WPS6136 Policy Research Working Paper 6136 Matching Reforms to Institutional Realities A Framework for Assessing Social Service Delivery Reform Strategies in Developing Countries Ariel Fiszbein Yasuhiko Matsuda The World Bank Human Development Network Office of the Chief Economist July 2012 Policy Research Working Paper 6136 Abstract The paper is based on the premise that effective practices.� The paper asks how the specific institutional institutional reforms for service delivery require a contexts of a given country limit options for service carefully-considered institutional “fitting� process delivery reforms, and therefore how a feasible set of as opposed to transplantation of “international best reform options may vary systematically across countries. This paper is a product of the Office of the Chief Economist, Human Development Network. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at afiszbein@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Matching Reforms to Institutional Realities: A framework for assessing social service delivery reform strategies in developing countries Ariel Fiszbein (Chief Economist for Human Development) Yasuhiko Matsuda (Senior Public Sector Specialist) World Bank1 JEL: D73, H11, H50 Sector Board: PSM 1 The views expressed do not necessarily reflect those of the World Bank. 1. Introduction Social service delivery is weak across the developing world. While there is substantial heterogeneity across regions and countries, the picture of failing services is a familiar one. Problems such as systematically high levels of absenteeism among teachers, doctors and nurses (see Figures 1 and 2); persistent rates of drug stock outs –particularly in rural health clinics; rates of leakage in public funds reaching the frontline health or school facilities of 50% or more; and the use of safety net programs as a source of political patronage are some extremes manifestations of those failures. Figure 1: Provider Absence Rate in Health (%) Papua New Guinea 19 Peru 24 Honduras 27 Uganda 37 Chad 17 Cameroon 5.6 Indonesia 40 India (UP) 40 Bangladesh 35 0 10 20 30 40 50 SOURCE: ROGERS AND KOZIOL (2011) 2 Figure 2: Provider Absence Rate in Education (%) Papua New Guinea 15.1 11 Honduras 14 14 Zambia 17 27 Senegal 30 8.1 Ghana 19.4 18 Lao PDR 8 19 India 25 16 Bangladesh 16 0 5 10 15 20 25 30 35 SOURCE: ROGERS AND KOZIOL (2011) But, more generally, developing countries have shown grave difficulties in improving the quality of service. Take the case of primary education. A majority of developing countries show appalling performance in a variety of learning assessments. Just as an example, in rural India 41% of children in class 5 cannot read at class 2 level, 22% of children in class 2 cannot recognize words and 58% of children in class 5 cannot divide 3 digits by 1 digit (Pratham 2008). It is not an exaggeration to say that school systems in many developing countries are becoming large factories producing what will be, de-facto, illiterate young people. Given this background, it is hardly surprising that service delivery reform has become almost a rallying cry among development practitioners. A quick review of the international experience shows that over the last two decades developing countries have 3 pursued a variety of reforms in the area of service delivery. Decentralization of the responsibility for social services to local governments, delegation of management responsibilities to communities, creation of special agencies to run high-priority programs (social funds, conditional cash transfers, etc.), contracting out to NGOs and non-state providers, and introduction of management reforms within public bureaucracies are examples of this trend. Slowly, evidence of the impact of such reforms is emerging. Two recent examples involve the case of Conditional Cash Transfers (Fiszbein and Schady, 2009) and school accountability reforms (Bruns, Filmer and Patrinos, 2011). A key message of all such reviews is the well-known argument that „context matters‟. Quite often, though, it is not clear in which way it matters –making the statement more a source of frustration than a helpful reminder of the need to consider country conditions. This paper is based on the premise that effective institutional reforms for service delivery require a carefully-considered institutional “fitting� process as opposed to transplantation of “international best practices.� We seek to ask how the specific institutional contexts of a given country limit options for service delivery reforms, and therefore how a feasible set of reform options may vary systematically across countries. In Citizens, Politicians, and Providers (Fiszbein 2005) we reviewed the experience with service delivery reforms in Latin America. We first mapped (see Figure 3) the range of reforms in service delivery (both for social and infrastructure services) using the accountability framework first developed in the 2004 World Development Report (World Bank 2003). The WDR presented a by-now familiar accountability framework focused on 4 three key accountability relationships in the service delivery chain: between users and providers, between citizens and policymakers, and between policymakers and providers. The long route of accountability, long because it comprises two separate legs, links citizens through voice (or politics) to policymakers and policymakers in turn to service providers through their managerial ability to supervise and monitor providers (the compact). The short route of accountability links citizens more directly to service providers, representing the ability to monitor providers and hold them accountable (client power). Service delivery failures happen when one or both of the routes of accountability are weak. We then discussed the range of institutional options for strengthening accountability relations among citizens, politicians/governments, and service providers, and explored a set of country-specific conditions, such as the quality of government bureaucracy, the strength of the rule of law, and the quality of political representation that may condition applicability of these reform options. In this paper we explore in more depth the overall concept of fitting reforms to institutional settings. We do so by focusing on the critical issue of delivery of social services in decentralized settings. 5 Figure 2: A Network of Influence and Accountability Mechanisms. In both federal and unitary countries, the delivery of social services (health, education, safety nets) typically involves roles for both the central and local governments. Financing of social services typically involves a high share of central funds, often through a combination of programs and activities executed by agencies of the central government and grants to local governments. Even in highly decentralized settings, human resources (the largest budget item for both health and education in all countries) are often managed under policy settings that involve some level of national regulation. National curricula and learning assessment systems or national medical protocols and professional certification systems often coexist with local practices and rules. As is widely recognized, decentralization is no panacea but, under the right conditions, it can be a good ground for innovation and experimentation. At the core of successful 6 decentralization is a combination of improvement of the compact between levels of government and strengthening of local accountability to all citizens. But elements that make decentralization work are numerous and institutionally demanding for a weak state. These elements would include, depending on the country and sector, simpler and more transparent use of inter-governmental transfers, the enforcement of hard budget constraints for local governments, clear definitions of responsibilities (including through legal and/or administrative instruments), and the strengthening of the capacity of central government to monitor and audit sub-national governments as well as the development of evaluation systems. It may also require specific instruments for intervention (by national authorities or federal bodies) when service delivery failures threaten the well-being of citizens. In many countries, there is great heterogeneity in the size and capacity of sub-national entities (e.g. municipal governments ruling over large cities and very small localities with dispersed populations) which are in very different position to deliver the various aspects of particular services. Recognizing that heterogeneity adds another important dimension to the argument against „best practices‟ and in favor of fitting reforms to institutional realities. In the next section we discuss in some detail the type of institutional characteristics of both central and local governments that we believe can provide the basis for an assessment of the range of reform strategies for social service delivery. We also use those characteristics to propose a simple typology that can be used to conduct such assessments. 7 In section 3 of the paper, we explore that typology by looking the experience of a range of countries with reform strategies for health, education and safety net programs. Finally, we conclude with a set of suggestions for future research. 2. Fitting service delivery to institutional characteristics In a decentralized setting, institutional options for delivery of services that in theory could be delivered efficiently by sub-national (local) governments depend on a range of institutional characteristics at the levels of both the central state and local governments. This section develops a simple heuristic framework to facilitate systematic considerations of the institutional fitting process. The framework is not intended to be deterministic. The presence of the particular variables we single out does not guarantee the suggested institutional arrangement will yield positive results or vice versa. Nor is the exercise here intended to capture full subtlety and complexity that is necessarily a part of a complex social process such as the operation of a service delivery arrangement. The framework distills what we consider minimally essential for a particular type of institutional arrangement to have a reasonable probability of “working.� The framework is built around the variations in capabilities and incentives of the central and local governments in a decentralized (e.g., federal) setting. The levels of capabilities and incentives for the two levels of government determine the range of viable institutional options available for a given political jurisdiction (i.e., within a country between the national government and sub-national governments, or within a sub-national 8 territory between, say, the state and municipal governments). Since by definition we are examining the case of decentralized service delivery, the starting point is the sub-national level. The quality of decentralized service delivery depends first of all on the level of incentives and capabilities of the sub-national governments directly in charge of providing services to the population. Electoral Accountability and Local Service Delivery At the sub-national level, the single most salient characteristic is the extent to which the “long route� of accountability works (World Bank 2004, Ahmad et al. 2005). This largely depends on the efficacy of electoral accountability. Recent empirical work has shown that competitive elections can improve sub-national governments‟ accountability for service delivery. For example, both Faguet (2004) and Skoufias et al. (2011) show empirically that introduction of competitive local elections as part of the broader political and fiscal decentralization reforms in Bolivia (in the 1990s) and Indonesia (in the early 2000s), respectively, has let to local governments‟ greater responsiveness to citizen demands measured as expenditure allocation. Furthermore, Khemani (2001) offers evidence from India that voters tend to be more “vigilant� of sub-national government performance than national government performance. In her findings, the Indian voters cast their votes in state elections on the basis of the sub-national (state) governments‟ performance during the government‟s entire term whereas for the national elections, they tend to focus on the government performance only in the year preceding the election. The empirical relationships between elections and service delivery are ambiguous despite the findings of the work cited here. For example, in a more recent paper Khamani (2010) 9 finds evidence suggesting that Indian politicians ignore large demand for village infrastructure and emphasize employment and welfare transfers as means to buy votes. The risk of local capture is widely recognized in the literature as a potential downside of decentralization (e.g., Bardhan 2002). What matters, therefore, is not only the presence of elections but the quality of the elections as an accountability mechanism. Needless to say electoral accountability as a source of local government incentive is irrelevant in non-electoral regimes or in systems where incumbents are not allowed to run for re-election and at the same time political parties are so poorly institutionalized that they don‟t internalize the incumbents‟ electoral incentive as their own (i.e., they behave like they don‟t care about the next election). But this does not necessarily mean local governments in non-electoral regimes are not accountable for performance (Tsai 2007), but it does mean that non-electoral regimes are handicapped in terms of institutional options available to hold sub-national governments accountable for their performance. For example, in Vietnam, a non-electoral regime widely credited with impressive improvements in health outcomes over the 1990s (Lieberman and Wagstaff 2009), the decentralization with concurrent delegation of authority for facilities to charge user fees has led to an increase in inequality of de facto access to low cost, quality health care. In the face of limited access and rising costs of health care, which forces those who can afford to pay “informal payments� of considerable amounts, there appears to be limited pressure on the provincial governments that directly manage health facilities to control costs while improving access and quality (Forsberg 2011, Vasavakul 2009). Even in full electoral democracies, an additional assumption is that conditions exist for a relatively efficient electoral “market� where the principle of “one person one vote� holds 10 and where the electoral outcomes reflect the majority‟s preferences. If the majority are relatively poor, as they tend to be in developing countries, their preferences are more likely to be channeled in the direction of better public service delivery (rather than, say, lower taxes).2 In essence this is what the WDR2004 assumes as the political basis of the “long route� of accountability. As Keefer and Khemai (2005) have shown, however, these are stringent conditions that are rarely obtained in full in most real-life circumstances. To the extent such conditions materialize, the sub-national governments operating in such electoral environments are expected to invest their resources in improving public service delivery in order to respond to the demands of the majority voters in their respective jurisdictions. According to Keefer and Khemani (2005), at least three kinds of imperfection limit elections‟ efficacy in translating voters‟ preferences into adopted policies of the government. First is imperfect information, which limits citizens‟ abilities to assign credit or blame for a government action, which in turn heightens politicians‟ incentive to seek rents (since they are unlikely to be punished for this) rather than improve public services (for which they are unlikely to be rewarded at the ballot). Second is heterogeneity of the citizenry with respect to its economic and social background. A heterogeneous community is less able to agree on a set of policy priorities because of the divergent interests originating from their diverse social and economic needs. Furthermore, certain types of heterogeneity, such as ethnicity or religion, tend to lead citizens to cast their votes on the basis not necessarily of material interests in an 2 This obviously is a simplification. In those systems where the weight of indirect taxation (e.g., consumption tax) is high, the poor too may feel the tax burden as excessive and demand its lessening, especially if they don‟t see corresponding “return� on the side of public services. 11 immediate sense (e.g. better roads that benefit multiple ethnic communities coexisting in adjacent areas) but of pre-defined identities (i.e., ethnic or religious lines). Third is the lack of credibility of political competitors. Incumbents are expected to under-perform vis-à-vis voter expectations when they are not faced with a serious enough threat to be unseated. Even when there is a challenge to the incumbency, however, to the extent voters do not believe the challenger‟s promise to deliver some desirable performance in the event he is elected, the incumbent can afford to continue to ignore the general voters‟ preferences. When one or more of these imperfections are present, the effectiveness of elections as a mechanism of holding the incumbent politician accountable gets diluted considerably. 3 Central Government Influence: Incentives for Responsiveness Sub-national governments do not operate in a vacuum in a decentralized system. They operate within a policy and fiscal framework set by the central government (or legislature) and, in most cases, interact with the central government in the process of delivering specific services. The degree to which the central government plays an active role in this interaction depends on the extent to which the “long-route� accountability is operating effectively at the sub-national level on the one hand and on the extent to which the central government has influence over sub-national government behavior. 3 Another variable of concern is sub-national government‟s institutional capabilities. The higher the capabilities, the more institutional options are available. In other words, in any given sub-national jurisdiction at any given moment, the existing capabilities become an exogenous constraint. For the purpose of crafting an institutional strategy for service delivery improvement, however, it is possible to see the capabilities from a more dynamic perspective because under “right� circumstances, capabilities can be built over time. A key determinant is the government‟s own incentive to invest in capability building, which brings us back to the strength of electoral accountability as the primary driver. For an application of this argument in the case of Colombia‟s local governments, see Fiszbein (1997). 12 The central government‟s influence over sub-national governments is a function of two factors; its political responsiveness, or an incentive to improve social outcomes, especially among the poor who depend more on public services; and its stewardship capabilities to direct or induce certain behavior from sub-national governments. The political responsiveness to the electorate is a function of a variety of factors, an exhaustive discussion of which is not possible in this paper. For illustrative purposes only, some of the key variables highlighted in recent literature are structural in nature and include, for example, the degree to which society is stratified into different categorical (e.g., ethno-linguistic, religious) groups (Alesina et al. 2001, Heller and Malloney 2003). The more the society is stratified across different categorical groups, the less likely it is that the state will be expected to invest in broad-based public services. The tendency is likely to be accentuated in situations where minority groups are also the socially and economically disadvantaged that could benefit disproportionately from effective public service delivery (e.g., poor Haitian immigrants in Dominican Republic, ethnic minorities in the “highland� areas of the mainland Southeast Asia). Another set of variables is politico-institutional, such as the extent to which the country‟s governing coalition draws broad-based support from the general public, especially the lower-income segment, on the basis of a relatively coherent policy program (e.g., Brazil‟s Workers‟ Party) in which case the government may be under greater pressure to expand its public spending and improve public services, rather than from relatively narrow elite segments (e.g., Philippines), in which case the pressure on the government may be more to keep the size of the state limited and expand the realm of weakly regulated private economic activities. A simple, though by no means perfect, measure of the national 13 state‟s responsiveness to citizen demands for public services would be measures of public finance. Ceteris paribus, higher overall public spending and especially those with higher social spending are more likely to be a result of the central state‟s responsiveness to citizen demands for public services. In contract, the perennially low taxation and hence the level of public spending are likely to be reflective of the state‟s responsiveness to elite interests which depend less on public services for their welfare.4 Central Government Influence: Stewardship Capabilities The desire to improve service delivery is one thing but the ability to do so is another. In a decentralized setting, it is not sufficient if the central government was itself institutionally capable of effective service delivery. It must also possess the ability to direct or induce sub-national governments to improve service delivery within their respective jurisdictions. We call this stewardship. The stewardship capabilities mean abilities to set clear and coherent policies including objectives and standards, to monitor local government performance, and enforce the policy intent, including by use of positive incentives and sanctions (e.g., fiscal incentive, legal sanctions, party discipline). Countries vary considerably in terms of their central governments‟ stewardship capabilities. Simple quantitative indicators of stewardship capabilities are not easily available. But, some qualitative measures that capture aspects of these capabilities include the presence of a coherent sector policy and regulatory standards, the existence and use of performance information systems with availability of statistical data that allow the central government to measure sub-national government performance (at least indirectly), and the availability of discretionary budget funds which the central 4 For a discussion in the case of Latin America see De Ferranti et al (2004) 14 government could use as fiscal incentives for local governments.5 Equally important is the central government‟s ability to impose a hard budget constraint so that local governments‟ fiscal behaviors are sufficiently disciplined and incentives for cost- effective government operations are reinforced. Logically, four types of situation are conceivable based on the variations along the two dimensions of the national government‟s incentive to improve sub-national service delivery and its stewardship capabilities. In practice, however, a strong desire to improve performance without sufficient stewardship capabilities would be moot and the end result would be the same: that the central government‟s influence is low and sub-national governments cannot be induced into more effective service delivery. Similarly, possession of stewardship capabilities in the absence of a sufficient political desire to use them for the purpose of improved service delivery would mean low central government influence. Therefore, we can collapse the four logical possibilities into two practical scenarios of high vs. low central government influence: Central Government Central Government Central Government Responsiveness to Citizen Stewardship Capabilities Influence over Local Demands Service Delivery High High High High Low Low Low High Low Low Low Low 5 Stewardship has been recognized as a key function of well performing systems since the WHO introduced the concept in its 2000 report. Several of the dimensions mentioned above have become a core part of WHO‟s health systems assessments. See Travis et al. (2003). 15 3. Applying the Framework The diagram below depicts the spectrum (from high to low) of central government‟s influence over sub-national service delivery along the vertical axis and local government‟s electoral accountability along the horizontal axis. It shows four ideal type situations: strong local electoral accountability combined with strong central government influence (strong/strong); weak local electoral accountability with strong central government influence (weak/strong); strong local electoral accountability with weak central government influence (strong/weak); and weak local accountability and weak central government influence (weak/weak). In reality, it is more appropriate to treat these variations as continuum rather than discrete categories, but for the sake of clarity, the following discussion illustrates each scenario as a distinct ideal type with selected country examples. Figure 3: Central Government Influence vs. Local Accountability Central Gov‟t Influence Strong Local Accountability Strong Weak Central Government Influence Low Weak 16 Strong /Strong: When both the central government‟s influence and local government‟s electoral accountability are high, we could achieve the theoretically most efficient form of decentralized service provision and the widest range of viable institutional options. Under these circumstances, the central government‟s role would be best focused on setting a clear and efficient policy such as clear expenditure assignments under hard budget constraints, transparent and predictable funding arrangements, or an explicit service or performance standards. Local governments would be given relatively wide latitude to provide a range of services with minimum central government interventions and be held accountable by their own electorate. For certain services with high externalities, these local governments could engage in efficient co-production with the central government following a variety of schemes such as matching grants and outcome- based inter-governmental compacts. Real-life examples may include the relationship between Brazil‟s Federal Government and some of the reformist states such as Minas Gerais, which have embarked on ambitious public management reforms, including in the areas of public service delivery, where the State Governor signs explicit results agreements with each cabinet secretary in charge of a sector. The reform started in 2003 under the then Governor Aecio Neves who faced the state administration heavily in debt with limited effectiveness in service delivery. Since the Federal Government had imposed tight control over sub-national fiscal behavior through the Fiscal Responsibility Law and detailed regulatory control by the powerful Federal Treasury, reform-minded states such as Minas Gerais had few options but to initiate a rigorous program of fiscal adjustment and efficiency improvements. The latter incentive was a part of the motivations for the state 17 government‟s concerted efforts to implement results-based management reforms across a range of sectors including education and health. Minas Gerais is one of the largest states in Brazil which has historically yielded a high level of influence in the national political scene. When elected, Governor Neves broke with the traditional style of politics that heavily relied on patronage (Hagopian 1996) and “sold� his government as one focused on results and delivery of public goods and services within the financial constraints. A dominant motivation for this choice appears to be his frequently-rumored presidential ambition for which a successful record of governing his own state and strong electoral support from the local voters would be a valuable political asset. The emergence of Governor Neves (and other reformist governors and mayors) in Brazil coincided with the gradual democratic maturation at the national level. Since the disastrous episode of the early 1990s when the elected president was forced to resign amidst accusation of corruption, Brazilian politics has evolved into a situation where major political parties, the incumbent Workers‟ Party and the major opposition (Social Democratic Party), have repeatedly competed on their appeals to the electorate on the basis of past policy performance as well as future programmatic promises. The “culture� of fiscal responsibility has taken hold. Although manifestations of poor governance such as high-level corruption continue to surface, the effectiveness of electoral accountability at the national level appears as good as in any middle-income country. In terms of the central (i.e., federal) government‟s influence over sub-national governments in Brazil, the most salient aspect is fiscal. Brazil boasts of a rigorous Fiscal 18 Responsibility Law which is jealously enforced by the National Treasury. The tight fiscal discipline forces those local governments intent on addressing citizen demands for more public goods and services to resort to measures to improve efficiency rather than an easier option of “throwing money at problems�. Of course, not all Brazilian states and municipalities have adopted serious reforms as Minas Gerais has. This, in our view, is a function (at least partly) of the extent of local electoral accountability. Weak/Strong: When local governments‟ electoral accountability is limited but the central government‟s influence is high, the top-down control from the center has to substitute for the citizens‟ inability to hold their local governments accountable. This may be a case where local governments are either unelected (as in decentralized authoritarian regimes) or where political market failures render elections an ineffective means for holding the elected local governments accountable (many States in Mexico, Brazilian municipalities in general). In these situations, something akin to the application of the subsidiarity principle is necessary in the realm of accountability. The “strong� central government could inject tight control of local governments through means such as earmarked transfers, or input or output-based inter-governmental compacts. The central government could also allow entry of non-state service providers, either for- or not-for-profit, within a well-regulated market (in the case of for-profit entities) or with well-crafted performance contracts (in the case of not-for-profit entities). Real-life examples of this pattern include use of earmarked federal transfers to improve access to (better quality) health and education services in countries such as Mexico (Seguro Popular in health), Brazil (FUNDEF in education), and Argentina (Plan Nacer in health). In all these cases, a Federal government seeks to influence the performance of 19 sub-national governments through the use of financial incentives and a strong monitoring of results, typically around a specific sector outcome. The Brazilian State of Ceará – as an example of a “central� government vis-à-vis the municipal governments within the state territory – has taken this approach a step farther. Ceará is a relatively poor state in the Northeast of Brazil and has traditionally been dominated by patronage politics. With the transition from the military rule in the late 1980s, however, a group of local entrepreneurs captured the state politics with a modernizing perspective (Borges 2006). For more than two decades since then, the state government has pursued a range of innovative service delivery reforms that involved using its leverage to influence actions and decisions by (relatively weak) municipal governments (Tendler 1997). In 2008, the State Government of Ceará introduced a new, performance-linked fiscal transfer to the municipalities within its territory, linking the amounts to be transferred to a set of service outcome indicators in education (student test scores in Math and Portuguese, student grade completion rates), health (infant mortality) and environmental management (presence of an appropriate waste disposal system) in each municipality. In Brazil, the central government influence over resource allocation is relatively strong in many instances because the federal constitution and other legislation prescribes distribution formulae of a number of revenue sources and also mandate sub-national governments to allocate certain proportions of their revenues to specific sectors such as education. One such example is the federal law that requires states to transfer 25% of the state value-added tax collection to municipalities. The federal law defines the distribution 20 formula for 75% of the total transfer and allows each state to decide how to distribute the remaining 25% based on a state law. Furthermore, the federal constitution requires each municipality to allocate 15% of its net current revenues to education. What the State Government of Ceará has done is to build on these existing legal requirements which merely focus on fund allocation (as inputs for education service delivery) and incorporate quality dimensions in the distribution formula, and hence introduce an outcome orientation in the financial incentives for the municipalities (Holanda 2011). Strong/Weak: When the central government is “weak� in the way defined here, the options become more limited. If electoral accountability is reasonably effective at the local level, one option would be to introduce “radical devolution� or “quasi- independence� to allow maximum freedom of the local governments to provide services and let the local electorate hold them accountable directly with the minimum of central government involvement. These “local islands of excellence� exist in a number of countries (e.g., Naga City in the Philippines), though they are not always sustainable or replicable. But to the extent local governments (even strong ones) cannot operate completely free of central government influence, there will be inevitable efficiency losses in this arrangement. For example, inchoate central governments, through various agencies that do not coordinate with each other, often require local governments for duplicative or even contradictory reporting requirements, or inconsistent policy directives. If the expenditure assignment is relatively unclear and if the central government intervenes rather arbitrarily in the provision of services that the local governments are delivering, this could obscure the responsibility for specific services and thus dilute accountability. 21 From a policy perspective, a sensible option in these situations is to limit, or if possible reduce, the extent of constraints imposed by the central government and by the overall architecture of the country‟s intergovernmental relations. A secondary option would be to facilitate dissemination of local innovations and promote replication of successful local governance experiences among other local governments with strong enough motivations to improve their performance. Weak/Weak: Finally, when both the central government is “weak� and local governments are not accountable to their local electorate, the choices are even more limited. One option would be to rely heavily on a community-driven approach as in many fragile and post-conflict situations (e.g. the early days of EDUCO en El Salvador), on non-state providers, especially NGOs, which operate under relatively simple contracts (e.g., as in Guatemala‟s health sector), or on organizational enclaves (e.g., FONCODES in Peru in the early 2000s). Where the private sector is active, it is also possible to see unregulated growth in private provision, though with little to no quality control (e.g., education in Pakistan, health in India). A well-known example of a weak/weak situation is Cambodia. Until the peace accord in 1991, the central state in Cambodia was significantly weakened by war and internal conflict. Still one of the poorest countries in Asia, the Cambodian state suffers from limited technical and fiscal capacities to influence service delivery at the sub-national level. Local governments are even weaker. Elections of commune officials took place for the first time only in 2007 and the second elections are scheduled in 2012. If these 22 elections are genuinely competitive, they might contribute to improved local accountability. But the country‟s poverty combined with the strong patronage orientation of the national politics and the relatively strong control of the national political party could dampen the effectiveness of elections as a means for holding commune officials accountable for performance. Finally and perhaps most importantly, communes are given only limited service delivery responsibilities, and therefore, their improved accountability would at best have only indirect effects on service delivery at the local level (Smoke 2008). Against this institutional backdrop, Cambodia has become famous for a number of innovative approaches to service delivery, especially in health where NGO contracting has been used extensively with positive results. Several international NGOs were hired by the government (with donor assistance) to oversee delivery of primary healthcare and management of provincial hospitals. The selected health operational districts and provincial hospitals would sign a performance contract with the NGO and the latter would hold the providers accountable with a combination of regular performance monitoring and incentive pays (Bhushan et al. 2007). Similar approaches have been attempted successfully in Central America following civil war as mentioned above. 23 4. Conclusions The paper is based on the premise that effective institutional reforms for service delivery require a carefully-considered institutional “fitting� process as opposed to transplantation of “international best practices.� It proposes a simple framework to consider the range of reform options available to countries depending on the characteristics of central and local governments. The framework suggests four ideal types of institutional options. When both the central government influence and local accountability are strong (“effective federalism�), the range of institutional options can encompass some of the most complex and sophisticated management approaches such as multi-sector performance contracting, as implemented in the State of Minas Gerais in Brazil. When the central government influence is strong but local accountability weak, the former would be in a position to substitute for the latter‟s weak accountability to the local electorate with accountability to the central government through means such as output-based grants (“accountability substitution by central government). When the central government influence is weak, which typically obtains when the central government‟s own political accountability is weak, but when some local governments operate under relatively strong local accountability (“islands of excellence�), desirable interventions would be to reduce policy distortions the central government might impose on local governments and promote learning among reform- oriented local governments. Finally, when both central government influence and local accountability are weak, other non-governmental actors such as NGOs, local communities themselves, or international donors would have to step in (“accountability 24 substitution by non-governmental actor�). Figure 4 maps a few of the examples discussed in the text to the proposed categories. Figure 4: Influence, Accountability and Institutional Options Central Government Influence High “Accountability Substitution by Central Gov� Brazil-state/Education Ceara-Municipalities “Effective Federalism� Brazil-Minas Gerais Local Electoral Accountability Low High “Islands of Excellence� “Accountability Substitution e.g., Naga, Philippines by NGO/Community� Cambodia/Health EDUCO – El Salvador Low We foresee two directions in which this line of research could evolve. The first direction involves deepening the understanding of the characteristics of central and local governments, including through a better definition of indicators. The second, and possibly more important, dimension implies the application of the proposed framework to a larger set of country and sector cases to continue testing its robustness and applicability. 25 References Ahmad, Junaid, et al. (2005). Decentralization and Service Delivery. World Bank Policy Research Working Paper 3603. Alesina, Alberto et al. (2001). Why Doesn’t the US Have a European-style Welfare System? NBER Working Paper 8524. Bardhan, Pranab (2002). “Decentralization of Governance and Development.� Journal of Economic Perspectives, Vol. 16, No. 4, pp. 185-205. Borges, André. (2006). “State Government, Political Competition and School Decentralization: Comparative Lessons from Brazil.� Paper delivered at the 2006 Annual Meeting of the American Political Science Association. Bruns, Barbara, Deon Filmer and Harry Patrinos (2011) Making Schools Work: New Evidence on Accountability Reforms, Washington DC, World Bank. Faguet, Jean-Paul (2004), “Does Decentralization Increase Government Responsiveness to Local Needs? Evidence from Bolivia.� Journal of Public Economics, 88. 867-893. Bhushan, Indu, et al. (2007). Contracting for Health: Evidence from Cambodia. De Ferranti, D., G. Perry, H. G. Ferreira, and M. Walton (2004). Inequality in Latin America: Breaking with History? World Bank, Washington, DC. Fiszbein, Ariel (ed.) (2005), Citizens, Politicians and Providers: The Latin American Experience with Service Delivery Reform. Washington, DC, World Bank. Forsbereg, Le Thanh (2011). The Political Economy of Health Care Reform in Vietnam. Oxford-Princeton Global Leaders Program, Working Paper. Fiszbein, Ariel (1997). “The emergence of Local Capacity: Lessons from Colombia, World Development, vol.25, No7. Hagopian, Francis (1996). Traditional Politics and Regime Change in Brazil. Cambridge University Press. Heller, Patrick and James Mahoney (2002). The Resilience and Transformability of Social Inequality in Latin America. Background paper prepared for David de Ferranti et al. (2003). Inquality in Latin America and the Caribbean: Breaking with History? World Bank Latin American and Caribbean Studies. Holanda, Marcos C. (2011). Should Results Indicators Induce or Define Budget Allocations? Memeo. Keefer, Philip and Stuti Khemani (2005). "Democracy, Public Expenditures and the Poor." World Bank Research Observer. 20:1, p. 1-28 (Spring). 26 Loevingsohn, Benjamin and April Harding (2005). “Buying Results: A Review of Developing Country Experience with Contracting.� The Lancet 366(9486): 676-681. Skoufias, Emmanuel, et al. (2011). Electoral Accountability, Fiscal Decentralization and Service Delivery in Indonesia. World Bank Policy Research Working Paper 5614. Khemani, Stuti (2001). Decentralization and Accountability: Are Voters More Vigilant in Local than in National Elections? World Bank Policy Research Working Paper 2557 Khemani, Stuti (2010). Political Economy of Infrastructure Spending in India. World Bank Policy Research Working Paper 5423. Rogers, Halsey F. and Margaret Koziol (2011). Provider Absence Surveys: A Guidance Note. World Bank. Washington DC. Pratham (2008), Assessment Survey Evaluation Research: Annual Status of Education Report 2007 (Rural), Mumbai. Smoke, Paul and Joanne Morrison (2008). Decentralization in Cambodia: Consolidating Central Power or Building Accountability from Below? Andrew Young School of Public Policy International Studies Program Working Paper 08-36. Travis, Phyllida, Dominique Egger, Philip Davies and Abdelhay Mechbal (2003). “Towards Better Stewardship: Concepts and Critical Issues� in Christopher J.L. Murray and David B. Evans (Eds.) Health Systems Performance Assessment: Debates, Methods and Empiricism, WHO, Geneva: Tsai, Lily L. (2007). Accountability without Democracy: Solidarity Groups and Public Goods Provision in Rural China. Cambridge University Press. Vasavakul, Thaveeporn (2009). Corruption in the Health Sector: Management of Service Delivery and Impact on Poverty Reduction in Vietnam. Policy Discussion Paper at the 6th Anti-Corruption Dialogue in Vietnam, Hanoi. World Bank (2003), Making Services Work for Poor People: World Development Report 2004. Washington, DC., World Bank. 27