IFPRI Discussion Paper 01039 December 2010 In Pursuit of Votes The Capture of the Allocation of Local Public Goods by the Central State in Ghana Leah Horowitz Nethra Palaniswamy Development Strategy and Governance Division INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE The International Food Policy Research Institute (IFPRI) was established in 1975. IFPRI is one of 15 agricultural research centers that receive principal funding from governments, private foundations, and international and regional organizations, most of which are members of the Consultative Group on International Agricultural Research (CGIAR). PARTNERS AND CONTRIBUTORS IFPRI gratefully acknowledges the generous unrestricted funding from Australia, Canada, China, Denmark, Finland, France, Germany, India, Ireland, Italy, Japan, the Netherlands, Norway, the Philippines, South Africa, Sweden, Switzerland, the United Kingdom, the United States, and the World Bank. AUTHORS Leah Horowitz, International Food Policy Research Institute Program Coordinator, Development Strategy and Governance Division Nethra Palaniswamy, The World Bank Notices 1 Effective January 2007, the Discussion Paper series within each division and the Director General’s Office of IFPRI were merged into one IFPRI–wide Discussion Paper series. The new series begins with number 00689, reflecting the prior publication of 688 discussion papers within the dispersed series. The earlier series are available on IFPRI’s website at http://www.ifpri.org/publications/results/taxonomy%3A468. 2 IFPRI Discussion Papers contain preliminary material and research results. They have been peer reviewed, but have not been subject to a formal external review via IFPRI’s Publications Review Committee. They are circulated in order to stimulate discussion and critical comment; any opinions expressed are those of the author(s) and do not necessarily reflect the policies or opinions of IFPRI. Copyright 2010 International Food Policy Research Institute. All rights reserved. Sections of this material may be reproduced for personal and not-for-profit use without the express written permission of but with acknowledgment to IFPRI. To reproduce the material contained herein for profit or commercial use requires express written permission. To obtain permission, contact the Communications Division at ifpri-copyright@cgiar.org. http://www.ifpri.org/publications/results/taxonomy%3A468� iii Contents Abstract v Acknowledgement vi 1. Introduction 1 2. Formal Institutions of Local Government and the Targeting of Local Public Goods in Ghana 3 3. Capture of Local Governments through Informal Institutions 6 4. Research Design 8 5. Informal Institutions of Local Government Targeting Local Public Goods in Ghana 11 6. Case Studies 18 7. Conclusion 21 References 23 iv List of Tables 1. Selected district statistics 9 2. Access to services in the study districts 10 3. District A 19 4. District B 20 List of Figures 1. De facto decentralization in Ghana 4 2. De facto decentralization in Ghana – formal and informal institutions 17 v ABSTRACT Decentralization reforms have been the overwhelming response to failures in the targeting of public resources by the central state in developing countries. The policy debate on decentralization typically revolves around several a priori hypotheses on how the design of formal institutions of local government, such as electoral rules, affects accountability in the provision and targeting of public goods. Yet a growing body of research suggests that many rules that structure political incentives and policy outcomes are informal. Indeed, it is widely acknowledged that informal rules such as legislative norms and clientelism can strongly influence political behavior and policy outcomes. This evidence makes a compelling case for both the impact of these informal institutions on political incentives and their role in complementing formal institutions and shaping the status quo when formal venues are absent or weak. How much do these formal institutions matter? How do they shape policy outcomes? In particular, do they merely substitute for weak or absent formal institutions or do they exist alongside and dominate these formal rules and institutions? In this paper, we examine the effect of informal institutions on decentralized public-resource allocation in Ghana. The decentralization policy debate in Ghana, as elsewhere, typically focuses on the role of formal institutions of local government in the targeting of local public resources. Through a comparative case study of two districts in northern Ghana, we argue that informal institutions, grounded in the rationale of partisan politics of the central state, are the key determinants of decentralized public-resource allocation outcomes. In particular, we show that this political rationale is expressed through an informal model of vote buying, and this vote buying is dictated by a national political agenda. Our findings suggest that ignoring this informal institution is likely to undermine the current efforts to reform decentralized public-resource allocation in Ghana. Keywords: local government, informal institutions, decentralization, local public goods vi ACKNOWLEDGEMENT We are a deeply indebted to Kipo Jimah for sharing his insights on the District Assembly in Ghana and for his expert facilitation of interviews. We are also grateful to Regina Birner and Shashi Kollavalli for invaluable guidance and comments. Emelia Guo, Fatahi Abdullai, and Sala Sulemana provided outstanding research assistance. 1 1. INTRODUCTION Failures in the provision and targeting of public goods and services have long been a concern in the developing world. Analyses of direct spending by central government programs in several countries consistently show that such spending favors regions that are more politically important for the central government (Khemani 200). Given the importance of public services to poverty alleviation and development (Besley and Ghatak 2004; World Bank 2004 such mistargeting is expected to have serious developmental consequences. The failure of central governments to equitably distribute resources across regions and social divisions has been rooted in failures of accountability to these groups. Decentralization, defined herein as the devolution of political decision making power to local governments, has been the overwhelming response to failures of centralized governance in developing countries. A priori it is supposed that local politicians’ accountability to a smaller number of voters for electoral success and local constituents’ greater information about the performance of their representatives dampen the incentives to target resources to disproportionately benefit certain communities or specific social groups. Following this logic (or pressured by donors who do), states in a growing number of countries have shifted a considerable amount of resources to be distributed at the discretion of local governments. Local political and administrative actors in many places thus play a newly significant role in affecting the targeting and allocation of public goods. While these decentralization reforms hold the potential to address these targeting failures, it is also widely acknowledged that formal rules of governance, implemented as they are by weak institutions tend to be weak in developing country contexts (Bardhan 2002). The limits of these formal rules of governance could then limit the potential of these reforms. The debate on the effect of decentralization reforms typically revolves around the question of how the design of formal institutions of local government affects concepts such as accountability and democracy in the provision and targeting of public goods (Ayee 2005, Besley et al 2004, Crook 1994, Crook and Manor 2008). However, evidence from some countries suggests that these formal institutions do not explain political incentives and therefore the subsequent policy outcomes (Desposato 2006). In China, for example, local officials compensated for the state’s inability to raise revenues and provide public goods by mobilizing resources through the temple and lineage associations of which they were a part (Tsai 2007). In the Indian state of Karnataka, traditional village leaders bolstered the authority of newly elected local council members in some instances, while competing with them in others (Ananthpur 2002). Hence informal institutions can complement or substitute for ineffective formal institutions as in the case of China, or they can define, depending on the context, the incentives to comply with formal rules or subvert new formal structures of the state as in the case of Karnataka. Indeed, these examples are part of a growing body of evidence that suggests that many rules” that structure political life are informal and that these rules are created, communicated, and enforced outside formal channels (Helmke and Levitsky 2004). This evidence on informal institutions makes a compelling case for considering both the impact of these informal institutions on political incentives and their role in complementing formal institutions and shaping the political outcomes when formal institutions are absent or weak (Helmke and Levitsky 2004). The key questions then are as follows: How much do these informal institutions matter? How do they shape policy outcomes? In this paper, we examine how informal institutions of decentralized public-resource allocation work (if they do) in the case of Ghana. The decentralization policy debate in Ghana, as elsewhere, focuses on the role of formal institutions of local government in the targeting of local public resources. Through a comparative case study of two districts in northern Ghana, we identify the informal institutions that are pertinent to public-resource allocation outcomes, and we examine how these institutions influence the decentralized targeting of public resources. Ghana, with its relatively new class of political leaders and decentralized institutions of local government, presents a particularly interesting case to study. While the decentralization reforms that have been adopted are far from complete, they have shifted a significant amount of resources to local governments or District Assemblies (DAs) along with the devolution of what appears to be legitimate 2 political power to local representatives (Crook and Manor 1998). DAs, therefore, now serve as the node for the within-district targeting of village- or community-level public goods. Comparing two districts with identical formal rules of decentralized targeting allows us to isolate the effect of informal institutions while holding the effect of the formal rules constant. The majority of work on policies of decentralized provision and targeting of public goods has hitherto focused on the Indian case and more specifically on a formal rule of mandated representation. What emerges is that the performance of the formal mandate in Indian local government is highly context specific and varies across different implementation strategies, institutions of accountability, and one could hypothesize, informal institutions (Palaniswamy and Krishnan 2008; Ban and Rao 2008a, 2008b; Besley, Pande, and Rao 2007; Besley and Ghatak2004; Bardhan and Mookherjee 2006; Chattopadhyay and Duflo 2004).Even within this growing literature, there is nearly no evidence on the form and nature of informal institutions that simultaneously determine policy outcomes. In Ghana, the decentralization policy debate has focused largely on how the formal structures of local government affect the process of decentralization (for example, is it participatory or democratic?). It rarely considers what shapes the outcomes of decentralization, as measured by resource allocation to needy populations, and completely fails to consider the critical impact of informal institutions—either preexisting ones or those created alongside new formal rules—on outcomes. That informal institutions matter to the targeting of local public goods is evident from the story told by our case studies. We argue that informal institutions, grounded in the rationale of partisan party politics of the central state, are the key determinant of decentralized public- resource allocation. In particular, we show that this political rationale is expressed through an informal model of vote buying and that this vote buying is dictated by a national political agenda. Our findings also suggest that ignoring this informal institution is likely to undermine the current efforts to reform formal institutions of decentralized public-resource allocation in Ghana. How institutions of local government should be structured so as to ensure a fair and efficient use of public resources is a central issue in the debate on the role of decentralized governance in improving the quality of government in the developing world (Besley et al. 2004). This paper contributes to this debate in two important ways. First, from a theoretical perspective, it contributes to the scarce empirical evidence on the determinants of decentralized public-resource allocation outcomes in developing country contexts and to the literature that tests for the impact of informal institutions. Second, from a practical standpoint, it contributes to answers to questions of how institutions that allocate local public resources should be reformed and strengthened in the Ghanaian context. The rest of this paper is organized as follows. Section 2 describes the formal institutional context within which local public goods are allocated in Ghana. Section 3 proposes a model of capture of this public-good agenda through informal institutions of local governments. Section 4 describes the research design, and sections 5 and 6 discuss the results of our case studies. Section 7 offers concluding remarks. 3 2. FORMAL INSTITUTIONS OF LOCAL GOVERNMENT AND THE TARGETING OF LOCAL PUBLIC GOODS IN GHANA In 1988, facing domestic and international pressure to democratize, the military administration of Jerry John Rawlings and his Provisional National Defense Council initiated Ghana’s current decentralization program. These decentralization reforms, which were institutionalized in the Constitution of the Fourth Republic four years later, created a multitiered local-government structure. The upper tier of this government is the District Assembly (DA), the highest political authority in the newly demarcated districts. The DA serves as a legislative body and is made up of elected and appointed members. It includes an executive unit that is meant to incorporate under 22 deconcentrated line departments (of specific ministries of government). The legal framework for local governments assigns the DA with 86 total functions, most of which are related to local service provisions including those related to health, education, agriculture, public works, communications, and public safety (Ayee 1995. The DA also has responsibility for district planning and budgeting and the authority to collect taxes, tender contracts, and pass bylaws. In practice, however, instead of a broad formal mandate for the responsibility of specific services, the DA is assigned the narrower task of providing and allocating (targeting) local public goods within the district. For example, in the case of education, the DA is tasked with providing specific public goods relevant to education, such as school buildings. The overall responsibility for the provision of education, however, lies with several actors, most of whom are either outside the local government, are not accountable to the local government or both. While DAs have some funds that can be used entirely at their discretion, it is often the case that the use of these funds is biased toward more visible physical infrastructure investments, and "soft investments," such as school quality, are most often not a preferred use of public resources. Political incentives that are stacked against such “invisible” investments offer a clear explanation for such a pattern (Keefer and Khemani 2005). Providing local public goods—that are located in particular towns or villages (henceforth, called communities)—is, therefore, the primary mandate of the DA. Therefore, DAs are the key providers of community-based public infrastructure such as schools, clinics, latrines, roads, and water services, which garner clear political credit. Local public goods provided by the DA are largely financed from two sources. The first source, the District Assembly Common Fund (DACF), is a central government grant that is allocated to districts based on a parliament-approved formula. The second source is comprised of donor funds that are devolved, typically through specific projects, directly for the district. In our two study districts, the DACF accounted for between (29% to 51% with an average of 38% ) one-total district expenditures while donor funds (from multilateral and bilateral donors) accounted for (31.8% to 54.2%, average 41%). of total district expenditures. Together, these two funds account for at least 75% of total district expenditures1 The DA is formally headed by the district chief executive (DCE), a presidential appointee with far-reaching political and administrative influence within the district. The Constitution of Ghana mandates the DCE to chair the Executive Committee of the DA, which sits atop the DA subcommittee system (Ayee 1995; Crook and Manor 1998; Crook 1994). As such, the DCE must approve all decisions made by the elected and appointed membership of the DA. As the chair of the District Tender Board, the DCE plays a significant role in awarding contracts. As the chief executive officer, the DCE signs off on all expenditures, correspondences, and public statements of the DA administration. Indeed, the formal rules that mandate the DCE as the head of the district accord overwhelming powers to this nonelected and . The majority of the DACF funds are not tied to use, and the DAs decide on the questions of what public goods to provide and where to locate them. However, donor funds are typically tied to use, and the DA merely makes the decision of where to locate them. The task of targeting local public goods, therefore, rests primarily with the DA. 1 The remainder of district expenditure comes from special grants from the central government( 4% to 23% with an average of 14% ) and internally generated funds (1% to 14% with an average of 5.7%). 4 politically partisan head of the local government; the amount of discretion the DCE wields over management of DA affairs and its resources cannot be overstated. The DCE also sits atop the legislative and the bureaucratic institutions of the DA. The legislative institution of the DA includes both elected members who represent specific electoral areas and appointed members. Two-thirds of all members are elected, and appointed members are intended to be prominent citizens of the district drawn from different walks of life. Formally, this legislature is supposed to make and approve all decisions made in the DA. Yet, because of structural imperfections of the decentralization system that grants power over management and budgetary decisions to the executive office of the DCE, this legislative institution of local government exercises close to no control over the affairs of the DA (Ayee 1995, Ayee 1996, Crook and Manor 1998; Crook 1994). The formal rules also envisaged that the elected members of the DA would serve as locally rooted advocates for development in the communities that they represent. These elected members were, and are, for the most part genuinely “representative” of their electorates. A large majority of them fulfilled official injunctions to work closely with their communities in eliciting needs and in representing these needs to the DA (Ayee 1990, Crook 2004, Crook and Manor 1998). It is also widely acknowledged that the DCE dominates the bureaucratic institutions of local government. This domination is manifest in the routine subversion of planning mandates with which the bureaucracy is formally tasked. The de facto devolution of powers, both political and administrative, to local actors is then significantly more limited than that envisaged by the formal rules. This de facto form of decentralization is depicted in Figure 1, where the solid lines represent the de facto rules that that govern the allocation of local public resources and the broken lines represent the ineffective nature of the formal rules of public resource allocation. Figure 1. De facto decentralization in Ghana Source: Authors, based on key informant interviews The decentralized provision of public goods typically calls for the devolution of power away from the hitherto all-powerful central state. Yet, de facto, decentralization in Ghana is characterized by the domination of the formal institutions of local government—both political and bureaucratic—by an appointee of the central government. This dichotomy between the powers that should be devolved and that are devolved is, however, not particular to the Ghanaian case. In the 1980s, in Kenya the form of 5 administrative decentralization was closely tied to the political ambitions of President Moi (Barkan and Chege 1989). More recently, the state legislature in the Indian state of Karnataka tried to pass a bill that sought to limit the powers currently accorded to local governments in the state(India Together,2009). That political considerations of the central government play a role in designing de facto devolution to local governments is not surprising. Indeed, Devarajan, Khemani, and Shah (2008) call this phenomenon of limited devolution to local governments “partial decentralization,” and they suggest that this phenomenon could, in part, reflect an attempt by higher-level governments to hold on to political power, financial power, or both. Khemani (2010) argues that such partial decentralization is selected by politicians in the face of increasing participation by swing voters in elections and that this choice facilitates the vote buying imperatives of the central government. Vote buying, in this model, is manifest in the provision of targeted and clientelistic transfers to swing voters; such vote buying supersedes the importance of providing more broad-based public goods. In this model, the political capture of local governments is the result of the political imperatives created by swing voters; however, it has implications only for the design of formal institutions of local government. In particular, it results in the creation of formal jurisdictions of local government that are both grant dependent and small. Formal rules, which govern the design of how local public goods are provided and the institutions of local government, then take precedence in this explanation of the political economy of decentralized public-good provision. The relevance of informal institutions to political behaviors and outcomes, however, suggests this exclusive focus on the formal presents, at best, an incomplete explanation for observed policy outcomes. Indeed, Helmke and Levitsky (2004) argue that a key motivation for creating informal institutions emerges when certain goals—such as the centralized control of public service provision—are not considered publicly acceptable. The inconspicuousness of informal institutions allows the pursuit of these unacceptable goals. The questions that emerge are the following: What are informal institutions of local government that are associated with this political economy of decentralized public-good provision? How do these informal institutions matter to the observed public-good provision and targeting outcomes? Indeed, if informal institutions trump formal rules, as they do in several instances (See Helmke and Levitsky (2004) for examples), then an exclusively formal explanation can be both misleading and incomplete. We propose a model where the political capture of local governments takes place through informal institutions. 6 3. CAPTURE OF LOCAL GOVERNMENTS THROUGH INFORMAL INSTITUTIONS We propose a model of political capture of the public-good provision agenda of local governments, and we argue that this capture takes place through informal institutions. Even when the responsibility for service provision has been formally transferred to local governments, central governments may still have incentives to assume responsibility for service provision; that is, they are motivated when citizens reward them for such provision. In a system of partial decentralization, citizens continue to place their expectations of service delivery with national governments. Indeed, citizens are more likely to punish national politicians rather than local ones for inadequacies of service delivery (Devarajan, Khemani, and Shah 2007). The strength of these political incentives, however, depends on the relevance and strength of identity- based voting when citizens vote based on fixed social identities (Keefer and Khemani 2005). When identity-based voting is widely prevalent, politicians are less likely to be judged on their record on public service provision. When such voting is less relevant, and voters tend to be swing voters, there are clear political incentives to capture the public-service provision agenda of local governments (Khemani 2010). How does such capture take place? The question of how, or the mechanism through which such capture takes place, has hitherto focused on formal institutions. In contrast, we argue that informal institutions are more likely to shape, enforce, and mediate such capture. Indeed, informal institutions have been known to shape the performance of formal institutions in unexpected ways, mediate the effect of formal rules, and create and strengthen incentives to comply with formal rules. In this context, ignoring these institutions could even lead to misleading explanations for observed policy outcomes (Helmke and Levitsky 2004).While formal rules and institutions of local government could reflect political capture (as in Khemani 2010), we argue that an exclusively formal explanation of such capture is, at best, partial. Following Helmke and Levitsky (2004), we define informal institutions as “socially shared rules, usually unwritten, that are created, communicated, and enforced outside of officially sanctioned (formal) channels.” In this paper, vote buying is defined as the strategic targeting of local public goods so as to garner political credit (votes) from the group that has access to the public good (Diaz Cayeros and Maglioni2003). In particular, it does not refer to clientelistic and other private transfers targeted to individual voters.2 While the formal rules stipulate that elections to the District Assembly (DA) be nonpartisan, it is widely acknowledged in Ghana that these elections are, in fact, partisan and that the local machineries of political parties play a critical role in the elections. National elections in Ghana are also closely contested. Indeed, studies of the empirical determinants of voting show that Ghanaian elections are unpredictable and that the large majority of voters are swing voters who do not vote, on account of ethnic or other identities, on fixed party lines. In particular, the vast majority of Ghanaian voters do not identify through fixed identities with either of the two major political parties, the National Democratic Congress (NDC) and the New Patriotic Party (NPP). (Bossuroy 2007; Fridy 2007). As Fridy (2007) points out, the vast majority of Ghanaian voters view the major political parties- the NPP and the NDC as representative of Asante and Ewe (two major ethnic groups that typically align on fixed party lines) interests respectively. Therefore, the majority of these voters do not themselves vote as a block; instead they base their evaluations of the Asante and Ewe parties ultimately on issues other than ethnicity. This makes the electoral game for votes a closely fought one. The control over the provision of local public goods could be a very valuable tool in any attempt to win these elections. At the time of this study, such clientilistic transfers to voters were limited within the scope programs implemented by both the district governments in Ghana. In this context, vote buying, on account of political imperatives, can take two forms. The first form of vote buying is manifest in the agency of the central government that is built into the formal design of decentralization policy (as is typical in a system of partial decentralization). The second form 2 Such vote buying, through pork barrel politics, is indeed a strategy used by governments all over the developed and developing world. 7 takes place through the informal institutions that are manifest in party machineries and that either operate alongside or dominate formal institutions of local government. Both forms of capture, through the formal and the informal institutions, are determined by the electoral logic facing the central government. Following this logic, central politicians are likely to court electoral support through the strategic targeting of the provision of public goods. This implies that patterns of public good provision vary across the country and that these decisions depend on the nature of the electoral market in place. In regions of established support for a political party, where citizens vote based on fixed party identities (ethnic or otherwise), governments are likely to continue to court support through the provision of goods to the these core support groups. In regions where voters are not already so committed, they are more likely to hold politicians accountable for inadequacies in service delivery. Service provision then tends to be broader based in these swing regions relative to the regions of core support. Closely contested national elections, however, mean that to win elections, national politicians need to target both core and swing voters. Hence, national politicians faced with partial decentralization and with electoral competition have incentives to capture the provision of local public goods. In systems with entrenched political parties, party machineries are likely to naturally aid in this capture. The incentives that underlie both formal and informal institutions through which vote buying is implemented lie in the nature of electoral markets, and party machineries are likely to be pertinent to both institutions. Even the formal institutions of local government—be it the creation of small grant-financed jurisdictions (see Khemani 2010) or the formal appointment of a central agent as the head of the district as in Ghana—are likely to require a mechanism through which vote buying can be mediated and enforced. For example, even when jurisdictions are dependent on the central government grants, the knowledge of how these grants should be allocated across jurisdictions is likely to come from institutions associated with the central government, such as parties and their machines, rather than the head of the state and the executive alone. However, political parties have not been accorded a formal role in local government in most decentralization reforms adopted by developing countries. This implies that the political party, or other informal institutions of a similar nature, cannot find a public profile through the formal institutions of local government. We hypothesize that within a context of a system with contested elections and strong party machineries, political capture is likely to take place through informal institutions of vote buying. We also hypothesize that such vote buying through informal institutions is likely to be the dominant determinant of targeting local public goods. Following Helmke and Levitsky's (2004) classification of informal institutions, we consider vote buying as a competing informal institution, where a competing informal institution is defined as an institution that structures incentives in ways that are incompatible with the formal rules. While the divergence from intended policy outcomes is not surprising per se, and while this can be inferred from the formal agency of the central government in local government affairs, the questions of how and why the divergent outcomes are obtained are of critical importance. We argue that informal institutions constitute a critical part of the answer to why. We now examine, through a comparative case study, whether these hypotheses hold in the case of Ghana. 8 4. RESEARCH DESIGN Ghanaian districts, with their differences in the nature of political competition (electoral markets) and the associated differences in political incentives, provide a perfect opportunity to test our hypotheses on the informal institutions of vote buying. In particular, the differences in the nature of political competition, or electoral markets, across districts allow us to test if the provision of public goods is strategically targeted as a model of vote buying through informal institutions would predict. Examining the variation in patterns of public service allocation within the same country also allows us to hold constant the force of the formal rules of governance in general and those of decentralized targeting of public goods in particular. This then allows us to isolate the role of informal institutions associated with electoral markets in determining outcomes. It is also widely acknowledged in Ghana that these elections are, in fact, partisan and that the local machineries of political parties play a critical role in the elections. Indeed, there is extensive anecdotal though little systematic evidence on the importance of political parties in both local and national elections. The presence of partisan local politics alongside differences in the nature of electoral markets presents us an ideal setting within which the role of informal institutions of vote buying might be tested. The research reported here was conducted by the International Food Policy Research Institute (lFPRI) during February and March 2008 in two neighboring districts in the northern region of Ghana. A multiple-case embedded research design was used, meaning that data was not pooled across the sample but rather clustered into cases (Yin, 2004). The primary unit of analysis was the individual service- delivery case (for example, a school, borehole, or electricity line). However, because each individual case of resource allocation is embedded in and predicted\by political institutions that operate at the community, electoral area (EA), and District Assembly (DA) levels, data used in the analysis of each case were collected at each of these levels. An EA is defined at the sub district level. Each DA in Ghana comprises several EAs, each of which is represented by an elected member of the DA (referred to as the “assembly member” or “assembly person”). This assembly member typically represents more than one village or community. The case studies covered 10 communities in six EAs in two districts. A larger IFPRI survey on decentralization and rural service delivery in Ghana ,that was conducted in March 2008 ,defined the sampling strategy for the district pairs chosen. This survey covered 3 district pairs3 In each district, the case studies profiled the district chief executive (DCE), (several) bureaucrats of the local government, and three assembly persons. Each of the assembly persons represented the chosen EAs. These EAs were selected based on their definition by multiple sources as "high-performing" as well as consideration of heterogeneity in gender, tenure, age, strategy, and religion. The profiles included information on individual cases of service delivery and the determinants of each of these individual cases. The major data collection techniques were key informant interviews, including home stays (6 districts in total) in three different agro-ecological zones of Ghana. The chosen district pair in the northern region presents us with a very interesting variation: one district is widely identified as a district that votes based on fixed party lines while the other district in the pair is the quintessential swing voting one. Three EAs were sampled in each district. These EAs were also a subset of a random sample of assembly people covered by the 2008 FPRI decentralization and rural service delivery survey. 4 Because our primary unit of analysis, the individual service-delivery case, is located at the level of the community, we also collected data at the community level. Where the sampled assembly people represented more than one community, two communities were studied: the assembly person’s hometown , focus group discussions with community opinion leaders, and community meetings. Secondary data on the provision of local public goods was also collected from the Electoral Commission and the DA. 3 The district pairs were chosen to illuminate the effect of different DA administrations within similar physical, social, and financial environments. The inclusion of one district headed by a female district chief executive was also a factor in the selection. 4 interviewers stayed in the homes of respondents for several days at a stretch and observed and interviewed them. 9 and either the biggest or the farthest community. The biggest communities yielded insights about assembly people’s accountability to locations with high service pressures and the farthest communities revealed intra-EA dynamics created by spatial dispersion in rural areas. Our study districts—called districts “A” and “B” (in order to preserve anonymity of the respondents)—offer two models of electoral markets, each of which define a different model of vote buying. Of our two districts, B has a long history of supporting one political party, while district A is known to be a swing voting district. Indeed, the sitting member of parliament from this district does not come from the current ruling party. Our districts also differ ethnically and politically. District A is one of the most ethnically diverse in the region, and the dominant and most populous ethnic group is long known as one that does not identify ethnically with either of the two major parties. District B, in contrast, is more ethnically homogenous and has historically been tied to a single party, with this fixity being rooted in more fixed and inflammatory ethnic identities. The dominant (and most populous) ethnic group in the district has been characterized as lifelong supporters of a single party, and this calcified identity is based on their sub-ethnic group. Table 1 presents basic demographic information, ethnic fractionalization indices, and information on the local government structures from the two districts. Table 1. Selected district statistics District A District B Area 17,317 km2 2,389 km2 Population 76,702 (2000) 132,880 (2000) Population density 8/km2 55/km2 Poverty head count ratio 57.2 83.5 Ethnic fractionalization index 0.66 0.19 Number of area councils 6 12 Number of electoral areas 20 42 Unit committees inaugurated 74 percent (2007) 7 percent (2007) Number of communities 102 220 Total funds received in 2007 1,903,267 GHS 2,014,748 GHS Internally Generated Funds 7.6 percent 12.1 percent Funds per capita 24.8 GHS 15.1 GHS Source: Authors compilations from 2003 QWIC surveys and DA data. Note: No census has been conducted since the District A was bifurcated: so population estimates for district B are an estimate. The value of the funds received in 2007 includes an estimate for the Heavily Indebted Poor Countries (HIPC) funds- this estimate is based on the average of the preceding two years. Our study districts neighbor each other in Ghana’s disadvantaged northern region and are predominantly agricultural, with more than 85 percent of the population categorized as rural. District A boasts the lowest poverty rate in the region (Table 1). With a substantial population reduction when part of the district was carved off to create a new one, its total funds available per capita are also more than twice that of its more populous neighbor. Nonetheless, district A still remains a large district, and this size defines the service delivery challenges for the DA. Mobility challenges caused by low road density and low access to services centered in the district capital, such as secondary schools, hospitals, and agricultural extension are particular constraints. The DA of district B serves a much larger and poorer population. More than 80 percent of the district residents fall below the poverty line, and the per capita expenditure is much lower than in the neighboring district. Yet in recent years, total revenues in district B have roughly equaled those of its neighbor. 10 Nevertheless, as Table 2 shows, district B has similar or better access to major infrastructure, notably roads and social services, compared to district A. One reason for this may be its much closer proximity to and linkages with the regional capital; many staff commute daily from there to their jobs in the district. However, providing clean drinking water remains a severe challenge in the southern half of the district, where the water table is too low for tapping. Additionally, Guinea worm -a parasitic disease that is contracted from open water sources - is endemic to the region, and while district A has eradicated this parasite, most open water sources in district B remain contaminated. Our study districts, therefore, represent a case where access to services does not have a clear relationship with poverty. Indeed, the district with better services is poorer. Table 2. Access to services in the study districts District A District B Region Poverty head count ratio 57.2 83.5 - Access to safe water source (2007) 53.2 54.1 54.4 Access to sanitation 8.1 19.5 23.5 Road density .04 .15 - Access to roads 48 77.1 - Access to electricity 14.3 18.4 28.8 Access to health facilities 41.3 54.2 35.0 Access to primary school 80.8 96.7 80.1 Source: Authors compilations from 2003 QWIC surveys and DA data. Note. All access figures refer to the percentage of households in the district who have access to the given public good or service. 11 5. INFORMAL INSTITUTIONS OF LOCAL GOVERNMENT TARGETING LOCAL PUBLIC GOODS IN GHANA In this section, we draw on the results from our case studies to demonstrate that informal institutions are an empirically relevant and important determinant of vote buying in Ghana. Vote buying influences the targeting of local public goods in three important ways. First, it influences this distribution through the incentives faced by the district chief executive (DCE. These incentives then result in the exercise of the formal mandate of general oversight of the affairs of the District Assemblies (DAs) in rather far-reaching ways. Second, vote buying influences the distribution of the local public goods through the informal domination by the DCE of both the legislative and bureaucratic institutions of the DA. Third, vote buying rather paradoxically make the informally disempowered elected members of the DA work as true advocates on behalf of their constituents. We now examine each of these informal institutions. Vote Buying through the Office of the District Chief Executive “I am a politician. I am here to implement the programs of my government. My job is to work for my party, to help my party do well in the elections.” —District chief executive, C14 There is a common consensus in Ghana that the work of the DAs is tinged by partisan considerations, largely because of the clear political allegiance of the district’s most important actor. Per the constitutional statute that created it, the DA system is officially nonpartisan (Secs. 248.1-2, Constitution of Ghana).5 To win votes and increase party penetration, however, the DCE must use DA resources to woo communities. The following two observations point to the mechanics of electoral markets in play. A planning officer noted: Yet at the helm of the local government sits an unconcealed partisan. The DCE is appointed by the President of the Republic of Ghana and can be removed by him/her at will. In fact, individuals interested in the post apply directly to the party machinery: Party executives at the constituency level handle recruitment, while the regional minister, minister of local government, and office of the president vet, interview, and nominate applicants. This selection process means that DCEs are first and foremost accountable to the national party machinery of the ruling party, a fact well documented in the literature on Ghana’s decentralization (Ayee 2003, Ayee 1996). As an explicitly partisan actor, the DCE faces two types of political pressures. First, as a political appointee of the ruling party, one of the DCE’s key performance measures is his/her ability to increase the penetration of the ruling party in the district. As one DCE opined, “At the end of the day, if you are there and you don’t make an impact for the party, your stay might as well have been useless.” Second, DCEs are often major contenders for the parliamentary seat in their district, taking advantage of their proximity to the constituency relative to the Accra-based member of parliament (MP). Both these political pressures make the DCE a key player in electoral markets for national political gains. Which community gets what is politically based. Because our current government has its supporters in some communities, okay. It will prioritize giving projects to such communities that support the current government more than those that do not support. Every government wants to stay in power. And that actually determines the siting of projects more than other factor. 5 Secs. 248.1-2: Candidates seeking election to a District Assembly (or lower local government unit) may not affiliate with any political symbol, and no political party can endorse, sponsor, or campaign for a candidate.6 A8,A9,A10,A11,A14,A15,A17,A27- Respondent numbers, assigned to preserve anonymity of respondents. 12 In contrast, several DCEs argued that if there was any bias in their allocation decisions, it was, in fact, in favor of opposition communities. These competing observations lay out the two underlying political rationales of local electoral markets. It follows that the DCEs, therefore, need to buy votes, and how they do this depends on the nature of the electoral market. If a DCE presides over a district with several swing voters then the allocation of public goods needs to be more broad based. In a politically unpredictable region, public goods clearly need to reach a wider range of communities as targeting core support is no longer a sufficient strategy to win elections. If, however, the district has several voters who vote, for whatever reason, along fixed party lines, then public goods need only be targeted to those communities that offer their loyal support. Therefore, while public-resource allocation decisions in core voting regions might resemble a quid pro quo or patronage, these decisions are in effect determined by the underlying political logic of electoral markets. This logic finds expression through the exercise of wide- ranging informal authority by the central government through the office of the DCE. Indeed, vote buying is the ubiquitous response to this logic, and it emerges as a critical determinant of public resource allocation in the status quo. The incentives of the DCE to act in the genuine interest of the party are driven by political aspirations. Indeed, the tussle between the aspirations of the DCE to be an MP and the aspirations of the sitting MP have been well documented (Ayee, 2003). The thoughts of a DCE planning to run for MP in the nearest national election will turn toward campaign finance, which hinges on doling out favors, especially to wealthy party activists. This means that the DCE’s own political aspirations provide the pressure to channel resources to instrumental individuals. The DCE, as chair of the Tender Board, has a strong influence over who work on DA contracts. Where contractors are also district-party power brokers, our evidence suggests that the DCE may give them undue preference to win their support. In the district where contractors in the district did not have close ties to the DCE’s party, there were fewer party-based pressures to favor certain contractors over others. Indeed, several people6 In both the political imperatives and contracting pressures facing the DCE, therefore, political parties play a critical role in brokering the status quo. Indeed, despite the formal nonpartisanship in the DA, partisan political imperatives are the driving force behind the capture of the targeting of public resources. in the district noted that “[DCE] party people are not contractors, except one man in community x”. In the district where the candidate was running for MP in the upcoming elections and where [DCE] party had a strong presence, it was reported that the party machinery played a key role in handing out contracts and therefore locating public goods. Indeed, a number of observers also pointed out that the need to favor particular contractors meant that the parts of the district where these contractors had a dominant presence would be favored in the provision of public goods. Hence, our evidence shows that the strength of the party machinery and the instrumental value of this machinery to the DCE are critical in linking campaign finance pressures to the targeting of community-based public goods. In particular, a weak presence of the party machinery in the district reduces such campaign finance–induced biases in the targeting. Party machineries and their instrumental value to DCEs, therefore, play a critical role in connecting the dots from the political rationale defined by electoral markets to vote buying in the allocation of local public goods. Indeed, the political rationale is consistent with the incentives of the key political actors in the DA—the DCEs. The fulfillment of the latters’ current and future aspirations hinge upon how useful they are to the goal of increased party penetration. Vote Buying through the Domination of Political Institutions of Local Government I thought the truth came from assemblyman. But now that I'm going to assembly I'm thinking that the [assemblymen] don't have truth. No, you must crucify yourself before your people can get the good services. All I am is a glorified beggar.—Assembly member, B10 6 A8,A9,A10,A11,A14,A15,A17,A27- Respondent numbers, assigned to preserve anonymity of respondents. 13 Despite the formal assignment of power for the within-district distribution of local public goods to the institution of the DA, it is widely known that the DCE makes the final decision on this distribution (Ayee 1995; Crook and Manor 1998; Crook 1994). However, we find that despite their limited powers, assembly members are not passive observers in the DA. Rather, the most effective of them act as lobbyists, working vigorously to attract resources to the communities within their constituency. Indeed assembly members lobby all people for all things- -including public goods and services- that they lack. Why would these assembly members, who neither exercise control over the resources of the district nor receive any meaningful remuneration, go to such lengths to represent their constituents? They lobby because they feel accountable, in some sense, and find their position desirable. Our evidence suggests that the desirability of this position seems to be closely tied to the avenue (and associated incentives) for upward mobility provided by the party machinery. Several assembly members in our sample cited that the desire for upward mobility was the driving force that led them to activate incentive structures to win resources in a game of pork-barrel politics controlled by others.7 However, they too do not know the processes by which they finally merit a project. Indeed, an institutional structure that keeps them ignorant of allocation criteria naturally incentivizes pork barrel politics as the only action they can take on behalf of their constituents. . Ultimately therefore, individual political loyalties and social networks also come into play. How this happens is complicated. In a bid to strengthen their chances of success, assembly persons draw heavily on their political connections. In the lobbying process, the instrumentality of party chairmen on a ruling is particularly evident. There is evidence that at least some assembly members are, in fact, known party activists and that almost all identify with a particular party, at least privately. We find that the rewards from this game of pork barrel depend, in large measure, on the strategic compulsions of the electoral politics in play. Assembly members in our sample repeatedly reported that they deployed electoral politics in their lobbying strategies by playing into deciding actors’ extant motivations. For example, knowing DCEs are concerned with securing the votes of their EAs, assembly members may highlight the political case for investing assembly resources in their area. Alternatively they may send the party chairman to do so. 8 The formal rule of a nonpartisan local government dictates that members must be strategic in how they publicly wield their party identity to best position themselves to extract public goods from the partisan DCE.9 Assembly members also exploit their social networks in nonparty spheres of influence. Instances of influential community members, such as members of the local chief’s house, unit committee members, As one assembly member said, “Where the government is in power, you should have a bit of a close relationships with the [ruling party] [ward] chairman” (assembly member, A4). Even opposition activists at the community level acknowledged the expediency of exploiting the potentially fuller access to the DCE enjoyed by ward party chairmen: “If you want to go to assembly, seeing that NPP is in power, then they will not listen to any problem unless we consult the NPP chairman. So he will lead us to assembly. If they know you are opposition, they won’t heed you, but if the chairman leads people, they will listen” (assembly member, A48). 10 7 In this role they also act as the key intermediaries between community and state. and local businesspeople, themselves lobbying on behalf of the assembly person at the DA, are frequently cited. It is no surprise that extant political imperatives are drawn upon to dominate the institution and the public-good allocation function of the DA. It is, however, imperative to note that the 8 One MP admitted to sponsoring known National Democratic Congress party members to run for DA positions so that they could increase the partisan voting bloc, which was especially relevant during the confirmation of the DCE. 9 However, a majority of our sample preferred to portray themselves as nonpartisan, often citing the statute that stipulates it. One assembly person even recounted a verbal argument engaged with a party executive when he was accused of being a member of the opposition party in front of the DCE. For these actors, keeping secret their sympathy for the opposition maintains their ability to lobby the DCE on other grounds. In contrast, having a clear opposition party identity will likely close fruitful avenues to the DCE. While relevance of revealing party identity is unclear, our evidence on the location of public goods suggests that it is electoral logic that determines which communities receive public goods and that the role of political parties are intricately tied to this logic. Indeed, our case studies also show that a key lobbying strategy adopted by assembly people is, in fact, to play into the extant motivations to increase party penetration in the district. 10 The unit committee is a unit of local government that sits at the level of the community. 14 elected members of the assembly—who adopt lobbying as a ubiquitous strategy to bring public goods to their communities—are incentivized to do so by the avenues for upward mobility provided through political parties. Vote buying imperatives dominate the legislative institution of the DA, and this vote buying is mediated, enforced, and implemented by party machines. Indeed, rewards by party machineries to assembly members—through party position or instrumental support for their personal political aspirations—provide these members with a potent incentive to lobby. It should also be noted that these rewards are handed out by the party and are closely tied to factors that are instrumental in their efforts to win support at the national polls. Vote Buying through the Political Accountability of Elected Members of the District Assembly “As a newly established assemblyman, it is not easy. The people are always on [him] and [he] does not want a failure.” —Newly elected assembly member, A52 “If you don’t lobby, if you just keep quiet, you might get, but at the time they feel like giving it to you. And if you want to wait until they feel like giving it to you, you may not get anything.” —Assembly member, A30 In 1998, Crook and Manor argued that the paradox of genuine political representation by the assembly people, combined with their lack of political power to bring resources to their communities, increasingly make this position undesirable. That the large majority of the members in the sample did not want to stand for re-election was argued to be representative of this negative view of the office. Yet we find that the position of an assembly member continues to be a contested and coveted one. Almost all elections in both sample districts had more than one candidate contesting elections. Indeed, most elections had more than one candidate, and the winning margins in several instances were small. Voters also saw the DA as legitimate and worthy of one’s participation. The election turnout in all the EAs in our sample exceeded the district average of 50 percent, with the number as high as 74 percent in some EAs. Assembly members, in some sense, also felt and were held accountable. While the residential proximity of the assembly member enabled the constituents to exert pressure on him/her, these constituents were also widely aware of the limited powers of the assembly members in the DA. As a result, their communities held them accountable for their “efforts” to bring resources, as evidenced by letters written to the DA and taking other members of the community along to the DA at the time that the lobbying request was being made. In fact, one local observer referred to assembly members as the “doorstep governors” of their communities. Indeed, this political accountability of assembly members plays a critical role in generating incentives for their lobbying efforts. Our case studies show that assembly members lobby anyone who could influence the allocation of resources to their community. Knowing that the DCE is concerned with securing the votes of their EAs means that DCEs present their first port of call for assembly persons. Drawing on party loyalties is also something that all members, vested as they are in these party machineries, do. The second port of call for assembly men is the door of nongovernmental organizations (NGOs), that sometimes implement public projects outside the purview of the DA. While the NGO public-goods agenda is largely predetermined, often times there are choices of location involved. Assembly members actively seek out and lobby these NGOs to bring these public goods to their communities. From the point of view of the assembly person, therefore, it makes sense to adopt a scattershot approach to lobbying, and to therefore lobby anyone and everyone who could matter. The case of a junior high school (JHS) in one of the study communities explicates this scattershot lobbying process. The assembly person first identified the need for a JHS in his community through his own observation as resident and parent. He then lobbied for the individual pieces necessary for the whole, drawing on pre-existing social networks, persuasion, and persistence to gain success. He approached the 15 District Office of the Ghana Education Service (GES) to apply for official recognition and the assignment of teachers in his community. As a teacher, he had maintained good relationships with supervisors and former teachers in the GES district office He decided to “let them all be aware” about his community’s need for a school and told them “they should fight, always be on the director…so they also pushed their hands” and he was able to get the school.11 Elected members of the DA, therefore, have strong incentives to lobby for public resources. This lobbying in turn reflects, in part, the strong political accountability of these members to their constituents. He also successfully lobbied the GES for classroom furniture. To secure funding for the school block, he lobbied both the director of education and the DCE. In particular, he made himself a nuisance of the DCE, visiting him anytime he was at the DA, which was at least weekly to pick up his paycheck. He also tried to activate the DCE’s sympathies as a former assembly member, reminding him that “as a newly established assemblyman it’s not easy. As this story demonstrates, one major strategy assembly people use to strengthen their lobbying is to exploit existing social capital. Networked relationships are particularly important. As this story demonstrates, these grassroots lobbyists must target the entire array of actors who can help their case, including the DCE, DA staff, MPs, line departments, NGOs, and fellow assembly people, especially the presiding member and Executive Committee of the DA. To do so, they may activate loyalties or call on old favors deriving from membership in schooling, employment, religion, ethnic group, hometown, or other networks. For example, one assembly person (respondent A30) attributed both her electoral success and the public goods she was able to secure early in her tenure to the fact that the DCE and her sister had been classmates and maintained a close relationship. The assembly person depicted above drew on professional connections to push his hand at the relevant line department. Vote Buying through the Capture of Bureaucratic Institutions of Local Government “Sometimes the nature of your problem will let you get the project.”—Assembly member, B8 In several instances—such as in the construction of boreholes or the allocation of funds for fighting malaria—formal rules envisage that bureaucratic targeting by need, loosely defined, will determine within-district allocation of resources. However, these targeting mandates are uneven in their presence and implementation. While the directors of departments are thought to be most influential in proposing project selection, ultimately the DCE—as both chair of the Executive Committee and chief executive—must approve the project. As discussed above, the DCE can all too readily subvert bureaucratic targeting if it suits his/her other needs. However, to the extent that DCEs feel that their performance on development metrics affects the sustainability of their job, they have incentives to support this type of targeting for two reasons. First, they have reason to believe that the adequate implementation of needs-based projects can result in the flow of more resources to the district. In one of the study districts, the DCE referenced a major donor-funding modality in this respect. Project money for the District Wide Assistance Program was time bound, and the DCE felt that if she distributed contracts along partisan lines, she would likely be pressured to hire less skilled contractors, who, in doing shoddy or incomplete work may not meet the donor’s time line. This could impede future funding. Because this happened in three of the districts in the northern region in 2007, she saw these sanctions as having real teeth, and she felt this was important because the District Assembly Common Fund (DACF) allocation to the district was too small on its own. The DA relies on donor money to deliver development, and in turn, to reflect the DCE’s competence in this realm. She felt that as a DCE “you must deliver; otherwise you cannot remain at your position.” Second, as the president’s man or woman in the district, the DCE is implicitly (or explicitly) held responsible for the performance of central-government flagship initiatives in his/her area. This metric means that the DCE must steer the bureaucracy to make investments that will help the area conform to 11 He was told that if he collected the names of all the students who would comprise the incoming class, they would consider his case. 16 regional or national statistical trends. Indeed, in a few service-delivery cases in this study, the overriding logic appeared to be need-based targeting. The case of the Wantugu borehole is illustrative. Not only were women there spending hours daily searching for water during the dry season, but the community also had one of the highest incidences of guinea worm in the northern region. The community had a break of luck when one of the NGOs in the district drilled a borehole outside of town and hit a major aquifer. Seeing the high water yield, the DA’s Water and Sanitation Team sought funds for the mechanization of the borehole. After several donors with sanitation or infectious disease missions signed onto the project with tied funds, the DA staff informed the assembly member that he would be receiving the project and that he would need to organize community contributions of cash and labor. In the Wantugu borehole project, the motivating logic for project distribution appeared to be developmental and referent to the needs of beneficiary populations. Wantugu, in fact, presented a classic case of a problem that was visible and on the radar of the national government. For example, in 2008, the national press media covered the declining incidence of guinea worm in the northern regions of Ghana where it is endemic, citing star and lagging districts (for example, (Ghana News Agency,2008). Northern DCEs thus had a political interest in achieving this developmental outcome. This political interest resulted in the motivation to steer their DA to deliver its water infrastructure in a way that would best achieve this outcome. Enrollment in the NPP government’s National Health Insurance Scheme, School Capitation Grants, National Youth Employment, and a school feeding program are other examples of cases where the DCE’s political interest motivated the implementation of need-based targeting mandates. Of course, the DCE does still accrue political credit through the disbursement of projects whether or not they were actually targeted by need. Hence targeting mandates that are both visible and deemed to be of national importance are implemented. Mandates not deemed to be of importance are routinely subverted, and this subversion is often follows the logic of national politics. Once again, the imperatives of electoral politics and the political aspirations of the DCE determine how his/her discretion is used, and this, in turn, determines which planning mandates will be implemented and how. That bureaucrats are, for the most part, passive observers of this capture in not surprising. Although the DCE does not directly hire and fire the bureaucrats, he/she is their boss and could likely lobby the relevant central ministry in Accra to get them moved. Bureaucrats in the DA are likely to follow the DCE’s demands, especially in the absence of competing demands or sanctions from elsewhere. Hence the formal rules that stipulate some degree of need-based targeting of public goods are held hostage at the discretion of the DCE and on the basis of the vote buying imperatives. In summary, a vote buying model, characterized by informal institutions, defines the key political incentives that determine the targeting of local public goods. Importantly, the model of vote buying is not defined by the design of formal institutions alone. This landscape of political incentives includes both formal and informal institutions, as can be seen in Figure 2. It can been seen in this figure that the determinants of targeting when viewed exclusively through a formal lens (as in Figure 1) are starkly different than when the informal institutions are included as in Figure 2. 17 Figure 2. De facto decentralization in Ghana – formal and informal institutions Source: Authors, based on key informant interviews. Note: DCE refers to District Chief Executive. It is also instructive to note that the effect of informal institutions is not uniform. In particular, while three of the four informal institutions of local government in Ghana lead to divergent policy outcomes; one institution—that of political accountability of assembly members—actually leads to an outcome envisaged by the formal reforms. We use Helmke and Levitsky’s (2004) framework of the effect of informal institutions on policy outcomes to identify the nature of four informal institutions. The powers of the office of the DCE represent a competing institution where the wide-ranging powers of the DCE structure incentives in ways that are incompatible with the formal institutional rules. The domination of the legislative and bureaucratic institutions of local government, in turn, defines an accommodating institution. In these accommodating institutions, incentives of partisan politics combined with the powers of the DCE create incentives for members of these two institutions of local government to act in a manner that alter the substantive effect of formal rules without directly violating them. Indeed, both these sets of institutions are created by actors who dislike outcomes generated by formal rules but are unable to change or openly violate them. In contrast, the political accountability of assembly members emerges as a substitutive institution, where local partisanship helps achieve what the formal institutions were designed to achieve but failed to do. We now use our case studies to examine if the patterns of public good provision in the EAs in our sample match the predictions of the model of vote buying through informal institutions. 18 6. CASE STUDIES While it is clear that a partisan political rationale defines informal institutions through which logic of vote buying obtains, it remains to be seen if the pattern of public resource allocation reflects this rationale. We now use our case studies in two northern Ghanaian districts to examine if the informal institution of vote buying is the key determinant of the targeting of local public goods and to examine whether the nature of the electoral markets shapes the political rationale and subsequent patterns of public goods allocation in the status quo. To do so, we examine data on the allocation of local public goods across the electoral areas (EAs) in our districts. Table 3 presents the results from district A, the swing voting district, and Table 4 presents the results from district B, the core voting district. Ghana has two major political parties that have been in close elections over the last two decade—The National Patriotic Party (NPP) and the National Democratic Congress (NDC). The NPP held power from 2004 through 2008 and was in power at the same of this study in March 2008 We first examine the results from district A (Table 3). The data on party affiliation shows that the district is indeed a swing district. One of the three electoral areas (EA), EA3, elected a member of parliament (MP) from the opposition party, though the margin of victory was very small, and EA3 voted for an MP from the current ruling party, even though it was identified as being a weakly dominant area for the opposition, NDC. EA2 provides an interesting contrast: It is traditionally a stronghold for the NDC party. The tables show that the allocation of public goods across these District Assemblies (DAs) closely follows the voting patterns in the EA. Indeed, EA3, which voted for the ruling party (for MP) despite not being strong supporters of this party, got a plethora of public goods, ranging from several boreholes to a clinic. EEA1, which the opposition party won by small margin, was also awarded public goods. Notably, this included a small check dam—a high value and coveted investment in the semiarid regions of northern Ghana. EA2, which stands out as a core voting EA (for the opposition) within a largely swing voting district presents an interesting contrast. Indeed, this EA received almost no public goods, despite the fact the assembly members lobbied for several. The only project granted to this EAcame from a donor funded program, which was not implemented through the DA. The table also shows evidence of the intense lobbying for projects by these elected members of the DA. The results from district B (Table 4) present a stark contrast. Notably, all EAs in this district were core voters for the NDC in all parliamentary elections. Surprisingly, one of the DA members identified as belonging to the ruling NPP. It should be noted that within this core voting district, one part of the district was widely identified as being the home region of the DCE. Not surprisingly, the EA (EA4) that was located in this home region of the DCE received a very valuable public good: piped water to all communities in the EA. It is also interesting to note that the EA who had voted for the opposition party in a core voting district (EA5) received no public goods at all. Once again, there is strong evidence of lobbying from assembly members. In this district, however, it is interesting to note that the assembly members from the non-home regions lobbied far more than the member from the home region. In summary then, these tables suggest that the pattern of public good allocation closely tracks electoral imperatives and that the affiliations to political parties and voting patterns play a critical role. Indeed, swing district A got many more public goods overall than core voting district B. We argue that this is a result of the fact that granting these public goods in district A represented a way in which community allegiances and subsequent electoral support might be courted. The political imperatives in district B, in contrast, offered no such incentives. 19 Table 3. District A Source: Case studies. Note. C1 and C2 represent communities 1 and 2, respectively; an MP is a member of parliament; NDC and NPP are the political parties of the National Democratic Congress and the New Patriot Party, respectively; an NGO is a nongovernmental organization; pop refers to population. Electoral Area 1 Voters: 482 2 communities Electoral Area 2 Voters: 2,153 5 communities Electoral Area 3 Voters: 931 6 communities ELECTION 2006 -- Turnout: 61.4 percent 3 candidates 2006 – Turnout: 57.5 percent 5 candidates 2006 – Turnout: 75.3 percent 4 candidates PARTY AFFILIATION Electoral Area • NDC for MP • Small margin: swing voting community NDC stronghold NDC dominant, NPP for MP PARTY AFFILIATION (of assembly member) Not publicly partisan NDC NPP COMMUNITY Community 1 Pop: 665 Community 2 Pop: 890 Community 1 Pop: 3,871 Community 2 Pop: 434 Community 1 Pop: 1,363 Community 2 Pop: 251 PROJECTS GRANTED 1. C1-C2 Dam 2. Road 1. C1 market (Community Based Rural Development Program) 1. C2 Clinic 2. Boreholes: Most of the communities 3. Grinding mills: All but one community 4. Tractor from assembly to pick up produce and to plow PROJECTS LOBBIED FOR /FROM EXTERNAL SOURCES Lobbied for 3-unit classroom block School feeding center in C1 1. Lobbied for teachers’ quarters in C1 2. School block in C2 Boreholes 1. Shea butter processing machine Solar panels Roofing for schools— NGO Beekeeping 20 Table 4. District B Source: Case studies, Districts A and B. Note. C1 and C2 represent communities 1 and 2, respectively; hh refers to households, VIP is a Ventilation Improved Pit toilet , UNDP is the United Nations Development Program), UNICEF is the (United Nations Children’s Fund), and HIPC is the (Heavily Indebted Poor Countries Initiative) Electoral Area 4 Voters: 3,303 8 communities (Home region of the DCE) Electoral Area 5 Voters: 1,058 2 sections of a community Electoral Area 6 Voters: 3,166 10 communities ELECTION 2006—Turnout: 57.0 percent 2 candidates 2006—Turnout: 47.0 percent 3 candidates 2006—Turnout: 43.6 percent 2 candidates PARTY AFFILIATION OF Electoral Area NDC stronghold NDC stronghold NDC stronghold PARTY AFFILIATION (of assembly member) NDC NPP NDC COMMUNITIES Community 1 Pop: 2,505 Community 2 Pop: 26 hh Community 1 Pop: 3159 Community 1 Pop: 640 PROJECTS GRANTED 1. Piped water system to all communities 1. Electricity poles provided by the district assembly PROJECTS LOBBIED FOR /FROM EXTERNAL SOURCES 1. Organized community labor to dig a dam 1. Sanitation assistance (VIP) garbage removal truck 2. Loans for women from HIPC fund earmarked for women 3. Community labor— cleaned up KVIP 1. Wires to provide electricity, repair of water pipes in 4 communities, road 2. UNDP repaired pipes 3. UNICEF provided wires 4. School built through European Union micro projects 21 7. CONCLUSION In conclusion, our case studies of the targeting of local public goods in two northern Ghanaian districts tell a story of how national political imperatives capture the provision of community-based public goods. This capture, while reflected to some degree in the design of formal institutions of local government, is primarily implemented through informal institutions of vote buying. These informal institutions of vote buying, in turn, are embedded in the imperatives of partisan politics. The targeting of local public goods in the current status quo is then tied to the nature of these electoral markets. Moreover, this targeting is mediated and enforced by the local machineries of political parties. Following this electoral rationale, the swing district in our sample received several local public goods, and several communities in the district shared in this allocation. In the core voting district, public good provision was much more limited, and this provision resembled a clientelistic transfer to the home region of the district chief executive (DCE). Our case studies also tell a story of an informal institution that depends in no small measure on the political aspirations of individual agents. The desire of DCEs to gain the ticket for member of parliament and the desire of assembly members to move up a very similar political ladder are critical elements of this institution. Indeed, such aspirations—set within this partisan and politically competitive environment—are central to the informal institution of vote buying. While the divergence from intended policy outcomes on account of these informal institutions is not surprising per se and while this can be inferred from the formal agency—that is, the central government in local government affairs—the questions of how and why the divergent outcomes were obtained are of critical importance. When viewed through the lens of formal accountability alone, this divergence would be attributed to imperfections in local political accountability, and a commonly recommended solution would be to make this central agency politically accountable in some way. While formal accountability is weak or absent, the targeting of local public-resource allocation is not merely characterized by this weak or absent formal accountability. Instead, this targeting follows the logic of the informal accountability embedded in the institution of vote buying, and this institution defines a complex and potentially perverse set of accountability relations. Indeed, while the central agent does not govern through the imperatives of local political accountability, he/she follows vote buying imperatives defined by these informal institutions. With partial decentralization alongside a strong informal-party presence in local government affairs, the political aspirations of local government actors who represent the agency of the central government are always likely to be tied to incentives provided by these party machineries. Making the formal institution of the central agency or agent accountable, while ignoring the presence and persistence of informal institutions that run counter to local accountability, is therefore unlikely to effect any change in formal accountability. Indeed, we argue that in these contexts, the same partisan political incentives are likely to characterize the central agency within the local government— regardless of whether this agent is formally elected or nominated. This finding finds resonance in a current policy debate focused on the question of making the position of the DCE an elected one. One prominent group, led by donors and multilateral agencies, argue that making these position elected is a precondition for local accountability. Local observers of the system, however, argue that such election, amidst sharply partisan national politics, would undermine a fair flow of resources to districts. In particular, they argue that home districts of the DCE of the ruling party will get a disproportionately large share of the transfers from the central government. Indeed, our results suggest that such national political partisanship has even larger implications and that it shapes the informal institutions that, in turn, determine local targeting outcomes. Given partial decentralization, alongside the role of party machineries and partisan politics, it could well be argued that an elected head of the local government will face the same political incentives as the appointed DCE currently does. More generally the proponents of formal accountability rarely consider the fact that the micropolitics engendered by vote buying could engender incentives that run counter to broad-based accountability in the provision of public goods, even with a formally politically accountable actor. 22 Our case studies also point to the presence of a critical, and often missing element, of several decentralization reforms—that of genuine political representation of villages or communities. Assembly people in our sample, though effectively without power to effect public goods provision, constantly elicit information on community needs and are indeed the doorstep governors of these communities. They then exploit extant political motivations of vote buying, draw on their social networks, and use their political capital to bring development projects or community-level public goods to their constituents. The incomplete nature of the formal rules governing subdistrict allocation, combined with the overarching power of the DCE, however, restricts them from making the case for political relevance of their communities through the formal institution of the DA. Instead, these members lobby anyone who could matter. A formal subdistrict allocation formula, which does not exist at present, could lay the basis for a drawing upon this extant political accountability, even within a partially decentralized system. The success of such a formula will, however, clearly depend on increasing the de facto power of the assembly people relative to the DCE. Given partisan local politics, reducing the relative powers of the DCE in affecting the targeting of local public resources might be done more effectively through increasing the power of elected assembly members. One way of doing so would be to implement a formula-based transfer to all electoral areas (EAs). This formula too, might face implementation problems on account of partisan local politics, as previous studies have shown (see Banful 2008). This suggests that a formula would require third party monitoring. Overall, these results suggest that any impetus to institutional change must take into account the informal institution of vote buying. Specifically, in the presence of this institution, reforming the formal rule of local government is unlikely to wield any influence on outcomes. Finally, collecting data on allocation of public goods—at the levels of both EAs and communities within these EAs—would make this public resource allocation more visible. This visibility, in turn, could improve the incentives that DCEs face to achieve need-based performance metrics, even though these incentives might be restricted to politically relevant public-goods programs. In the absence of such political relevance, leveraging the fact that the political careers of local government actors are built through the use of district funds can be useful. For example, in the case of district B, donor-fund modalities were cited as a reason for the quick execution of public goods construction. Building on this example, making the flow of funds to the district—whether from the central government or from donors—contingent on the achievement of some measure of needs-based allocation could effect an improvement in the targeting of local public goods. 23 REFERENCES Ananthpur, K. 2002. Rivalry or synergy? Formal and informal local governance in rural India. Working Paper, Sussex, U.K. Institute of Development Studies. Ayee, J. 1990. The implementation of the 1988-89 District Assembly election in Ghana. Africa Insight, Volume 20 , Issue 3, pages 169-175_____. 1995. Financing sub-national governments in Ghana: The District Assemblies Common Fund. Regional and Federal Studies 5(3): 292–306. _____. 1996. The measurement of decentralization: The Ghanaian experience. African Affairs, Volume 95,No. 378, pages 31-50 _____.2003. “Local government, decentralization and state capacity in Africa”. In Critical Perspectives on Politics and Socio-Economic Development in Ghana.Edited by Oluwu,D and Wunsch,J. Ban, R., and V. Rao. 2008a. Is deliberation equitable? Evidence from transcripts of village meetings in rural India. Development Research Group, The World Bank,Washington DC. Mimeographed. _________. 2008b. Tokenism or agency? The impact of women’s reservations on village democracies in South India. Economic Development and Cultural Change 56: 501–530. Banful, A.B. 2008. Do institutions limit clientelism? A study of the District Assemblies Common Fund in Ghana. IFPR discussion paper 855. Washington, D.C.: International Food Policy Research Institute. Bardhan, P. 2002. Decentralization of government and development. Journal of Economic Perspectives 16(4): 185– 205. Bardhan, P., and D. Mookherjee. 2006. Pro-poor targeting and accountability of local governments in West Bengal. Journal of Development Economics 79 (2): 303–327. Barkan, J. D., and M. Chege. 1989. Decentralising the state: District focus and the politics of reallocation in Kenya. Journal of Modern African Studies 27(3): 431–453. Besley, T. and M. Ghatak. 2006. Public goods and economic development. In Understanding Poverty, eds. A. Banerjee, R. Benabou, and D. Mookherjee, Oxford University Press. Besley, T., R. Pande, and V. Rao. 2007. Just rewards? Local politics and public resource allocation in South India. Development economics papers 49. Suntory and Toyota International Centres for Economics and Related Disciplines, London School of Economics. Besley T., R. Pande, L. Rahman, and V. Rao. 2004 The politics of public good provision: Evidence from Indian local governments. Journal of the European Economic Association 2 (2–3): 416–426. Bossuroy, T. 2007. Voting in an African democracy: Does only ethnicity rule?.Mimeographed.Chattopadhyay, R., and E. Duflo. 2004. Women as policy makers: Evidence from a India-wide randomized policy experiment. Econometrica 72(5): 1409–1444. Constitution of the Republic of Ghana, 1992.Crook, R. 1994. Four years of the Ghana District Assemblies in operation: Decentralization, democratization and administrative performance. Public Administration and Development 14: 339–364. Crook, R., and J. Manor. 1998. Democracy and decentralization in South Asia and West Africa. Cambridge University Press,Cambridge,UK. Desposato, S. 2006. How informal electoral institutions shape the Brazilian electoral arena. In Informal Institutions and Democratic Politics, ed. G. Helmke and S. Levitsky, Johns Hopkins University Press. Devarajan, S., S. Khemani, and S. Shah. 2007. The politics of partial decentralization. World Bank,Washington DC. Mimeographed. Diaz Cayeros, A., and B. Maglioni. 2003. The politics of public spending: Part 1—the logic of vote buying. Background paper for World Development Report 2004, World Bank,Washington DC. 24 Fridy, K. 2007. The elephant, umbrella, and quarrelling cocks: Disaggregating partisanship in Ghana’s fourth republic. African Affairs 106: 423. Ghana News,(2008, May 22 ), “Headway in fight against guinea worm”. Retrieved from http://www.ghanadistricts.com/districts1on1/tolonkumbungu/?arrow=nws&read=15265 Ghana News Agency (2008,February 4 ).“ Ghana records lowest ever Guinea worm cases in 2007”. The Statesman. Retrieved from http://www.thestatesmanonline.com/pages/news_detail.php?section=1&newsid=6082Helmke, G., and S. Levitsky. 2004. Informal institutions and comparative politics: A research agenda. Perspectives on Politics 2: 725–740. Keefer, P., and S. Khemani. 2005. Democracy, public expenditures and the poor: Understanding political incentives for providing public services. The World Bank Research Observer 20(1), pages 1-27. Khemani, S. 2004 Political cycles in a developing economy: Effect of elections in Indian states. Journal of Development Economics 73(1): 125–154. ________. 2010. Political capture of decentralization: Vote buying through grant-financed local jurisdictions. World Bank, Washington DC. Mimeographed. Palaniswamy, N., and N. Krishan. 2008. Local politics, political institutions and public resource allocation. IFPRI discussion paper 834. Washington, D.C.: International Food Policy Research Institute. Reddy, N .2009. “Once bitten, twice not shy”. India Together. Available at http://www.indiatogether.org/2009/jul/gov-prbillkar.htm Tsai, L. 2007. Solidarygroups, informal accountability, and local public goods provision in China. American Political Science Review 101(2): 355-372 World Development Report, 2004. “Making Services Work for Poor People”, World Bank, Washington DC. Yin, R. 1994. Case Study Research, Second. Ed. Thousand Oaks: Sage http://www.ghanadistricts.com/districts1on1/tolonkumbungu/?arrow=nws&read=15265� http://www.thestatesmanonline.com/pages/news_detail.php?section=1&newsid=6082� RECENT IFPRI DISCUSSION PAPERS For earlier discussion papers, please go to http://www.ifpri.org/publications/results/taxonomy%3A468. All discussion papers can be downloaded free of charge. 1038. An econometric investigation of impacts of sustainable land management practices on soil carbon and yield risk: A potential for climate change mitigation. Edward Kato, Ephraim Nkonya, Frank Place, and Majaliwa Mwanjalolo, 2010. 1037. Effects of inclusive public agricultural extension service: Results from a policy reform experiment in Western China. Ruifa Hu, Yaqing Cai, Kevin Z. Chen, Yongwei Cui, and Jikun Huang, 2010. 1036. Food as the basis for development and security: A strategy for Yemen. Clemens Breisinger, Olivier Ecker, Jose Funes, and Bingxin Yu, 2010. 1035. Purpose and potential for commodity exchanges in African economies. Shahidur Rashid, Alex Winter-Nelson, and Philip Garcia, 2010. 1034. Investigating economywide and household-level impacts of sector-specific shocks in a poor country: The case of avian flu in Ethiopia. Gezahegne Ayele, Ekin Birol, Xinshen Diao, Dorene Asare-Marfo, Devesh Roy, and Marcelle Thomas, 2010. 1033. Market and climatic risks and farmers’ investment in productive assets under the Second Fadama Development Project in Nigeria. Hiroyuki Takeshima and Futoshi Yamauchi, 2010. 1032. Factors affecting the choices of coping strategies for climate extremes: The case of farmers in the Nile Basin of Ethiopia. Temesgen Tadesse Deressa, Claudia Ringler, and Rashid M. Hassan, 2010. 1031. Intellectual property rights, private investment in research, and productivity growth in Indian Agriculture: A review of evidence and options. Deepthi Kolady, David J. Spielman, and Anthony J. Cavalieri, 2010. 1030. Examining relationships between customary and state institutions in Ghana’s decentralized system. Cory Belden, 2010. 1029. How safe is my food? Assessing the effect of information and credible certification on consumer demand for food safety in developing countries. Ekin Birol, Devesh Roy, and Maximo Torero, 2010. 1028. Rehabilitating agriculture and promoting food security following the 2010 Pakistan floods: Insights from South Asian experience. Paul Dorosh, Sohail Malik, and Marika Krausova, 2010. 1027. Neighbor effects in the provision of public goods in a young democracy: Evidence from China. Claudio A. Agostini, Philip H. Brown, and Xiaobo Zhang, 2010. 1026. Opportunities and challenges of community-based rural drinking water supplies: An analysis of water and sanitation committees in Ghana. Yan Sun, Felix Asante, and Regina Birner, 2010. 1025. “Made in China”: Crisis begets quality upgrade. Jianqing Ruan and Xiaobo Zhang, 2010. 1024. Overview of the agricultural input sector in Ghana. Marika Krausova and Afua Branoah Banful, 2010. 1023. Efficacy and adoption of strategies for avian flu control in developing countries: A Delphi study. Ekin Birol, Dorene Asare-Marfo, and Yorbol Yakhshilikov, 2010. 1022. Do differences in the scale of irrigation projects generate different impacts on poverty and production?. Andrew Dillon, 2010. 1021. Crop price indemnified loans for farmers: A pilot experiment in rural Ghana. Dean Karlan, Ed Kutsoati, Margaret McMillan, and Chris Udry, 2010. 1020. Institutional and public expenditure review of Ghana’s Ministry of Food and Agriculture. Shashidhara Kolavalli, Regina Birner, Samuel Benin, Leah Horowitz, Suresh Babu, Kwadwo Asenso-Okyere, Nii Moi Thompson, and John Poku, 2010. 1019. Building capacities for evidence and outcome-based food policy planning and implementation: The example of the Comprehensive Africa Agriculture Development Programme. Ousmane Badiane, Sunday P. Odjo, and John M. Ulimwengu, 2010. 1018. Modeling the global trade and environmental impacts of biofuel policies. Antoine Bouët, Betina V. Dimaranan, and Hugo Valin, 2010. 1017. Analyzing nutritional impacts of policies: An empirical study for Malawi. Olivier Ecker and Matin Qaim, 2010. http://www.ifpri.org/publications/results/taxonomy%3A468� INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE www.ifpri.org IFPRI HEADQUARTERS 2033 K Street, NW Washington, DC 20006-1002 USA Tel.: +1-202-862-5600 Fax: +1-202-467-4439 Email: ifpri@cgiar.org IFPRI ADDIS ABABA P. O. Box 5689 Addis Ababa, Ethiopia Tel.: +251 11 6463215 Fax: +251 11 6462927 Email: ifpri-addisababa@cgiar.org IFPRI NEW DELHI CG Block, NASC Complex, PUSA New Delhi 110-012 India Tel.: 91 11 2584-6565 Fax: 91 11 2584-8008 / 2584-6572 Email: ifpri-newdelhi@cgiar.org mailto:ifpri-addisababa@cgiar.org� Abstract Acknowledgement 1. Introduction 2. Formal Institutions of Local Government and the Targeting of Local Public Goods in Ghana 3. Capture of Local Governments through Informal Institutions 4. Research Design 5. Informal Institutions of Local Government Targeting Local Public Goods in Ghana Vote Buying through the Office of the District Chief Executive Vote Buying through the Domination of Political Institutions of Local Government Vote Buying through the Political Accountability of Elected Members of the District Assembly Vote Buying through the Capture of Bureaucratic Institutions of Local Government 6. Case Studies 7. Conclusion References RECENT IFPRI DISCUSSION PAPERS