Distributive politics and spatial equity: the allocation of public investment in Chile

Distributive Politics and Spatial Equity: The Allocation of Public Investment in Chile DOI:10.1080/00343404.2017.1309013 Felipe Livert Economics and Business Faculty, Alberto Hurtado University, Erasmo Escala 1835, Santiago 8340539, Chile. Tel: (+0056) 28897366. E-mail: [email protected] Xabier Gainza Department of Applied Economics I, University of the Basque Country UPV/EHU, Avda. Lehendakari Aguirre 83, Bilbao 48015, Spain. Tel: (+0034) 946013639. E-mail: [email protected] Abstract This paper analyzes the influence of electoral concerns on investment distribution from the central government to Chilean municipalities. Drawing upon panel data, the paper shows that investment is mediated by pork-barrel and political budget cycles, as well as favouring the relatively better-off areas. Estimations also reveal that resources are channelled to the municipalities where the vote margin in local elections is larger whereas national results are not relevant, indicating that local governments’ lobbying capacity is prioritized over national electoral interests. Based on these results, the implications for regional governance and for reducing the margin of arbitrary allocations are discussed. Política Distributiva y Equidad Espacial: La Asignación de la Inversión Pública en Chile Resumen This is an Accepted Manuscript version of the following article, accepted for publication in Livert, F., & Gainza, X. (2017). Distributive politics and spatial equity: the allocation of public investment in Chile. Regional Studies, 52(3), 403–415, https://doi.org/10.1080/00343404.2017.1309013. It is deposited under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/bync-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. Este artículo analiza la influencia de factores electorales en la distribución de la inversión desde el gobierno central a los municipios chilenos. A partir de datos de panel, el artículo muestra que la inversión está condicionada por una política de corte electoralista y por el ciclo político, además de favorecer las áreas más prósperas. Las estimaciones también revelan que los recursos se dirigen a los municipios donde el margen de victoria en las elecciones locales es mayor mientras que los resultados nacionales no son relevantes, lo que indica que la presión de los gobiernos locales es decisiva frente a los intereses electorales nacionales. A partir de estos resultados, se discuten las implicaciones para la gobernanza regional y para reducir el margen de una asignación arbitraria. Keywords: Distributive politics; pork-barrel; political budget cycle; equity-efficiency trade-off; regional governance; Chile JEL: D72; H50; H72; R58 Introduction How is public investment allocated? Is it distributed in response to social equity or economic efficiency criteria, or mediated by political considerations? Economic and regional development literature has traditionally analyzed the role of public investment as a trade-off between equity and efficiency (Richardson, 1979). This debate has recently intensified with the contributions of the new economic geography literature, which state that spatial agglomerations of economic activity may benefit national growth and, thus, policies that seek to reduce regional economic inequalities may in fact be nationally inefficient (Martin, 2008). Intermingled in this debate stand the questions about the spatial organization of the state and whether decentralization may favour a more efficient and balanced spatial pattern or, conversely, be a further source for spatial inequalities and inefficiencies (for a theoretical discussion, see Rodríguez‐Pose and Gill (2005); for empirical evidence of the effects of devolution on regional disparities Rodríguez-Pose and Ezcurra (2009) and Lessmann (2012)). Beyond efficiency and equity considerations, in the past two decades a burgeoning literature on distributive politics has documented the significance of partisan and electoral concerns for public resource distribution in many countries around the world (Golden and Min, 2013; Kramon and Posner, 2013). The underlying hypothesis is that politicians are motivated by the desire to retain public office and, consequently, elected officials allocate specific types of goods to specific groups of constituents in electoral districts at specific times in the electoral cycle (Golden and Min, 2013, p.78). Political interference includes manipulating fiscal variables along the electoral cycle to convince voters that the politicians have recently been doing ‘an excellent job’, i.e. the political business cycle (Rogoff, 1990), showing political favouritism for culturally constructed population subgroups (Kramon and Posner, 2013) or benefitting particular areas to maximize their re-election chances, i.e. ‘pork-barrel’ (Tavits, 2009). This paper analyzes the existence of political concerns in the distribution of investment funds from the central government to municipalities in Chile, and its implications for territorial cohesion. In particular, it drives attention to three aspects of distributive politics. First, the existence of electoral concerns in investment distribution is explored and contrasted with the capacity of municipalities to generate revenues and to formulate investment projects. The aim is to test whether public investment, when mediated by pork-barrel and political business cycle considerations, pursues efficiency by concentrating on the relatively better-off areas, at the expense of spatial equity. Second, the paper explores how grant allocation varies along the electoral cycle in both, municipalities that are ruled by political parties of the central government’s ruling coalition, and rival municipalities. Third, it examines whether the central government transfers resources to the municipalities where the vote margin in local and national polls is larger or, conversely, where competition is stronger. The purpose is to determine not only whether core or swing municipalities are benefitted, but also if distribution follows national government’s interest or, on the contrary, is due to municipal governments’ capacity to put pressure on the central government, encouraged by their electoral power. Based on these results, the paper explores the implications for area governance and the mechanisms to reduce the chances for a political bias on investment fund allocations. The analysis draws on panel data of public investment funds from the central government to the 345 Chilean municipalities over the period 2004-2014. The period covers three local polls and three national polls, in which the winning coalition was from both political branches, two left-wing and one right-wing. The political variables considered are municipal and national electoral results, whereas socioeconomic data includes municipal revenues per capita, municipal staff expenses per capita, population and the percentage of people below the national poverty line. The econometric analysis relies on fixed-effects and generalised method of moments (GMM) (Arellano and Bover, 1995; Blundell and Bond, 1998) to control for potential endogeneity. This paper adds three novelties to the literature on the relationship between distributive politics and regional development. Firstly, it analyzes how the burden of using investment for electoral concerns is distributed along the electoral cycle and between allies and rival local governments. Other studies of which the authors are aware focus on the variations of fiscal transfers or taxes in election years, but how the effects of this fiscal manipulation are distributed along the term in office and across localities with different political alignment remains understudied. Another original contribution is that it is tested whether political influence is exercised to benefit the electoral prospects of the central government or, conversely, it is due to local governments’ lobbying capacity in a bottom-up process. This result is particularly important because understanding the nature of the political game is fundamental for the success of any policy meant to reduce the margin for discretionary allocations. One last novelty is that the paper explores, for the first time, the existence of political influence on investment grant transfers to Chilean municipalities. Previous analyses focused on the influence of economic variables for presidential elections (Cerda and Vergara, 2007) and on the access to publicly funded benefits from the perspective of voters (Calvo and Murillo, 2012) but, to the best of the authors’ knowledge, no study has revealed the influence of political factors in public investment distribution in Chile. The existence of a political bias is relevant because, being Chile a highly unequal country, the goal of balancing socio-spatial differences may be undermined if redistribution is mediated by electoral concerns. Beyond this particular case, the paper also discusses the institutional mechanisms to cope with political distortions, contributing in this way to the debate about their implications for regional development. The remainder of the paper is organized as follows. In the next section the theoretical framework is built from the literature on distributive politics. In section three, the case study is characterized in terms of Chile’s socio-spatial characteristics, the multi-level governance system and the sub-national finance mechanisms. Section four presents the data and the methodology for the empirical analysis. In section five the results are explained. The paper concludes by contextualizing the results and discussing their implications for local governance. Distributive politics and regional development In broad terms, distributive politics refers to how public authorities confer geographically concentrated benefits while diffusing costs across voters (Weingast, Shepsle and Johnsen, 1981). This definition, though, may be too broad, since it includes pork-barrelling as well as resource allocation in lagging areas, which may be a political goal under territorial cohesion criteria. To narrow the conceptual framework, Stokes et al. (2013) differentiate among distributive strategies according to their programmatic vs. non-programmatic nature. For a distribution to be programmatic, rules of distribution have to be formalized and public, and they have to shape the distribution of resources, whereas in nonprogrammatic distributions the criteria for allocations are not rendered public. Within non-programmatic distributions, two of the most spread political distortions are concentrating resources in particular geographical areas with electoral motifs and manipulating the timing of the fiscal variables. Political business (or budget) cycle (PBC) is the term used to identify the variations on budget expenditure or taxes along the electoral cycle. The existence of a PBC has been widely studied by macroeconomists, showing that monetary and fiscal policies tend to be expansionary before an election (Willett and Keil, 2004). Empirical analyses have demonstrated the existence of a PBC in several countries around the world, but the magnitude differs across countries. This is so because PBCs depend on the institutional conditions under which incumbent political leaders have greater ability and incentives to manipulate policy variables along the electoral cycle (Alt and Rose, 2006). Although the bulk of the literature on PBCs has used national-level data and macroeconomic variables, other studies have focused on the local level. In fact, Veiga and Veiga (2007) argue that the hypothesis of the PBC is better suited to the local level because the economic evaluation of local service delivery, such as education, infrastructures or amenities, is easier than the economic assessment of national services such as defence, legal institutions or income redistribution. Evidence of manipulating fiscal variables at the local level has been found in both, developed and developing countries. In France, Foucault, Madies and Paty (2008) found that local governments increased all categories of public spending in pre-electoral periods. Likewise, opportunistic behaviour has been reported for Portugal, but Veiga and Veiga (2007) suggest that mayors tended to reduce taxes and increase expenditure on highly visible items shortly before elections, whereas expenditures remained the same or even decreased for the less visible items. Along the same line, Drazen and Eslava (2010) showed for Colombian localities that politicians targeted infrastructure spending prior to elections while other types of expenditure, such as interest payments, transfers to retirees, and payments to temporary workers, fell in election years. In Brazil, over the period 19802005 a decrease in the fiscal surplus occurred in election years because local expenditure increased while local tax revenues and investment declined (Sakurai and Menezes-Filho, 2011). Empirical analyses on arbitrary spatial allocation of resources are based on two different formal models. For the ‘core supporter model’ politicians will channel resources to support groups because the optimal strategy for risk-averse candidates is to redistribute to core supporters (Cox and Mc Cubbins, 1986). Assuming that swing groups are riskier investments, Cox and Mc Cubbins (1986) predict that politicians will invest little in opposition groups, somewhat more in swing groups, and the most in their support groups. By contrast, Dixit and Londregan (1996), building on Lindbeck and Weibull (1987), state that if political parties are equal in their abilities to allocate redistributive benefits, they will woo groups that are most willing to switch their votes. Alternatively, if parties differ in their ability to target redistributive benefits to different groups, they will favour their own core constituencies (‘machine politics’). Empirical evidence on ‘pork-barrels’ supports both the core and the swing voter hypothesis or, to be more precise, core vs. swing ‘districts’ hypothesis, as research overall is based on data from electoral units rather than individual voters (Golden and Min, 2013). Moreover, there is no pattern of tactical distribution depending on the electoral system, the grant allocation scheme, the scale, the spending type or the geographical context. Analyses of grant allocations to core areas include advanced democracies such as the U.S. (Ansolabehere and Snyder, 2006), Greece (Rodríguez-Pose, Psycharis and Tselios, 2016) or Scandinavia (Tavits, 2009), as well as new and developing democracies like Albania (Case, 2001), Mexico (Costa-I-Font, Rodriguez-Oreggia and Lunapla, 2003) or South Africa (Kroth, 2014). On the contrary, evidence on targeting swing areas has been reported for countries of the Global North (i.e. Australia (Denemark, 2000), Sweden (Johansson, 2003), England (John and Ward, 2001), South Korea (Kwon, 2005), Canada (Milligan and Smart, 2005), Portugal (Veiga and Pinho, 2007)) and the Global South (i.e. Ghana (Banful, 2011) and Brazil (Brollo and Nannicini, 2012) (see Table A1 in the Appendix for a selection of the empirical research). Several studies have analyzed the implications of electoral distortions for regional development, placing them within the equity-efficiency trade-off. Castells and Solé-Ollé (2005) found that in Spain efficiency criteria played a limited role, while infrastructure needs and political factors mostly explained the geographical distribution of infrastructure investment. In Turkey, Luca and Rodríguez-Pose (2015) concluded that regional investment is motivated by politics and efficiency, as the state concentrates resources in relatively better-off areas rather than distributing them to the poorest areas. In Germany, the councils with the same political affiliation as the state received more grants, and efficiency considerations appeared to be less important than redistributive concerns (Kemmerling and Stephan, 2002). Last, Cadot, Röller and Stephan (2006) found evidence of electoral motifs for regional transport infrastructure investment in France, despite its economic returns being low. As they summarize graphically: roads and railways are not built to reduce traffic jams: they are built essentially to get politicians re-elected (Cadot, Röller and Stephan, 2006, p. 1151). The Chilean institutional context Chile is a highly centralized country, in spite of some mild steps towards decentralization since 2002. Spatially, it is organized in three government tiers. The national government is the major institution as it controls the different ministries, sub-secretaries and departments, and the governance of lower tiers. At the regional scale, the regional government has limited autonomy because it represents national government’s interests and it has no revenue raising capacity of its own. The regional government is also the administrator of the National Fund for Regional Development (NFRD). Given the dependence of the regional government from the national government, the Chilean governance framework works in real terms as a two-tier system (national and local). At the local level, comunas are legally autonomous, but in practice decision-making capacity and autonomy are limited as local authorities are highly dependent on central funds. Local budgets are financed through their own sources of revenue, a horizontal transfer system and the central government grant mechanisms, such as the NFRD. Comunas generate their own revenues through commercial licenses (34% of the total), property taxes (32%), circulation permits (11%) and other fees (OECD, 2013). The high Models (2) and (3) were intended to estimate how investment is distributed along the term of office, but while Model (2) captures public investment allocations to coalitionand non-coalition mayors in election years, Model (3) identifies whether there are differences in the years after local election. According to the estimations, in election years investment increased 7.6 percent (Model 2), whereas in the subsequent year after the municipal election investment dropped by 9 percent and two years after the municipal election the decrease was 8 percent (Model 3). Although investment fluctuates along the term in office countrywide, there are significant differences between areas depending on their political alignment. Because Models (2) and (3) work with interaction terms between the variables year and coalition, the following equations were calculated to capture the interaction terms, and thus reflect the time-effect for the different municipalities: The estimated equation from Model (2) is: Estimated per capita investment = 17.61 + 0.0860(coalition) + 0.0757 (y0) + 0.0641 (y0 x coalition) + γ Zit + σi + τt + uit (6) The estimated equation from Model (3) is: Estimated per capita investment = 17.69 + 0.15 (coalition) - 0.0895(y1) - 0.0833 (y1 x coalition) + γ Zit + σi + τt + uit (7) Equation (6) captures the performance of investment in election years. In non-coalition municipalities, investment increased by a factor of 0.076 in election years (the y0 coefficient), whereas in those aligned with the central government, investment was raised by a factor of 0.14. From Equation (6) too, the different levels of investment that would result depending on the budget cycle (municipal election year vs. non-election year) and the mayor’s political affiliation were estimated (Table 3). Apart from significant differences depending on their political alignment, these results indicate that investment is more stable in non-coalition comunas because it increases less in election years and decreases less in non-election years. [Insert Table 3] Table 3. Expected values of investment according to mayors’ political affiliation and year Note: the expected values of investment were adjusted to thousand pesos of 2014 and these are found to per capita level Based on Model (3), Equation (7) displays the interaction between mayors’ political membership and the subsequent year after the election. This equation shows that the burden falls mostly on localities ruled by mayors of political parties in the ruling coalition at the national level since, given the interaction terms, the net effect in coalition municipalities is -0.17 and -0.09 in non-coalition municipalities. So, while the level of public investment decreased in the years after municipal elections without political distinction, comunas with coalition mayors were particularly responsible for ‘footing the bill’. Finally, it was estimated whether investment is directed to core or swing areas and whether national or local election results were taken into account. Models (4) and (5) included vote margin as an indicator for core or swing municipalities, but while Model (4) considers municipal election results, Model (5) introduces national election data. The municipal vote margin of the mayors aligned with the central government’s coalition appears to be relevant, whereas the national vote margin is not. Thus, the central government concentrates more investment in core comunas, but only municipal electoral prospects are considered when distributing resources. These results suggest a bottom-up political influence as funds are channelled to areas where coalition mayors have a bigger vote margin in local polls, while national election results are not relevant. The rationale may well be that distributive politics is mediated by local governments’ ability to obtain more funds, encouraged by their electoral results. In addition, investment distribution to coalition and non-coalition localities depending on the municipal vote margin was estimated (see Figure C1 in the Appendix). The coalition variable changes both the intercept and the slope of the municipal vote margin, indicating that municipal vote margin increases investment in coalition localities, whereas is slightly reduced in non-coalition localities. In order to test for possible endogeneity, Table 4 presents GMM results for the five models. The consistency of system-GMM relies on two hypotheses: instrumental variables must not be correlated with the error terms, and a negative first-order autocorrelation (AR1) in residuals may be observed, but no second-order autocorrelation (AR2). The Hansen test indicates that instrumental variables are valid, while the ArellanoBond tests for AR1 and AR2 show no second-order serial auto-correlation, thus indicating valid GMM estimations. Moreover, these are overall consistent with FE results for the five models, since all political and socioeconomic variables except municipal revenues appeared to be significant in the GMM estimations. [Insert Table 4] Table 4. System-GMM estimation results Discussion and conclusions Like other countries of the Global North and South, in Chile investment grant allocations are mediated by electoral concerns, and not only by equity or efficiency goals. The analysis above shows the existence of pork-barrel politics because the municipalities with mayors belonging to a party ruling the national government receive 10 percent more investment funds than the rest of the municipalities. These results align with other studies that argue that when local (or regional) governments are ruled by the party in charge of the central government, they obtain more benefits than those others controlled by the opposition (Bertelli and John, 2010; Solé-Ollé, 2013). Additionally, elected officials consider timing since on election years investment is up to 10 percent greater. The burden of such a political business cycle is distributed unevenly across time and space: the fiscal excesses of election years are paid mainly during the subsequent two years and, although investment increases on election years countrywide, it is mainly coalition municipalities that benefit from the political cycle’s fluctuations. Subsequently, coalition comunas have to pay mainly for this higher level of investment, whereas in non-coalition comunas it remains more stable. This result highlights the importance of taking a dynamic and spatial perspective in the analysis of political business cycles, an issue not adequately addressed in the literature. Other studies have focused on the opportunistic behaviour in election or pre-election years, but understanding how fund distribution varies throughout the electoral cycle and between allies and rivals is necessary for appropriate institutional responses to cope with timing in the manipulation of fiscal variables. Some of these are discussed below. One of the key issues to understand the nature of the distributive game is determining whose interests are furthered by a discretionary fiscal allocation. In Chile, municipal election results are related with investment distribution, whereas national election results are not relevant. Investment is channelled to coalition municipalities with a higher vote margin in municipal elections, pointing to core comunas as the main beneficiaries of such investment. Our estimations, thus, side with the empirical evidence supporting a tactical distribution in favour of core areas. Tavits (2009) argues that targeting core municipalities is the most likely strategy to be followed because voters require less stringent cognitive capacities since, if the ruling party is the same on the local and national level, there is no confusion about whom to reward. In a similar vein, Brollo and Nannicini (2012) state that if voters are not able to distinguish the source of transfers and political credit spillovers occur in favor of municipal governments, aligned municipalities receive more transfers. Nevertheless, rather than suggesting a conservative strategy by the central government to protect itself from the cognitive asymmetries of the electorate, for the authors this result highlights local governments’ ability to put pressure on central policy-making. The central government decides on disbursements, but municipal electoral results give local politicians a stronger capacity to exercise influence in the competition for grants. The importance of lobbying is reinforced by the fact that in every model staff expenditures are positively related with investment. In this sense, the hypothesis would be that comunas with strong mayors and larger bureaucratic structures have a greater capacity to formulate sound proposals and to put pressure on disbursers. The latter is a particularly relevant contribution of this paper. In the academic literature remains understudied whether the distribution of national grants is mediated by central governments’ interests or local governments’ power. However, determining if fund allocation responds to top-down or bottom-up pressures is fundamental to design efficient control mechanisms and has significant consequences for regional development. In a centralized state moving slowly towards devolution, local governments’ lobbying capacity is a further distortion, particularly when territories are highly heterogeneous and may therefore have different, perhaps opposing, interests (Rodríguez‐Pose and Gill, 2005). Such distributive pattern penalizes the territories not aligned with the central government, but it could also lead local governments benefitted from political favouritism to a fiscal trap because the incentives for prudent local budgets are suppressed (Psycharis, Zoi and Iliopoulou, 2015). Apart from the particular concerns of the electoral game, the analysis yields important implications for territorial cohesion. Chile has the third-highest Gini coefficient on income distribution in South America (ECLAC, 2014) and also registers the secondhighest level of territorial disparity among OECD countries (OECD, 2013). Under these socio-spatial conditions, spatial equity seems a rather relevant policy to aim for and yet, investment is concentrated in municipalities with higher municipal revenues and higher local staff expenses. Determining the particular causes for such a regressive distributive pattern lies beyond the scope of this paper, but it may well be due to an efficiency bias in investment planning. Investment favours the relatively better-off because concentrating resources in these areas is usually more efficient from a cost-benefit perspective, despite increasing territorial disparities. As such, the NFRD is meant to be a regional redistributive fund but this function is clearly limited, especially if it is mediated by political influence. Different policy procedures and instruments could narrow the incentives for arbitrariness and lead to a more equitable spatial distribution. First, if non-programmatic allocations are fed by the absence of formalized and public rules of distribution, policy reforms should be oriented to increase transparency and accountability. Indeed, the level of influence of electoral cycles on the fiscal balance depends on the degree of transparency (Alt and Lassen, 2006). Increasing transparency and accountability mainly requires political will, although overcoming politicians’ lack of enthusiasm for such reforms may be a rather difficult obstacle. Second, mechanisms for midand long-term investment planning could be considered. The actual design of the NFRD allows local governments to rely on investment grants to finance operational expenditures (OCDE, 2013), but this is a pitfall that distorts its strategic nature and increases the margin for discretionary distribution. In addition, the design of the fund generates “December fever”, since it has to be disbursed before that fiscal year finishes (December 31st) in order not to be penalized in subsequent years (Tarschys, 2003). In this context, the introduction of an independent planning office for regional investment would help with the recovery of such a strategic perspective. For this body to operate alongside electoral purposes it is not only necessary to separate the political sphere from the administrative sphere, it is likewise necessary to set up stable finance mechanisms. Multi-year budgeting could be a valuable fiscal policy because it can give local governments greater certainty about future funding, which enables them to plan and manage their expenditure more effectively. Although multi-year budgeting is not free of political clout, it could decrease political influence if the time frame of the budgeting period and investment planning differs from the term of office. Besides, consideration can be given to the use of fixed multi-year ceilings, similar to the United Kingdom or Sweden (Hawkesworth, Melchor and Robinson, 2012). A third policy change should go in the direction of designing institutional mechanisms to reduce the gap between comunas. From the authors’ perspective, one of the most striking results of the above econometric analysis is the importance of municipal revenue and staff expenses for investment distribution. Local public employees’ capacity to design sound proposals is essential because selection is based, inter alia, on the proposals’ technical feasibility. Horizontal transfer mechanisms should be improved in order to reduce income disparities. Increasing the contributions to the equalisation fund would level the provision of public goods between areas, while reducing differences in municipalities’ capacities to formulate projects if they are used to improve local bureaucracies’ skills and resources. An alternative method for resolving this problem is exploring the mechanisms by which different comunas could temporarily join up when applying for investment grants. Local collaborations would not only lead to agglomeration economies by sharing staff resources and technical expertise, they would reinforce the strategic perspective of investment planning too if, beyond each comuna’s interests, the wider context were considered. Let us not sound naïve and pretend that these policy recommendations will remove every political obstacle. The political bias is a constitutive element of the distributive game because politicians are office-motivated and the electorate enjoys receiving benefits, even at the expense of inefficiencies imposed on the majority. As a result, any attempt to erase completely the electoral influence will be unsuccessful. However, these policy proposals would limit the scope of unjustified political discretion and lay the basis for a more equitable fiscal distribution. Endnotes 1Coalition and vote margin variables change over time and across municipalities. Year of the term of office variable changes over time, but it is constant across municipalities because elections are held the same day in all municipalities. Acknowledgements We are grateful to Joachim Wehner and Rafael Hortala-Vallve for their help on an earlier draft of the paper, and Tomasz Mickiewicz and two anonymous referees for their comments and suggestions through the review process. We are also in debt to Susan Andrade for her generous assistance copy-editing. The usual disclaimers apply. References  Alesina, A. (1989). Politics and business cycles in industrial democracies. Economic Policy, 4, 55-98. doi: 10.2307/1344464 Alt, J. E. and Rose, S. (2006). Context-Conditional Political Budget Cycles. In C. Boix and S.C. Stokes (Eds.) The Oxford Handbook of Comparative Politics (pp. 845-867). Oxford: Oxford University Press. Alt, J. E. and Lassen, D. D. (2006). Transparency, political polarization, and political budget cycles in OECD countries. American Journal of Political Science, 50, 530550. doi: 10.1111/j.1540-5907.2006.00200.x Ansolabehere, S. and Snyder, J. M. (2006). Party control of state government and the distribution of public expenditures. The Scandinavian Journal of Economics 108, 547-569. doi: 10.1111/j.1467-9442.2006.00470.x Arellano, M. and Bond, S. (1991). Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations. The Review of Economic Studies, 58, 277-97. doi: 10.2307/2297968 Arellano, M. and Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 68, 29-51. doi: 10.1016/0304-4076(94)01642-D Banful, A. B. (2011). Do formula-based intergovernmental transfer mechanisms eliminate politically motivated targeting? Evidence from Ghana. Journal of Development Economics, 96, 380-390. doi: 10.1016/j.jdeveco.2010.08.012 Bertelli, A. M. and John, P. (2010). Government checking government: how performance measures expand distributive politics. The Journal of Politics, 72, 545-558. doi: 10.1017/S002238160999082X Blundell, R. and Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87, 115-143. doi: 10.1016/S03044076(98)00009-8 Brollo, F. and Nannicini, T. (2012), Tying your enemy's hands in close races: the politics of federal transfers in Brazil. American Political Science Review, 106, 742-761. doi: 10.1017/S0003055412000433 Cadot, O., Röller, L. and Stephan, A. (2006). Contribution to productivity or pork barrel? The two faces of infrastructure investment. Journal of Public Economics, 90, 11331153. doi: 10.1016/j.jpubeco.2005.08.006 Calvo, E. and Murillo, M. V. (2012). When Parties Meet Voters: Assessing Political Linkages Through Partisan Networks and Distributive Expectations in Argentina and Chile. Comparative Political Studies, 46, 851 - 882. doi: 10.1177/0010414012463882