WPS5350 Policy Research Working Paper 5350 Political Capture of Decentralization Vote-Buying through Grants-Financed Local Jurisdictions Stuti Khemani The World Bank Development Research Group Human Development and Public Services Team June 2010 Policy Research Working Paper 5350 Abstract A recent trend in decentralization in several large and for political support. Decentralization, in this model, diverse countries is the creation of local jurisdictions is subject to political capture, facilitating vote-buying, below the regional level--municipalities, towns, patronage, or pork-barrel projects, at the expense of and villages--whose spending is almost exclusively effective provision of broad public goods. There is financed by grants from both regional and national anecdotal evidence on local politics in several large governments. This paper argues that such grants-financed countries that is consistent with this theory. The paper decentralization enables politicians to target benefits to explores its implications for international development pivotal voters and organized interest groups in exchange programs in support of decentralization. This paper--a product of the Human Development and Public Services Team, Development Research Group--is part of a larger effort in the department to examine incentives and accountability for the delivery of public goods. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at skhemani@ worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Political Capture of Decentralization: Vote-Buying through Grants-Financed Local Jurisdictions Stuti Khemani Development Research Group, The World Bank * I thank participants at the World Bank's South Asia Region Decentralization Seminar, and at the Decentralization Task Force meeting of the Initiative for Policy Dialogue at Columbia University. Disclaimer: The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors, and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. 1. Introduction Decentralization in the developing world has involved the political selection of the design of local jurisdictions by central and regional governments. In many countries, especially in the African and South Asian regions, and in poor provinces in Latin America, the governments' choice of decentralization instruments results in the creation of very local village-level governments, with low capacity, almost no access to own tax bases, and spending and decision-making powers restricted to the administering of higher-tier grants for local infrastructure and beneficiary selection of welfare schemes. In contrast, decentralization to municipalities or cities with revenue-raising potential has been stalled, with revenue powers being retained by central or regional governments.1 Also, in the case of local public services such as village schools, even as some institutions for community participation are created, teacher management powers remain with central or regional governments. Devarajan et al (2009) have characterized such decentralization as "partial", and have examined its implications for local service delivery. This paper addresses the question of why countries choose to decentralize partially. In one view, even grants-based decentralization to rural local governments is a significant institutional reform to address political agency problems in developing countries, particularly of government responsiveness to poor citizens (World Bank, various; Bardhan, 2002).2 This paper examines grants-financed decentralization from an alternate perspective, as a strategic political choice which enables the targeting of benefits to select constituents, at the expense of broad public goods. In this model, decentralization is subject to political capture, with local spending targeted to vote- buying, patronage, or pork barrel projects. The tension in the model is between clientelist politics (targeted transfers to buy votes) versus the broad public interest--clientelist 1 In India, for example, decentralization has largely taken the form of charging locally elected village governments, the Gram Panchayats, with beneficiary selection of poverty alleviation schemes; while municipal governments with access to tax bases and user charges are denied the power to set these, and local revenue-raising power is retained by state governments (Mohanty et al, 2007; Sahsranaman, 2009). 2 Governments in developing countries pursuing decentralization reforms also describe these as designed to promote greater participation of poor citizens in public decision-making, and to strengthen citizen ability to exact accountability from their public agents. For example, these views were provided in publicly available conference papers by the initiator of Bolivia's Ley de Participacion Popular (Sanchez de Lozada, 2000), and the implementer of India's Panchayati Raj Act (Aiyer, 2009). 2 transfers may indeed be targeted to the poor, but only to some of them or only in small amounts, at the expense of broad public goods for larger numbers.3 The paper therefore explores a political economy explanation for cases where decentralization remains partial long since the initiation of reforms. It argues that partial decentralization is selected and maintained by politicians in the face of increasing participation by swing voters, because it enables them to win elections by dividing the strength of swing voters, and continue to provide targeted benefits to core supporters. Decentralization in this theory helps prolong clientelistic political competition at the expense of broad public goods platforms, even when elections become more broad-based and contested. This risk of "political capture" of decentralization is to be distinguished from that of elite capture analyzed by Bardhan and Mookherjee (2000, 2005, 2006a). Under political capture, even when local spending is better targeted to poor citizens than centrally or provincially allocated spending, such local targeting enables central or regional political parties to buy votes and win elections even as they reduce general government allocations to broad public goods. Much of the evaluation of impact of moving from centralized to decentralized resource allocation has focused on whether the benefit incidence of decentralized spending is more likely to be on poorer households (Araujo et al, 2008; Bardhan and Mookherjee, 2006; Galasso and Ravallion, 2005; Mansuri and Rao, 2004). This paper does not compare allocations by central and local agents, instead examining the political choice to decentralize and its implications for consolidated government tax and spending decisions. The theoretical framework is derived from the literature on political determinants of fiscal policy, where political parties consist of organized interest groups that derive benefits from holding office and from club goods targeted to their group (Alesina and Perotti, 1995; Persson and Tabellini, 2000; Persson et al, 2007). Within this political economy framework, it is difficult to find technical (non-political) conditions under which any governing party would choose to decentralize, even when a benevolent 3 Indeed, theoretical and empirical studies of "vote-buying" have concluded that such transfers are targeted to the poor (Harding, 2008; Stokes, 2005; Brusco et al, 2004). 3 planner would for reasons of heterogeneity of preferences across local settlements. This is because any governing party would have to give up revenue-raising authority to jurisdictions where rival parties could come to power and thereby exercise control over the allocation of those revenues. Again, within this political economy framework, even if a benevolent planner manages to gain leadership and initiate decentralization reforms, the implementation of those reforms would be subsequently subject to political capture. Political incentives for decentralization can be created in this framework under assumptions of political advantages from mobilization of votes through local elections. Political selection of decentralization, or political capture of benevolent decentralization in this model would be characterized primarily by overwhelming or exclusive dependence on grants for local government spending and little or no devolution of revenue raising authority. Local government spending decisions would be characterized by welfare transfers to poor swing voters in exchange for their vote. Inter-jurisdiction grants distribution would be influenced by political affiliations of local incumbents. Changes in political participation in elections would have implications for the size and distribution of local government grants. When higher tier political incumbents face more demanding "swing" voters, decentralized delivery of welfare programs may be used for more effective targeting of the "cheapest" swing votes, at lower levels of progressive taxes and public goods than in the absence of decentralized vote-buying. That is, this framework can create another type of "swing voter's curse", in which greater demands by them for public goods are thwarted through greater decentralized vote buying. Decentralization policies are therefore determined in this model by politicians to facilitate vote mobilization for their political party to win office, akin to gerrymandering districts in electoral politics. The fiscal characteristics of decentralization implied by this political economy framework are in fact increasingly common in many developing countries. Large amounts of public resources for anti-poverty programs are being provided as grants to rural local governments who have the authority to select beneficiaries of local projects (Ravallion, 1999; Alderman, 2002; Mansuri and Rao, 2004; Bardhan and Mookherjee, 4 2005, 2006b). In several decentralized countries in Africa, South and East Asia, more than 90 percent of local government expenditures are financed by grants from higher tiers (Choudhuri, 2006; World Bank, various).4 Additionally, studies in a range of countries as diverse as Albania, Brazil, India, Mexico, Peru, the Philippines, Sweden, and the United States, find evidence that higher tiers of governments strategically distribute grants across local jurisdictions to favor their political supporters (reviews in Boex and Martinez- Vazquez, 2005; and Khemani, 2006). There is some available evidence, mostly anecdotal, from a range of countries, including Brazil, India, Nigeria, Pakistan, and the Philippines, on the salience of vote- buying, patronage, and violence in local elections (Bardhan et al, 2008; Cheema and Mohmand, 2008). In the political capture model presented here, and in this anecdotal evidence, the capture of public resources for private benefits need not be restricted to the elite; even poor and traditionally disadvantaged voters may demand private goods from local politicians (such as jobs, subsidies, cash and in-kind transfers) instead of broad public goods (such as quality health and education). Indeed, the critical assumption required to derive the result of political capture of decentralization for greater clientelism at the expense of broad public goods, is that the pivotal swing voters in local jurisdictions provide strictly greater vote returns when allocated private transfers instead of broad public goods. Lizzeri and Persico (2004) use completely opposite assumptions, of the increasing value of the public good to pivotal voters, to explain the extension of suffrage by English elites in the mid-late 1800s. They argue and provide some evidence that in England, the increase in the value of urban public goods following the industrial revolution (public health infrastructure such as sewerage, waterworks and paved roads), led to a majority of the franchised elite pushing for reforms to extend the suffrage so that political parties would have stronger incentives to deliver these public goods. They show that following suffrage reforms, spending by municipal corporations on public health infrastructure increased substantially, while spending on welfare transfers to the poor was reduced. That is, a diametrically different 4 Even this number, 90 percent, masks the true extent of grants dependence of rural local governments which is closer to 99 percent when some cities that do raise significant resources from own revenues are excluded from the calculation. 5 conclusion than that provided here of local spending on targeted transfers. It has implications for innovation in current institutional interventions to improve accountability for public goods in developing countries by influencing the demands of pivotal voters. In modeling decentralization as an endogenous political decision, this paper differs from traditional theories of the benefits of decentralization predicated upon the natural existence of local jurisdictions with intra-community common preferences and inter-community heterogeneity (Tiebout, 1956; Oates, 1972). However, there are surprisingly few formal models of the political choice to decentralize powers and resources to lower level jurisdictions. Available models of endogenous decentralization, to the best of the authors' knowledge, are all consistent with the original insight of Oates (1972) that decentralization arises from regional heterogeneity of preferences. Oberholzer-Gee and Strumpf (2002) model state legislative decision-making for decentralization to districts within a state, and obtain as an equilibrium greater likelihood of decentralization in those states where there is greater heterogeneity of preferences across districts (on the policy issue being considered for decentralization, in this case liquor licensing in the USA). Cremer and Palfrey (1996), Lockwood (2004), and Redoano and Scharf (2004) examine the role of decision-making rules, national referenda versus legislative voting, majority and unanimity rules, in the selection of the degree of centralization, with preferences for decentralization distributed regionally. There is other work exploring decentralization to local governments as a political strategy, focusing on incentives of national politicians to bypass regional governments that pose a political threat. Dickovick (2007) analyzes decentralization to municipalities in Peru, Brazil and South Africa as a strategy adopted by national politicians to weaken intermediate levels of government. O'Neill (2003) argues that political parties devolve greater resources when support for them is more secure in local than in national elections, citing evidence from Bolivia, Colombia, Ecuador, Peru, and Venezuela. However, these works do not draw specific implications for the characteristics of decentralization and ensuing local politics as derived here. Finally, there are several descriptions in the literature of decentralization as a political tool to accommodate ethnic or regional conflict within countries, but most of these relate to the creation of larger province-type 6 jurisdictions, with significant local autonomy, and not to the creation of fiscal authority at the level of villages or towns, the tier for which implications are drawn in this paper (Panizza, 1999). The theory offered in this paper of the political selection of jurisdiction boundaries has policy implications for international development assistance. Technical assistance and lending programs typically focus on addressing vertical and horizontal fiscal "imbalances", spring-boarding from the observation that new local governments are not devolved "sufficient" grants to fulfill the expenditure responsibilities assigned to them. This paper suggests that when jurisdiction boundaries are deliberately chosen to keep them grants-dependent for political targeting, focusing policy largely on increasing grants to existing jurisdictions for greater local spending can exacerbate patronage politics. It can fuel people's evaluation of local governments on the basis of funding and projects that can be garnered from above, at the expense of inter and intra-jurisdiction competition on the basis of competency to deliver broad public goods with scarce resources. Instead, the political analysis suggests strengthening other strategies that focus on identifying institutional and governance interventions to undercut patronage incentives and enable voters to mobilize to demand broad public services. The next section provides a new theoretical framework for the choice of decentralization of public spending to local jurisdictions, financed by grants, to serve electoral objectives. Section 3 discusses some evidence on decentralization experiences in select countries and regions that are consistent with the theory. Section 4 discusses the implications of the theory for international development assistance and explores the potential of local governance interventions to overcome local patronage politics. Section 5 concludes. 2. The theoretical framework The theoretical framework here to analyze political selection of decentralization is drawn from a large literature on the political economy of fiscal policy choices (Alesina and Perotti, 1995; Persson and Tabellini, 2000; Persson et al, 2007). A basic premise in these models is that voters organize themselves into interest groups, represented by political parties, to influence public policy for their group's benefit. Khemani and Wane 7 (2009) introduced into these models another type of voter, the poor and unorganized voter, who only receives benefits from public spending programs (not from a party representing their interests being in power). The paper first adapts Khemani and Wane's (2009) centralized political decision-making model to the problem of choice between targeted transfers to buy votes versus allocating to the broad public good, and then examines its implications for the decision to decentralize. Consider a population with 3 groups indexed by j. The welfare of group j's member is given by: W j (c, g ) c j H ( g ) W( ) is an increasing and concave function, c j is private consumption, and g is a general public good such as village-wide infrastructure provision. Two of these groups are economic elites, controlling different economic resources, paying taxes, and organized into political interest groups (political parties) to gain private rents from office and public spending. The third group is poor, with income normalized to zero (and hence does not pay taxes), is not organized into a political group and therefore votes solely on the basis of individual benefits received from government spending. Let the 2 elite groups organized into political parties be designated P1 , and P2 , and the third group of unorganized voters be designated as the swing voters S . 5 The size of each group is denoted by j , for j P1 , P2 , S . For j S , c j y j r j . That is, the consumption of members of political parties consists of their disposable income, the difference between initial income y j and the income tax , plus a private rent r j targeted towards group members from public resources. For j S , y S 0 , and W s f s H (g ) . That is, members of the 5 The characterization of this group as swing voters is consistent with several different characterizations in the literature of what it means to be "swing". First, swing voters in this model are indifferent between voting for or against an incumbent government, based upon comparing the benefits of economic policy against a reservation utility (as in Persson et al; 2006). Second, swing voters in this model are not ideologically attached to political parties, and vote only the basis of evaluating general public policies of incumbent governments (as in Lindbeck and Weibull, 1987; Dixit and Londregan, 1995, 1996; and much of the political science literature). Third, some models define swing voters as those whose ballot ultimately determine the outcome of elections (as in Feddersen and Pesendorfer, 1996), which also happens in the equilibrium in this model. 8 unorganized swing group do not have any income, cannot be taxed, and receive welfare only from government spending on public goods and any welfare transfers f s targeted to their group. Raising taxes is costly for the government. It may need to allocate resources for enforcing the payment of taxes and prevent tax evasion for instance. We assume that when the government imposes an amount in taxes it only collects ()* i.e. the cost associated with this level of taxation is (1-())*. The inefficiency of the tax system is captured by ( ) which has the following usual properties for an inverted-U Laffer curve for tax revenues: 0 ( ) 1 , ( ) 0 , and ( ) 0 for 0 y j , j . The tax rate at which revenues are maximized is given by max . Correspondingly, the maximum amount of public good that can be provided in this economy is given by g max max . The general public good g , the rents to political party members, r j , and the private transfers to swing voters, f s , are financed one-to-one with tax revenues, and satisfy non-negativity constraints g 0 and f j 0 j . The government budget constraint is therefore given by: jS j r j s f s g R ( ) where R( ) ( ) j is the government's tax revenue when the tax rate is . jS The politics. The two political parties P1 , and P2 compete during elections to win a majority of votes, and choose the policy vector p [ , g , f j ] . Any member of the political parties, j for j P1 , P2 can be a candidate. The political party that wins office chooses policies to maximize the welfare of its members (core constituents) subject to a re-election constraint (to be derived below), in addition to the budget constraint specified in equation (2) above. The objective function of party j in government is given by: max [ y r j H ( g )] B j j j { ,g, f } B represents the rents or benefits (outside of fiscal policy) from holding office, which are shared equally among all members of the party. Note that swing voters and opposition party supporters do not enter the government's objective function. Any 9 transfers to them would be determined by their role in the re-election constraint to be derived below. Political parties in this model are therefore both partisan and opportunistic. They cater to the interests of their core constituents once in office, but also choose policies to try to win elections and gain office. The core-supporters j of political party j, always vote for the party to which they are attached, consistent with the party's objective of maximizing the welfare of its supporters subject to re-election and budget constraints. The unorganized or swing voters cast their ballot for the incumbent government if their welfare under government policy, H ( g ) f s is higher than or equal to a reservation utility parameter ; otherwise they vote for the opposition. The swing voters' reservation utility parameter is distributed on the support [, ] , with density f and cumulative F which are common knowledge.6 When the incumbent implements the policy p [ , g , f j ] it can expect to receive S F ( H ( g ) f s ) of swing voters' ballots. The remaining swing voters S (1 F ( H ( g ) f s )) will punish the incumbent and vote for the opposition. The incumbent government party j G therefore receives G S F ( H ( g ) f s ) of the votes. The reelection constraint is given by: S G F (H (g ) f s ) or H ( g ) f s F 1 ( ) 2 S Where G ( G G ) is the difference in size of the core constituents of the two political parties. is the minimum swing votes the incumbent government needs to win re-election. The political distortion. In the absence of political considerations, a utilitarian social planner would solve only the public goods and taxation problem, setting all targeted transfers to zero. The public good optimization problem is therefore: max H ( g ) j [ y j ] { , g , f j } jG jS We assume that 0 i.e. the swing voters expect to be at least as well off as under laissez-faire with no 6 taxation and no public goods or spending targeted to them. 10 subject to g R ( ) ( ) j , g 0 , and 0 y . jS The solution is familiar, given by: H ( g * ) [ ( ) ] 1 , R ( * ) g * The 1 level of public goods supplied in the utilitarian optimum equates marginal social benefit of the public good to its marginal social cost of production. The utilitarian government raises just enough taxes to finance the production of the public good, spending nothing on targeted transfers to organized interest groups or individual swing voters.7 When political considerations are taken into account, the incumbent political party will deviate from this optimum and aim to provide targeted transfers to its core constituents, and perhaps also to some swing voters to win elections. The extent to which it is able to capture public benefits for its core supporters depends upon its prospects for gaining public office which in turn depends upon how demanding swing voters are, or the function F() chosen by nature.8 We now derive the preferred levels of the tax, the public good, and targeted rents of an incumbent political party, denoted by *G , g *G , r *G respectively, or the policy package incumbents would implement if their re-election constraint were ignored or not binding. A government's optimization of the welfare of its core constituents consists of a two-step procedure. First, it chooses the tax rate that maximizes the disposable income of its constituents to whom all tax proceeds are redistributed:9 max y j ( ) , where G j G . { } jG jS 7 If the objective of the social planner is redistribution, to equalize welfare or to guarantee a minimum level of welfare for the poorest group, then private transfers to the poorest may emerge in equilibrium depending on the gains to welfare from public goods versus private transfers. 8 In our model, if voters could cooperate and choose the distribution of reservation utilities (as in Persson et al. , 2000) they will set it in a way to extract the maximum amount of public good from whatever government is in place. This assumption is, however, unrealistic since it requires a high level of information. It is also inconsistent with our modeling of swing voters as unorganized, and therefore unable to coordinate. We instead consider atomistic voters who cannot communicate or cooperate. Each swing voter will then independently set her reservation utility or will be assigned one by nature. 9 The simultaneous use of inefficient transfers to and taxes on the same group was a puzzling feature of African fiscal policies highlighted in Bates' (1981) classic study of political economy. In our model, transfers are redistributive, to lower the net tax burden on the incumbent political group. Taxes across groups are uniform, the underlying rationale being to avoid additional distortions to revenue generation from economic decision-making by the rival groups. 11 The first order condition for the above maximization, where *G denotes the preferred tax rate, satisfies G [ ( *G ) *G ( *G )] 1 or R ( *G ) G . Second, the government chooses that optimal level of the public good that would be financed by its constituents alone, equating the marginal benefit and costs of the public good accruing only to its constituents, or H ( g *G ) 1 G . Note that this level of public good is lower than the utilitarian social planner's optimum. Finally, it redistributes the total tax revenue, net of public good spending, amongst its constituents, giving no specific transfers to the members of the opposition party or to swing voters. The government's constituents thus receive *G rj r *G [ R ( *G ) g *G ] G as specific transfer, which gives them a consumption *G level of c j y *G r *G , j G . The other groups not represented in the government jG do not receive any rents i.e. r * 0 , and have a final consumption of jG c* y *G .10 The incumbent government can choose its preferred policy package when its preferred level of public good is sufficient to win the minimum swing votes needed for electoral victory, or F ( H ( g *G )) . It would also choose this when re-election is entirely out of reach, which happens when swing voters are too demanding and the minimum swing votes required for victory cannot be obtained even by spending all revenues on them at the maximum tax rate max That is, governments anticipate defeat in an upcoming election, and implement predatory policy (their most preferred policy). Incumbent government's constrained choice under re-election. Suppose that the distribution of reservation utilities is such that F ( H ( g *G )) F ( H ( g max )) . The government can then either increase the level of the public good or provide targeted transfers to swing voters until it receives the minimum swing votes . The decision to re- allocate resources away from the preferred allocation, in which the core constituents of We assume throughout the paper that R ( ) g *G *G 10 j . This assumption means that the preferred tax rate of the government generates at least enough revenue to pay for its preferred level of public good. This assumption puts an upper bound on the inefficiency of taxation. If we do not impose it, the government's optimal tax rate does not change but the amount of public good provided becomes g *G Max{H 1 (1 G ), R( *G )} which may entail some rationing. 12 the incumbent party receive private rents, to the public good versus the private transfer would depend on the relative expected vote returns from the two types of instruments. The incumbent would attempt to find the cheapest possible strategy to win the minimum swing votes needed. The paper examines decentralization as a strategic political instrument in this context. The first critical assumption we make to build the argument is that the vote return from shifting resources at the margin to targeted transfers to some swing voters is greater than the vote return from increasing the public good, at the preferred public good level of the incumbent. That is, F ( H ( g *G ) f s ) F ( H ( g *G ) f s ) g f s If the incumbent party chooses to seek re-election, it would provide private transfers to swing voters just up to the point where it meets the reelection constraint. The constrained optimal amount of private transfers to swing voters for a government seeking re-election is therefore given by f s* where F ( H ( g *G ) f s* )) . The government would attempt to finance the provision of these transfers, by diverting tax revenues collected through its preferred rate *G from rents for its constituents. Assuming sufficient revenues, the central politician would continue to keep the tax rate at *G , **G provide g *G , but would reduce the targeted rents to its constituents to r j r *G f s* . As mentioned at the outset in the paper, Lizzeri and Persico (2004) use completely opposite assumptions, of the increasing value of the public good to pivotal voters in post-industrial revolution England, to explain how suffrage reforms led to substantial increases in spending by municipal corporations on public health and sanitation infrastructure. The micro evidence on voting behavior in poor countries, however, suggests that small private transfers provided within a quid pro quo arrangement may bring surer vote returns to parties than broad provision of public goods. Finan and Schechter (2010) explore the role of social preferences such as reciprocity, the psychological pressure to return a favor received from someone, in shaping electoral strategies of vote buying. Combining experimental games that measure the strength of reciprocal feelings among individuals with survey data on vote-buying 13 offers made to individuals in the country of Paraguay, they find that political agents target vote buying to individuals with greater feelings of reciprocity. A large literature on get-out-the-vote campaigns in the US finds that personal or direct contact with households are the most persuasive in getting them to vote on election day, compared to other strategies that are more broadly targeted, such as distributing pamphlets (DellaVigna and Gentzkow, 2009). This argument might logically extend to comparing the marginal vote-returns from spending on broad public goods versus privately targeted transfers--broadly provided public goods might not generate the same strength of reciprocal feelings in individual voters as do private, one-to-one transfers. Even when an incumbent government provides broad public goods, voters that are targeted with private benefits from a rival party might still vote against the incumbent. Political studies of vote-buying, largely concentrated in Latin America and in poorer constituencies in the US, have provided theoretical arguments as well as empirical evidence that low-income or poor voters are more likely to be targeted with vote-buying offers, and more likely to comply (Harding, 2008; Stokes, 2005; Brusco et al, 2004). Others have argued for countries like India that since the poor live at or near subsistence levels they are more likely to reward with a vote even a small welfare transfer made directly to their household rather than broad performance in the delivery of public goods (Varshney, 2000; Acharya, 2004). Their private voting action may therefore be at odds with their publicly declared preferences for public goods, such as when they answer questions about priority problems for the village where they invariably mention village infrastructure (Banerjee et al, 2007). Bardhan et al (2008) included a secret ballot in their survey of rural households in West Bengal to examine the correlates of voting for the incumbent Left Front political parties. They find that an individual household's vote for the incumbent party is significantly correlated only with private and recurring benefits received, in the form of personal help in times of need or benefits from welfare schemes, and not with the availability of local infrastructure. Decentralization under heterogeneity of preferences? Within this political economy framework, would a central political incumbent choose to decentralize the 14 authority to raise revenues under the standard technical rationale for decentralization-- heterogeneity of preferences over public goods? Suppose there are n localities such that H n (g ) varies across n, that is the marginal utility from the public good varies across localities. Under the assumption that g is divisible into g n (such as schools, health centers etc.), benevolent decentralization by a central social planner would involve the creation of n local governments, and decentralization of decision-making to these local governments over g n and n . Benevolent local governments would in turn select local levels of taxation to finance locally preferred levels of the public good. However, such decentralization involves giving-up taxation rights by the central incumbent, and risks greater tax rates and redistribution to rival interest groups by rival political parties that come into power at local levels. For such decentralization to happen, it would need conditions on the likelihood of the ruling political party controlling local governments, as in the analysis of O'Neill (2003). Decentralization under local vote mobilization? Now suppose local elections enable the identification of local leaders with social networks within which vote buying is more effective than through centralized "party machines". This is the second critical assumption for the arguments offered in this paper. Consider the following form of decentralization--1) announcement of the decentralization of local grants, D n (0, R( )) , determined for each locality at the discretion of the central government for poverty alleviation programs for which the beneficiaries will be identified by locally elected governments; 2) local elections occur to select the candidate who will subsequently determine f nj , j 1,2, S ; the outcome of the first local elections is determined by a coin toss; 3) the locally elected leader receives D n , and allocates it to f nj to maximize the utility of his/her core supporters subject to a budget and re-election constraint.11 11 Tax decentralization is not an option because it would lead to loss of central incumbent control over the revenues collected by opposition party local governments. 15 First, under decentralization, in equilibrium D n nG 0 for those localities where the citizen-candidate who wins elections belongs to the opposition party.12 To determine grants to the localities where its party affiliates win office, the central incumbent party goes through the following exercise--it first chooses its unconstrained (by re-election) most preferred tax rate and public good level; it then estimates how many swing votes it can get with its preferred level of public goods, and how many more it needs to win re- election. If it has sufficient swing votes at its preferred policy, there is no benefit from decentralization. If the swing votes are not sufficient, F ( H ( g *G )) , the central politician could provide grants D * D n [ F 1 ( ) H ( g *G )] to its affiliated localities to buy votes nG through direct transfers to those swing voters, *s ,whose reservation utility is given by F 1 ( ) * H ( g *G ) . These swing voters cast their vote in the next elections (both local and central) in favor of the party of the local incumbent who provides them with [ F 1 ( ) H ( g *G )] transfers f s .13 *s Assuming sufficient revenues, the central politician would continue to keep the tax rate at *G , provide g *G , but would reduce the targeted rents to its constituents to DG rj r *G D *G . What would the local politicians belonging to the group G do upon receiving grants D n ? Since their objective is the same as the incumbent political party G, to maximize the utility of core constituents subject to a re-election constraint, they will spend the entire grant on buying the swing votes needed for their party to win office both locally and centrally.14 12 The actions of the state government of Andhra Pradesh in India in the late 1990s have been interpreted in this light--because the elected leaders of village governments were not aligned with the state ruling party, the state party bypassed these local governments and gave program transfers directly to other forms of community organization, user committees, that it was better able to control (Powis, 2003; Manor, 2004). We assume at this point that the number of localities n is sufficient to contain the swing voters. 13 G *s 14 The local party leader receives the group-targeted transfers and shares in the general benefits of his/her party holding office. 16 If local elected leaders are more efficient and effective at identifying and winning the cheapest swing votes than local party operatives, then the local spending needed to win the minimum swing votes would be less than the additional transfers provided by party operatives (in the absence of decentralization): D *G f s* Decentralization of grants for beneficiary selection of targeted transfers would therefore be selected by the central political incumbent if it were cheaper to buy votes from the poorest voters through the advantages of local-election vote mobilization than through centralized party machines. Under such political capture of decentralization, it may well be that comparing central versus local targeting performance in the distribution of welfare programs shows that local targeting is more pro-poor; however, the cost of decentralization might lie in the ability of the political party to continue winning office with lower progressive taxation and investment in public goods than would have been possible in the face of the demands of swing voters. This framework therefore suggests a new type of "swing voter's curse" than is available in the literature (Robinson and Torvik, 2009; Feddersen and Pesendorfer, 1996). The more demanding swing voters become for broad public good, the less likely they are to receive their preferred levels because central politicians would use the means of vote buying in local elections to divide the strength of swing voters, targeting those that can be bought most cheaply, and win elections at lower levels of progressive taxes and public goods (than without well targeted vote buying). The framework may be extended to allow for heterogeneity in distribution of groups across localities, and to allow for other interest groups to emerge at local levels. This might lead to results on the fragmentation of jurisdictions. Depending on how groups are distributed geographically, the central politician might choose to further sub- divide existing local jurisdictions into smaller units where votes are easier or cheaper to purchase. The key intuition here is that beginning with a standard workhorse model of partisan policy objectives makes it difficult to find conditions under which a ruling party would choose to decentralize revenue raising authority. Decentralization would be consistent within this model under assumptions of differential political support for parties 17 in local versus central elections, or as grants-based decentralization driven by greater effectiveness of vote mobilization in local elections.15 3. Evidence on characteristics of local jurisdictions This section provides evidence from several countries around the world of characteristics of local jurisdictions that is consistent with the implications of a theory of political capture of decentralization explored above. The purpose here is to show that the proposed theory has empirical or practical relevance, rather than to formally test predictions. Table 1 provides some data on characteristics of local jurisdictions in selected countries. These countries have been selected on the basis of available reports on the nature of decentralization to local jurisdictions. There is little systematic data available on fiscal characteristics of local jurisdictions such as villages, municipalities, and districts. The main source of comparable cross-country data employed in the literature on sub- national finance is the Government Finance Statistics (GFS) database compiled by the International Monetary Fund (IMF); but although this provides reasonably reliable estimates of finances of states, regions, or provinces, it is widely acknowledged as not reliable for measuring finances of the lower level jurisdictions which are the focus of this paper. For this paper, detailed reports on local government finances were located for several large countries spread across four major regions of the world. There is clear evidence from three large countries, India, Indonesia, and Nigeria, where responsibility for public service delivery is being increasingly decentralized to local governments, that such decentralized spending is financed by fiscal grants from higher tiers. The average share of own-source revenue in the total revenues managed by local jurisdictions in these three countries is less than or barely equal to 10 percent. This average in fact masks even higher dependence on grants by the majority of local jurisdictions in these countries. In Nigeria, for example, a detailed survey of local government finances in the rural state of Kogi finds that own revenues are only 1-2 percent of total revenues managed by rural 15 Even with such local political differences, decentralization to locally elected governments might be trumped by other strategies of centralized targeting of local precincts of central electoral districts. 18 local governments (Khemani, 2006a). In another country included in Table 1, the Philippines, although the average share of own-revenues is 14 percent, the share in the median municipality is 10 percent, and more than a quarter raise less than 5 percent of their total income from local sources. Municipalities in Peru, the country with the highest share of own revenues among those included in Table 1, have been characterized by some analysts as having no discretion over local rate setting and tax collection, and effectively being fully financed by central grants (Ahmad and Garcia-Escribano, 2006).16 Even this relatively high share of own revenues, at 27 percent, is comparable to the very lowest of such shares among OECD countries. Ambrosanio and Bordignon (2006) provide statistics on the share of grants in local government revenues in the 20 OECD countries, which shows that only 3 countries had a grants share greater than 55 percent--Ireland at 76 percent, and the Netherlands and UK at 70 percent. This OECD experience shows that even in the face of theoretical arguments in favor of grants financed decentralization, in which taxes can be collected more efficiently at national levels and then distributed to sub-nationals as grants, grants dependence is not in the realm of 80 or 90 percent of local revenues as in the countries listed in Table 1. Within the developing world, countries in the Latin American region appear to have a smaller share of municipal spending financed by grants, but even these are higher than in most OECD countries. In Brazil and Mexico grants constitute 67 and 64 percent of municipal revenues (Burki et al, 1999). Grants dependence in these countries appears to go hand in hand with small size of local jurisdictions (in terms of population and area), and re-drawing of jurisdiction boundaries or growth in number of jurisdictions over time. Although the average population size of local jurisdictions is higher in Latin America than in Europe, this average appears to be masking large disparities across jurisdictions. Burki et al (1999) find that the vast majority of municipalities in the region have fewer than 15,000 residents. Burki et al (1999) also report shrinking in jurisdiction size, with the number of 16 Some have argued that Peru should not be considered a decentralized country at all. Indeed, prior to 2002 spending by municipalities hardly accounted for a significant share of government spending. However, since 2002, the share of municipal spending has risen to 13 percent and is expected to increase further. For the purposes of this paper, the issue of interest is that even as the country moves towards greater decentralization of spending, it chooses national grants to finance it. 19 municipios in Brazil increasing from 3000 to nearly 5000 in the fifteen years following the return to democracy in the country. In Venezuela, they report the number of municipios increasing from 202 to 330 within a decade. In a presidential address in Nigeria in 2003, then President Obasanjo indicated that state governments in the country were in the process of creating more than 500 new local governments, a 65 percent increase over the existing 774 listed in the country's 1999 Constitution.17 In a review of international evidence on the distribution of grants across jurisdictions, Boex and Martinez-Vazquez (2005) find an inverse relation between jurisdiction size and per-capita grants, that is, smaller jurisdictions tend to receive larger grants per capita across a range of countries. This pattern is consistent with the theory of political capture of decentralization offered here, and resonates with recent lessons from field experience. Examples have been provided in the Philippines of re-drawing of boundaries of municipalities and barangays (villages), or conversion of barangays into municipalities, to enable local political families to gain access to grants from higher tiers of government.18 Burki et al (1999) have also observed that demand for the creation of new jurisdictions is driven by the system of intergovernmental grants in many countries. Local requests for new jurisdictions typically have to be passed by the national or provincial legislatures, and the theory of political capture of decentralization is consistent with higher tier authorities granting these requests to facilitate vote buying. Although in many of the countries discussed thus far, a large chunk of grants distribution across local jurisdictions is determined by criteria set out in national constitutions, or a legal decree governing decentralization, there is evidence of significant political manipulation in targeting grants on the basis of local electoral characteristics. Reviews of international evidence are provided in Boex and Martinez-Vazquez (2005) and Khemani (2006b). Another striking feature in several countries is the provision of specific spending programs to local jurisdictions by higher-tier politicians, separate from and outside of the regular channels of intergovernmental grants. India, the Philippines, and Kenya have recently launched programs that are generally termed "constituency 17 Google search result on a national broadcast by President Olusegun Obasanjo in Abuja, Nigeria, on June 18, 2003, titled "On the Issue of the Review of the Structure of Governance in Nigeria". 18 Descriptions provided to the author during field visits to municipalities in the Philippines in October 2008. No documentation of such changes has been accessible to date. 20 development funds" (CDFs) which allow individual national or state legislators to provide funds for local public infrastructure investment. Most, if not all, of the literature on inter-governmental fiscal and political relations in the Philippines emphasizes the importance of congressional, gubernatorial, and even presidential pork-barrel projects for city and municipal politics. De Dios (2007) and Rocamora (2008) argue that one of the two main pillars of local political contests in the Philippines is the generation of funds and projects from higher tier governments.19 In on-going field work in the Philippines, in every municipality visited, respondents describe the importance of political affiliation of local mayors to provincial governors and national congressman in attracting spending programs to their jurisdiction (Khemani and Matsuda, 2008). In India, resources are transferred to local governments largely in the form of "schemes" for poverty alleviation, and the primary decentralized local responsibility is that of identifying beneficiaries. Such decentralization has been viewed in the literature as good policy design, in the face of solid evidence of informational advantages of local governments in appropriately identifying the poor or those that have faced particularly negative economic shocks (Alderman, 2002). However, the arguments in this paper imply that such schemes-based decentralization might lead to clientelist local politics. Analysis of perverse political incentives at local levels in India has focused overwhelmingly on the risk of "capture" of public resources by local elite for their own benefit, systematically excluding poor and disadvantaged people (Bardhan and Mookherjee, 2000, 2002; Baland and Platteau, 1999). The alternate hypothesis posited here is that even if such specified "elite capture" is not salient, and poor and disadvantaged groups are politically mobilized, their incentives are to demand short-term private benefits from local governments charged with beneficiary selection for poverty alleviation schemes. The existence of historical institutions of social inequality in India makes it a setting from which much evidence has been garnered on relative "capture" by local governments. We review this evidence below to argue that it is equally consistent with 19 The other being the control of illegal economic activity such as gambling and smuggling. 21 political mobilization of disadvantaged groups to demand private goods targeted to members of their "group", at the expense of broad public goods from which all group members would benefit. Besley et al (2004) focus on analyzing distribution of access to poverty alleviation schemes, a BPL (Below Poverty Line) card, by village governments (panchayats) in India. They find that legally identified disadvantaged groups, households belonging to the scheduled castes and tribes (SC/STs), are more likely to receive a BPL card and/or targeted home improvement schemes (toilets, drinking water, electricity, repairs) when the elected position of head of the village government, the Gram Pradhan, is reserved for members of SC/ST groups. They also find that a second institutional feature of decentralization in India--the requirement of village-wide meetings, the Gram Sabhas, to deliberate upon allocations of public funds reaching local governments--enables targeting of public benefits to disadvantaged groups. Specifically, they find that if a village is the kind of village that holds a Gram Sabha, then disadvantaged households are more likely to participate in it than are advantaged households, and they are simultaneously more likely to receive BPL cards. It is important to note that the impact of political reservations and Gram Sabhas is additional to overall targeting of BPL cards to disadvantaged households. That is, in general, in all villages a household that is SC/ST, or landless, or poor along other measurable dimensions, is more likely to receive a BPL card than upper caste or richer households, and they are even more likely to be thus targeted when a village has political reservations and holds Gram Sabhas. Besley et al (2004) interpret this as evidence of appropriate targeting of disadvantaged groups when political decentralization is accompanied by institutional mechanism (political reservations, Gram Sabhas) to combat entrenched inequalities. However, this evidence is equally consistent with the arguments in this paper that schemes-based or grants-financed local governments would emphasize the delivery of targeted private benefits to citizens. Bardhan et al (2008) contribute recent evidence from the state of West Bengal in India that voters cite short-term private benefits received from 22 their local governments as most important for their voting decision and support of incumbents, as opposed to longer-term policy initiatives taken by incumbents to promote general public goods. They also find little evidence of political marginalization or political exclusion of weaker socio-economic groups.20 Some evidence on incentives of politicians representing disadvantaged groups to exert effort on public service delivery to their constituents is provided by Keefer and Khemani (2009) who examine a CDF in India (mentioned above)--the Member of Parliament Local Area Development Scheme. This scheme entitles every member of the national parliament, elected from single-member constituencies, to substantial resources to spend on local public infrastructure in their districts. They find that dominant incumbents from districts that are reserved for SC/STs, that is, SC/ST politicians who have been elected for several consecutive terms, spend 14 percentage points less of their entitlement than other politicians. In short, dominant incumbents from reserved districts are not dominant because they exert great effort in providing public infrastructure to their constituents. Their dominance likely comes from other kinds of private identity-based services. Foster and Rosenzweig (2001) also provide evidence consistent with Bardhan et al's argument that voters place greater emphasis in local elections in India on short-term gains from public spending. They focus on 3 categories of public goods which together account for 73 percent of the activities of village governments in their sample in India-- roads, irrigation, and schools. They find that villages with democratically elected governments are more likely to provide more of all three public goods, but the largest effect is for irrigation, as calculated at the sample average, which is the service most likely to benefit the rural elite. However, in villages with a very high proportion of landless (much above the sample average) public investment shifts from irrigation to road construction (rather than education, which is unaffected by proportion landless), which suggests that capture by elites can be ameliorated when the numerical strength of the poor increases, but in a manner that might not be the most efficient for extending benefits to the poor. Roads built by village governments primarily benefit the poor, but largely by 20 The only groups with low participation identified by Bardhan et al (2008) are immigrants, women, and those with low education, not low caste groups or low income/wealth groups. 23 raising their (short-term) wages, as local road construction and improvement initiatives in India serve as employment programs for the landless poor. Education, which one expects to have the most profound effect on poverty over the medium and long-term, seems least affected by decentralization. New evidence from India specifically on decentralization of education services further illustrates this point (Banerjee et al, 2007). A central plank of public policy for improving primary education services in India is the participation of Village Education Committees (VECs) which were created in the 1990s, consisting of the head of the elected village government, parents, and public school teachers. In a survey of 280 villages in India's most populous and educationally challenged state, Uttar Pradesh, in March 2005, at least 10 years after the formal creation of this agency, Banerjee et al (2006) find that parents do not know that a VEC exists, sometimes even when they are supposed to be members of it; VEC members are unaware of even key roles they are empowered to play in education services; public participation in improving education is negligible, and correspondingly, people's ranking of education on a list of village priorities is low. Large numbers of children in the villages have not acquired basic competency in reading, writing, and arithmetic. Yet, parents, teachers, and VEC members seem not to be fully aware of the scale of the problem, and seem not to have given much thought to the role of public agencies in improving outcomes. That is, school failures coexist with local apathy to improving it through local public action. Between September and December 2005, an Indian NGO, Pratham, intervened in 195 of the 280 villages surveyed with different types of information and advocacy campaigns that communicated to village citizens the status of learning among their children, and the potential role that VECs and local governments could play in improving learning. The basic format of the interventions was to organize a village meeting on education, with the attendance of the head of the local village government and the head teacher of the village public school, the key members of the VEC, from whom the village community is urged to ask and receive basic information about local agencies in primary education. The issue raised most frequently in the village meetings, and about which people were most animated, was a government scholarship program intended to provide cash assistance to students from SC/ST groups. SC/ST parents complained that they were 24 not getting these scholarships, whilst teachers complained that parents inappropriately enroll under-age children, that can't and don't attend school, just to lay claim to the scholarships. The second issue that attracted attention was a new government mid-day meal program. Actual learning levels attracted the least attention, and the facilitators had a difficult time steering the conversation away from scholarships and school meals to the broader issue of learning. Follow-up surveys were undertaken in the same 280 villages in March 2006, 3-6 months after the information campaigns were implemented, and the most surprising fact emerging is that the campaigns did not lead to any substantial improvement in citizens' lack of knowledge of VECs. Less than 10 percent of citizens are aware of the VECs both before and after the interventions. We also find no effect on public school performance. This contrasts with a dramatic increase in private efforts to improve learning of lagging children in response to information provided--local youth volunteered to hold additional classes outside school, parents of illiterate children in particular chose to participate in these classes, and consequently the children made great strides towards literacy. However, we don't even have anecdotal evidence that these local volunteers were assisted in their efforts by local government structures--neither the village government head, nor the village public school teacher, nor any member of the VEC. Indeed, according to anecdotes provided by Pratham's facilitators in the field, the public school teacher and the Pradhan in some villages felt threatened by the volunteer activities and attention drawn to learning failures in public schools. 4. Implications for international development assistance Much of the reforms being pursued by donors in support of decentralization consist of providing greater revenues or transfers, and building capacity of local governments through training programs such as in public financial management. The arguments here suggest that these reforms on their own might not be efficient or effective in delivering improved public services for actual development outcomes, if the local jurisdictions being supported are politically captured. Greater funds devolved to local governments are more likely to go towards clientelist transfers rather than improvement of public goods. Capacity building can be a waste when local politicians have no incentives to develop technocratic skills for better service delivery. 25 One of the governance strategies being pursued in international development assistance to improve the impact of decentralization is termed community-driven development (CDD) in which communities are mobilized to identify their preferred projects and to participate in their implementation. The hypothesized governance impact of this strategy is greater social cohesion, and enhanced ability of citizens to demand and receive better public goods performance from their governments. Another is conditional cash transfers (CCT) where households are mandated to access public health and/or education services in order to receive cash transfers. The governance impact of this strategy is hypothesized to be an increase in citizen demand for public health and education services which in turn strengthens political incentives to improve the quality of these broad public goods. Although there are several evaluations planned, underway, or completed, on the impact of CDD and CCT programs on their directly targeted outcomes, such as project quality and beneficiary household-level welfare outcomes, none are designed sufficiently rigorously to assess their impact on local public goods or local political incentives (to the best of the author's knowledge).21 This is a serious gap in the policy research literature that should be addressed. Another governance intervention which has not been widely experimented with, that may be used in concert with CDD and CCT interventions or alone, is to promote yardstick competition on service delivery performance by regularly and systematically providing independent, credible, and objective information to citizens and public officials about relative performance across jurisdictions. Part of the problem of clientelist local politics might be lack of independent data and expert analysis available to civil society to judge relative performance across local jurisdictions. New empirical methodologies can be used to fill this gap by measuring service delivery outcomes that are representative at the level of local jurisdictions in feasible and cost effective ways. Once measured, the outcomes can be compiled as "municipal report cards" and disseminated to attempt to promote yardstick competition over performance improvements. 21 Labonne and Chase (2008) make an attempt to assess impact of a CDD project on local governance in the Philippines, but do not appear to have useful governance outcome variables. The variable on which they do find impact--number of village meetings organized by local officials, and household participation in these meetings--is a requirement for a village to access the CDD project, and hence automatically expected to be higher in CDD villages because of the nature of program implementation. 26 Brazil provides some examples of successful experiences with this type of governance intervention. Tendler (1997) describes how the politics of patronage in municipal governments in the state of Ceara in Brazil was tackled head-on through massive information campaigns by a state government that took office in 1987. The state government flooded radio airwaves with messages about how infant and child mortality could be drastically reduced through particular programs of municipal governments, thus bringing political pressure to bear upon the mayors to actually deliver basic health services. The state also created a new class of public health workers through a publicized recruitment effort that conveyed information to communities about the valuable role the workers could play in improving public health through community-wide effort. In only a few years coverage of measles and polio vaccination in Ceara tripled to 90 percent of the child population, and infant deaths fell from 102 to 65 per thousand births. The campaigns' success has been attributed to bringing a remarkable turnaround in the politics of the state--from clientelist and patronage-based to service-oriented (Tendler, 1997). More recently, Brazil has been the source of another innovative experiment in reducing local political rent-seeking by generating and providing credible information to citizens. In May 2003 the national government of Brazil launched an anti-corruption program based on the random auditing of municipal government expenditures by an independent public agency, and then publicly releasing audit findings on the internet and to media sources.22 New evidence from more than 600 municipalities covered by the audit suggests that the disclosure of information significantly and substantially reduced the re-election rates of mayors who were found to be corrupt (Ferraz and Finan, 2006). Furthermore, this impact was significantly more pronounced in municipalities with greater access to radio stations. More generally, there is a role for greater measurement and monitoring of citizen expectations and demands from government, what policy issues they choose to organize for, to identify new ways in which collective action can be nudged towards broad public 22 In Portuguese, this program is called Programa de Fiscaliza¸c~ao a partir de Sorteios P´ublicos, details of which is available from the following website: www.presidencia.gov.br/cgu. 27 goods. More work is needed to identify innovative institutional design or conditions under which political capture is reduced and accountability for local public goods increases. 5. Conclusion This paper has examined a theory of decentralization as a political choice of national or provincial leaders to combat growing demands of swing voters for broad public goods, and to continue targeting private benefits to organized interest groups. Decentralization to new local jurisdictions may be subject to political capture, enabling politicians to buy votes from some swing voters while continuing to provide targeted benefits to their core constituents and to under-provide broad public goods. Local governments in the political capture model are financed entirely by grants, and spend it on targeted transfers to poor voters in exchange for their vote, at the expense of local public goods. The paper provides examples from several large countries of the developing world of decentralization to small local jurisdictions, financed almost entirely by grants or spending programs received from higher tiers of government, as evidence of the empirical relevance of its theory. It also reviews evidence of the importance of clientelism in local politics, in these same countries where grants finance small local jurisdictions, which further supports the conclusions derived under a theory of political capture of decentralization. The aim of this analysis is to strike a note of caution for well-intentioned reformers who are attempting to implement participatory local government in political economy contexts that are inimical to organization of citizens for the broad public interest. By bringing to the fore the tension between citizen mobilization for private or group-targeted benefits versus the greater public interest, this work suggests the value of experimenting with other types of complementary governance interventions designed to overcome political constraints to broad public goods. 28 References Ahmad, E. and M. Garcia-Escribano. 2006. "Fiscal Decentralization and Public Sub- national Financial Management in Peru, IMF Working Paper, WP/06/120, IMF, Washington DC Aiyer, Mani Shankar, 2009. "Inclusive Growth Through Inclusive Governance," Conference Paper, Initiative for Policy Dialogue, Decentralization Task Force Meeting, Columbia University, New York http://www0.gsb.columbia.edu/ipd/pub/Aiyar_inclusive_growth.pdf Alderman, H. 2002. "Do local officials know something we don't? Decentralization of targeted transfers in Albania," Journal of Public Economics, 83(3): 375-404. Ambrosanio, M. F. and M. Bordignon, 2006. "Normative versus positive theories of revenue assignments in federations," In E. Ahmad and G. Brosio (Eds.) Handbook of Fiscal Federalism, Edward Elgar, Cheltenham, UK Baland, Jean Marie and Jean-Philippe Platteau, 1999. The Ambiguous Impact of Inequality on Local Resource Management." World Development 27(5): 773-88 Banerjee, A., R. Banerji, E. Duflo, R. Glennerster, and S. Khemani, 2010. "Pitfalls of Participatory Programs: Evidence from a randomized evaluation in education in India," American Economic Journal: Economic Policy, 2(1): 1-30. Banerjee, A., R. Banerji, E. Duflo, R. Glennerster, D. Keniston, S. Khemani, and M. Shotland. 2007. "Can Information Campaigns Raise Awareness and Local Participation in Primary Education?" Economic and Political Weekly, April 14, 42(15): 1365-1372 Bardhan, P., S. Mitra, D. Mookherjee, and A. Sarkar, 2007. "Local Democracy in Rural West Bengal: Political Participation and Targeting of Public Services", Working Paper, UC-Berkeley, Department of Economics Bardhan, Pranab, and Dilip Mookherjee, 2005. "Decentralizing antipoverty program delivery in developing countries", Journal of Public Economics, 89: 675-704 29 Bardhan, P. 2002. "Decentralization of Governance and Development." Journal of Economic Perspectives 16(4): 185-205. Bardhan, Pranab, and Dilip Mookherjee, 2000. "Capture and Governance at Local and National Levels." American Economic Review, 90(2): 135-39 Besley, Timothy, Rohini Pande, Lupin Rahman, and Vijayendra Rao, 2004. "The Politics of Public Good Provision: Evidence from Indian Local Governments." Journal of the European Economic Association, April-May 2004 Besley, T. and S. Coate, 2003. "Centralized versus decentralized provision of local public goods: a political economy approach," Journal of Public Economics, 87: 2611-37. Besley, T. and A. Case. 1995. "Incumbent Behavior: Vote-Seeking, Tax-Setting, and Yardstick Competition", American Economic Review, 85(1): 25-45 Boex, J. and J. Martinez-Vazquez. 2005. "The determinants of the incidence of international grants: a survey of the international experience", Working Paper 05- 09, Andrew Young School of Policy Studies, Georgia State University, Atlanta, Georgia Breton, A. 1996. Competitive Governments: An Economic Theory of Politics and Public Finance, Cambridge: Cambridge University Press Brusco, V., M. Nazareno and S. Stokes. 2004. "Vote Buying in Argentina." Latin American Research Review, 39(2):66-88. Burki, Shaid Javed, Guillermo Perry, and William Dillinger 1999. Beyond the Center: Decentralizing the State. World Bank: Washington, D.C. Chaudhuri, S. 2006. "What difference does a constitutional amendment make? The 1994 Panchayati Raj Act and the attempt to revitalize rural local government in India", in P. Bardhan and D. Mookherjee (Eds.), Decentralization and Local Governance in Developing Countries: A Comparative Perspective, MIT Press, Cambridge, MA 30 Coronel, S., Y. Chua, L. Rimban, and B. Cruz, 2004. The Rulemakers: how the wealthy and well born dominate Congress, Philippine Center for Investigative Journalism, Manila, Philippines Cremer, J. and T. Palfrey. 1996. "In or out? Centralization by majority vote", European Economic Review, 40, pp. 43-60. De Dios, Emmanuel. 2007. "Local Politics and Local Economy", In Arsenio M. Balisacan and Hal Hill (Eds.) The Dynamics of Regional Development ­ The Philippines in East Asia, Ateneo De Manila University Press, Manila, Philippines Dickovick, J. T. 2007. "Municipalization as Central Government Strategy: Central- Regional-Local Politics in Peru, Brazil, and South Africa," Publius: The Journal of Federalism, 37(1): 1-25. Ferraz, C. and F. Finan. 2008. "Exposing Corrupt Politicians: the effect of Brazil's publicly released audits on electoral outcomes," Quarterly Journal of Economics, 123(2): 703-745. Finan, F. and L. Schechter. 2010. "Vote-buying and reciprocity," Mimeo, Department of Economics, University of California--Berkeley, Berkeley, USA http://www.econ.berkeley.edu/~ffinan/Finan_VB.pdf Foster, A. and M. Rosenzweig 2001. "Democratization, Decentralization, and the Distribution of Local Public Goods in a Poor Rural Economy." Working Paper, Brown University, http://adfdell.pstc.brown.edu/papers/democ.pdf. Harding, Robin. 2008. "Vote Buying Across Africa." Mimeo. Department of Political Science, New York University Keefer, P. and S. Khemani, 2008. "Why do some legislators pass on "pork"? The role of political parties in determining legislator effort," American Political Science Review, Forthcoming Khemani, S. and W. Wane. 2008. "Populist Fiscal Policy", Policy Research Working Paper No. 4762, Development Research Group, The World Bank, Washington DC 31 Khemani, S. 2007. "Does Delegation of Fiscal Policy to an Independent Agency Make a Difference? Evidence from Intergovernmental Transfers in India," Journal of Development Economics, 82(2): 464-484 Khemani, S. 2006a. "Local Government Accountability for Health Service Delivery in Nigeria", Journal of African Economies, June 2006, 15(2): 285-312 Khemani, S. 2006b. "The Political Economy of Fiscal Equalization Transfers." In Fiscal Capacity Equalization and Intergovernmental Transfers, Edited by Roy Bahl, Jorge Martinez and Robert Searle, Springer Publications, New York, NY Labonne, J. and R. Chase. 2008. "Do community-driven development projects enhance social capital? Evidence from the Philippines," Policy Research Working Paper No. 4678, Development Research Group, The World Bank, Washington DC Lizzeri, A. and N. Persico, 2004. "Why did the elites extend the suffrage? Democracy and the scope of government, with an application to Britain's `Age of Reform'," Quarterly Journal of Economics, 119(2): 707-765 Lockwood, B. 2004. "Decentralization via federal and unitary referenda", Journal of Public Economic Theory, 6, pp. 313-337. Lockwood, B. 2002. "Distributive politics and the costs of centralization," Review of Economic Studies, 69(2): 313-37. Manor, J. 2004. "User Committees: A potentially damaging second wave of decentralization?" Special Spring 2004 issue of European Journal of Development Research, eds. J. C. Ribot and A.M. Larson, Volume 16, Number 1, Chapter 11 Mendes, Marcos, Rogério Miranda and Fernando Blanco, 2008. "Intergovernmental Transfers in Brazil," Mimeo, The World Bank, Washington DC Oates. W. 1972. Fiscal Federalism, New York: Harcourt-Brace Oberholzer-Gee, F. and K. S. Strumpf. 2002. "Endogenous Policy Decentralization: Testing the central tenet of economic federalism", Journal of Political Economy, 110, pp. 1-36 32 O'Neill, K. 2003. "Decentralization as an Electoral Strategy," Comparative Political Studies, 36(9): 1068-1091. Panizza, U. 1999. "On the determinants of fiscal centralization: theory and evidence", Journal of Public Economics, 74: 97-140 Persson, Torsten and Guido Tabellini 2000. Political Economics: Explaining Public Policy. Cambridge, MA: The MIT Press Powis, B (2003). "Grass Roots Politics and the "Second Wave of Decentralisation" in Andhra Pradesh." Economic & Political Weekly (June 28th) Redoano, M. and K. Scharf. 2004. "The political economy of policy centralization: direct versus representative democracy", Journal of Public Economics, 88, pp. 799-817. Rodden, J., 2002. "The dilemma of fiscal federalism: intergovernmental grants and fiscal performance around the world." American Journal of Political Science 46(3): 670-687 Sanchez de Lozada, 2000. Letter to Mr. Eduardo Aninat. http://www0.gsb.columbia.edu/ipd/pub/SanchezLozada.pdf Seabright, S. 1996. "Accountability and decentralization in government: an incomplete contracts model," European Economic Review, 40: 61-91. Sidel, J. T. 1999. Capital, Coercion and Crime: Bossism in the Philippines, Stanford University Press: Palo Alto, CA Stokes, Susan. 2005. "Perverse Accountability: A Formal Model of Machine Politics with Evidence from Argentina." American Political Science Review, 99(03):315-325. Tendler, J. 1997. Good Government in the Tropics, Johns Hopkins University Press, Baltimore, MD Tiebout, C. M. 1956. "A pure theory of local expenditures", Journal of Political Economy, 65(5): 416-24 Varshney, A. 2005. "Democracy and Poverty," In Deepa Narayan (Ed.) Measuring Empowerment, The World Bank, Washington DC 33 Varshney, Ashutosh, 2000. "Why have poor democracies not eliminated poverty?" Asian Survey Vol XL, No. 5, September/October 2000, 718-736 Weingast, B. 1995. "The economic role of political institutions: market-preserving federalism and economic growth", Journal of Law, Economics, and Organization, 11: 1-31. World Bank. 2008. Spending for Development: Making the Most of Indonesia's New Opportunities, Indonesia Public Expenditure Review, The World Bank, Washington DC World Bank. 2003. Restoring Fiscal Discipline for Poverty Reduction in Peru: A Public Expenditure Review, The World Bank, Washington DC 34 Table 1: Local Government Characteristics in Selected Countries Share of Local Own-Revenues in Local Jurisdictions Total Revenues ~500 districts, ~6000 blocks, India >230,000 villages 3.7% a (Rural Local Bodies) Indonesia 440 Districts and Cities 8.8% b Nigeria >770 Local Government 10% c Authorities Peru ~1700 district municipalities 27% d The Philippines ~1500 municipalities 14% e a. Source: Choudhuri (2006) b. Source: World Bank (2008) c. Source: Author's own calculations from data provided by Central Bank of Nigeria, 1999 d. Source: World Bank (2003), excluding the municipal area of the capital city Lima e. Source: Author's own calculations for municipalities from data provided by the Bureau of Local Government Finances, 2001-2005 35