data and WB ASPIRE database for other countries (most recent year available) l 21 Poverty and Social Assistance in the TCc Figure 22 - Targeting of Social Assistance Programs (%), by quintile, 2015, Beneficiaries (left) and Program Resources (Right) Source: SPO HBS 2015 for TCc data and WB ASPIRE database for other countries (most recent year available) scholarship program displayed a significantly Adequacy lower targeting performance than other The adequacy11 of social assistance transfers SA programs with only 34.9 percent of the in the TCc was high compared to selected beneficiaries coming from the poorest quintile. countries. Relative to the total pre-transfer Looking at the resources devoted to social income of all SA beneficiary households in assistance benefits, the same picture the first quintile, social assistance benefits emerged. The share of social assistance represented around 98 percent of said income, suggesting that social assistance was budget spent on the first quintile (i.e. the the main and often only income source for benefit incidence) was 56.4 percent in the TCc these households in the bottom quintile. This in 2015, as shown in Figure 21, which fared finding can be interpreted in two contrasting very favorably to a set of selected countries. ways. The first is that the SA system may The shares of the total SA budget delivered to have indeed been supporting households the fourth and fifth quintiles were only 7.4 and who were not able to generate income from 6.0 percent, respectively, which suggests that other channels and was therefore achieving its leakage was very low. objective. Alternatively, though, one could see the SA system as too generously supporting Once again, the best performing programs households, giving them an incentive were cash transfers which delivered around not to work and generate other income. 57 percent of their resources to the bottom Unfortunately, the data available does not quintile. As shown in Figure 22 (right-hand allow for a more definitive answer on this issue, side), the scholarship program also distributed as this would require detailed information on a large share of benefits to the bottom quintile, the determinants of labor participation among with 44.8 percent of its total budget reaching households, and an analysis of the labor force the households in the bottom quintile. survey (including wage data). 11 As a reminder, adequacy is defined here as the total transfer amount received by all beneficiaries in a given group as a share of the total “pre- transfer” welfare of beneficiaries in said given group. 22 l Poverty and Social Assistance in the TCc Figure 23 - Adequacy in Selected Countries (%) - All Social Assistance, by quintile Source: SPO HBS 2015 for TCc data and WB ASPIRE database for other countries (most recent year available) Figure 24 - Adequacy rate of SA programs (%) - by quintile, 2015 Source: SPO HBS 2015 Cash transfers constituted a large share of the 3.5 Profiles of Social Assistance total household budget of the beneficiaries. Beneficiaries The adequacy of cash transfers reached up to 100.5 percent in 2015 which suggests This section aims to assess whether social that those households had on average more assitance programs were and are aligned income from social assistance programs than with the profile of the poor in the TCc. It is from other sources. The scholarship transfers’ important to understand which households adequacy rate was much lower compared over-represented and which households are to cash transfers, with 55.4 percent for the left out of the social assistance system. This poorest quintile. information can in fact inform policymakers on how to modify and redirect programs and policies to ensure that households who are currently not receiving support can do so if deemed appropriate. l 23 Poverty and Social Assistance in the TCc Quite unusually, social assistance beneficiary Social assistance transfers were more likely households tended to be smaller and have to be delivered to poor households whose less dependents than poor households. heads are women or, as per its design, el- In 2015, based on HBS data, the average derly. Among the households in the poorest household size of households in the poorest quintile 23.4 percent were female headed, quintile was 3.6, while the average size of a while the same ratio was 40.1 percent among beneficiary household was only 3.1, closer to social assistance beneficiary households, indi- the average throughout the TCc. Moreover, cating that female headed households were the dependency ratio of households receiving well included in the social assistance system social assistance was significantly lower (0.45 in the TCc. However, while social assistance dependent person for each working-adult) transfers supported the basic needs of elderly compared to households in the poorest headed households (40.4 percent of benefi- quintile (whose dependency ratio was on ciary household heads were over 55 years of average 0.60). This suggests that some poor age), only 21.3 percent of the poorest quin- households with more members and higher tile had elderly household heads. This points dependency ratios were being left out of the to the fact that the social assistance system social assistance system. This is atypical, as favored women and the elderly (which is not in most countries in the world, larger families surprising given the design of the “Poor in (especially families with children) are normally Need” program for instance). The flip side of favored or targeted by the social assistance this argument is that, while 38.5 percent of system through minimum income schemes or the households in the poorest quintile had a separate ad hoc programs. Eligibility criteria household head aged between 25-39, only 23 for the “Poor in Need” program could thus percent of social assistance beneficiary house- be modified to ensure that poor families with holds were headed by someone in this age children or with dependents could receive group. The TCc’s social assistance system, in support, and the benefit structure of the other words, made it harder for poor house- program could be revised to favor larger holds headed by working age adults (men in families and families with children (this could particular) to receive support if needed. This suggests, once again, that the criteria of the be done by introducing clear equivalence “Poor in Need” program should be revised to scales on the basis of which benefits would transform it into a purely poverty-targeting vary. That is benefits granted to families would program (which can be accessed no matter increase – proportionally or not - as household the age, gender, or employment status) as op- size and types changes). posed to mostly a categorical benefit. Figure 25 - Beneficiaries of Social Assistance Programs, Household size (left) and dependency ratio (right) 24 l Poverty and Social Assistance in the TCc Figure 26 - Beneficiaries of Social Assistance Programs, Gender (left) and Age (right) of Household Head Figure 27 - Beneficiaries of Social Assistance Programs - Employment Status of Household Head Kaynak: SPO HBA 2015 Social assistance beneficiary household or that work is seen as an alternative to social heads were more likely to be out of the labor assistance recipiency. As discussed in the force compared to poor household heads. previous section, the “Poor in Need” program Around one third of household heads that should be modified to ensure that working received social assistance benefits reported is not a condition for exclusion, that working being out of the labor force, while this rate age adults can receive support, and that labor was 20 percent among those in the bottom income and benefit recipiency are not seen quintile. As said above, this could have been at odds. This can be achieved by introducing either because of the TCc’s social assistance a feature commonly known as a “labor mainly supports households with heads who income disregard” when assessing households are not able to work, or because the SA system incomes for benefit eligibility determination. created disincentives for work due to the high This feature, used in many programs around adequacy of the benefits; the current data the world allows part of the income from does not allow us to answer this question. labor of households applying for benefits not However, this fact does indicate that “the to be taken into account when determining working poor” may have not been covered as eligibility, thereby reducing the disincentive to they should be by social assistance programs, take on and retain employment. l 25 Poverty and Social Assistance in the TCc 3.6 The Effect of Social Assistance transfers was only one percentage point in 2015, while the poverty gap is reduced by just System on Poverty over 2 percentage points (from 29.1 to 26.7 Expectedly, social assistance transfers have percent). The impact of transfers on inequality a limited impact on poverty and inequality. was less than one percentage points, with Although, as described above, adequacy the Gini coefficient decreasing from 35 to levels are high, the overall low coverage of 34.1 percent in 2015 pre and post transfers. the social assistance system, and the fact that This finding points once again to the need to reducing poverty is not its explicit objective, increase the effectiveness of ‘public’ spending result in a limited impact of social assistance on social assistance programs, and to refocus transfers on poverty and inequality in the TCc. programs towards a poverty targeting The difference in relative poverty pre and post objective. Figure 28 - Simulated Impact of Social Assistance on Poverty and Inequality, 2015 Source: SPO HBS 2015 26 l Poverty and Social Assistance in the TCc CONCLUSIONS AND POLICY The heads of poor households had lower education compared to non-poor households. RECCOMMENDATIONS While net school enrolment rates were high This report profiles poverty and assesses the throughout the TCc, there was a significant performance of the social assistance system difference in tertiary education enrolment in the TCc. The aim of this report is to provide between the poor and the non-poor. Only information and evidence-based analysis half of the poor owned the houses in which to policy-makers on the determinants of they live in, while this share was 81 percent poverty in the TCc and on the effectiveness of among non-poor households. Similarly, car social assistance programs, with the ultimate and computer ownership were much lower for objective of putting forward recommendations poor families, 66 percent of the poor owned on how to improve them. a car and 50 percent of the poor owned a computer. Main Findings Social assistance programs in the TCc aim In the TCc, relative poverty - defined in to support individuals who are not able line with EUROSTAT methodology as the to work and are less focused on reducing proportion of people with an equivalized poverty. The SA system delivers transfers disposable income below 60 percent of the to those households that include either an median equivalized disposable income - elderly or disabled member or a student in was 22.2 percent in 2015. While (in absolute tertiary education. In addition to that, the numbers) most of the poor lived in the two TCc Administration provides free healthcare biggest provinces in the TCc (Gazimağusa / services for households already receiving SA Famagusta and Lefkoşa / Nicosia), the poverty benefits. The SA system essentially aims to rate was significantly higher in the Iskele / ensure that selected households have a level Trikomo region. of income sufficient to cover basic food needs. The poverty rate in the TCc was relatively high Eliminating poverty, or supporting those able compared to EU countries and to selected to work, was and is not a main objective of the countries within Europe and Central Asia for SA system. which data is available. While EU averages The total expenditure on social assistance were around 17 percent, poverty rate varied has been increasing both in nominal and real from 9.2 percent in Iceland to 25.4 percent in terms since 2014. However, as a percentage of Serbia. The TCc ranked among the countries the GDP, social assistance expenditure remains with the highest relative poverty incidence, low and below international standards, having after Serbia, Romania, Turkey and Latvia. dropped from 1.24 percent of GDP in 2014 to Income inequality in the TCc was high in 2015 1.05 in 2017. and above the EU28 average. According to The TCc’s SA system had both some strengths HBS 2015 data, the Gini coefficient was 34 and some weaknesses. In 2015, the targeting percent, 3 percentage points above the EU performance fared relatively well even by average. international standards, with around 42 Poverty in the TCc displayed some key percent of all SA beneficiaries belonging to the characteristics. Poor households were on bottom quintile (only 10.8 percent were from average larger and had a higher dependency the richest quintile), and around 56 percent of ratio relative to non-poor households. The total benefits reaching the poorest quintile. poor were more likely to work in low-skilled However, social assistance programs covered private jobs and less likely to work in the ‘public’ only a small share of the poor households, sector and in education. They were also more delivering benefits to only 22.7 percent of the likely to be unemployed and out of the labor households at the bottom end of the income force. Among the working population, the distribution. Expectedly, but worryingly, social poor were more likely to work in temporary assistance transfers have a limited impact on jobs, in casual jobs and to be self-employed. poverty and inequality. l 27 Poverty and Social Assistance in the TCc Adequacy levels were high enough to lift The “Poor in Need” program is an households out of poverty; in particular, cash unconditional cash transfer program aimed transfers constituted a large share of the at particular categories of people, with total household budget of the beneficiaries. some income testing elements. The program, This finding can be however interpreted in which is implemented by the social services two contrasting ways. The first is that the SA department, is designed mainly for elderly system may be indeed have been supporting people and is very complex in terms of households who were not able to generate eligibility criteria. Whether someone is or isn’t eligible varies in fact according to a long list income from other channels and was of categories and income criteria. The benefit therefore achieving its objective. Alternatively, structure of program is equally complicated though, one could see the SA system as too and unclear and provides beneficiaries with a generously supporting households, giving clear disincentive to work. them an incentive not to work and generate other income. Unfortunately, the data The “Poor in Need” program represents, de- available does not allow for a more definitive spite the shortcomings listed above, an ex- answer to this issue, as this would require cellent basis on which one can build upon detailed information on the determinants of to endow the TCc with a poverty targeted labor participation among households, and an benefit. The “Poor in Need” transfer, in its cur- analysis of the labor force survey (including rent structure and design, resembles in fact a wage data). Guaranteed Minimum Income (GMI) program which can be found in many EU countries. Social assistance beneficiary households tended to be smaller and have less depen- Policy Recommendations dents than poor households. They were also To complement the currently used relative more likely to be headed by women or elder- poverty measures, an absolute poverty line ly relative to poor households, or to be out should be constructed to monitor poverty and of the workforce compared to poor house- identify poor households. An absolute poverty holds’ heads. This suggests that many poor measure would define the minimum monetary households with more members and higher amounts that households would need to meet dependency ratios, or with working members their basic needs. Even though an absolute as heads (especially men) were being left out measure does not link households’ conditions of the social assistance system. This is unusual, to changes in inequality or takes into account as in most countries, larger families (especial- the overall unfair distribution of resources, ly families with children) are typically favored absolute poverty rates are less sensitive to or targeted by the social assistance system overall economic changes (as the threshold through minimum income schemes or sepa- is not directly linked to overall shifts in the rate ad hoc programs. income distribution). An absolute poverty The most common types of cash-based line would also allow to assess whether poor support are the “Poor in Need” program households’ incomes have increased through first and foremost, and disability transfers, time, and hence whether their situation has followed by education scholarships. The only improved in absolute terms. It would also sizeable in-kind noncontributory program is introduce a clear and transparent metric by the provision of free health services for the which poverty status would be determined. poor. Other in-kind programs are very rare. In To define and measure absolute poverty, and the TCc, there is no conditional cash transfer to track it through time, a purchasing power program in place to incentivize specific parity (PPP) index is required, which has not outcomes (i.e. education, health, employment been calculated for the TCc until today. etc.), or a program with the explicit objective To increase the efficiency of the resources of reducing poverty or protecting households devoted to social assistance, and to increase from shocks. the impact of programs on poverty indicators, 28 l Poverty and Social Assistance in the TCc existing programs would need to be reformed. Household budget survey should be The design parameters of some programs conducted more frequently to allow for should be reconsidered, moving away from evidence-based discussions on poverty categorical targeting and towards an explicit monitoring and poverty reduction. Poverty poverty reduction objective. This would allow measurement in the TCc is an important to achieve greater poverty reduction impacts challenge when designing policies for the poor within the current budgetary allocation. households. First, HBS is the only household However, given the overall low coverage of survey in the TCc and is conducted only every the TCc’s social assistance system, if the TC 6-7 years due to the high costs involved. administration aims to achieve the levels of Second, the household survey data is not developing and developed countries, the publicly accessible to scholars and researchers resources devoted to social programs may which does not allow evidence-based need to be increased. To allow for greater policy discussions. It is therefore extremely spending on social assistance programs, a challenging to monitor poverty and accurately rebalancing of ‘public’ budgets would be estimate the effects of policies and projects on needed to create additional fiscal space. poverty reduction. Most importantly, the “Poor in Need” could More detailed household survey is required serve as a strong starting point to design a to assess the performance of the SA system true poverty-targeting program in the TCc, to more accurately. Although there are only four better target poor households, and to better social assistance programs in the TCc, the HBS protect them in the fact of shocks and crisis. does not include detailed questions for each The program could be tweaked to include different social assistance transfer. In general, an explicit poverty reduction objective. social assistance transfers are reported in the Policymakers may wish to consider simplifying same question as social insurance transfers. its eligibility criteria, introducing clear For instance, disability benefit is reported equivalence scales, revising benefit amounts, together with contributory disability benefit, and reducing the implicit disincentives to work contributory sickness benefit and “Poor in that the current set up provides beneficiaries. Need” transfer. Collecting program-level data The program could be modified to ensure that through separate questions in the HBS can poor households with members of working age, enable policymakers, scholars and researchers which are currently excluded from accessing to better assess the performance of the SA benefits, can apply and receive support if they system and increase the effectiveness of those meet certain income criteria. The program type of programs in the future. should also be redesigned to ensure that The changes suggested above become even poor larger families, especially families with more important in the face of the recent children, could qualify for the benefit, as the sharp depreciation of the Turkish Lira. As absence, in the TCc, of a program with the ‘public’ revenues shrink further, making objective supporting families with children spending on social programs more efficient is unusual relative to the setup of social will be paramount. At the same time, assistance programs in most EU countries. designing program which can effectively target Working should no longer be an exclusion poor households and support them in the face condition for the program. In other words, of crises will be crucial towards weathering the program could be gradually transformed the macroeconomic shocks which the TCc is into a poverty-targeting tool, moving away currently undergoing. from categorical eligibility criteria and towards income and assets criteria. The program could be better monitored and communicated more clearly and simply to the population to ensure maximum take up. l 29 Poverty and Social Assistance in the TCc ANNEX: Gender differences in Figure A1: Poverty rates in female and male headed households, before and after poverty and well-being social assistance transfers As discussed in Section 2.4, poverty rates are comparable between households headed by men and women. This annex discusses this finding in greater detail and examines the extent to which the other indicators of well-being discussed in the main body of the report differ across men and women. Poverty and household incomes The limited differences in poverty rates across female and male headed house- hold, shown in Figure 8, partly disguise Demographics variation across female and male headed households in income sources and house- Households headed by men have more hold composition. Female headed house- members than those headed by women. holds have lower total incomes on average On average, households have 3 members, but also have fewer household members. while female headed households have 2.4 When income is put in adult equivalent or compared to 3.2 members in male head- per capita terms, female headed house- ed households. The difference is larger in holds have slightly higher disposable in- households with below-median income comes per adult equivalent or per capita levels, driven by changes in the size of than male headed households. This dif- male-headed households, while rises ference is partially a reflection of female to 3.6 members in below-median male headed households being more likely to re- headed households compared to 2.4 in fe- ceive social assistance transfers: although male headed households. Female headed 22 percent of households are headed by households are smaller and differ in their women, 40 percent of the households demographic composition: on average, receiving social assistance are headed by they have fewer children residing in them, women (Figure 26). The adequacy rate of typically have fewer individuals aged be- social assistance received was higher in tween 15 and 59, and have a greater share female-headed households compared to of individuals aged 60 and above. Depend- in male-headed households and, as share ency ratios in female headed households of household income, social assistance ac- are slightly lower than those in male head- counted for 14 percent in female headed ed households, driven by lower child de- households compared to 2.4 percent of in- pendency ratios and offset by higher el- come in male headed households. derly dependency ratios. The marital status of heads in male and female households varies considerably, with potential implications for well-be- ing. Female heads of households are also considerably more likely to be widowers than male headed household (31 percent versus 1 percent respectively) and are also 30 l Poverty and Social Assistance in the TCc more likely to be divorcees (14 percent in to tertiary education compared to earlier female headed households compared to generations. As a result, the average years 2 percent in male headed households). In of education completed has risen from addition, the share of never married heads just over 6 years among those born be- is higher in female headed households (17 fore 1949 to over 11 years for those born percent compared to 12 percent). While between 1980 and 1989. Alongside rising it is not possible to compare well-being years of education completed for both in these different structures due to limit- men and women, the gender gap that was ed sample size, the results indicate that highly visible for those born before 1949 divorced heads of households are dispro- closed significantly for those born after portionately found to live in poverty. 1960, and has remained minimal since Education this point. Some differences do however remain: while women born between 1980 The level and number of years of edu- and 1989 were slightly more likely to have cation completed has risen significantly completed tertiary than men, they were across generations. Those born after 1985 also slightly more likely to only have com- are significantly more likely to transition pleted primary education. Figure A2: Household size split by demographics (left) and dependency ratios (right), 2015 l 31 Poverty and Social Assistance in the TCc Figure A3: Education completion by year of birth (top) and average years of education by year of birth (bottom) Assets ever widen with respect to transportation There are limited differences across male ownership and to a much smaller degree and female headed households in housing with respect to technology/telecommuni- conditions. Both female and male head- cations (Figure A4). female headed house- ed households have good sanitation and holds are substantially less likely than male shower conditions in the TCC (Figure A4), headed households to own a car (61 per- and the differences across poor and non- cent compared to 81 percent) and the gap poor households in hot water access and widens slightly when other transportation house ownership depicted in Figure 14 are assets are considered. These differences in greater than the differences across male asset ownership could limit access to mar- and female headed households. Gaps how- kets and services. 32 l Poverty and Social Assistance in the TCc Figure A4: Housing conditions (top) and ownership of assets (bottom) in the TCc, 2015 Labor force participation between 30 and 49, during child-bearing A considerable gender gap in employment and raising years, and the gap subsequent- and labor force participation rates has ly declines as men and women move to- been noted in TCC: on average, 30.4 per- wards retirement. Participation rates of cent of women over the age of 15 are em- both men and women drop considerably ployed compared to 58.6 percent of men, after the age of 50. The figure also shows and 33.9 percent of women are in the la- the share of women who are not working bor force compared to 62 percent of men. but are ready to work – this includes the The low labor force participation rates for unemployed, who are actively seeking women partly reflect lower rates of partic- work, and also includes workers who re- ipation among older workers: when con- port being ready to work if a job opportu- sidering only 15 to 60 year olds, the labor nity opens up within the next two weeks. force participation rates rise to 39.9 per- The gender gap closes entirely for the cent for women and to 69.8 percent for youngest cohorts when this group is in- men (Figure A5). cluded, signaling a readiness to engage in opportunities that emerge among younger When participation is separated into women. 5-year age groups (Figure A6), it’s clear that the participation gap is widest for women l 33 Poverty and Social Assistance in the TCc Figure A5: Labor force status (top) and employment status among those that are employed (bottom) Figure A6: Employment rates by age and sex 34 l Poverty and Social Assistance in the TCc There are substantial differences across likely to be engaged in construction, man- working men and women in the type of ufacturing, wholesale and retail trade and work they do and the sectors they are en- in the transportation sector (Figure A7). gaged in. The majority of men and wom- Similarly, women are more likely to report en are employed as employees (Figure clerical and professional occupations while A6); men are however twice as likely to men are more likely to report those involv- be self-employed than women (8 percent ing trade and machine operator roles. The compared to 4 percent) and are also con- 18 percent raw gender labor income gap siderably more likely to be employers (3 widens when education, potential expe- percent versus 1 percent). rience, occupation and sectoral composi- Sector wise, women are far more likely to tion are considered, suggesting that the work in professional services, education wage gap continues despite human capital and health than men, while men are more endowments and work choices. Figure A7: Sector of employment by sex for those aged 15-60 l 35 Poverty and Social Assistance in the TCc REFERENCES 'Ministry of Labor and Social Protection'. (1989, 11 28). Regulation on Social Assistance. Retrieved from Social Services Department: http://shd.gov.ct.tr/Portals/1082/yasalar/72-1989.doc%20 Sosyal%20Yard%C4%B1mlar%20Yasas%C4%B1.pdf 'State Planning Organization'. (2016). Household Income Distribution. Lefkosa/Nicosia: State Planning Organization. 'State Planning Organization'. (2017). Statistical Yearbook 2016. Lefkosa/Nicosia: “State Planning Organization”. Bradshaw, J., & Mayhew, E. (2010). The Measurement of Extreme Poverty in the European Union. York: European Union. Demery, L. (2000). Benefit Incidence: a practitioner’s guide. Washington DC, USA: The World Bank. Haughton, J., & Khandker, S. R. (2009). Handbook on Poverty and Inequality. Washington DC, USA: The World Bank. Ravallion, M. (2016). The Economics of Poverty. New York: Oxford University Press. Tesluic, E., Pop, L., Grosh, M., & Yemtsov, R. (2014). Income Support for the Poorest. Washington DC: The World Bank. The World Bank. (2018). The State of Social Safety Nets 2018. Washington DC: The World Bank. 36 l