Report No 19599 Civil Service Reform A Review of World Bank Assistance August 4, 1999 Operations Evaluation Department Document of the World Bank Abbreviations and Acronyms AfDB African Development Bank AFR African Region CAS Country Assistance Strategy CEM Country Economic Memorandum CSR Civil Service Reform EAP East Asia and the Pacific Region ECA Europe and Central Asia ECSPE Poverty Reduction and Economic Management Sector Unit EDI Economic Development Institute ESAF Enhanced Structural Adjustment Facility ESW Economic Sector Work GIRL Governance and Institutional Reform Loan GOA Government of Albania ICR Implementation Completion Report IDB Inter-American Development Bank IDF Institutional Development Facility LAC Latin America and the Caribbean LLC Leadership Learning Center MDB Multilateral Development Bank M&E Monitoring and Evaluation MNA Middle East and North Africa NGO Non-Governmental Organizations NIE New Institutional Economics NIR National Institutional Reviews NLS Non-Lending Services NPM New Public Management O&M Operations and Maintenance OD Operational Directive ODA Overseas Development Agency OPR Operations Policy and Research P-A Principal Agent PAR Performance Audit Report PBD Personnel and Budget Department PER Public Expenditure Review PREM Poverty Reduction and Economic Management PSM Public Sector Management QAG Quality Assurance Group RVP Regional Vice Presidency SAL Structural Adjustment Loan SAR Staff Appraisal Report SAS South Asia SBA Stand-By Arrangement SDS Service Delivery Survey SPA Special Programme of Assistance SR Sector Report SSP Sector Strategy Paper UNDP United Nations Development Program USAID United States Agency for International Development WDR World Development Report Director-General, Operations Evaluation Mr. Robert Picciotto Director, Operations Evaluation Department Ms. Elizabeth McAllister Manager Mr. Ruben Lamdany Task Manager Mr. Navin Girishankar Peer Reviewers Mr. Brian Levy Mr. Nick Manning Mr. Rogerio Pinto Mr. Anwar Shah The World Bank Washington, D.C. 20433 U.S.A. Office of the Director-General Operations Evaluation August 4, 1999 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Civil Service Reform: A Review of World Bank Assistance Over the past two decades, Bank support for civil service reforms (CSRs) has served as a primary vehicle for removing institutional impediments to market-led development in client countries. Despite this accumulated experience, CSR continues to suffer from definitional, strategic, and operational ambiguities. In an attempt to clarify these issues, this OED review "unbundles" Bank assistance, evaluates its relevance and efficacy, makes recommendations based on lessons learned. It also presents initial feedback from borrowers on the value of Bank support to their reform efforts (see Annex 12). The study sample covers 124 loans (comprising 350 interventions) approved in 32 client countries, as well as ESW from a subsample of 11 countries, over the 1980-97 period. These interventions are unbundled into three classes-downsizing, capacity building, and institutional reforms. The study argues that the Bank broke new ground in the 1980s by encouraging governments to "do more with less." The strategy pioneered the notion that governments in poor countries could simultaneously limit CS growth and improve policymaking capacity and operational efficiency. Yet the policy was weakened by erratic assessment of bureaucratic dysfunction and insufficient emphasis on performance incentives. Moreover, in the 1990s, the Bank was slow to mainstream new thinking about institutions and the role of the state. The review concludes that Bank-supported CSRs were largely ineffective in achieving sustainable results in downsizing, capacity building, and institutional reform. Only 33 percent of interventions in completed projects and 38 percent of ongoing efforts achieved satisfactory outcomes. Downsizing and capacity building initiatives failed to produce permanent reductions in civil service size and to overcome capacity constraints in economic management. Nor did Bank-financed operations lead civil servants to "own" and adopt formal rules such as codes of ethics or minimize arbitrary action by politicians. Four factors undermined efficacy-the poor quality of information on CSR; the limited role afforded to strategic management and cultural change; the absence of checks and balances on arbitrary action; and a failure to appreciate labor market and institutional constraints. The Bank's approach was too technocratic; it relied on small groups of interlocutors within core ministries and promoted one-size-fits-all CSR blueprints in diverse country settings. The study recommends that CSR activities be "unbundled," monitored, and guided by upstream institutional assessments and labor market analyses in addition to budget scenarios. Capacity building interventions should be linked systematically to job descriptions and monitorable performance. In countries with comprehensive reform programs, the Bank should rely more on stand-alone projects, lending instruments that allow adequate time for implementation, field-based supervision to better monitor institutional change, and participatory processes to nurture broad reform constituencies in government, civil society, and the private sector. Finally, the Bank should explore the feasibility of promoting results-based management by supporting comprehensive reforms of this kind in three to five pilot countries. Contents Executive Sum m ary .............................................................................................................. i 1. Introduction .......................................................................................................................... 1 Unresolved Questions for an Em erging Priority.................................................................... Objectives and Organization of the Review...........................................................................2 2. A Fram ework for Evaluating Civil Service Reform .........................................................2 Unbundling Civil Service Reform ......................................................................................... 2 The Fram ework...................................................................................................................... 3 Sample Coverage and M ethodology......................................................................................4 3. The Anatom y of Bureaucratic Dysfunction....................................................................... 5 4. Bank Support for Civil Service Reform , 1980-1997 ......................................................... 8 Evolution of CSR Strategy..................................................................................................... 8 Strategy Implem entation: Civil Service Reform M easures....................................................8 Strategy Implem entation: Lending and Nonlending Services............................................. 10 5. Outcom e ............................................................................................................................. 12 Performance of CSR M easures............................................................................................ 13 6. Assessing Strategic Capacity and Relevance..................................................................14 Helping Governm ents Do M ore with Less.......................................................................... 15 Lim itations of "Doing M ore with Less" .............................................................................. 15 Eroding Relevance in the 1990 ........................................................................................... 18 7. Assessing and Understanding Eficacy ............................................................................ 20 Assessing Eff icacy ............................................................................................................... 20 Understanding Efficacy: Factors Shaping CSR Performance............................................. 20 Information on CSR Performance.............................................................................. 21 Strategic M anagem ent and Cultural Change.............................................................. 23 Coordination Arrangem ents, Checks and Balances...................................................23 Contextual Constraints...............................................................................................24 Tim e Horizons for Reform and Sequencing of CSR Classes..............................................25 This report has been prepared by a team led by Navin Girishankar and including Migara De Silva, Velislava Grudkova, and Poonam Gupta. Additional working papers were prepared by Douglas G. Hartle and Robert Klitgaard. Caroline McEuen edited the report. Betty Casely-Hayford and Brigitte Wittel provided administrative support. 8. Findings and Recommendations ....................................................................................... 25 Bibliography.................................................................................................................................27 Boxes 2.1 Elements of a Country's Institutional Endowment Affecting Civil Service Reform............4 3.1 An Example for Sri Lanka: The Role of the State in East Asian Economies........................7 6.1 The Value of Upstream Assessment and Analysis of State Institutions in Albania.............18 7.1 The Benefits of External Checks and Balances in Low-Capacity Environments................ 24 Tables 2.1 Classifying Bank Support for Civil Service Reform.............................................................. 3 4.1 CSR Interventions by Lending Type.................................................................................... 11 6.1 Bank Response Rates by Class of Intervention ................................................................... 16 7.1 Contrasting Technocratic and Systems Approaches to Civil Service Reform..................... 20 7.2 Factors Shaping CSR Perform ance...................................................................................... 22 Figures 2.1 T he Fram ew ork ...................................................................................................................... 3 4.1 Number of CSR Interventions by Class, 1980-97................................................................. 9 5.1 OED Outcome Ratings of CSR Interventions...................................................................... 13 Annexes 1.1 Composition of Economic and Sector Reports in Sample................................................... 30 1.2 Composition of Lending Operations in Sample................................................................... 30 2. Civil Service Wage Bill as a Percentage of Current Expenditures Over Time................... 31 3. Index of Bureaucratic Quality: Regional Averages (1991-1997)....................................... 32 4. Civil Service Reform Measures by Class, Region, and Time Period..................................33 5. Bank C SR Portfolio Com position........................................................................................ 34 6. CSR Performance by Class of Intervention and Region...................................................... 35 7.1 Relative Effectiveness by Lending Type............................................................................. 36 7.2 Outcome Ratings: Completed Interventions........................................................................36 7.3 Outcome Ratings: Ongoing Interventions............................................................................ 36 8.1 Select Countries Falling Short of Rationalization Targets.................................................. 37 8.2 Select Countries with Short Run Successes, Unsustainable Results.................................... 37 9. Percentage Change in Civil Service Wage Bill as a Share of GDP Over Time.................. 38 10. CSR Initiatives by Client Country and Institutional Quality, 1981-1997............................ 39 11. Indicators of Downsizing and Capacity Building Used in Selected ESW...........................40 12. Borrower Feedback on Bank Assistance for Civil Service Reforms...................................41 Attachment 1. R eport from C O D E .............................................................................................................. 43 Executive Summary Introduction 1. Over the past two decades, Bank support for civil service reforms (CSRs) has served as a primary vehicle for removing institutional impediments to market-led development in client countries. Despite its growing importance, CSR continues to suffer from definitional, strategic, and operational ambiguities. In an attempt to clarify these issues, this OED review "unbundles" Bank assistance, evaluates its relevance and efficacy, and makes recommendations based on the findings. 2. The study sample comprises 124 loans approved in 32 client countries, as well as economic and sector work (ESW) from a subsample of 11 countries, over the 1980-97 period. For relevance, the review evaluated substantive issues as well as upstream processes (ESW, SARs, and diagnostic assessments) that shaped Bank-client responses to bureaucratic dysfunction. For efficacy, it examined substantive and process-related downstream factors (supervision, project management, and evaluation) that influenced implementation. Combating Bureaucratic Dysfunction: Bank Support, 1980-97 3. Between 1980 and 1997, the Bank diagnosed three stylized forms of bureaucratic dysfunction that undermined the ability of governments to secure the fundamentals of adjustment and development. First, endemic overstaffing accompanied by unsustainable wage bills was identified from the early 1980s onward. Second, a combination of misaligned organizational structures, poor human resources, and inadequate incentives weakened administrative capacity to carry out core government functions. Finally, by the early 1990s, the Bank found that the credibility and accountability of state institutions strained under the weight of cumbersome CS rules, political interference, and cultures of nonperformance. 4. In the first decade of support, the Bank's strategy to combat bureaucratic dysfunction turned on the notion that governments could "do more with less." It supported downsizing measures to limit and cut CS size, while imposing hard budget constraints on wage expenditures. These cutbacks were typically complemented by capacity building initiatives designed to help governments "do more" --that is, implement difficult adjustment programs. By the early 1990s, the Bank added a third class of measures-institutional reforms such as intra-public sector regulatory reform and external checks and balances-in order to make governments "more transparent and accountable," in addition to being more efficient. Relevance of Bank Strategy 5. Given the conventional wisdom on the role of the state in the early 1980s, the Bank's strategy of "doing more with less" broke new ground. It pioneered the notion that governments in developing countries could simultaneously limit CS growth and improve policymaking capacity and operational efficiency. For the first time, the Bank subjected questions of bureaucratic quality to the test of fiscal performance, thereby providing the impetus for borrowing governments to improve the allocative and technical efficiency of personnel expenditures. A key instrument in this early strategy, Bank ESW significantly enhanced knowledge of public administration in developing countries. ii 6. The relevance of this strategy was hampered, however, by erratic assessment of bureaucratic dysfunction and insufficient emphasis on performance incentives. The objectives of lending operations did not adequately stress either the removal of wage disincentives or the strengthening of linkages between wages and basic performance criteria. Interventions seldom sought to influence nonwage incentives such as organizational culture. Moreover, the Bank has been slow in mainstreaming the new thinking about institutions and the role of the state, which gained prominence in the past decade. Efficacy of Bank Support 7. On average, only 33 percent of closed CSR interventions and 38 percent of ongoing efforts achieved satisfactory outcomes. Even when desirable, outcomes were often not sustainable. Downsizing and capacity building initiatives failed to produce permanent reductions in CS size and to overcome capacity constraints in economic management and service delivery. There was no evidence that civil servants began to "own" and follow formal rules such as codes of ethics in any meaningful way. As a result, institutional reforms could not substantially limit arbitrary action by bureaucrats or politicians. 8. Four factors undermined the efficacy of Bank-supported interventions: (i) the poor quality of information on CSR performance, (ii) the limited role afforded to strategic management and cultural change, (iii) the absence of checks and balances on arbitrary action, and (i;) a failure to appreciate key contextual constraints. The significance of these factors in explaining outcomes illustrates how the Bank's approach to CSR has been largely "technocratic." Rather than engaging CSs as dynamic systems that are influenced by multiple stakeholders, Bank operations relied on small groups of interlocutors within core ministries to design and implement one-size- fits-all CSR blueprints in diverse country settings. 9. Poor Information on CSR Performance. With the exception of fiscal data, standardized indicators of CSR performance were neither fully developed nor operationalized for monitoring and evaluation (M&E). As a result, implementation suffered from information asymmetries that prevented reforming governments from holding bureaucrats accountable to CSR objectives. For instance, managers granted accelerated promotions to raise salaries during wage freezes, which led to higher wage bills rather than savings. 10. Limited Role for Strategic Management and Cultural Change. Capacity building efforts were based on a narrow understanding of incentives as wage enhancements. They did not envision appropriate roles for strategic management (including the reorganization of work) and cultural change in strengthening incentives to perform. M&E often failed to acknowledge the detrimental impact of conventional Bank processes-such as the use of expatriate consultants and Project Implementation Units (PIUs)-on the credibility of standard operating procedures, as well as on the morale of civil servants. On a related point, the Bank's skepticism about promoting greater autonomy for public sector managers meant that CSRs often compelled bureaucrats in client countries to be efficient without empowering them to be innovative. 11. Absence of Coordination Arrangements, Checks, and Balances. Efficacy and sustainability suffered as borrowing governments carried out interventions without establishing the institutional arrangements necessary for coordination between government bodies or providing checks and balances to arbitrary action. For example, poorly defined authority between budgeting and personnel departments allowed overstaffing and the proliferation of ministries to continue, even after downsizing efforts. Similarly, in the absence of external checks on political interference, CS hiring remained a primary mode of dispensing patronage. Broad iii groups of stakeholders such as public employees, citizens, and private firms rarely participated in the process of defining and enforcing new rules and norms governing CS activities. 12. Failure to Appreciate Labor Market and Institutional Constraints. While all three contextual constraints-macroeconomic performance, labor market trends, and institutional endowment-significantly influenced CSR outcomes, the Bank relied mainly on budget scenarios to guide the design of CSRs. Neither ESW nor M&E used models of CS systems that adequately accounted for labor market and institutional trends (for example, projected demand for CS jobs or the level of political appointments in the CS) when elaborating reform scenarios. Conclusions and Recommendations 13. The review found that Bank-supported CSRs were largely ineffective in achieving sustainable results in downsizing, capacity building, and institutional reform. This was, in part, due to significant political difficulties in implementing CSRs. Yet the relevance and ownership of reforms were also weakened by a technocratic approach that failed to mainstream institutional analysis and develop a coherent framework for intervening in administrative systems. The Bank should take the following steps to promote lasting improvements in public sector performance. 14. For the purposes of monitoring and evaluation, the Bank should develop a system of categorizing CSR interventions in three classes based on their primary objective-downsizing, capacity building, or institutional reform. In addition, the Network should standardize performance indicators for each class of CSR in order to monitor impact. 15. Future CSR operations should systematically link capacity building interventions to job descriptions and monitorable performance of civil servants and their respective units 16. Bank-supported CSR interventions should be preceded by institutional assessments of administrative systems and analyses of labor market trends in addition to budget scenarios. 17. Where the Bank supports comprehensive reform programs, interventions should be designed as stand-alone projects and supported by lending instruments that allow adequate time for implementation (such as adaptable lending). The Bank should also employ participatory processes to nurture reform constituencies in government, the private sector, and civil society. 18. The Bank should use ongoing evaluations of CSR support by other multilateral and bilateral donors as a basis to more clearly define its strategic role in different contexts. Where it does not have a comparative advantage, the Bank should only play a supporting role. 19. Finally, the Bank should explore the feasibility of promoting results-based management by supporting comprehensive reforms of this kind in three to five pilot countries. 1. Introduction Unresolved Questions for an Emerging Priority 1.1 Over the past two decades, civil service reforms (CSRs) have been a primary vehicle for the Bank and its clients to remove the institutional impediments to market-led development. Between 1980 and 1997, the Bank supported 180 lending operations with CSR components and built a considerable body of knowledge on civil service (CS) issues through its economic and sector work (ESW). Demand for such lending and nonlending services is growing rapidly across developing and transition countries. In FY98 alone, 72 percent of country assistance strategies (CASs) highlighted public sector reform as a development priority, as opposed to only 39 percent in FY95. In collapsing states in southeastern Europe and Central Asia, CASs have gone as far as to identify "state building" as a key objective. 1.2 Despite its growing importance and considerable trackrecord, CSR is plagued by questions of definition, strategy, and operational effectiveness. The ambiguities that surround CSR hamper the Bank's ability to meet client demands. 1.3 Definitions: Who Are Civil Servants? What Is Civil Service Reform? In general, the Bank has not clearly identified which public sector employees are "civil servants," in part because they are defined differently across countries. Depending on regional or administrative conventions, "civil servants" can refer to military personnel, teachers, or health workers, as well as career bureaucrats.1 Rather than differentiating between elite cadres of careerists and other public employees, Bank operations have essentially considered "civil servants" to be those personnel (outside public enterprises) whose salaries are supported by the central government's wage bill. As it continues to support CSRs in diverse contexts, the Bank will require a standard nomenclature for public servants who are to be affected by such interventions. 1.4 The second definitional issue concerns the bundling of diverse lending and non-lending activities under the rubric of CSR without a formal Operational Directive (OD) or guidelines for good practice.2 From pay and employment reforms to intrapublic sector regulatory reform, CSR comprises a wider array of administrative measures than its name suggests.3 The hydra-headed nature of CSR is also the by-product of attempts to import innovations and reforms that have reshaped public sectors in countries of the Organization for Economic Cooperation and Development (OECD) over the past two decades. Driven by a rising tide of popular frustration with government performance and scarce public financial resources, countries such as New Zealand, Australia, the United Kingdom, and the United States adopted a variety of New Public Management (NPM) innovations such as "open" civil service systems, competition in hiring, performance contracting, outsourcing of service provision, output- and outcome-based measurement of agency performance.4 The radical character of these reforms raised the development community's expectations of what assistance can do to reshape the state in 'Schiavo-Campo et al., 1997, pp: 4-5. The Bank has codified good practice guidelines for technical assistance (TA), which covers a small subset of CSR issues related to capacity building. However, the manual's focus on CSR is coincidental rather than comprehensive. 'Nunberg, 1997. 4 Organization for Economic Cooperation and Development, 1995. 2 developing and transition countries. With these new expectations came the need to better define the range of CSR activities the Bank would support and their intended development objectives. 1.5 Strategy: Why Support CSRs? With the publication of WDR 1997, The State in a Changing World, the Bank acknowledged that state institutions significantly shape the development prospects of client countries.5 A growing body of empirical evidence supports the notion that institutional reform of the state significantly enhances the prospects for growth in countries with low institutional capability.6 Since CSR is only one method for reforming state institutions (e.g., service delivery reforms, anti-corruption initiatives), the Bank and its clients require greater selectivity in pursuing different types of public sector interventions in achieving the development objectives of country and sector strategies. 1.6 Operational Effectiveness: What Works? While the internal literature on CSR is replete with prescriptions, it is weak on systematic evaluations of Bank experience. Aside from this OED review and a 1997 self-evaluation by the Poverty Reduction and Economic Management Network (PREM), there has been little by way of comprehensive, cross-country evaluation of CSR. As a result, the Bank has found it difficult to establish standardized indicators for bureaucratic performance or to codify an OD or guidelines for good practice. Ultimately, the Bank's success in supporting the reform of complex public sector institutions will depend on its ability to identify and apply elements of good practice across regions. Success also requires rigorous measurement of CSR impact on state capability. Appropriate instruments, tools, and processes will be required to equip the Bank to help achieve the myriad objectives of CSR. Objectives and Organization of the Review 1.7 In an attempt to clarify these issues, the review sought to evaluate Bank strategy over the life of the portfolio; examine the efficacy of different classes of CSR; and identify key factors shaping CSR performance. Section 2 defines OED's approach to "unbundling" CSR measures, explains the evaluation framework, and describes the review's coverage and methodology. Section 3 presents Bank diagnoses of bureaucratic dysfunction across regions from 1980 through 1997. Section 4 outlines the evolution of Bank strategy along with the CSR measures and instruments used to respond to bureaucratic dysfunction. Section 5 describes implementation and outcomes, followed by analyses of relevance and efficacy in sections 6 and 7 respectively. Finally, section 8 summarizes key findings and makes appropriate recommendations to improve Bank support. 2. A Framework for Evaluating Civil Service Reform Unbundling Civil Service Reform 2.1 To define "civil servants," the review assumes country-specific definitions used in CSR operations. On the larger issue of unbundling CSR activities, OED devised a matrix for categorizing CSR measures according to the three overarching development objectives: (i) achieving fiscal discipline, (ii) enhancing policymaking capacity and operational efficiency, and Eggertson, 1990. 6 Chong and Calderon, 1997; Knack and Keefer, 1995; Mauro, 1995; Mauro, 1996; World Bank, 1998. 3 (iii) realizing good governance. In pursuing these objectives, the review found that the Bank supported three corresponding classes of CSR-downsizing, capacity building, and institutional reform (table 2.1). Table 2.1: Classifying Bank Support for Civil Service Reform Development Objectives Class of Civil Service Reform Fiscal Discipline Class 1. Downsizing the Civil Service Pay and Employment Reforms Size and Wage Bill Rationalization Wage Decompression Compensating Losers Policymaking Capacity Class 2. Building Administrative Capacity and Operational Efficiency Organizational Structures and Systems Streamlining Governmental Organizations Strengthening Intergovernmental Linkages Improving the Quality of Human Resources Training Recruitment Principal-Agent Relations/Reducing Information Problems Wage and Nonwage Incentives Providing Complementary Inputs/Technology Good Governance Class 3. Reforming Institutions Intrapublic Sector Rules and Regulations Broader Checks and Balances on Arbitrariness Anti-Corruption Measures Increasing Accountability of Bureaucrats to Principals 2.2 Each of the 124 lending operations in the sample included interventions from one or more of the CSR classes described above. All interventions were categorized by country, time period, and class in order to evaluate relevance and efficacy. Figure 2.1: The Framework Upstream The Framework n Processes 2.3 OED first evaluated the outcomes of all ongoing and completed CSR interventions on the basis of their Policy Content RELEVANCE Country Context achievement of stated objectives Dnz.MroeconuiPerformance (efficacy). After having set this "low Building Capacy EF CY n int Trent threshold" for rating CSR performance, Institutonal Reform - Institutional Endowment the review analyzed relevance-that is, the fit between policy content and country context (figure 2.1). Substantive policy Downstream content of CSRs was evaluated in light of Instruments and Processes 4 the established literature in public administration and the New Institutional Economics (NIE). Accordingly, CSs were considered to be systems of principal-agent (P-A) relations between managers (or senior bureaucrats) and civil servants, and between senior bureaucrats and politicians. The review then examined the extent to which CSRs fit country factors- macroeconomic performance, labor market trends, and institutional endowment (box 2.1).7 2.4 Finally, Bank processes and instruments were analyzed to judge both the relevance of CSR strategy and the efficacy of interventions (figure 2.1). For relevance, the review identified upstream factors (during policy dialogue, diagnostic assessments) that shaped the Bank's strategic capacity to appropriately respond to bureaucratic dysfunction. In understanding efficacy, it highlighted downstream factors (during supervision, monitoring, and evaluation) that explain performance. Box 2.1: Elements of a Country's Institutional Endowment Affecting Civil Service Reform Political Institutions * Legislative and executive institutions * Formal and informal relationships between politicians and public administration Social Institutions (as they relate to the civil service) * Custom and informal, well-accepted norms that restrain individual and collective action * Character of contending social interests, including the role of ideology * Business, NGOs, labor organizations, and civic associations Sources: Adapted from Levy and Spiller (1996) and Pinto and Mrope (1994). Sample Coverage and Methodology 2.5 The review did not cover a uniform sample for lending and non-lending services. For Bank lending, the review evaluated 124 operations in 32 client countries (Annex 1.1). These 124 loans were drawn from a universe of 180 approved between 1980 and 1997 with at least one CSR component. Selection criteria for the predominantly African sample included frequency of lending (especially serial CSR projects), intensity of interventions, and significant independent borrower experience with governmental reform. Staff appraisal reports (SARs) and Institutional Development Facility (IDF) grants were primary sources for evaluating Bank diagnoses, strategy, recommendations, and upstream processes. Ratings and analyses of outcomes and downstream processes relied on Implementation Completion Reports (ICRs) and Performance Audit Reports (PARs) for closed operations as well as Form 590s, supervision reports, and Annual Report of Portfolio Performance (ARPP) ratings for ongoing operations. For nonlending services (NLS), OED analyzed ESW from a sub-sample of 11 countries drawn from the 32-country lending sample based on intensity of CSR lending, independent borrower interest in CSR, depth of Bank- country policy dialogue, and regional distribution (Annex 1.2). Analyses of lending and ESW were supplemented by extensive interviews and a joint seminar with Bank PSM staff in PREM, the Regional Vice Presidencies (RVPs), and the World Bank Institute (formerly EDI). 2.6 OED faced significant difficulties in finding reliable data on the fiscal, capacity, and institutional aspects of CSR. In addition, data on Bank processes such as supervision were generally not collected for CSR components of loans. Where available, the review relied on ' Lindauer and Nunberg, 1994; Stevenson, 1992. 5 standard OED, Bankwide (including OPR and PBD), and IMF databases, as well as country teams, for time-series data on traditional pay and employment issues. For data on bureaucratic quality, the review relied on PREM indicators that date back to 1991. 2.7 Time constraints limited this review to a desk study. But, the main findings provide a basis for the design of in-depth, country case analyses. Country case studies could more readily construct counterfactuals to determine the value-added of Bank support. 3. The Anatomy of Bureaucratic Dysfunction 3.1 During the 1980-97 period, Bank diagnoses across client countries revealed the complex anatomy of bureaucratic dysfunction, with its wide-ranging problems such as overstaffing, distorted wage incentives, and weak governance institutions. These problems increasingly constrained the development prospects of client countries, particularly those in the Africa Region (AFR). 3.2 Bank ESW and SARs were the primary vehicles for diagnosing dysfunction and analyzing its causes. In that sense, they are also reflective of the Bank's evolving framework for the role of the state in developing and transition economies. As the depth and breadth of its analysis improved over the 1980s and 1990s, the Bank increasingly acknowledged the links between state capability and economic development. It would ultimately conclude that a client country's framework for the role of the state significantly shaped its chances for sustained growth. 3.3 Fiscal Costs ofBloated Bureaucracies. From the early 1980s onward, the fiscal .onsequences of endemic overstaffing posed a major threat to the success of Bank-sponsored adjustment programs. By the early 1990s, the Bank had diagnosed CS size as unrestrained in 53 percent of client countries (and 77 percent of AFR countries).8 ESW, particularly in AFR and SAS countries, attributed unrestrained CS growth to patrimonialism, ideology, and interest group pressure. Patron-client relationships placed demands on elected officials and bureaucrats to hire large numbers of casual or temporary workers. University students formed a vocal special interest group that pressured governments to continue historical guarantees of CS employment upon graduation. These forces established patterns of CSR hiring that reinforced populist beliefs in government as the employer of last resort. 3.4 The adverse impact of overstaffing on wage bills in recurrent budgets soon became difficult to ignore, especially in AFR and LAC (Annex 2). Upon being hired, public sector employees typically obtained repeated wage hikes, accelerated promotions (or grade creep), and a variety of non-monetized allowances (e.g., housing, transportation), all of which placed upward pressures on recurrent expenditures. Since they were often not transparent, allowances and fringe benefits carried hidden costs that could not be easily accounted for in budget scenarios. Excessive wage bills, in turn, crowded out expenditures on operations and maintenance (O&M); dwindling O&M expenditures undercut the provision of even the most basic supplies, such as stationery, preventing civil servants from carrying out routine functions. 8 De Silva and Gupta, 1997. 6 3.5 By identifying the fiscal costs of dysfunction, the Bank underscored the pervasive problems facing functionally overextended states. Analogous to structural reforms of the economy, borrowing governments required CSRs to reshape central administrations to fit the needs of market-led growth. 3.6 Capacity Constraints. Along with bloated bureaucracies, borrowing governments typically lacked the administrative capacity to carry out those core functions necessary for successful adjustment. Distortions within government bodies-misaligned organizational structures and systems, poor human resources, and poorly defined P-A relations-hampered policymaking capacity and operational efficiency. 3.7 In the early 1980s, the Bank found that organizational structures and systems were not aligned with the functional requirements of ministries in 25 percent of sample countries. Over the next decade, 66 percent of sample countries suffered from organizational problems of this nature. Specifically, governments in AFR, LAC, and SAS allowed for indiscriminate proliferation of cabinet ministries and their line agencies, often at the expense of predictable policymaking and efficient operations. Typically, the number of ministries and agencies within specific functional policy areas (e.g., health or agriculture) grew rapidly, resulting in significant duplication of effort and jurisdictional overlap. Poor intergovernmental linkages affected pay and personnel management in central ministries, as well as coordination between central and sectoral ministries or local governments. From the 1980s through the early 1990s, the Bank found these intergovernmental linkages to be particularly weak in LAC and AFR. 3.8 Another major constraint on administrative capacity was the poor quality of human resources. Over half of all sample countries were in urgent need of specific technocratic skills (e.g., budgeting and economic analysis) during the mid-1980s and early 1990s. Skills shortages were partially a function of governments' inability to recruit and retain high-caliber civil servants; the Bank and its clients confronted this problem in AFR, EAP, ECA, and LAC. The migration of qualified personnel abroad was particularly evident in AFR and LAC countries. 3.9 Difficulties in recruiting and retaining qualified staff were largely attributable to poorly defined Principal-Agent (P-A) relations and attendant information problems. P-A problems arose from inadequate "contractual arrangements"-both formal and informal-between managers (the principals) and civil servants (their agents). For instance, civil servants in a third of all client countries were severely underpaid for the services they provided. Wage disincentives such as low or falling real wages, severe wage compression, and low relative wages in comparison with the private sector were endemic in these countries.9 In AFR, real wages typically fell below subsistence levels for lower-level staff. Disincentives further undermined performance by encouraging moonlighting and other income-generating activities such as petty corruption during work hours. Wage compression, identified in AFR, LAC, and SAS during the mid-1980s, sapped morale among technical and managerial staff, who were also underpaid in relation to their private sector counterparts. 3.10 During the 1987-93 period, the Bank found that CS grading and classification systems were opaque-complicated by an excessive number of wage brackets-in 18 percent of AFR and 20 percent of LAC countries. This, in turn, encouraged managers to accelerate promotions of CS ' Wage compression ratios or decompression rates are the relative wages of higher- to lower-graded civil servants. The fair comparison rule measures wages in the public sector relative to those in the private sector. 7 cohorts as a way of increasing salaries during official wage freezes. More important, grade levels and salaries were rarely linked to job descriptions, and almost never tied to performance. Poor pay-performance linkages were diagnosed in nearly 30 percent of AFR countries during the 1987-93 period and in 20 percent of LAC countries after 1993. Such P-A problems were often rooted in information asymmetries because managers lacked accurate information about the performance of civil servants. During the mid-1980s, weak information systems in 40 percent of client countries exacerbated problems of management and performance measurement. 3.11 Implicit in diagnoses of capacity constraints was the view that adjustment itself, to be successful and sustainable, required a capable state. The rationalization of public administration was a necessary but not sufficient condition for carrying out demanding adjustment programs. CSR would have to address the ancillary issues of public sector capacity if client countries were to successfully achieve structural reform of their economies. 3.12 Institutional Weakness Inside and Outside Government. In addition to fiscal crises and capacity constraints, a third form of bureaucratic dysfunction was the poor quality of governance institutions. In the mid-1980s and early 1990s, the Bank became aware that cumbersome and incongruous intra-public sector rules and regulations undermined transparency and accountability across 36 percent of countries in AFR, 33 percent in MNA, 25 percent in EAP, and 20 percent in LAC. Confronted by a web of formal rules and regulations, civil servants in many client countries often did not know what exactly was expected of them. Since enforcement of appropriate rules and restraints was weak, compliance by civil servants suffered, as evidenced by high rates of absenteeism. In addition, arbitrary action by civil servants created increased uncertainty for private sector firms in AFR, LAC, and SAS. 3.13 In countries emerging from long periods of state interventionism, civil servants continued to harbor a "mistrust" of business and civil society. PREM indicators for bureaucratic quality, which are based on surveys of private sector firms, do reveal downward trends in state credibility from 1991 to 1997 across all regions except SAS and MNA (Annex 3). Borrower governments were advised that significant shifts in CS ethos and standard operating procedures were needed to improve the credibility of state institutions in the eyes of private firms and citizens. 3.14 Even in the 1990s, the Bank only Box 3.1: An Example for Sri Lanka: The Role of sparingly assessed checks and balances on the State in East Asian Economies interactions between bureaucrats and their principals-that is, politicians and citizens. In 1996, the Bank advised the Government of Sri Lanka ESW noted systemic corruption as a major to consider the following characteristics of East Asian administrative problem only in Sri Lanka and states in creating conditions favorable to rapid growth Uganda. It more frequently cited problems of and development: "informality"-exacerbated by overly complex * A cadre of economic technocrats insulated from "ifomaiy"excrbte y vel cmpe narrow political pressures legal and regulatory systems-in LAC.10 By arrep letcivl sre * A reputable civil service the late 1990s, uncertainty in relations between * Sound legal and regulatory environment the CS and politicians due to political * Formal institutions to share information and win interference became apparent in a small support from the business community. number of AiFR and EAP countries. The need Source: Sri Lanka, "Country Economic Memorandum," 1996 for stronger "watchdog" organizations, such as I o Informality is the degree to which production and exchange are governed by informal rather than formal rules. 8 Offices of Auditors General or nongovernmental organizations (NGOs), was cited more broadly in 18 percent of AFR and 20 percent of LAC countries during the late 1980s and early 1990s. 3.15 In most cases, the rationale for a governance approach to CSR was not clearly articulated. A notable exception was in Sri Lanka, where the Bank argued that state credibility and administrative capacity significantly shape prospects for growth. It cited the East Asian experience as evidence of the macroeconomic benefits of developing a reputable CS (box 3.1). 3.16 Overall, OED found that the Bank encountered a complex and diverse matrix of bureaucratic dysfunction across regions and time periods. At the same time, countries also faced problems peculiar to their region, such as the high incidence of informality in LAC or wages below subsistence level in AFR. Notably, problems prevalent across sample countries (for example, excessive wage bills from overstaffing) were especially widespread and acute in AFR. 4. Bank Support for Civil Service Reform, 1980-1997 Evolution of CSR Strategy 4.1 Throughout the 1980s, the Bank's CSR strategy was based on the notion that governments could "do more with less." It supported downsizing measures to limit and cut CS size while imposing hard budget constraints on wage expenditures. These efforts were typically complemented by efforts that aimed to remedy distorted systems and incentives in government organizations, and thereby build the state's capacity to undertake the core economic management functions necessary for adjustment. With growing pressure to adopt a governance approach to CSR, the Bank added a third class of measures in the 1990s. These institutional reforms sought to make the state "more open and accountable," rather than simply more efficient. 4.2 With the publication of WDR 1997, The State in a Changing World, the Bank unequivocally acknowledged the importance of good governance in ensuring efficient and effective public sectors. In FY99, PREM began drafting a Public Sector Strategy Paper (SSP), with an eye towards operationalizing the institutional or governance approach to public sector reform. The SSP focuses on changing the "rules of the game" for policymaking, service delivery, and regulation. It envisions CSR as one of many options (others include anti-corruption, judicial reform, and public expenditure reform) available to countries to enhance the quality of state institutions. Strategy Implementation: Civil Service Reform Measures 4.3 In line with its evolving strategy, the Bank supported three classes of CSR--downsizing, capacity building, and institutional reform--over time (figure 4.1).11 Each class of CSR included a combination of diagnostic tools and operational measures (Annex 4). 4.4 Downsizing the Civil Service. The 1987-93 period was a watershed for downsizing. Over two-thirds of downsizing interventions were initiated in these years for a predominantly "Figure 4.1 disaggregates the third class of CSRs-institutional reforms-in order to distinguish between intrapublic sector regulation and more recent governance-focused interventions such as external checks and balances (para. 4.9). 9 African clientele. In AFR, as well as other regions, CS censuses helped prepare downsizing efforts by documenting real employment levels in nearly a quarter of client countries. Subsequently, size reductions were implemented through voluntary early retirement packages, hiring freezes, natural attrition, and strict enforcement of the retirement age. Concurrently, the Bank financed schemes to compensate the "losers " in downsizing efforts. For the decade following 1987, lump-sum severance packages were the most broadly supported form of compensation in over 40 percent of client countries, particularly in AFR and LAC. Less frequently, the Bank also assisted the transition of exiting civil servants into private labor markets through job training and redeployment programs. 4.5 Size reductions were often accompanied by wage policy reforms. Eighty-four percent of these measures sought to contain wage expenditures in the recurrent budget. Wage bill caps and targets based on detailed medium-term budget scenarios were prevalent in 40 percent of APR client countries. Other measures included wage freezes and FIgure 4.1: Number of CSR Interventions by Class, 1980-97 salary containment measures, 35 - and monetization and 30 cGevemsce rationalization of allowances 12lntns-sector Regulations 25 UCpaftvBuidinia and benefits. A second, less rL- frequent, component of wage 20 --- - -- policy reform was wage 15 decompression, which involved suspending automatic, across- 1o . the-board wage hikes (especially s - - for lower-level staff) and limiting increases in the 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 199 1991 199 1993 1994 1995 199 1997 minimum wage. 4.6 Capacity Building. Capacity building efforts comprised the broadest range of CSR measures supported by the Bank. Typically these efforts sought to identify and capitalize on opportunities to redesign organizations, improve human resources, and clarify P-A relations. Assessment tools to diagnose capacity problems included organizational audits, functional reviews, and manpower audits, virtually all of which were used by AFR. clients. The Bank's operational agenda to redesign organizations most often supported vaguely defined "capacity building" initiatives in core ministries. Relatively evenly distributed over the life of the portfolio, these initiatives were most heavily concentrated in AFR (50 percent), followed by LAC, ECA, EAP, and MNA. As part of organizational redesign efforts, the Bank also financed initiatives to strengthen intergovernmental linkages that affected personnel management in AFR and LAC. Measures typically included efforts to improve systems or procedures related to budgeting, as well as coordination between central and sectoral ministries (or central and local governments). 4.7 Measures to improve the quality of human resources comprised short and medium-term training initiatives (for example, workshops, seminars, and career development programs), especially in AFR, EAP, and LAC. In 18 percent of AFR countries, longer-term training programs sought to re-orient the skills and ethos of civil servants to meet the needs of a market economy. In addition to training, recruitment measures aimed to mobilize scarce skills and improve the overall attractiveness of CS employment. Forty-five percent of these measures were short-term interventions (for example, hiring consultants) designed to fill immediate skills shortages in AFR. The other half of recruitment support had a longer-term, predominantly 10 African focus: the Bank helped governments resettle expatriates in government posts, identify and classify qualified staff, and administer competitive examinations for entry into the CS. 4.8 A third set of capacity building measures aimed to clarify P-A relations and associated information problems. Wage and nonwage incentives were enhanced to compensate civil servants appropriately for the terms of their employment. Typical wage enhancements comprised selective salary supplements for high-priority posts in AFR, LAC, and EAP. Notably, most wage interventions were limited to studies and reviews of the real value of allowances and benefits. In a small number of countries, mostly in AFR, the Bank sought to clarify job classification and grading systems by streamlining the number of grades between priority and lower-skilled posts. Performance-oriented measures such as merit-based promotion and performance evaluation were undertaken almost exclusively in AFR countries during the 1987-93 period. In addressing informational problems, the Bank financed complementary inputs such as computers to improve information management. About 25 percent of these interventions helped build information systems to support payroll and personnel linkages. 4.9 Reforming Institutions. Despite some experimental tools developed by staff, the Bank did not operationalize a diagnostic agenda to diagnose elements of country institutional endowment that influence CSR performance. But, the Bank did support the revision of intrapublic sector rules that directly and indirectly affected civil servants (for example, redrafting CS labor codes, staff regulations, and bidding arrangements). A second set of interventions- those with a more deliberate governance focus-aimed to strengthen external organizations that monitored and limited arbitrary action. Measures to enforce external checks and balances included strengthening the judiciary, enhancing the public's access to information, and establishing independent oversight bodies. 4.10 Supporting Implementing Agencies. Some of the Bank's capacity building efforts were designed to help core ministries (such as the Ministries of Finance or Ministries of Public Service) to coordinate CSR efforts more effectively. This marked the extent of Bank support for building governments' strategic capacity to plan and consolidate public sector reforms. Strategy Implementation: Lending and Nonlending Services 4.11 Lending-Trends and Allocation. Between 1980 and 1997, the Bank supported 180 lending operations with CSR components. The sample for OED's review covered 124 of these loans, which contained a total of 350 interventions. Over the period of review, the majority of loans (and therefore interventions) were committed to AFR countries.12 By the 1990s, countries in LAC (the second largest recipient), as well as in EAP, ECA, and MNA, became increasingly important clients (Annex 5). Fifty-five percent of operations were supported by adjustment lending, primarily through structural adjustment loans (SALs). Of the remaining investment operations, a third consisted of technical assistance loans (TALs). The majority of downsizing (68 percent) and institutional reforms (58 percent) relied on adjustment lending, while capacity building efforts were split between adjustment and investment operations (table 4.1). 12 Bank data sources did not collect data on commitments of CSR components of operations. As a result, OED relied on the total number of non-stand alone projects to analyze lending trends. 11 4.12 Most CSR interventions were Table 4.1: CSR Interventions by Lending Type designed and implemented as Class of CSR Interventions, 1980-1997 components of larger operations. From Total Adjustment Investment the late 1980s, the Bank did support a Downsizing 85 58 27 growing number of stand-alone (68%) (32%) operations focused exclusively on CSR Capacity building objectives. All 24 stand-alone projects Organizational structures 93 49 44 in the portfolio were investment (53%) (47%) operations, with commitments totaling Quality of human resources 67 27 40 $492.06 million. This was partially (40%) (60%) responsible for the shift in the ongoing Principal-agent relations 48 21 27 portfolio toward investment lending (44%) (56%) across classes. While most closed stand- Reforming institutions 57 32 25 alone operations went to AFR countries, (56%) (44%) the composition of ongoing ones Total 350 187 163 indicates increased activity in LAC. (53%) (47%) 4.13 Innovations in Instrumentation. In recent years, the Bank has attempted to modify disbursement mechanisms in order to allow for greater flexibility during implementation. For instance, SALs to Mali and Mauritania have been restructured to include floating tranches, which ostensibly provide greater flexibility in the implementation of CSR conditions. In addition, it has financed new upstream assessment tools through the Institutional Development Facility (IDF). Between FY93 and FY97, 43 grants totaling $10.77 million supported diagnostic work on the role of the state and ancillary issues such as training, CS rules, and independent oversight institutions. Most grants supported analyses of institutional reform (42 percent) or capacity building (39 percent) issues respectively, particularly in countries with little previous CSR experience. Accordingly, countries in EAP (including China and Indonesia) received over half of total IDF disbursements. It is also worth noting that new instruments such as Adaptable Program Lending (APLs) and Learning and Innovations Loans (LILs) have been developed to support pilot projects that facilitate innovation and learning. Since they are relatively new, these adaptable lending instruments were not part of the study's sample. 4.14 Economic and Sector Work. The Bank's primary vehicle for NLS, ESW comprised Country Economic Memoranda (CEMs), Public Expenditure Reviews (PERs), and sector reports (SRs) issued over the period of review. Of these, CEMs and PERs were the most exhaustive in their treatment of CSR issues. ESW was not only conducted preparatory to lending, it was also used to evaluate past CSRs, particularly downsizing efforts in AFR, LAC, and SAS. 4.15 Aid Coordination and Cofinancing. Several multilateral and bilateral donors such as the Asian Development Bank (ADB), Department for International Development (DFID), United Nations Development Program (UNDP), and the U.S. Agency for International Development (USAID) have supported CSRs over the past two decades. While reliable data on the aid commitments of these donors were not available, discrete issues regarding partnerships were identified. OED found significant support among donors for formal arrangements to coordinate assistance for CSR. A noteworthy example is the CSR Working Group of the Special Program of Assistance (SPA) for Africa, of which the World Bank, IMF, DFID, the Netherlands, Germany, France, and other partners are active participants. Chaired by the UNDP in 1995, the group developed guiding principles for donors seeking to support CSRs in AFR. Some members, such as the Netherlands, have used the guidelines as a benchmark in evaluating CSR in select 12 countries. Similar ongoing and recently completed evaluations by other members (IMF, International Labour Office, UNDP) should provide a more definitive basis for assessing comparative advantage in supporting CSR.13 4.16 On the question of comparative advantage, interviews with key partners revealed broad support for the Bank's continued leadership on the fiscal aspects of CSR, given its conceptual and operational strengths in economic analysis and fiscal adjustment. Alternatively, the UNDP and bilaterals (for example, EU Phare and USAID) were considered to be well-positioned to lead on human resource development and governance, particularly in politically sensitive settings. The viability of this division of labor may be tested as the Bank takes on a more active role in analyzing and combating corruption in client countries. 4.17 The role of the IMF as a key partner for the Bank on CSR requires particular attention and continuous evaluation. The Fund was often involved in implementing various wage policy and tax reform components of Bank-supported CSRs. More important, it shaped the enabling environment in which the Bank functioned. Nearly 61 percent of Bank-supported CSRs were preceded by Fund agreements that required some form of CS downsizing. Country case analyses are required to determine whether the short-term, fiscal objectives of Fund-supported stabilization programs influenced the sequencing of Bank-supported CSRs (for example, compelling the Bank to support downsizing and postpone longer-term institutional reforms). 4.18 Cofinancing of Bank operations provided another opportunity for development partners to work together in supporting CSRs. A total of 35 donors provided cofinancing in the portfolio. Of these, UNDP, ODA (now DFID), USAID, and the IMF were the most active. Prominent regional donors included the African Development Bank (AfDB), the Government of France, and the EU in AFR, ADB and the Government of Japan in EAP, and the Caribbean and Inter- American Development Banks (IDB) in LAC. The participation of regional and bilateral donors as cofinanciers of specific classes of CSR was indicative of the regional character of bureaucratic dysfunction. For instance, the dominance of AfDB and the Government of France as cofinanciers of downsizing efforts reflects the high incidence of overstaffing in AFR. The IDB was prominent in institutional reforms, indicating the willingness of LAC countries to address such issues. 4.19 Participation. The use of participation as an approach to intervention was largely absent from CSR interventions. On the contrary, participation was discouraged due to fears of policy capture by organized interests such as public sector unions. As a result, the Bank endorsed participatory approaches to CSR only in Uganda, where the government had taken the lead in introducing a consultative approach to downsizing. Similarly, the only country where the Bank supported "voice" as an objective of CSR was Uganda, where service delivery surveys (SDSs) were used to monitor the efficiency of public expenditures. 5. Outcomes 5.1 Most CSR operations were components of larger projects. Standard OED project ratings (including outcomes, sustainability, and institutional development impact) wet e not specific to " The ECA Region noted that formal coordination mechanisms alone did not guarantee collaboration and synergy on the ground. 13 CSR performance, except on the 24 stand-alone projects approved after 1987 (Annex 7). Accordingly, OED disaggregated interventions in completed and ongoing operations by CSR class and then rated outcomes vis-A-vis their stated objectives. On average, only 33 percent of completed CSR interventions and 38 percent of ongoing ones were found to be satisfactory. Downsizing efforts outperformed capacity building and institutional reforms in the closed portfolio. In the ongoing portfolio, capacity building interventions were more likely to achieve satisfactory outcomes than dowinsizing or institutional reforms. CSR performance in the most active regions was satisfactory 29 percent of the time in AFR and 39 percent in LAC. Performance in the ongoing portfolio in AFR and LAC improved to 45 percent and 43 percent respectively (Annex 6 and 7). Performance of CSR Measures Figure 5.1: OED Outcome Ratings of CSR Interventions Interventions in Completed Projects 5.2 Downsizing the Civil Service. Nearly half of all completed downsizing interventions and 31 percent of ongoing ones resulted in satisfactory outcomes (figure 5.1).14 When supported by Orgiiimdcm_and_soxam adjustment lending, such interventions performed marginally better. Size reductions proved successful as short-term oommes measures in 40 percent of countries. However, actual cutbacks were not sustainable for several reasons. Typically, the retrenched were rehired or the numbers of ghost workers on payroll increased after downsizing. In some cases, the health and education Interventions in Recently Completed and Ongoing Projects workforce-classified as civil servants- U Y.AY . was expanded to meet sectoral demand for social services (Annex 8).15 5.3 By and large, compensation 01oxii; m a packages (such as, voluntary retirement schemes) overestimated the opportunityhe costs to retrenched workers of leaving the7 CS. Aside from their considerable fiscal cost, retrenchment packages were poorly managed. Governments often failed to properly screen candidates, disseminate information about compensation schemes, or track severance financing. Retraining programs to facilitate the transition of civil *2saisfactory plackof %&E ousatisfactory " In figure 5.1, capacity building interventions are disaggregated into their constituents-organizations and structure, quality of human resources, and principal-agent relations. " As with other classes of CSR, OED evaluated downsizing purely in terms of its stated objectives-reduction in numbers of civil servants. In the future, both downsizing objectives and ex post evaluations should account for the changing human resource needs of governments as well. In some cases, 'rightsizing' may require cutbacks in areas such as licensing, along with significant increases in others such as health, education, and financial management. Haltiwanger and Singh, 1997. 14 servants to the private sector were not implemented. Finally, the poverty and gender impact of downsizing, particularly on lower-level staff, was rarely monitored. Nor were they taken into account in the design of packages. 5.4 While client countries have experienced difficulties meeting wage bill targets, central government wage expenditures as a share of GDP were reduced in nearly 50 percent of client countries (Annex 9). In other countries, the high cost of voluntary retirement and the ineffectiveness of wage bill ceilings ate away at the potential savings expected from downsizing. Finally, wage decompression, designed as part of pay and employment reforms, were not sustainable. 5.5 Building Capacity. Only a third of capacity building efforts achieved satisfactory outcomes as the Bank and its clients faced difficulties monitoring and sustaining organizational design, human resources development, and improved P-A relations. Like size reduction, organizational restructuring, such as consolidation of ministries, took the form of short-term ventures that typically gave way to further proliferation and duplication over time. New budgeting systems were not properly routinized. 5.6 Little information was available on the impact of training efforts to improve human resources. Recruitment strategies to attract skilled staff through wage hikes were marginally effective assuming wage and decompression levels were maintained. However, medium- and long-term efforts, such as the resettlement of expatriates and the creation of Senior Executive Services and enclaves, proved counterproductive. The latter offered new channels for the continuation of patrimonialism. 5.7 Only a third of completed interventions aiming to improve P-A relations were satisfactory. Of ongoing operations, which relied more heavily on investment lending, nearly half achieved their objectives (Annex 7). Selective salary adjustments not only generated resentment, but were also vulnerable to adverse macroeconomic trends such as hyper-inflation. Other measures to improve the terms of civil servants' employment -monetization of benefits, regrading and job reclassification, and performance-orientation-were rarely implemented. 5.8 Reforming Institutions. Typically, intrapublic sector regulation (CS codes, for example) were revised and passed as law without being routinized in government agencies. The high percentage of institutional reform interventions with little or no monitoring and evaluation (M&E) illustrates the difficulty in measuring changes in rules and behavior. Finally, initiatives to establish autonomous oversight bodies failed to provide sustained and credible checks and balances on the activities of civil servants. In some cases, these autonomous oversight bodies created increased opportunities for abuse. 5.9 Coordinating Reforms. Attempts to designate a central agency to coordinate CSRs were mixed. Again, the scarcity of reliable information prevented adequate M&E of the effectiveness of central implementing agencies. 6. Assessing Strategic Capacity and Relevance 6.1 Section 6 examines the Bank's upstream work to evaluate the extent to which it did the "right things" in addressing bureaucratic dysfunction. This includes both substantive issues and 15 upstream processes (ESW, SARs, diagnostic tools) that shaped the Bank's strategic capacity to design relevant CSRs. Any evaluation of relevance should take into account the dynamic policy environment in which the Bank formulated its CSR strategy. Between 1980 and 1997, three factors conditioned the policy environment for CSR-(i) the fiscal imperatives of adjustment programs; (ii) the compelling example of the NPM in the OECD; and (iii) conceptual advancements in the NIE, public choice, and managerial economics, with applications to policy. Helping Governments Do More with Less 6.2 Given the existing state of knowledge about CSR in the early and mid-i 980s, the Bank's strategy of "helping governments do more with less" was relevant. The early CSRs' pointed focus on achieving the fiscal and economic management objectives of adjustment provided the original economic rationale for Bank involvement in the reform of the core public sector. More broadly, this early strategy pioneered the notion that governments in poor countries could simultaneously limit CS growth and improve policymaking capacity and operational efficiency. In so doing, it expanded the scope of Bank-country policy dialogue in three ways. It established the first indicators for bureaucratic quality, provided fiscal impetus for enhancing efficiency, and improved empirical knowledge of public administration in poor countries. 6.3 First Indicators ofBureaucratic Quality. In the mid-1980s, the development community was focused on disengaging the state from productive activities. The prevailing view of the role of the state was exceedingly minimalist.16 In this context, the Bank's early concerns about the fiscal costs of overstaffing first made bureaucratic quality a legitimate development issue. Wage bill data provided a straightforward way of measuring bureaucratic quality; this soon became standard practice in ESW and appraisals. 6.4 Providing the Impetus for Enhancing Efficiency. The fiscal focus of Bank support also provided the impetus to examine the allocative and technical efficiency of personnel expenditures. This represented an attempt to create conditions analogous to those which precipitated NPM reforms in the OECD.17 6.5 Improving Empirical Knowledge ofPublic Administration. Finally, Bank analyses conducted in preparation for early CSRs considerably enhanced the state of empirical knowledge on public administration in poor countries. Specifically, Bank ESW in the early 1980s helped dispel the conventional view that in AFR, "too many government employees [were] getting paid too much."18 Instead, it established that AFR governments actually suffered from severe wage erosion and compression in the midst of rampant overstaffing. These findings paved the way for the first generation of pay and employment reforms, which continued into the next decade. They also demonstrate the potential long-term impact of ESW and other forms of upstream assessment in shaping the larger development agenda in client countries. Limitations of "Doing More with Less" 6.6 In spite of breaking new ground, the relevance of the Bank's strategy was hampered in two ways. First, irregular application of diagnostic tools weakened the Bank's strategic capacity 16 Lindauer and Nunberg, 1994, p. 1. 1 NPM reforms typically emerged in response to fiscal crises and the imposition of hard budget constraints. Nye et al., 1997. * Lindauer and Nunberg, 1994, pp. 4-5. 16 to design relevant interventions. Second, interventions were not sufficiently focused on enhancing incentives to perform. 6.7 Irregular Assessments. Diagnostic tools such as CS Table 6.1: Bank Response Rates by Class of Intervention (Percent) censuses, functional reviews, and Right No Shots in strategic audits were often poorly Class of CSR Measures response response the dark sequenced and erratically applied, particularly in AFR. This Downsizing attenuated linkages between ex Pay and employment ante assessment and project Size reduction 57 14 29 design, as evidenced by the 30 Wage decompression 68 23 9 percent of size reductions that Compensation of losers 35 6 58 were "shots in the dark," and 14 Capacity Building percent that constituted missed Organizations and structures to reform (see table Streamlining 84 3 13 opportunities Linkages 68 2 30 6.1). One reason for the Human resources irregularity of diagnostic work Training 53 5 42 was the choice of lending Recruitment 56 9 35 instruments. Reliance on fast- Information/P-A relations disbursing adjustment lending for Salary supplements 65 3 32 CSR gave the Bank and its clients Job classification 46 15 39 a short time horizon to identify Performance oriented 57 7 36 opportunities for downsizing, Measures capacity building, or institutional Inputs and technology 54 5 41 reform. Without sufficient time to Reforming Institutions diagnose bureaucratic Intrapublic sector regulation 42 2 56 dysfunction, selectivity suffered: Increasing accountability Anti-corruption measures 50 0 50 interventions supported by Voter control 27 9 64 adjustment lending registered a Independent institutions 50 13 37 higher percentage of "shots in the dark" (40 percent) than those Note: OED calculated the Bank's response rates for different classes of CSR interventions in order to gauge its strategic capacity to respond to diagnoses of bureaucratic dysfunction. A supported by investment lending "right response" denotes an intervention that is appropriate to a Bank diagnosis (for example, (20 percent). downsizing in response to overstaffing). "No response" indicates a missed opportunity to reform. Finally, a "shot in the dark" is an intervention that is not responding to any particular diagnosis. This approach, originally used in OED's review of financial sector reform, does 6.8 The Bank did not not identify those cases in which the Bank's diagnoses were inaccurate. When diagnoses of establish a system of regular, dysfunction are inaccurate, responses may be "right" (as defined above), but irrelevant. institutional assessment upstream of CSR programs or projects. This would have allowed country teams to build knowledge on CS issues and track institutional trends over time. As anecdotal evidence from Albania suggests (box 6.1), the exigencies of unexpected downsizing were easier to manage when relevant and timely information on state institutions, usually developed in ESW, was accessible. 6.9 At the same time, high-quality ESW alone did not guarantee improvements in Bank response rates on specific interventions. The strongest linkages between ESW analyses and lending objectives were in Uganda.19 Yet cutbacks in that country were still undertaken in the ' OED examined the extent to which ESW analyses and recommendations were closely linked to the objectives of lending operations. The strongest links were observed in Uganda; the weakest were in Laos, Peru, and Sri Lanka. 17 dark because existing ESW analyses were either insufficiently detailed or CS censuses funded under projects were improperly sequenced after retrenchments. This suggests that broad ESW analyses can guide the overall strategy of CSR and project identification. But, project-specific assessments (CS censuses, functional reviews) need to be routinized and integrated into appraisal and design processes in order to prevent shooting in the dark scenarios. 6.10 Weak on Incentives and Performance. Four lacunae in the Bank's approach to incentives undermined its ability to help governments "do more" with less. First, the Bank responded to endemic problems of migrating talent and poor performance caused by low real wages only 46 percent of the time. In some cases, this lack of response was the result of deliberate decisions to sequence macroeconomic stabilization before wage enhancements, since the success of the latter depends on controlling inflation. In other cases, the portfolio's poor response rate on low real wages can be explained by political resistance to wage decompression, scarcity of reliable information on CS pay and personnel systems, and internal biases in the Bank that gave priority to fiscal savings. 6.11 Second, the fair comparison rule was unsatisfactorily applied to both private and donor sectors in determining the true reservation wage of qualified civil servants. Even though wages in the donor sector were more appropriate comparators to those in the CS, they were not included in fair comparison analyses in ESW.20 Nor did the Bank's analyses use a standard methodology for making these comparisons. For instance, in AFR, EAP, and MNA, it compared average wage differentials over time while in LAC, MNA, and SAS it compared wages for specific skills sets (accounting for example). 6.12 Third, the Bank was particularly reluctant in supporting stronger linkages between wages and job descriptions or more important, performance. CSR interventions rarely supported even the most basic forms of performance evaluation, much less merit-based pay or promotion. While empirical work on performance-based compensation in poor countries is weak, anecdotal evidence suggests that such measures can result in marked improvements in organizational performance.21 Prevailing skepticism about performance orientation in developing countries on the part of the Bank and other MDBs may be premature. More systematic evaluations of these experiments will provide much-needed insight into the viability of the NPM in countries where formal Weberian bureaucracies have still not taken hold. 6.13 Finally, CSRs adopted a narrow view of incentives: myriad factors motivating government employees were conflated to the singular question of wage level. Interventions did not seek to foster organizational learning, workplace transformation, or cultural change, all of which are prominent in the literature on strategic management.22 Other issues highlighted by research in sociology--esprit de corps, the power of personal commitment, and the social value of civil servants' work-were also omitted from Bank strategy.23 20Lienert and Modi, 1997. 21 Klitgaard, 1997. 22 Tendler, 1998. 23 Kim, 1993. 18 Box 6.1: The Value of Upstream Assessment and Analysis of State Institutions in Albania In 1996, the Bank's ECSPE had prepared a background paper on pay and employment issues in Albania, which was intended to be published as part of a 1997 CEM or PER. The paper contained detailed scenarios of public sector wage levels and wage bill levels, as well as analyses of grading systems and employment structure based on a computer-based model of personnel-pay linkages. It also identified overstaffing in the Ministries of Agriculture, Public Works, Culture, Youth, and Sport, as well as three tiers of regional and local government. Despite its strategic importance, the ECSPE paper was not formally disseminated, because both the 1997 CEM and PER were canceled. In the wake of the 1997 collapse of pyramid schemes and the ensuing civil crisis, the Government of Albania (GOA) committed itself to downsize 10 percent of central government staff by the end of 1998 as part of an IMF stabilization and recovery program. Specific targets of 1,000 cuts within a month and 10,000 in the following three months were established under the Fund's agreement. Unfortunately, this timetable left no room for conducting functional reviews and other types of strategic audits of government organizations. Without any established civil service census data or functional reviews of ministries, the GOA faced the prospect of conducting major administrative surgery virtually in the dark. As the GOA scrambled to meet the conditions of the Fund's program in late 1997, the once forgotten ECSPE paper provided the only rational basis for the GOA to negotiate the "rightsizing" of ministries with the Fund. Source: World Bank, "Albania Country Assistance Review," 1998. Eroding Relevance in the 1990s 6.14 In the early to mid-1990s, the Bank's "doing more with less" formula remained fixated on the budgetary, organizational, and compensatory symptoms of dysfunction, instead of addressing its institutional causes. Even when the ESW had identified them, lending operations did not seek to remedy the institutional causes of overstaffing or the proliferation of ministries. For instance, the policy dialogue often attributed overstaffing to patrimonialism (Sri Lanka, Uganda), popular beliefs about the role of government as an employer of last resort (Jamaica, Kenya, Senegal, Sri Lanka), and interest group pressures (Kenya). Yet downsizing interventions rarely attacked the informal patron-client relationships that often drive CS growth. Nor did they support upstream efforts to monitor populist views of government, educate the broader public, or help governments negotiate with organized interests opposed to reforms. 6.15 On the one hand, the prevailing policy environment certainly slowed the Bank's response to the institutional causes of bureaucratic dysfunction. Open discussion and systematic analysis of patronage and corruption were virtually nonexistent. In addition, some task managers indicated that the "approval culture" and review process for CSR operations tended to discourage innovations in engaging the institutional and political factors. 6.16 On the other hand, the reasons for the Bank's eroding relevance in the 1990s was more likely conceptual than contextual. Although its own CSR literature had noted the value of fields such as transactions costs analysis, the NIE, public choice theory, and managerial economics, the Bank was slow to operationalize the new thinking on the economics of institutions.24 This new thinking had enabled NPM reformers in New Zealand and other OECD countries to identify more efficient and mutually binding arrangements between senior managers and civil servants in policymaking, service delivery, and regulation. It also helped establish more transparent and accountable arrangements between bureaucrats and their political principals. A growing body of " Schiavo Campo, 1994. 19 empirical research suggested that institutional quality could be rigorously monitored and compared at the country level. 6.17 The operational implications of these conceptual advancements for CSR were twofold. First, standard assessments of state institutions could be devised to better fit good practice CSRs to local contexts. Second, "feasibility analyses" based on analysis of political institutions could improve selectivity by ascertaining the likelihood that CSRs would be adopted and sustained. 6.18 Localization of CSR. The academic literature across disciplines indicates significant diversity in institutional endowment across countries, even those within the same region. For instance, administrative systems in AFR countries typically derive from either Anglophone or Francophone traditions, which in turn influence bureaucratic culture, policymaking processes, and social interactions between civil servants and citizens. Elements of country institutional endowment also share similarities within the same region. An obvious example is the oft-cited "crisis of statehood" in AFR, which has its roots in the colonial practice of arbitrarily drawing national boundaries without regard for historically established ethnic or social affiliations.25 By the early 1990s, the Bank's own literature acknowledged these local and regional similarities in public administration and recommended that CSRs account for them.26 6.19 Yet there was little in the CSR strategy for either ECA or LAC that distinguished it from the approach used in AFR. Nor did the Bank accommodate its CSR strategy to the level of institutional quality in individual countries. Annex 10 illustrates how the Bank used the same CSR "blueprint"-with its "downsizing-capacity building-institutional reform-downsizing" sequence-across countries regardless of their administrative system or level of institutional quality. For instance, two distinct AFR countries, Uganda (Anglophone, high quality) and Cameroon (Francophone, low quality), undertook downsizing and capacity building in the late 1980s, followed by institutional reforms in the early to mid-1990s. By the mid-1990s, both countries repeated another wave of downsizing, because the earlier efforts to cut CS size were not sustained. The recurrence of this "blueprint" in different country and regional contexts reveals the need for better knowledge of institutional endowment during strategy formulation. 6.20 Feasibility Analyses. It was only in FY99 that PREM started to develop standard, upstream mechanisms by which political "realities on the ground"-bureaucratic politics and the broader political economy of reform-influence decisions to support CSRs. During the 1980-97 period, neither political analyses nor existing tools such as social assessments were used to build upstream knowledge of patronage systems, custom, and other social factors. Often interventions were undertaken without analyses of the likelihood that coalitions would form to overturn reforms. Notably, policy reversals often frustrated downsizing interventions, and sometimes completely undid their effects (as evidenced by rehiring).27 25 World Bank, 1997; Davidson, 1993. 26 Dia, 1995. 27 Haggard and McCubbins, 1997. 20 7. Assessing and Understanding Efficacy 7.1 Having examined the upstream determinants of relevance in Section 6, the section analyzes downstream factors that shaped the efficacy of Bank support. OED relied on supervision reports, ICRs, PARs, and ESW analyses of past CSRs for this part of the review. Assessing Efficacy 7.2 Overall, Bank support was not effective in achieving and sustaining desirable CSR outcomes. Downsizing and capacity building efforts in particular were unable to produce permanent reductions in CS size, while simultaneously overcoming capacity constraints. There was no evidence that civil servants and politicians began in any meaningful way to "own" and follow formal rules such as codes of ethics. As a result, institutional reforms were unable to substantially limit arbitrary action by civil servants or political interference in CS management. Understanding Efficacy: Factors Shaping CSR Performance 7.3 Efficacy hinged on the Bank's ability to reinforce incentives for senior bureaucrats and politicians to carry out CSR-that is, to downsize, enhance capacity, and refrain from arbitrary action. From this vantage point, OED identified four key variables that shaped CSR performance during implementation-(i) the quality of information on CSR performance, (ii) the role afforded to strategic management and cultural change, (iii) underlying coordination arrangements or checks and balances, and (iv) contextual constraints. The Bank's tendency to overlook these four variables was indicative of a technocratic rather than systems approach (see table 7.1). Table 7.1: Contrasting Technocratic and Systems Approaches to Civil Service Reform Technocratic Approach Systems Approach * View CS narrowly as a bureaucratic entity * View CS as a dynamic system of P-A relationships within the core public sector. that are mediated through multiple variables * CSR as unilateral policy interventions affecting * CSR as attempts to influence multiple variables to discrete issues such as the wage bill, skills, etc. impact P-A relationships within the CS system * Assistance without a framework for intervening * Assistance based on a framework for intervening in administrative systems that acknowledges the Bank's role as an active participant in the process of institutional change * Reliance on a small circle of interlocutors * Reliance on a broad array of stakeholders inside within central implementing agencies and outside government 7.4 Technocratic support was not based on a conceptual framework for intervening in administrative systems. The Bank overlooked the ways in which its own interactions with government officials-mediated through processes, tools, and instruments-influenced the P-A relationships within CS systems. In other words, the Bank did not recognize its own role as a participant in the process of institutional change. CSRs were conceived as unilateral policy actions that aimed to achieve long-term objectives by engaging only a few interlocutors within core ministries. By contrast, a systems approach would afford greater self-awareness on the part 21 of the Bank; increased attention to downstream processes; and, ultimately, more avenues for achieving desirable CSR outcomes. 7.5 The impact of each variable on incentives for bureaucrats and politicians to carry out CSRs is described below and summarized in table 7.2. Information on CSR Performance 7.6 With the exception of data on discrete fiscal issues, the generally poor quality of information on CS systems posed a major obstacle to reform efforts. The absence of standard indicators on CSR performance bred information asymmetries during implementation. Resultant problems of moral hazard and adverse selection made it difficult for borrowing governments to hold senior bureaucrats and other civil servants accountable to reform objectives. 7.7 Lack of Standard Indicators for Bureaucratic Performance. Monitoring and evaluation (M&E) of CSR performance-especially on institutional reforms and ongoing downsizing efforts-was uniformly poor. ICRs and PARs did not report CSR outcomes for 17 percent of closed and 25 percent of ongoing interventions. This can be attributed in part to inadequate supervision. While data for CSR components were not available, OED found that supervision intensity on closed stand-alone projects (discounting outliers) fared poorly in comparison with Bankwide averages. In addition, the Bank failed to operationalize portfoliowide performance indicators to aid in monitoring, supervision, and ex post evaluation. It is worth noting that ESW in AFR, LAC, and SAS did compensate by analyzing past Bank support along with data CS size, wage expenditures, and capacity building (Annex 11). These evaluative analyses in ESW, however, did suffer from various shortcomings. 7.8 First, lags between the completion of operations and publication of ESW meant that NLS was inadequate for "real-time" M&E. Second, indicators used in ESW were generally input measures such as wage expenditures or the number of personnel trained. Monitoring of agency outputs or sectoral outcomes of CSRs was absent. Third, these indicators were not standardized across countries: CS data in some countries was far more advanced than in others, thereby limiting opportunities for cross-country comparisons. Ultimately, efforts to build evaluation capacity on CS issues (for example, support for information systems to monitor pay and personnel data) were often de-linked from the underlying incentives of bureaucrats to collect and share information. This further weakened monitoring of downsizing and capacity building, most of which were already low-priority components of SALs. 7.9 Poor M&E were even more conspicuous in institutional reforms. To some extent, this can be explained by the inherent difficulties in observing and evaluating changes in the "rules of the game." Yet the Bank did not systematically pursue alternative techniques or methods (for example, attitude surveys or participatory M&E) for monitoring institutional change within the core public sector. There is no evidence of increased reliance on field-based supervision in order to better capture behavioral or cultural change within the CS. 7.10 Moral Hazard and Adverse Selection. Asymmetric information between civil servants and senior bureaucrats (and the executive) during the implementation of downsizing and capacity building efforts prevented reforming governments from holding bureaucrats accountable to CSR objectives. Moral hazard on the part of anti-reformist bureaucrats went unchecked, because the Bank as well as implementing bodies lacked reliable information on CS size. In some AFR and 22 LAC countries (Guinea, Jamaica, Uganda), ministries removed ghost workers rather than actual workers during retrenchment exercises. In other cases (Ghana, Senegal), gross rather than net reductions were reported as the rehiring of retrenched workers went unreported. Without information on real changes in the CS workforce, the Bank was unable to determine whether clients had met downsizing conditionalities. Table 7.2: Factors Shaping CSR Performance Contextual Constraints Strategic Checks and -Macro Labor Institutional Information Management Balances Markets Endowment INCENTIVES TO BUREAUCRATS To leave the civil service (downsizing) V / V / To join the civil service (capacity building) V V To perform (capacity building) V V / To abstain from arbitrary action (institutional reforms) VV INCENTIVES TO POLITICIANS To hire and fire (downsizing) To hold bureaucrats accountable to policymakers (capacity building) VV To abstain from political interference (institutional reforms) V/ V 7.11 In the absence of clear linkages to monitorable CSR performance criteria, arrangements between managers and civil servants were also not mutually binding or reciprocal. Rewards were not granted in exchange for observable improvements in performance. As a result, wage enhancements, grade decompression, and training were subject to opportunism on the part of bureaucrats. For instance, some managers were able to use rewards for purposes other than their stated objectives (for example, the use of accelerated promotions in Ghana, Kenya, and Senegal to raise salaries during wage freezes). These dynamics exacerbated the very problems that CSRs sought to correct-most notably, excessive wage expenditures. 7.12 The second deficiency, adverse selection, skewed attempts to encourage civil servants to retire early and compensate retrenched civil servants. Voluntary retirement programs such as those in Senegal and Sri Lanka precipitated the exit of qualified and competitive civil servants, while their less-effective counterparts stayed behind.28 Severance packages typically 28 DEC research on public sector downsizing attributes adverse selection in voluntary retirement programs to the problem of unobservable performance. In such cases, poor performers, who stand to lose the most by leaving, elect to stay. Barring immediate improvements in performance measures for civil servants, DEC recommends self-selection mechanisms that provide incentives for the best workers to reveal their opportunity costs of staying or their "worker type" (lazy or hard-working). Examples include auctions in which workers indicate (under seal) the amount required for them to leave, or the option for civil servants to switch from permanent to higher-paid fixed-term contracts. Rama, 1997. 23 overestimated the compensation needed to encourage civil servants to leave (Benin, Ghana, Senegal). In both cases, Bank operations exaggerated the opportunity cost to civil servants of staying in government service. Strategic Management and Cultural Change 7.13 Capacity building efforts were based on a narrow understanding of incentives as wage enhancements. They did not allow for the use of strategic management (including the reorganization of work) and cultural change as important ways of strengthening incentives to perform. This allowed the Bank inter alia to avoid acknowledging the ways in which its own approach to intervention (that is, use of processes, tools, instruments) affected the incentive environment in government agencies. For instance, CSR interventions between 1980 and 1997 rarely identified and encouraged potential "change agents" to lead institutional and culture change within government agencies.29 Nor were ICRs or PARs particularly attuned to the impact of conventional Bank implementation strategies-heavy reliance on consultants or the establishment of financially endowed PMUs, for example- on the credibility of standard operating procedures and the morale of civil servants. 7.14 In addition, CSRs afforded public sector managers in borrowing governments a very limited role in entrepreneurship and leadership. Therefore, in its application, the Bank's "do more with less" strategy essentially compelled managers to be efficient, without adequately empowering them to be effective. Skepticism about granting managerial autonomy in the public sectors of developing countries is based on concerns among development policymakers that increased flexibility may increase corruption, patronage, and arbitrary action. By contrast, NPM reforms in the OECD typically combined fiscal austerity with greater autonomy for managers to innovate and improve the ways government does business. Coordination Arrangements, Checks and Balances 7.15 Bureaucratic dysfunction often persisted or recurred even after the implementation of CSRs because underlying checks and balances on hiring and performance and related coordination arrangements within the core public sector were not sufficiently supported by Bank interventions. Recent and ongoing Bank research suggests that problems of overstaffing and the proliferation of ministries result from coordination problems between budgeting and personnel departments within the core. Poorly defined authority over the hiring of civil servants induces senior managers to bargain against both departments to maximize their budgets by increasing either establishment levels or personnel budgets. 7.16 Downsizing efforts also were rarely buttressed by external checks on politicians for whom CS hiring offered a primary means for dispensing patronage. For instance, Bank insistence on public reporting of hiring practices in selected countries would have helped create broader support for merit-based hiring. For the most part, the Bank did not recommend external checks and balances on service delivery agencies, particularly in sectors such as health, education, and the environment. When deployed as complements to Bank-supported restructuring (for example, the creation of enclaves and autonomous agencies), voice and public reporting encourage stakeholders to apply pressure on civil servants to perform. In some cases, " Schiavo-Campo, 1996. 24 these accountability mechanisms also provided "institutional-light" options for ministries or autonomous agencies to better match their role to their limited capabilities (see box 7.1).30- Box 7.1: The Benefits of External Checks and Balances in Low-Capacity Environments: Contrasting Bank Support for Indonesia's BAPEDAL and Uganda's Investment Authority External checks and balances improve the efficacy of restructuring and capacity building efforts by providing alternatives for government agencies to carry out core functions even in the midst of resource and capacity constraints. Mechanisms such as public reporting and report cards provide the necessary incentives for civil servants not only to be efficient, but also accountable to constituents. The benefits of opening the doors of agencies with weak capacity were apparent in Bank support for Indonesia's autonomous environmental management agency, BAPEDAL. Under its Program for Pollution Control, Evaluating, and Ratings, BAPEDAL publicized ratings of polluter compliance to emission standards. This provided an impetus for communities, NGOs, and private investors to apply direct pressure on polluters to comply. It also generated significant demand for regular dissemination of rigorous ratings by BAPEDAL. By contrast, the Bank-supported Uganda Investment Authority (UIA) relied on traditional hierarchical management to carry out its activist agenda of investment and export promotion. It soon became apparent that UIA had become overly ambitious, burdened by redtape, and unaccountable to constituents such as the private sector, NGOs, and the news media. Even though its role outstripped its institutional capability, UIA did not enlist the efforts of external partners to promote investment. Nor did it use public reporting or voice mechanisms to solicit feedback on its own performance. Source: De Silva and Gupta, 1997. 7.17 Although findings on the relative effectiveness of lending instruments for institutional reforms were inconclusive, it was clear that neither investment nor adjustment operations were designed to enlist broad groups of stakeholders in changing norms within the CS or public expectations of government. Nor did operations actively support dissemination of information on CS performance, which could help reform constituencies hold governments accountable to CSR objectives. ECA's ongoing attempts to use workshops, surveys, and public reporting to combat corruption in Albania, Georgia, and Latvia may provide useful examples of participatory, nontechnocratic ways of supporting institutional reforms.31 Contextual Constraints 7.18 Macroeconomic and Labor Market Conditions. ICRs and PARs acknowledged that macroeconomic and labor market conditions regularly undermined the outcomes and sustainability of pay and employment reforms. Inflationary pressures, in particular, eroded real wages, despite efforts to decompress and enhance salaries. This indicates that macroeconomic stabilization and adjustment were prerequisites for sustainable downsizing and capacity building. Alternatively, macroeconomic crises exacerbated problems of bureaucratic dysfunction: the falling standard of living encouraged qualified personnel to migrate abroad. Once they left, client countries such as Ghana, Jamaica, and Uganda found it especially difficult to induce skilled personnel to return (para. 7.15). 7.19 Labor market dynamics eroded attempts to attract highly skilled staff, while simultaneously cutting levels of government employment. Steady increases in the number of new job entrants, partially a product of population growth (Kenya) generated strong interest 3 Paul, 1998; World Bank, 1997. 31 PREM, 1998. 25 pressures to increase CS employment. Governments also faced the ancillary problem of wage differentials between the CS and competing sectors (the donor sector, domestic and international private sectors), which further encouraged the flight of senior staff with training in economics and finance. 7.20 While nearly all ICRs and PARs noted the significance of macroeconomic conditions, evaluations of only 15 percent of projects mentioned unfavorable labor market conditions as a factor shaping CSR outcomes. This represented an apparent disconnect with evaluative work in ESW for LAC, AFR, and SAS, which gave more weight to unfavorable labor market factors in undermining CSRs. Neither ESW nor M&E integrated traditional budget scenarios with analyses of labor market trends. More comprehensive reform scenarios would have provided a means for setting establishment numbers in accordance with both the functional requirements of government and the projected supply of qualified workers. 7.21 Institutional Endowment. Thirty-six percent of operations noted the importance of borrower ownership. ICRs and PARs took "verbal expressions of support" at face value as credible indicators of commitment. Although M&E was silent on institutional endowment, the literature indicates that social and political factors significantly influence administrative reform. For instance, in countries where formal political appointments are relatively deep, attempts to base retrenchments or wage hikes on performance and merit criteria will likely meet resistance from political patrons and their clients. Moreover, issues of ethnicity, ideology, and gender are almost always at play in CSR. Yet analyses of the political economy of institutional change, including the social calculus of CS staffing (for example, ethnic and gender diversity) were not used either in reform scenarios or in ex post evaluation. Time Horizons for Reform and Sequencing of CSR Classes 7.22 The time required to reform public sector institutions varies significantly with circumstances. The time horizons for Bank-supported interventions were predetermined by the choice of lending instrument. Outcome ratings suggest that institutional reforms fared better with time-intensive investment lending, while downsizing and capacity building efforts benefited from the shorter intervals afforded by adjustment lending. 7.23 The sequencing of different classes of CSR is another important area in which the Bank and its clients are still learning. Theory suggests that checks and balances, which make governments more "open and accountable," will also compel them to be cost-efficient-to "do more with less." Anecdotal evidence from "success stories" (Uganda) also indicate that the traditional sequence of CSR interventions (para. 6.18) should be reversed so that initial reform of institutions increases the likelihood of sustainable efforts in downsizing and capacity building. 8. Findings and Recommendations 8.1 The review found that Bank-supported CSRs were largely ineffective in achieving sustainable results in downsizing, capacity building, and institutional reform. This was, in part, due to significant political difficulties in implementing CSRs. Yet the relevance and ownership of reforms were also weakened by a technocratic approach that failed to mainstream institutional 26 analysis and develop a coherent framework for intervening in administrative systems. The Bank should take the following steps to promote lasting improvements in public sector performance. 8.2 The objectives of "bundled" CSR activities were poorly defined. These ambiguities have weakened diagnoses of bureaucratic dysfunction, strained strategic linkages to larger development objectives, and delayed timely measurement of CSR performance. For the purposes of M&E, the Bank should develop a system of categorizing CSR interventions into three classes based on their primary objective--downsizing, capacity building, or institutional reform. In addition, the Network should standardize performance indicators for each class of CSR in order to monitor impact. 8.3 The failure to systematically link capacity building efforts to monitorable performance criteria represents a marked gap in the current strategy. Future CSR operations should systemically link capacity building interventions such as training to job descriptions and monitorable performance of civil servants and their respective units. 8.4 Budget scenarios used to guide downsizing interventions rarely incorporated labor market analyses or institutional assessments. The selectivity of interventions and complementarity of different classes of CSR were undermined by the absence of comprehensive models of CS systems to guide operations. To enhance selectivity and better integrate CSR classes in project design, operations should be prefaced by institutional assessments of administrative systems and analyses of labor market trends in addition to budget scenarios. 8.5 The Bank often assumed that the instruments, processes, and time used for adjustment programs were appropriate for all classes of CSR, including institutional reforms. In addition, it frequently designed CSR interventions as components of larger adjustment operations, which contributed to their low-priority status. Finally, the absence of a participatory approach in either design or implementation undermined ownership of CSRs by stakeholders inside and outside government. In countries where the Bank supports serial CSR projects and comprehensive reform programs, interventions should be designed as stand-alone projects and supported by lending instruments that allow adequate time for implementation (for example, investment or adaptable lending for institutional reforms). The Bank should rely more on field- based supervision to better monitor institutional change. It should also employ participatory processes to nurture reform constituencies in government, the private sector, and civil society. 8.6 Some progress was made in establishing formal arrangements among donors that support CSRs. Yet the comparative advantage of the donors in supporting different classes of CSR on the ground remains an open question. The Bank should use ongoing evaluations of CSR support by other multilateral and bilateral donors as a basis to more clearly define its strategic role in different contexts. Where it does not have a comparative advantage, the Bank should abstain or play a supporting role. 8.7 Developing and transition countries have limited experience with the New Public Management (NPM). The development community urgently requires an empirical basis for supporting NPM reforms-including the granting of managerial autonomy, performance-based compensation, and results-based management-in weak institutional settings. The Bank should help to pilot comprehensive NPM reforms in three to five countries that are committed to developing results-based civil services. These pilots, which can be supported by the new adaptable lending products, should be monitored closely to identify good practice. 27 Bibliography Boston, Jonathan, ed. (1991). Reshaping the State: New Zealand's Bureaucratic Revolution. New York: Oxford University Press. Campos, Ed, and Sanjay Pradhan (1996). "The Impact of Budgetary Institutions on Expenditure Outcomes: Binding Governments to Fiscal Performance," Public Economics Division and Finance and Private Sector Development Division, Policy Research Department, Washington, D.C.: World Bank. Chong, Albert, and Cesar Calderon (1997). Empirical Tests on the Causality and Feedback between Institutional Measures and Economic Growth. College Park, MD: IRIS Center, University of Maryland. Davidson, Basil (1993). The Black Man's Burden: Africa and the Curse of the Nation State. New York: Random House. De Silva, Migara, and Poonam Gupta (1997). "A Comprehensive Assessment of Civil Service Reform in Bank-Supported Lending," OEDCR, Washington, D.C.: World Bank. Dia, Mamadou (1995). Africa's Management in the 1990s and Beyond: Reconciling Indigenous and Transplanted Institutions, Washington, D.C.: World Bank. (1996). "Civil Service Reform: The African Experience," in Civil Service Reform in Latin American and the Caribbean. World Bank Technical Paper Number 259, Washington, D.C.: World Bank. Eggertson, Thrdinn (1990). Economic Behavior and Institutions. Cambridge, U.K.: Cambridge University Press. Girishankar, Navin (1997). "A Review of Civil Service Reform in World Bank Economic and Sectoral Work," OEDCR, Washington, D.C.: World Bank. Girishankar, Navin, and Migara De Silva (1998). Strategic Management for Government Agencies. Discussion Paper No. 386, Washington, D.C.: World Bank. Haggard, Stephan, and Matthew D. McCubbins, eds. (1997). "Political Institutions and Determinants of Public Policy: An Introduction," in Political Institutions and the Determinants ofPublic Policy: When Do Institutions Matter? San Diego, California: University of California. Haltiwanger, John, and Manisha Singh (1997). "Cross Country Evidence on Public Sector Retrenchment," Shrinking Smartly: The Public Sector Retrenchment Nexus, Washington, D.C.: World Bank. Haque, Nadeem UI, and Stephen Sheppard (1998). "Public Sector Efficiency and Fiscal Austerity," Working Paper, Washington, D.C.: International Monetary Fund. Heaver, Richard, and Arturo Israel (1987). Country Commitment to Development Projects. Discussion Paper No. 4, Washington, D.C.: World Bank. Johnson, Ronald N., and Gary Libecap (1994). The Federal Civil Service System and the Problem ofBureaucracy: The Economics and Politics ofInstitutional Change. Chicago, IL.: University of Chicago Press. 28 Kim, Hyung-Ki (1993). The Japanese Civil Service and Economic Development: Lessons for Policy Makers from Other Countries," Economic Development Institute, Washington, D.C.: World Bank. Klitgaard, Robert (1997). "Cleaning Up and Invigorating the Civil Service," Public Administration and Development. 17:487-509. Knack, Stephen, and Philip Keefer (1995). "Institutions and Economic Performance: Cross- country Tests Using Alternative Institutional Measures," Economics and Politics 7(3), 207- 227. Leonard, Dutch, & Peter B. Zimmerman (1994). SAMODEL for South African Public Service, Quattro Pro-based Model, Kennedy School of Government, Cambridge, MA: Harvard University. Levy, Brian, and Pablo Spiller, eds. (1996). Regulations, Institutions, and Commitment: Comparative Studies in Telecommunications. New York: Cambridge University Press. Lienert, Ian, and Jitendra Modi (1997). "A Decade of Civil Service Reform in Sub-Saharan Africa," Working Paper, Washington, D.C.: International Monetary Fund. Lindauer, David, & Barbara Nunberg, eds. (1994). Rehabilitating Government. Regional and Sectoral Studies, Washington, D.C.: World Bank. Mauro, Paulo (1995). "Corruption and Growth," Quarterly Journal ofEconomics, CX 681-712. (1996) "Effects of Corruption on Growth, Investment and Public Expenditures," Washington, D.C.: International Monetary Fund. North, Douglass (1990). Institutions, Institutional Change and Economic Performance. New York: Cambridge University Press. North, Douglass, and John Wallis (1986). "Measuring the Transaction Sector in the American Economy, 1870-1970," in Long-term Factors in American Growth, Vol. 51 of the Income and Wealth Series, Stanley L. Engerman and Robert E. Gallman, eds., Chicago, IL: University of Chicago Press. Nunberg, Barbara (1995). Managing the Civil Service: Reform Lessons from Advanced Industrialized Countries. Discussion Paper No. 204, Washington, D.C.: World Bank. (1997). Re-thinking Civil Service Reform: An Agenda for Smart Government. PREM, Washington, D.C.: World Bank. Nunberg, Barbara, and John Nellis (1990). Civil Service Reform and the World Bank. Discussion Paper No. 161, Washington, D.C.: World Bank. Nye, Joseph, Philip Zelikow, and David King, eds. (1997). Why People Don't Trust Government, Cambridge, MA.: Harvard University Press. Organization for Economic Cooperation and Development (1995). Governance In Transition: Public Management Reforms in OECD Countries. Paris: OECD. Ostrom, Elinor (1997). "Using Public Choice Theory to Help Bank Staff Assess Clients' Institutional Capacity," LLC Presentation, April 17, 1997, Washington, D.C.: World Bank. Paul, Samuel (1998). "Making Voice Work: What Happened to the Report on Bangalore?" PREM Seminar Series, March 4,1998, Washington, D.C.: World Bank. 29 Pinto, Rogerio, and Angelous J. Mrope (1994). Projectizing the Governance Approach to Civil Service Reform: An Institutional Environment Assessment for Preparing a Sectoral Adjustment Loan in the Gambia. Discussion Paper 252, Washington, D.C.: World Bank. Public Sector Group, PREM Network (1998). "The World Bank: Addressing the Challenge of Reforming Public Institutions," Draft Sector Strategy Paper, Washington, D.C.: World Bank. Rama, Martin. (1997). "Efficient Public Sector Downsizing," Shrinking Smartly: The Public Sector Retrenchment Nexus, Washington, D.C.: World Bank. Schiavo-Campo, Salvatore, ed. (1994). Institutional Change and the Public Sector in Transitional Economies. Discussion Paper No. 241, Washington, D.C.: World Bank. (1996). "Reforming the Civil Service," in Finance and Development. Vol. 33, No. 3, Washington, D.C.: IMF and IBRD. Schiavo-Campo, Salvatore et al. (1997). Government Employment and Pay: A Global and Regional Perspective. WPS No. 1771, Washington, D.C.: World Bank. Scott, Graham C. (1996). Government Reform in New Zealand. October 1996, Washington, D.C.: International Monetary Fund. Stevenson, Gail (1992). "How Public Sector Pay and Employment Affect Labor Markets," WPS 944, Population and Human Resources Department, Washington, D.C.: World Bank. Tendler, Judith (1998). Good Government in the Tropics. Baltimore: Johns Hopkins Press. UNDP (1995). Guiding Principles on Civil Service Reform, Civil Service Reform Working Group of the Special Programme of Assistance for Africa (SPA). World Bank (1992). Governance and Development. Washington, D.C.: World Bank. (1997). World Development Report 1997 The State in a Changing World. New York: Oxford University Press. (1998). Albania: Country Assistance Review. Report No. 18030, Operational Evaluation Department Washington, D.C.: World Bank. (1998). Assessing Aid: What Works, What Doesn't, and Why. World Bank Policy Research Report, Washington, D.C.: Oxford University Press. (1998). "New Frontiers in Diagnosing and Combating Corruption," PREMNotes, No. 7, Washington, D.C.: World Bank. (1998). Sri Lanka: Recent Economic Developments and Prospects. Report No. 17761, Washington, D.C.: World Bank. 30 Annex 1 Annex 1.1 Composition of Economic and Sector Reports in Sample Region Time Period Total (Year of Publication) Reports 80-86 87-93 94-97 AFR 15 26 1 42 EAP 2 2 1 5 ECA 2 5 1 8 LAC 10 13 5 28 NINA 2 5 1 8 SAS - 7 5 12 Total 31 58 14 103 Annex 1.2 Composition of Lending Operations in Sample Region Time Period Total (Year of Approval) Loans 80-86 87-93 94-97 AFR 10 48 23 81 EAP - 4 5 9 ECA 4 1 2 7 LAC 6 8 4 18 MNA 1 3 3 7 SAS - 1 1 2 Total 21 65 38 124 Annex 2. CS Wage Bill as a Percentage of Current Expenditures Over Time Country Region 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 Algeria K [N.-\ 54.4 53.0 37.8 40.5 39.8 42.1 40.4 38.7 Benin .\FR . 80.9 77.1 77.4 77.0 71.9 63.5 51.9 Burkina Faso AFR 53.7 57.4 59.2 48.9 48.5 46.7 47.5 47.9 Cameroon AFR 50.0 52.7 53.1 50.0 49.0 50.6 38.1 31.3 27.2 29.5 CAR AFR .. . .. . .. o5.9 58.7 51.3 53.9 45.1 48.2 45.6 50.0 42.7 45.3 53.4 China EAP .. 12.6 13.5 14.0 14.4 13.6 Congo AFR 38.4 44.0 41.9 39.3 34.9 46.3 49.9 51.1 39.3 35.3 32.8 31.1 Dominica L.AC .. .. .. .. 59.6 56.5 54.8 56.7 58.1 56.8 Ecuador LAC 418.2 49.6 52.4 41.7 46.3 Gambia, The AFR .. .. . 21.9 29.3 31 1 32.0 32.9 28.2 32.4 33.7 33.7 Ghana .FR . .. 22.7 37.7 43.1 42.4 41.4 .19.7 39.9 38.7 42.7 32.8 32.0 33.9 32.9 32.1 Guinea .FR .. .. 20.6 17.8 2Y.8 32.3 33.9 36.8 43.4 47.4 46.8 47 3 Guinea-Bi.sau AFR 29.3 25.0 25.8 26.7 24.0 22.7 18.3 18.2 15.2 21.9 Guyana L C 27.8 33.4 33.9 41.0 43.9 48.7 Indonesia EAP 27.7 28.o 29.3 2Q.3 Jamaica LAC .. .. .. ,, .. . .. .. .. 38.3 3o.5 24.5 23 7 40.4 32.3 35.5 37.5 Kenya AFR .. .. .. 37.8 32 0 29.6 40.9 37.4 Lao PDR EAP 40.4 49.1 50.9 44.8 37.2 42.0 44.0 44.0 47.3 .. Nadae.iscar .XFR 11.7 11.3 14.7 47.4 48.0 383 367 30.9 29.7 31.1 M11i FR . .. .. 48.1 48.0 41.3 42.9 41.0 33.9 34.2 51.7 48.3 46.4 31.7 35.6 30.9 M1auntania AFR 25.4 25.0 30.3 29.4 24.6 26.1 30.6 26.3 24.4 18.3 28.4 loldo%3 EC. 34.6 23.8 14.5 Pert 1AC . 24 0 27.5 27.9 26.7 30.1 31.9 28.8 31 2 20.3 2n.h 27.1 27.4 26.4 26.9 26.7 Sencoal AFR 49.1 50.7 51.5 50.0 505 4(.) 55.8 57.9 53.0 46.7 49.7 52 U 51.0 Sierrm Leone AFR I1 S 23 6 24.9 26.5 28.4 30.0 27.8 33.5 Sri Lanka SAS 23.2 25.7 23.1 22.6 22.4 25.1 Togo .AFR 44.5 46.2 45.3 34.8 42.0 39.2 42.0 Tunisia NIN.-\ . .. . . 1.0 1.2 1.2 1.1 Turke% E( \ . . . . . .. .. 27.0 27.0 35.o 43.0 45 h 47.7 44.2 35.4 33.1 28.9 i t nJa 13.1 18.4 28.1 15.2 17.4 15.0 17 3 12.1 19.5 14.8 19.4 21.6 27 5 .0.2 Vieinnm ENP 124 lit 11.2 12.7 11.1 1emen. Rep N .NA. .. .. . . .. 63.6 61.4 62.5 61.8 55.7 57.1 47.2 30 5 Source: IMF Staff Country Reports, IMF Statistical Appendix, Unified Surveys * The quality of data was questionable as evidenced by high year-to-year variability in wages bills as a percentage of current expenditure. In some cases, data were estimates generated by staff working on country teams. * The shaded areas indicate periods of Bank-supported CSR intervention. CD 32 Annex 3 Annex 3. Index of Bureaucratic Quality: Regional Averages (1991-1997) 4.00 .3.50 3.00 2.50 EAP _XECA -LAC MNA 2.00 . 1.50- 1991 1992 1993 1994 1995 1996 1997 Source: DEC/PREM Ratings 33 Annex 4 Annex 4. Civil Service Reform Measures by Class, Region, and Time Period Class Measures Interventions Time Period (FY) All AFR % total 80-86 87-93 94-97 Downsizing Diagnostic tools: * CS census 9 8 89% V * CS registries 2 2 100% / Operational measures: Size rationalization * Voluntary early retirement packages 29 21 72% / / * Hiring freezes 28 22 79% / / * Natural attrition 16 15 94% / / * Strict enforcement of the retirement age 13 9 69% V / * Removal of ghost workers 8 8 100% / * Limitations on the renewal of temporary contracts 4 3 75% / * Ending automatic recruitment 4 4 100% / * Termination of poor performers 1 1 100% Wage rationalization * Meet wage bill targets 15 13 87% / V * Direct decompression 1 1 100% / = Monetizing and reducing allowances 3 3 100% / V Compensating losers * Lump-sum compensation packages 16 9 56% V V * Job training programs 13 10 77% / * Redeployment programs 7 7 100% V * Facilitating integration into the private sector 9 8 89% V Capacity Building Diagnostic tools: * Organizational audits 10 8 80% / * Manpower audits 8 7 88% V * Reorganization studies 4 4 100% V Operational measures: Organizations and structures * Enhancing ministerial capacity 28 14 50% V V V * Developing indicators for organizational performance 6 5 83% V * Autonomization and decentralization of management 4 2 50% V V Intergovernmental linkages * Strengthen institutional links 6 5 83% V * Improvement in systems and procedures 19 14 74% V * Enhance coordination between central and sectoral ministries / central and local governments 3 3 100% V * Introducing performance, program and forward budgeting 8 4 50% V V Training - Workshops, seminars and study fellowships 24 12 50% V * In-service training and career development programs 20 16 80% V Recruitment * Short-term recruitment to fill immediate skills shortages 4 4 100% V - Skills mobilization and expatriate resettlement 7 7 100% V * Recruiting from parastatal organizations 1 0% V * Identifying and classifying qualified staff I 1 100% V * Competitive examinations and greater autonomy in hiring 6 3 50% V Incentives * Salary adjustments and competitive wage hikes 7 2 29% V - Review of allowances and benefits 11 11 100% V Performance orientation * Introducing merit-based promotions 8 8 100% V * Establishing performance evaluation systems 8 8 100% V Complementary inputs and technology * Technological inputs 4 2 50% V V * Information systems for payroll management 60 50 83% V Reforming Institutions Diagnostic tools: Operational measures: Intra-Public Sector Regulation - Revision of personnel management procedures 40 26 65% V * Establish new code for competitive bidding 7 3 43% V Governance (Checks and Balances) * Judicial reform 4 4 100% V * Enhancing the availability and quality of information 10 6 60% V * Strengthening independent government oversight bodies 8 6 75% V 34 Annex 5 Annex 5. Bank CSR Portfolio Composition Lending Period (FY) Region 1980-86 1987-93 1994-97 Total AFR 11 48 22 81 14% 59% 27% EAP 0 4 5 9 0% 44% 56% ECA 4 1 2 7 57% 14% 29% LAC 6 8 4 18 33% 44% 22% MNA 1 3 3 7 14% 43% 43% SAS 0 1 1 2 0% 50% 50% TOTAL 22 65 37 124 18% 52% 30% 35 Annex 6 Annex 6. CSR Performance By Class of Intervention and Region (Percent Satisfactory) AFR EAP ECA LAC NINA SAS Average Rating for CSR Class Completed Inier%entions DwAnsizin- 46 (1 MI 5m 0 46 Capaciiy Building Organizations and structures 21 so 100 50) 0 31 Quality of human resources 21 1 10 43 0 1 30 Principal-agent relations 20 lul 17 5( ) 30 Refomung Institutions 2-7 0 (1 25 i- 1.1 22 O%erall A%erage Rating for Region 29 17 8) 39 10 0 Ongoing and Recently Completed Intermentions Do., n< Lzing 3S . 1 33 31 Capacity Building Or-anizanons and structures 4-' 40 50 60 I 0 44 Qualir, of human resources 5 411 1 1(11 ) 52 Principal-agent relations 35 U U l 47 !Refornung Instituions 4u 0 0 0 19 'Oerall Average Rating for Region 45 27 14 43 17 36 Annex 7 Annex 7.1 Relative Effectiveness by Lending Type Outcome Satisfactory Poorly evaluated Lending Type Adjustment Investment Adjustment Investment Ongoing and Recently Completed Interventions Downsizing 44% 24% 22% 53% Capacity Building 43% 48% 21% 7% Reforming Institutions 40% 13% 40% 56% Total 43% 37% 25% 25% Completed Interventions Downsizing 49% 30% 16% 0% Capacity Building 41% 15% 17% 11% Reforming Institutions 19% 33% 33% 22% Total 40% 19% 19% 11% 7.2 Outcome Ratings: Completed Interventions Across All Projects In Stand-alones Total SAT PE Total SAT PE Downsizing 59 46% 14% 5 40% 0% Capacity Building Organizations and structures 61 31% 18% 10 0% 10% Quality of human resources 44 30% 14% 9 0% 0% Principal-agent relations 33 30% 9% 7 14% 0% Reforming Institutions 36 22% 31% 4 25% 0% Interventions 233 34 Intervention SAT Average 33% 12% Intervention PE Average 17% 3% Outcome 61% 40% Sustainability 38% 0% Inst. Dev. 16% 10% 7.3 Outcome Ratings: Ongoing Interventions Across All Projects In Stand-alones Total SAT PE Total SAT PE Downsizing 26 31% 42% 8 25% 50% Capacity Building Organizations and structures 32 44% 16% 12 58% 8% Quality of human resources 23 52% 4% 10 50% 10% Principal-agent relations 15 47% 7% 8 63% 0% Reforming Institutions 21 19% 52% 7 14% 57% Interventions 117 45 Intervention SAT Average 38% 44% Intervention PE Average 25% 22% Development Objectives 89% 86% Implementation 84% 79% Management Performance 89% 86% SAT - satisfactory PE - poorly evaluated 37 Annex 8 Annex 8.1 Select Countries Falling Short on Rationalization Targets Country Years Targeted Actual Cutbacks Cutbacks Benin 1991-94 8,000 5,000 Cameroon 1995-96 Unknown 22,000 Lao PDR 1993-95 15,660 13,700 8.2 Select Countries with Short Run Successes, Unsustainable Results Country Year Percent Reduction Reasons for Recurring Overstaffing in CS Size The Gambia 1987 30% Increase in health and education workforce Ghana 1987-93 50% Not known Uganda 1990-94 50% Ghost workers Senegal 1987-90 Unknown Rehiring graduates 38 Annex 9 Annex 9. Percent Change in Civil Service Wage Bill as a Share of GDP Over Time Country 1985-87 1995-97 Percent Change Africa Region Benin n.a n.a n.a Burkina Faso 7.40 6.10 -17.6 Cameroon 8.50 7.80 -8.2 CAR 6.30 5.90 -6.3 Congo n.a n.a n.a. The Gambia 5.60 5.90 5.4 Ghana 4.40 5.70 29.5 Guinea 4.10 4.50 9.8 Guinea-Bissau 4.60 2.90 -37.0 Kenya n.a n.a n.a. Madagascar 4.30 3.80 -11.6 Mali 6.10 6.10 0.0 Mauritania 8.40 9.80 16.7 Senegal 8.40 6.70 -20.2 Sierra Leone 2.60 4.10 57.7 Togo 8.80 10.00 13.6 Uganda 1.50 2.20 46.7 Other Regions Algeria 9.20 10.40 13.0 China n.a n.a n.a Dominica 15.10 15.50 2.6 Ecuador n.a n.a n.a Guyana 4.10 4.50 9.8 Jamaica 7.40 4.10 -44.6 Indonesia 3.20 2.60 -18.8 Laos 5.90 6.10 3.4 Moldova n.a n.a n.a Peru 2.60 2.60 0.0 Sri Lanka 4.80 5.10 6.3 Tunisia 9.40 10.20 8.5 Turkey 7.30 7.80 6.8 Viet Nam 1.60 2.30 43.8 Yemen n.a n.a n.a Annex 10. CSR Initiatives by Client Country and Institutional Quality, 1981-1997 Country Downsizing Capacity Building Reforming Institutions Pay and Employment Structures and Systems Human Resources information and P-A Relations Regulation Accountability Congo 2.04 87,94,96 87,94 7 94 87 Cameroon 2.24 89, 94, 94, 96 89.,94 89, 94, 96 89 89 ag 89, 94 89, 94 89, 94, 96 89, 99, 94, 96 96 Guinea 2.63 86,a88,8 86 86,8 asaas s8,88 85,3a, a B,88 88 88 88 88, 88 Yemren 2.63 96 96 85. 89 85, 89 89 85, 89, 96 85 89 Ecuador 2.69 95 95 85,95,95,95 95.95 85.95 as ss, 95, 95, 95 Guinea-Bissaui 2.72 87, 89 89,92 8S, 89, 92 88, 92 88 Togo 2.77 95 90 90,95 95 90a95 Senegal 2.79 86, 87, 90 80, 86, 87,90, 94 83, 86-88, 90 86, 87, 90 86-88, 90 88, 90 87, 88, 90 90 83, 87, 88, 90 87, 88, 90 CAR 2.81 S6, s, S,a90 &s, a, as, 90 as, Bs, Bs 81,86,88 86,88, s ,88,90 86,n 8,90 8s,8s,90 91, 86, ss, ss, 90 90 Sierra Leone 2.82 92, 94, 95 92-94 92 93 93 92M9 93 94 93, 94 94 Madagascar 2.86 88,91 88 88 BB,97 88,97 88,92,97 88 BS,92 88.97 Jamaica 2,89 91 85 83, 85, 88, 91, 97 84, 85, 88, 91, 97 83, 84, 85, 88, 91 85 84, 91, 97 84, 91 85, 91, 97 97 97 97 Dominica 2.99 87 87 87 87 87 Algeria 3.01 95,,96 95 9o 9o ao 9o, W Kenya 3.10 89, 95, 96 96 9s, 96 95 9s 95 96 96 96 % Moldova 3.13 94 94 Gambia 3.14 86, 89 86, 89 86,8 8g 9 86 89 89 89 86,8B9 Mauritania 3.15 87,87,96 87,96 87,96 87 87 0 Benin 3.15 89,91,95 89,91,95 91 91, 91, 94,95 94, 95 91 91 89 99, 94 Laos 3.6 91 89,91 89, 91 91 89 96 91, 96 96 91,96 91,96 91 Turkey 3.17 81-84 81-84 81-84, 92, 96 96 81-84. 92 81-84 84 92, 96 96 92 Mali 3.18 90,94 90,94,96 90 82, 9, 90, 96 82,89 82.89 89 90 2, 89, 99 Guyana 3.24 93 93 90,90,93 93 90 93 90 93 90 93 Burkina Faso 3.28 91-94 91,994 91,9 91,92 91,92 94 92 91 91,92 92,94 Uganda 3.31 87, 9o, 91, 93, 94 90, 94 97, 91 83, 87, as, 9o, 91, 94 93-95 83, as, 93, 95 as 9o, 91, 93-95 95 as, 93 90, 93, 95 ss 88 Vietnam 3.40 94 94 9499 95 Sri Lanka 3.40 0 90 90, 96 9o 90 90 0 30 Peru 3.60 92, 93 82 82 92 92 82,92 93 Ghana 3.65 3, 87, 87, 91, 95 83, 87, 87, 89, 95 89,91,95 87,89,91 87,89,91 87, 89, 91, 91 87 91 89,491 89 89, 91,91 9 Tunisia 3.75 70 China 3.82 93,09,95956995,9 95 95 9 95 Indonesia 3.86 95 95 92,95 92 *Countries are rank