89 PSP Discussion Paper Series 19708 March 1996 Opportunistic Government or Sustaining Reform? Electoral Trends and Public Expenditure Patterns in Peru, 1990-1995 Carol Graham The Brookings Institute Cheikh Kane The World Bank March 1996 Poverty and Social Policy Deparment Human Capital Development The World Bank PSP Discussion Papers reflect w*ork in progress. They are intended to make icssons emerging from the current work program available to operational staff quickly and easilv, as well as to stimulate discussion and comment. They also serve as the building blocks for subsequent policy and best practice papers. The views expressed here are those of the author-s and should not be attributed to the World Bank or :ts Board of Executive I)irectors or the countries they represenlt. Opportunistic Government or Sustaining Reform? Electoral Trends and Public Expenditure Patterns in Peru, 1990-1995 Abstract This article analyses the effects of discretionary public expenditures on electoral outcomes during a period of economic reformn in Peru. We exarnine the political consequences of a significant increase in public expenditure targeted to poor groups. and analyze the effects of those expenditures on voting patterns and government responses to those voting patterns. We attempt to gauge the effects of these expenditures on the political sustainability of reform, and on political behaviour more generally. The analysis departs from the traditional political economy literature, which explains govermnent behaviour from the point of view of macroeconomic variables prior to elections. and focuses primarily on aggregate variables and on industrialized countries. It is also distinct from the literature on the political economy of reform, as it analyzes electoral rather than interest group response to government policies, and focuses on micro rather than macro-level variables. We posit that with the decline of parties in Peru since 1990, voter behaviour more clearly fits a "retrospective" voter model, in which electoral outcomes are determined by voters' evaluations of incumbent performance rather than by other factors, such as partisan influence. We examine the role of one indicator of government performance: discretionary public expenditures. We conclude that government response to voting patterns - in the form of the allocation of discretionary public expenditures - was clearly influenced by electoral trends. Yet because the vote of poor rural areas had increased in relative importance with the diminished influence of urban- based political parties, this resulted in an unprecedented reallocation of public resources to poor and previously marginalized rural areas. Increasing the relative influence of poor rural voters may not have been the government's intent. Yet the allocation of a significant part of the public expenditures through a demand-based social fund enhanced organizational potential and raised expectations among sectors of the population previously marginalized from government attention, which may make them better able to demand and sustain public expenditure shifts in their direction in the future. The effects of the expenditures on voting patterns were less straightforward than the effects of voting were on expenditure allocations, meanwhile. While in many areas discretionary expenditures seem to have swayed the vote more strongly in favor of the government and therefore continued reform, that influence was rarely decisive. And while the outcome of expenditure allocations favored the poor, it is not clear that a similar outcome would occur if the experience was replicated in other countries, precisely because of the discretionary nature of the expenditures, as well as the rather exceptional political context in Peru. The article concludes that the positive effects that such expenditures can have in sustaining reform and in enhancing equity must be considered in light of the longer term benefits of channeling the same expenditures through permanent public social institutions such as line ministries. I Opportunistic Government or Sustaining Reform? Electoral Trends and Public Expenditure Patterns in Peru, 1990-1995' Carol Graham, The Brookings Institution Cheikh Kane, The World Bank There is increasing consensus on the fundamental principles underlying economic policy reforms. In addition, a series of recent comparative studies have increased our understanding of the political conditions necessary for launching such reforms.2 Yet our understanding of the factors that make reforms sustainable over the longer term is far less developed. There is a wide and unresolved debate on the role of institutions, politicians, interest groups, and the popular sectors. The influence of such groups tends to be marginal in the formulation and initial implementation of policies, a process which involves an insulated group of technocrats. As the reforms proceed, the opposition of different societal groups to specific policies may have some impact, yet is less critical to the success or failure of the adjustment program than is overall 3 economic performance. The primary strength of these groups is retrospective and collective: they can vote reforming governments out of office. This suggests that the role of elections - and therefore voter behaviour - is a critical factor in sustaining economic reforms over the longterm. Voters can, and at times do, reverse economic reform programs.4 Yet they also can play a major role in making programs more sustainable by legitimating their continuation through the ballot box. There are many instances where voters opt for the maintenance of reform policies even in cases where the government that initiated the program is defeated, such as in Bolivia (1989), Chile (1989), and, more arguably, Spain (1981).5 ' The authors would like to thank Alan Angell, Michael Coppedge, George Graham, and Carol Wise for helpful comments. 2 For a description of the so-called Washington Consensus, see John Williamson, The Political Econlomy of Policy Reform (Washington. D.C.: Institute for International Economics, 1993). The most comprehensive survey of the conditions necessary for launching reforms under democratic auspices is Stephan Haggard and Stephen B. Webb, Voting for Reform: Democracy. Adjust- ment. and Political Liberalization (New York: Oxford University Press and the World Bank, 1994). A good review of recent literature of the politics of reform is Barbara Geddes, "The Politics of Economic Liberalization", Latin American Research Review, Vol.30, No.2, 1995. 3 See Geddes (1995). The role of various interest groups is discussed by Joan Nelson in 'Poverty, Equity, and the Politics of Adjustment" in Stephan Haggard and Robert Kaufmanm, The Politics of Adiustment: International Constraints. Distributive Conflicts and the State (Princeton: Princeton University Press, 1992). 4 Voters reversed major privatization legislation in Uruguay in 1992, for example. Party and union structures vary across countries and will determine the capacity that specific interest groups and the general public have to carry out such reversals. For detail on the Uruguay case, see Fernando Filgueira and Jorge Papadopulos, 'Putting Conservatism to Good Use? Long Crises and Vetoed Alternatives in Ururguay", Mimeo, Northwestern University and University of Pittsburgh, 1995. s Karen Remmer uses empirical evidence for the 1980's to demonstrate that elections have actually enhanced the adoption of orthodox economic policies in Latin America. See Karen L. Remmer, 'The Political Economy of Elections in Latin America, 1980-91", American Political Science Review, Vol.87 (July 1993), pp.393-407. Barbara Geddes, meanwhile, building on Nelson's (1992) work, found that even in countries where real wages fell, often dramatically, fewer than half the initiating governments were defeated in the first election after the beginning of reforms. Barbara Geddes, 'How Politicians Decide Who Bears the Cost of Economic Liberalization" in Ivan Berend, ed., Transition to a Market Fconomy at the End of the Twentieth Century (Munich: Sudosteuropa-GeselIschaft, 1994). The content of the reformn program - ie. the balance between stabilization objectives and those of redistribution or compensation - is likely to affect electoral outcomes. particularly if the post-stabilization recession is severe or long.6 And while there may be trade-offs between redistribution and growth, recent empirical analysis emphasizes the contribution that certain kinds of public expenditure can make to political sustainability as well as to growth.7 Early political economy analysis emphasized the influence of electoral cycles on govern- ment policy at the macroeconomic level. Yet little has been done at the micro-level which disaggregates government policy - such as public expenditure - and its effects on electoral outcomes. More recent analysis has focused on the role of particular interest groups, on the incentives and behaviour of bureaucrats and politicians responsible for reforn, and on how differences in institutional structures affect the behaviour of such groups. There has been little analysis of how voters respond to the different government policies that emerge from particular political and institutional structures during reform. This article provides a more detailed analysis of public expenditure and voter behaviour during economic reform. In particular it will analyze the effects of discretionary government social expenditure on electoral outcomes, and provide an initial basis for further comparative research. The primary objective of this article is to evaluate the political effects of a significant increase in public expenditure targeted to poor groups, many of whom were previously marginal- ized from such state benefits. More specifically, we will attempt to analyze the effects of these expenditures on voting patterns, as well as on government responses to those patterns. Did the recipient groups exercise their votes differently - and possibly more effectively - once their perceived stakes in government policy increased? How did the govermnent respond to changes in voting patterns? Will these changes enhance the ability of such groups to exercise political choice in the longer term? Did the expenditures increase political support for the continuation of economic reforms? Did these expenditure trends have positive effects on political development by giving stronger voice to previously marginalized groups? This is particularly important in the Peruvian context, where the traditional mechanisms for guaranteeing political accountability - such as the judiciary, the legislature, and political parties - are extremely weak, and election-, are one of the few channels through which the public can effectively influence the government.8 6 The role of compensation policies in enhancing the political sustainability of reform is discussed in a comparative context in Carol Graham, Safety Nets. Politics. and the Poor: Transitions to Market Fconomies (Washington, D.C.: The Brookings Institution, 1994). 7 Alberto Alesina and Roberto Perotti note differential effects of redistributive policies. On the one hand, fiscal redistribution can reduce the propensity to invest by increasing the tax burden on investors. On the other, the same policies may reduce social tensions and therefore create a sociopolitical environment which is more conducive to investment and growth. See the article by Alesina and Perotti in "The Political Economy of Growth: A Critical Survey of the Recent Literature", World Bank Economic. Review, Vol.8, No.3, 1994. Nancy Birdsall, David Ross, and Richard Sabot challenge the traditional view that there are trade-offs between redistributive and growth objectives. They cite the experience of the East Asian economies with high levels of investment in primary education, for example, as a means to enhance both equity and growth objectives, and contrast the records of those economies with several in Latin America, whose record was much weaker in both growth and inequality terms. See their paper entitled "Inequality and Growth Reconsidered", presented to the American Economics Association, March 1994, 8 For example Catherine Conaghan notes that despite the prolific and vibrant free press in Peru, it has remarkably little influence on the policy debates within the government. She attributes this weak influence in part to the lack of the key institutions through which the press would normally exercise influence, political parties in particular. See Catherine Conaghan, "Public Life in A second objective of the article is an attempt to distinguish between opportunistic or "populist" government behaviour and coherent strategies for sustaining economic reform.9 The scant evidence that we have about the political effects of compensation during reform suggests that focusing resources on organized interest groups who are in opposition to reform is rarely an effective means to enhance its sustainability. Yet the same evidence suggests that it is possible to build new -coalitions in support of reform by reaching groups previously marginalized from state benefits.10 A plausible explanation in the Peruvian case is that what may indeed have been politically driven "opportunistic" behaviour may also have had positive effects on the political sustainability of reform by increasing the relative political weight of the vote of previously marginalized groups and giving them an increased stake in the reform process at the same time. The analysis of the effects of discretionary government social expenditure on voting behaviour will allow us to explore this hypothesis further. The article also attempts to distinguish the effects of public expenditures on voting behaviour from the effects of other significant variables, such as the resurgence of economic growth and consequent reductions in poverty, and the unexpected defeat of a violent guerrilla movement (Sendero Luminoso) in mid-1992.1" We posit that since 1990 voting behaviour in Peru fits a "retrospective" voter model, ie. electoral outcomes are determined by voters punishing or rewarding incumbent performance rather than by other factors such as partisan influence. We posit that "performance" hinged on three major trends: economic performance, terrorism, and discretionary public expenditure.'2 the Time of Alberto Fujimori", Paper presented to the Annual Meeting of the Latin American Studies Association, Washington, D.C., October 1995. 9 Kenneth Roberts interprets political trends under the Fujimori government as a new brand of populism which, unlike traditional populism associated with statist and redistributive policies, is compatible with neoliberalism and fiscal constraints. Instead of universal redistributive policies, such as price controls and subsidies, politicians such as Fujimori use populist political tactics and selected targeted policies to deliver material benefits to particular groups. As under "traditional" populism, institutional- ized forms of political representation have broken down. For an excellent analysis, see Kenneth Roberts, "Neoliberalism and the Transformation of Populism in Latin America: The Peruvian Case", Worid Politics, Vol.48, No. 1, October 1995, pp.82-116. i Indeed, utilizing at least some compensatory resources to reach more needy groups who were previously marginalized from state benefits may be a more effective strategy for building support for reform. For comparative evidence, see Graham (1994). ii GDP increased by a remarkable 19% between 1990 and 1994. The poverty headcount ratio fell, meanwhile, from a high of 55% of the population in 1991 to 50% in 1994. The poverty gap, meanwhile, feel from 21% of the poverty line to 17.5%. For detail see Edgardo Favaro and Donna MacIsaac, 'Who Benefited from Peru's Reform Program? Poverty Note", Mimeo, Latin America and Caribbean Region, The World Bank, Washington, D.C., June 1995. 12 In a cross-country analysis of party fragmentation, Michael Coppedge finds that the stronger party identification is, the less economic performance affects voting behaviour. Our analysis suggests that a slight variation of this phenomenon applies to Peru: as party identification weakened, economic performance had stronger effects on voting behaviour. See Michael Coppedge, District Magnitude, Economic Performance, and Party-System Fragmentation in Five Latin American Countries", Comparative Political * , forthcoming. Catherine Conaghan notes that in Peru since 1990 there has been a marked increase in the presence of economists and technocrats in public and political life, and a decrease in visibility of party politicians. See Catherine Conaghan, "Stars of the Crisis: The Ascent of Economists in Peruvian Public Life", Paper presented to CEDLA Workshop, Amsterdam, September 14-16, 1995. 4 While we consider the first two trends as the determining variables, this paper explores the relative importance of public expenditure shifts in influencing electoral behaviour.'3 In Peru in 1990, political stalemate and government incompetence in a context of hyperinflation and guerrilla violence resulted in voters rejecting not onlv the governing APRA party and its "heterodox" economic policies, but all traditional political parties. This created a political opening for sweeping economic reform, and also dramatically changed a political context which had largely been domninated by a small, party-based elite with weak ties to most rural regions of the country. At the same time, the increase in importance of the independent vote versus that of party lists may have given more weight to votes outside Lima, although some observers contend that that new system has given more weight to Lima votes and gives Fujimori undue advantages.14 The extent to which the new system responds to the traditionally neglected voters outside Lima has implications for the politics of reform, and more generally for the manner in which democracy develops in Peru. Giving more weight to voters outside Lima may not have been Fujimori's intent, as he may well have wanted to centralize power.15 However, measures related to the economic reform program, such as the demand-based social fund and the reallocation of municipal funds to departments other than Lima and to smaller size municipalities, took on a life of their own through their outreach to poor and remote regions. In an analogous fashion, Fujimori's discrediting of traditional political parties also took on a momentum of its own by giving more weight to the "independent" vote, a vote which demonstrated itself no more conmmitted to Fujimori than to the established parties, as the discussion of the electoral results will 13 To help overcome the limitations of examining isolated factors, the authors are currently conducting a multivariable analysis of the variables affecting voter behaviour, in an attempt to capture the interaction between them. 14 Until 1993 voters in Peru voted for party lists rather than for particular candidates, and then indicated a preferential vote for a particular candidate. This meant that candidates had no local links, responsibility, or representation. The 1993 Constitution eliminated assured congressional representation by department and established one national level voting district. Voters are still allowed to indicate two of their preferential votes for Congress, but only from one list. Single district models with one representative make it easier for voters to judge who is responsible for performance and outcomes. Yet rural interests are likely to suffer in single district PR models if the rural population is in the minority. The effects are as yet unclear in Peru, where the rural population is n= in the minority, but has traditionally been less politically active, informed, and articulate. The erosion of panies and the rise of a number of independent fronts may be giving more weight to individual votes, ie. number and location of votes mav matter more now, as people are less likely to vote along partisan lines. Critics of the 1993 changes contend that eliminating assured representation has given Lima more electoral weight, and that not allowing voters to indicate preferences on more than one list favors large parties, but neither outcome is evident in the 1995 results. For detail on the electoral system, see David Scott Palmer, 'Peru's 1995 Elections: A Second Look", LASA Forum, Vol.XX'VI, Number 2, Summer 1995. For detail on the effects of represenation formulas, see Philip Keefer, 'The Policy Effects of Political and Electoral Institutions in Developing Countries", Mimeo, Policy Research Department, The World Bank, Washington, D.C., August 1995. 15 One author's interviews with some of Fujimori's advisors in the Finance Ministry suggest that while he may have not been aware of the unintended consequences of his policies, they were much more so. 5 demonstrate. '6 Finally and not unrelated, some aspects of the reform program, such as streamlining the regulatory process and reforming the tax svstem, have introduced elements of transparency into public sector operations, which has democracy enhancing elements, even at a time when traditional democratic procedures, such as congressional power of investigation and the judicial system, remain very weak.17 Traditionally poorer departments constituted Fujimori's initial political base.'- The benefits of economic reform were less visible in these departments; and their support for Fujimori decreased by 1993. As a general trend, these departments became the focus of discretionary public expenditures from that point on, and helps explain the changes in government expenditure that occurred, particularly in the post 1993 period. However, there were also some "outlier" cases, where major increases in discretionary expenditures were made to departments which had higher average incomes and better social indicators than the national average, suggesting additional political factors influencing those allocations. These trends, and their role in determining electoral outcomes - and indirectly in sustaining economic reform - are explored in the final section of the article. To describe voting patterns we utilize electoral data from national elections in 1990; constituent assembly elections in November 1992; municipal elections in January 1993; a consti- tutional referendum in October 1993; national elections in April 1995; and municipal elections in November 1995. An important caveat in the analysis of voting patterns is differences in the kinds of elections held (presidential, constitutional, municipal) and changes in the electoral laws with the passage of a new constitution in 1993. To describe patterns in discretionary public expenditure we focus on expenditures made through three distinct mechanisms: the emergency social fund (Foncodes), municipal transfers, and a school building program (INFES). All of these expenditures are highly discretionary in nature and in theory are targeted to the poorest in society, hence those groups that were least likely to benefit from "normal" channels of public expenditure. Foncodes was selected because of its high visibility as a program for the poor. Despite its demand-based mechanism for 16 This is interesting in light of Kenneth Robert's proposition that targeted benefits may create stronger clientelist bonds than universal benefits. See Roberts (1995). While the reallocation and targeting of public expenditures may have had clientelist and political objectives, it does not seem to have been effective at creating a loyal, clientelist-based political base. i7 A positive relationship between the logic of economic reform and that of democracy has been noted by Larry Diamond. who posits that at a certain level of development, sustained economic growth requires an open political system that can guarantee the rule of law. See Larry Diamond and Marc Plattner 's introduction in Diamond and Plattner, eds., Economic Reform and Democracy (Baltimore: Johns Hopkins University Press, forthcoming). A slightly different point is made by Barro, who uses empirical analysis to establish a more general relationship between democracy and growth. Barro finds that at early levels of development, too much political freedom tends to have negative effects on growth. At later levels of development, however, democracy is growth enhancing, in part for some of the same reasons that Diamond notes. Barro concludes, however, by emphasizing the positive effects that growth have on democracy, and not the other way around. See Robert J. Barro, 'Economic Growth in a Cross Section of Countries", Ouarterly Journal of Economics, 106, No.2, May 1991. pp.407-433. 18 Roberts (1995) identifies three distinctly populist features of the Fujimori government: a heterogenous constituency with widespread lower class support, personalistic leadership style,and the absence of institutionalized political mediation. 6 allocating funds, Foncodes expenditures remained highly vulnerable to government discretion, both in terms of overall amounts and in terms of locational priorities. INFES expenditures are also highly discretionary: they are extra-budgetary and managed by the Ministry of the Presidency rather than the Ministry of Education. Municipal transfers are also included because, following the loss of the 1993 municipal elections, the government changed the allocation mechanism. Moreover, this mechanism can be altered each year by the executive. The purpose of the analysis is to deterrnine the criteria for expenditure allocation across departments; to identify any changes in expenditure allocations related to electoral outcomes and upcoming elections; and to determine the extent to which these trends explain voting patterns in the context of economic reform in Peru. Discretionary expenditures must be considered in the context of aggregate trends in public expenditure and their political effects. Overall public social expenditures in Peru had dropped dramatically in the late 1980's, as had expenditures targeted to the poor. Beginning in 1993 and certainly in 1994, overall expenditure levels increased markedly, as did those targeted to the poor through the social fund.19 A question which is not the subject of the analysis but is important to raise is to what extent discretionary public expenditures are made at the expense of other essential but less visible expenditures, such as in basic education and health. A related question is whether discretionary expenditures, which tend to be more visible, are more effective in influencing voter bahaviour, and therefore a more useful tool for enhancing the political feasibility of reform, at least in the short term, than are expenditures which are channeled through the permanent public social institutions such as line ministries. Finally, there are limits to the comparability of the Peruvian case. Firstly, the Fujimori goverrrnent suspended the constitution and closed the legislative and judicial branches of government in April 1992, replacing them with new ones after the 1993 referendum. While the elections held after the 1992 "autogolpe" were deemed generally free and fair by outside observers, the political dynamic was obviously altered significantly and electoral conditions were far from perfect. Secondly, the challenge from Sendero Luminoso, and its sudden defeat when the government captured its leader in September 1992, also altered the political dynamic: by giving the population a criterion to evaluate the government by which was at least, if not more important, than economic performance.. Thirdly, in 1994 a major and unexpected increase in government revenues resulted from a $2 billion over-bid in the privatization of the national telephone company, giving the government an unexpected source of discretionary public funds. 19 In 1991 and 1992, targeted social expenditures were 0.5 and 0.4% of GDP respectively. Spending on health, for example, had fallen to a low of $5 per capita by the early 1990's, while the average for the region was $46. This trend was gradually reversed with the initiation of economic growth and the improvement of the fiscal situation. In fiscal year 1996, the government planned to spend 40.4% of the total budget on social sxpenditures, in contrast to 2o and 30% in 1994 and 1995 respectively. The share going to health rose from 4.07% in 1991 to 6.17% for 1996, and education's share rose from 5.45% to 7.86%. Targeted social expenditures increased dramatically, meanwhile, with 22.6% of the total budget going to the Ministry of the Presidency to cover the costs of the regional governments and the targeted social expenditure programs. For detail see 'Peru: Public Expenditure Review", World Bank, Washington, D.C., June 1994; and 'Ministerio de la Presidencia Concentra 22.6% del Presupuesto", El Comercio, 7 October 1995. 7 The Literature This article aims to contribute to the existing literature on political economy in general (the "new political economics"), as well as to the literature on the political economy of reform. There are several models of political economy cycles that are relevant to this analysis and highlight its potential value-added.20 The traditional opportunistic model attempts to explain government behaviour from the point of view of macroeconomic variables prior to elections, and focuses primarily on aggregate variables and on industrialized countries.21 In contrast, our analysis focuses on the role of discretionary public expenditures in influencing electoral outcomes during reform. It also focuses on a developing country, while much of the theoretical political economy literature has focused on industrialized nations. More recently, the rational actor approach, which also focuses on macro-variables, introduced the concept of rational voter behaviour rather than adaptive expectation. Our analysis does not incorporate any ex ante assumptions about voter behaviour, although it will (indirectly) consider the relevance of the approach to patterns of voting behaviour in Peru. Finally, the traditional partisan model hinges on the different weights that relevant political parties place on economic outcomes, such as unemployment. While our analysis also focuses on the key role played by differential distributional outcomes, we focus on the role of fiscal transfers and, implicitly, their impact on income distribution rather than the usual focus on the Phillips curve trade-off (inflation- unemployment). In addition, we focus on the behaviour of individual voters rather than on the role of parties, as the role of the latter in influencing outcomes in the elections held during the period under study was marginal at best. The literature on the political economy of reform has focused more on the conditions necessary for launching reforms rather than those necessary for sustaining them, as well as on the role of particular interest groups in opposing or supporting economic reforms. Several studies based on empirical evidence find no correlation between regime type (authoritarian or democratic) and performance in economic reform. They also suggest that newly elected governnents have some distinct advantages over incumbents in initiating reform programs. Launching reform tends to be easier in contexts where the pre-reform crisis is severe, for example as evidenced by inflation at hyper or near hyper levels, and results in a broad societal consensus that the statns quo is unsustainable. Some degree of bureaucratic insulation for key policymakers, meanwhile, can also play an important role in favor of reform. In contrast, reform is more likely to be delayed in contexts where party systems are fragmented or highly polarized, where political and/or economic influence are distributed very unequally, and where there is a great deal of uncertainty about the outcome of reform.22 Finally, several studies suggest that 2 0 This review draws heavily from a review of literature by Alberto Alesina, "Elections, Party Structure and the Economy", Memo, Harvard University, Boston, August 1993. 21 For details on veterans programs see Keech and Pak, "Electoral Cycles and Budgetary Growth in Veterans' Benefit Programs", American Journal of Political Science, Vol. 33, 1989, pp.11-12. 22 See, for example, Alberto Alesina and Allan Drazen, 'Why Are Stabilizations Delayed", The American Economic Review, Vol.81. 1991. In the case of Brazil, for example, Cheikh Kane and Jacques Morrisett suggest that because the burden of inflation falls on the middle class rather than on wealthier - and probably more politically powerful groups - necessary stabilization policies 8 interest groups and/or popular opposition have remarkably little ex ante influence on the policymaking process." Their primary influence seems to be collective and retrospective - ie. judging performance ex post through elections. To date analysis of electoral outcomes and reform has focused on macro-variables. Our analysis is a departure in its focus on eleztoral rather than interest group response to government policies, and in the focus on rnicro-level rather than macro-level variables. The analysis of micro variables, ie. specific programs targeted to key constituencies, may be a more effective tool for identifying opportunistic expenditure cycles than is analysis of rmacro variables. This may be even more relevant in developing countries, where public understanding of government policies and macro-outcomes is often limited. In addition, opportunistic government expenditure behaviour is more likely to be prevalent in young democracies. Several studies have focused on such variables, such Tufte's study of the relationship between fiscal transfers and elections, and Graham's work on the effects of compensatory programs on voting 24 behaviour during reform. 4 This article focuses on the role of discretionary public expenditures. The existence of both a reform program and repeated elections in Peru may also yield some insights into how such government behaviour affects the sustainability of reform. Initial Conditions: The Effects of the 1990 Elections The disastrous performance of Peru's oldest and most established political party., the APRA, in its 1985-90 tenure in power dramatically changed the face of Peruvian politics. Firstly, it discredited the traditional political establishment, and parties in particular.25 Secondly, by dramatically reducing partisan criteria in the voting process, it opened the door for independent candidate Alberto Fujimori, as well as a host of other independents. Fujimori's rise to power signified a political revolution of sorts, as it gave voters alternatives which, at least in the short term, were virtually free of partisan loyalties. An implicit hypothesis of this article is that this change made the electoral process in Peru more responsive to the traditionally marginalized rural population, and therefore more broadly representative, not necessarily were delayed for several years. See Kane and Morisett, 'Who Would Vote for inflation in Brazil", Policy Research Working Papers, #1183, Latin America and Caribbean Region, The World Bank, Washington, D.C., 1993. 23 See Nelson (1992) and Geddes (1995). 24 For detail on these studies, see Alesina (1993), Graham (1994); and E. Tufte, Political Control of the Economy (Princeton: Princeton University Press. 1978) 25 A 1995 survey found political panies last on the list of institutions that Peruvians trusted, with only 3.6% of people outside the capital and 2.3% of people in Lima indicating faith in political parties. For detail see Fransisco Sagasti et al., Democraciay Buen Gobierno: Agenda Peru (Lima: Editorial Apoyo, 1995). 26 The debacle of the APRA and Fujimori's rise to power are discussed in detail in C. Graham, Peru's APRA: Parties. Politics. and the Flusive Quest for Democratic Consolidation (Boulder: Lynne Rienner Publishers, 1992). Throughout Fujimori's term, and certainly by 1995, a host of independents established important political presence. These included Ricardo Belmont (the highly popular mayor of Lima) and his Obras front, Alejandro Toledo and CODE-Pais Posible, Fernando Caceres and Frenatraca in Arequipa, and Rafael Rey and the conservative Renovacion movement backing him. 9 because it was an objective of Fujimori's, but because the previous system had been dominated by a party structure that was largely elitist and Lima-based, and by a voting system which emphasized party lists rather than local representation. Fujimori's rise to power stemmed from three distinct phenomenon. The first was the extent to which the Peruvian state had deteriorated over time, and by the 1980's was incapable of meeting the increasing demands upon it for services such as basic education, health, and protection from crime. The second was the extreme nature of the economic and socio-political crisis, and the extent to which the public perceived that not just the APRA but all traditional political parties were part of a non-functional state system. The third phenomenon was the nature of the 1990 political campaign. Mario Vargas Llosa, the nation's best known novelist and an unlikely political condidate, was by far the front-runner until three months prior to the elections. Yet the public's perception of Vargas Llosa as of the elite, as well as his formal links to the parties of the traditional right through the Fredemo alliance, were political liabilities in a racially divided and increasingly impoverished society. This was exacerbated by Fredemo's campaign style and slogans, which were far more effective at appealing to the aristocracy than to either the mestizo or Indian populations. Equally damaging was the blatant display of large sums of money going to finance the Fredemo campaign, which underscored its nature as an alliance of the 27 rich. Finally, while most of the population expected some sort of stabilizing economic measures after the elections, Vargas Llosa's constant references to an upcoming shock hardly generated voter support. Fujimori, the former rector of the Agrarian University, entered the campaign just three months prior to the first electoral round as a political unknown. His campaign message was vague and simple: work, honesty, and technology. His electoral front, Cambio 90, had its primary base of support in the small and medium-sized business community, as well as in the small but active evangelical community, which was active in generating popular support in shantytowns and other low-income areas. F Fujimori's vague platform, as well as his being of Japanese rather than traditional European origin, seemed to have a broad appeal, particularly among lower income groups. Although Vargas Llosa won the first electoral round, with 32.7% of the vote versus Fujimori's 29.1%, APRA's 22.6%, and the United Left's 5.2%, it was clear that Fujimori would have the support of the APRA and the left in the second round, which he took with 62.5% of the vote over 37.5% for Vargas Llosa. [See Table 1] 27 The Fredemo campaign spent approximately $11 million, which was at least 10 times what was spent by any of the other campaigns. GDP per capita, meanwhile, fell by 25% in 1989-1990. For detail see Graham (1992). 28 Peru is primarily Catholic, and evangelicals make up only about 4% of the total population, with some estimates as high as 8%. 10 Table 1 President Elections 1985-19951 1990 1990 1985 First Round Second Round 1995 ACCION POPULAR 7.2 APRA 52.1 22.6 4.2 CAMBIO 90 29.1 62.5 63.7 CODE 11.9 FREDEMO 32.7 37.5 Is 4.8 IU 5.2 IZQUIERDA UNIDA 26.7 OBRAS 2.0 2.4 A. TOLEDO 3.4 J. PEREZ 22.4 R.D. 1.7 OTROS 2.1 2.2 ABSENT VOTES 8.8 22.8 20.1 31.0 NULOS 7.3 7.8 8.0 VICIADOS BLANCOS 5.0 1.7 5.4 The 1990 percentages have been adjusted to account for the new method of calculating votes introduced in 1993, based on valid votes only, rather than total votes cast. * For detail on these results by department, please see Tables 1.1-1.4 in the Statistical Annex Source. JNE 11 Fujimori had no organized base of support in Congress, nor any coherent plan for governing. His original cabinet was exclectic and multi-party in nature, and combined members of the traditional right, the left, and independents. After a meeting with the heads of the IMF, IDB, and World Bank, Fujimori seems to have become convinced of the need to adopt an orthodox economic strategy. In August 1990, the government implemented a shock stabilization program, which was successful at curbing hyperinflation, but for the first two years also resulted in deep recession and sharp further increases m poverty and unemployment.-9 Regard- less, Fujimori retained an surprisingly high degree of popular support, and met little resistance to his economic program from Congress.30 The opposition in Congress was much more concerned with the government's human rights record in the fight with Sendero, and its giving of an increasingly free reign to the military than it was with economic measures. Yet Fujimori's authoritarian style soon alienated those in his Cabinet who were not personal loyalists, and also resulted in deteriorating relations with Congress. On April 9, 1992, Fujimori closed down the Congress and the Judiciary, and declared a period of national reconstruction. In the weeks following the "coup", and in response to international pressure, he called for elections for a Constituent Assembly, to be held in November 1992, and for a referendum on the new constitution in October 1993. While Fujimori's measures were criticized strongly abroad, they were very popular with the Peruvian public, and his opinion ratings soared after the coup. One reason for this is that the Congress, like the political parties, was still perceived to be part of the discredited "old" system in much of the public mind. Shortly thereafter, in an intelligence coup in September 1992, the government captured Abimael Guzman and the top ranks of leadership of Sendero Luminoso, thereby decapitating the movement. And by 1993 the economy began to recover, growing at 6.3% in 1993, and then booming at 12.9% for 1994. Most Peruvians credited Fujimnori with "saving" Peru from the depths of crisis. While the positive economic results of Fujimori's "revolution" are already evident, its effects from a political point of view are less clear. The following analysis of voting patterns and public expenditures since 1990 seeks to evaluate how voters responded to these events. In particular, it posits that the increase in importance of independent votes and votes outside Lima, which has traditionally domninated the attention of governments and politicians, may have forced the government to be more responsive to the concerns of voters outside the capital. We seek to evaluate the extent to which such a response occurred, and its role in contributing to the political sustainability of economic reform. 29 The most extreme increases in poverty occurred from 1985-1990, due to the APRA government's poor economic management. but also increased due to the recession and price increases rresulting from the stabilization measures. Poverty in Lima. for example, was 17% of the population in 1985, and rose to 54% by 1990. By 1994, however, this figure had fallen back to 49.6%, largely as a result of renewed growth. For detail see Paul Glewwe and Gillette Hall, "Poverty and Inequality During Unorthodox Adjustment: The Case of Peru 1985-90", Living Standards Measurement Survey, Working 86 (Washington, D.C., 1992); and "Peru 1994: El Perfil de la Pobreza", Cuanto, Diciembre 1994. 3 0 Most measures could be implemented by executive decree, with Congress only having a retroactive review power. Even then, few measures of an economic nature were questioned. I 12 Voting Behaviour: 1990-1995 We classify the factors that influence voting behaviour into three main categories: economic performance, party affiliation, and the provision of public goods/services, including non-discretionary ones such as security, and discretionary ones such as transfers. This list is not exhaustive, but allows us to isolate variables for which data are available, such as discretionary public expenditures and trends in number of terrorist attacks, that presumably affected voting behaviour in Peru over the five year period. Issues such as the canditate's popularity are important, but outside the scope of our analysis. A theme framing the analysis is that with the decline of parties, voter behaviour became more "retrospective" in nature, ie. voters "reward" or "punish" incumbents based on their performance in power, rather than voting along either partisan or clientelist lines. While voters in Peru have voted unpopular incumbents out of office in the past, they never rejected the entire political establishment in the extreme manner that they have since 1990. Prior to 1990, while there was volatility in support among parties, they remained relevant political actors.31 The decline in support for parties since then, meanwhile, has accelerated as economic performance has improved during the course of the Fujimori government. Economic performance under the 1985-90 Garcia regime was extremely poor: inflation reached a four digit level, per capita GDP fell, and the exchange rate premium reached its highest historic level. The number of terrorist incidents per hundred thousand persons reached an annual average of 14.2, against 8.1 during the preceding Belaunde administration. Not surprisingly, the APRA fared poorly in the first round of 1990 elections. Indeed, it is remarkable that it was able to attain as much support as it did: 22.6% of the votes, close to the one-third average that the party traditionally obtained in Peruvian elections. Much of this support can be attributed to the traditional hard core APRA loyalists; not surprisingly APRA received the highest levels of support in 1990 in the "Aprista" districts of the north, such as Piura, Tumbes, and La Libertad. APRA fared less well in other parts of the country, particularly those with high levels of terrorism: Ayacucho, Junin, Pasco, and Huancavelica (not surprisingly, all of these are sierra districts traditionally neglected by public expenditure as well as by political parties).32 (See Figure 1 for a codification of departnents by location and income levels relative to the national average). 3 1 In Peru, there has traditionally been volatility in partisan loyalty, particularly on the center and right. However, until 1990. the APRA controlled a traditional one-third of the vote in virtually all elections is was allowed to run in. The virtual disappearance of parties since then stands in contrast to some other Latin American countries, where party loyalties remain stronger. A study conducted in Mexico, for example, found that individual loyalty to a party is very strong; that the greater the belief that the PRI would get stronger, the less likelihood of voting for an opposition party; the more likely a voter was to have voted PRI in a prior election, the less likely he/she would support an opposition party; and the greater the belief that the economy would improve if a party other than the PRI would gain power, the greater likelihood of voting for an opposition party. The study concludes that the findings are inconsistent with the 'retrospective" voting model, where voters reward or punish incumbent performance, and therefore there are strong linkages between performance on the economic reform front and voting behaviour. See Dominguez McCann, "Shaping Mexico's Electoral Arena: The Construction of Partisan Cleavages in the 1988 and 1991 National Elections", American Political Science Review, March 1995. It would be useful to have similar results for Argentina, for example, where the incumbent party has a strong traditional following, and has performed well on the reform front and been re-elected. 3 2 This contrasts sharply with the 1985 results, where APRA won in all but three departments. By 1990, APRA, s support declined dramatically in the departments with high rates of terrorism. [See Table 1.1] 13 Figure 1 Departments bv Location/Income Levels' COAST/LOWLAND SIERRA JUNGLE ABOVE Arequipa Loreto AVERAGE Lima Madre de Dios Moquegua Tacna AVERAGE Ica Tunin* Ucayali* La Libertad Pasco* Lambayeque Pinra Tumbes BELOW Ancash* Amazonas AVERAGE Apurima* San Martin Ayacucho* Cajamarca Cuzco Huancavelica* Huanuco* Puno Average income levels for Callao were not available. 2 Departnent income levels are based on ratio of department to GNP per capita in 1992. For breakdown, see Table 4. *Indicates departments with high rates of terrorism Vargas Llosa won the first electoral round, with 32.7% of the vote, but was short of the required majority. Fujimori took 29.1 % and fared best in the departments with the highest terrorism rates - those where APRA fared the worst. In Junin, for example, Fujimori won 52.5 % of the vote, versus 29.5% for Fredemo and 8.8% for APRA. In the second round Fujiimori also received high levels of support in the departments hardest hit by terrorism: Ayacucho - 81 %, Junin - 67.4 %, Pasco - 61.5 %, and Huancavelica - 84 %. There is evidence that ethnic characteristics played some role during the second round of elections, in part, no doubt, the result of the polarized tone of the electoral campaign. 33 3 3 Another contention worthy of attention is that religion became a factor during the second round. In particular and somewhat ironic, the Catholic Church felt sufficiently threatened by the surge in support for a movement backed by evangelicals that it openly endorsed Vargas Llosa, thereby backing an agnostic against a Catholic (Fujimori). [See Graham, 1992]. The data do not show Fujimori support during the second round being strongly affected by religion. Nevertheless, in the department where Fuijmori fared the worst, Ucayali, Catholics represented the smnallest proportion of the population (77%) of all departments. In Piura, where Catholics account for 94% of the population, Fujimori had his highest score against Vargas Llosa (87%). 14 While party affiliation lost importance at an overall national level in 1990, and support for Fujimori no doubt reflected a general dissatisfaction with the traditional party system, parties remained a factor influencing electoral outcome. It is relevant to examine the extent to which APRA's support for Fujimori in the second round contributed to his victory. If one adds APRA's and Fujimori's percentages in the first round, and compares that total to Fujimori's percentages in the second round, the totals match almost perfectly in four departments: Callao, Pasco, San Martin, and Ucayali. Support for Fujimori in traditionally strong "Aprista" departments was also relatively high, reflecting APRA's official endorsement. The departments where Fujimori's final vote by far exceeded the additional support through APRA's endorsement are the ones with the highest terrorism rates. In Ayacucho, for instance, Fujimori obtained 25.2% of the votes during the first round and APRA 12.5%. By the second round, Fujimori received 81 % of the votes. Similarly, in Huancavelica, with a final Fujimori vote of 84%, the corresponding gap was 41.3 percentage points. In areas where terrorism was intense, APRA's endorsement may have helped, but falls short of explaining the magnitude of Fujimori's support. The extent to which the role of parties diminished further by 1992/1993 and even more so by 1995 is remarkable. Two most established parties, APRA and Accion Popular, boycotted the November 1992 constituent assembly elections, in part of out of fear that they would fare poorly. Fujimori's Nueva Mayoria-Cambio 90 front took 44 of the 80 seats, while the next highest winner, the PPC, took only 8 and a new front representing parties of the left only 4.34 The rest of the seats were taken by small independent fronts.35 In the January 1993 municipal elections, independent candidates took the majority of provincial and department capitals including Lima. APRA and AP each won only four department capitals, and each received less than 15% of the national vote (each received just over 12%). As in 1990, support for established parties was particularly low in the poor and remote regions of the sierra, which tended to have high levels of terrorist violence as well. The increasingly evident decline of parties did not translate into direct support for Fujimori and his Nueva Mayoria-Cambio 90 Front, however, which only received 2.8% of the national vote. Cambio did not present candidates in all departments, and won only one department capital, suggesting the limits to which it was able to operate as a national level front.36 Instead 48% of the nation's vote went to a variety of indepen- dents.37 34 APRA is the American Popular Revolutionary Alliance; AP is the Popular Action party; Nueva-Mayoria/Cambio 90 is the New Majority/Change 90 alliance, and PPC is the Popular Christian Party. 3 5 Peruvian National Electoral Commnision, November 1992. 3 6 For detail, see John Crabtree, 'The 1995 Elections in Peru: The End of the Line for the Party System". University of London, Institute of Latin American Studies, Occasional Papers No.12, London, 1995. 3 7 An interesting result in these elections in term of the retrospective voting model is that the highest electoral support went to mayors who were re-elected in districts where they were percieved to have good performance records. This occurred at the provincial level as well as the provincial - Chiclayo, Cuzco, Tacna, Arequipa, and Callao - as well as at the district level - Miraflores, Chorrillos. San Isidro, San Luis, and Surquillo. [Apoyo, S.A., 'Resultados de las Elecciones Municpales: Informe de Opinion", Febrero 1993.] 15 The results of the subsequent October 1993 constitutional referendum, which Fuiinmori won by a small margin. were also far from a complete endorsement of Fujimori and came as a surprise to most observers. While Fujimori won in the capital city, Lima, he lost in most other departments outside the capital. [Table 1] There are several plausible explanations for this outcome. One is that while Lima was beginning to see the benefits of economic reform, the rest of the country had not. Another is that the "no" vote in the referendum reflects a protest: vote against Fujimori's performance rather than support for political alternatives. Unlike the municipal elections, where there were several independents providing viable alternatives at the local level (many with good perfornance records), at the national level there was no alternative to Fujimori in 1993, nor was the referendum structured to provide one. In addition, the opposition's campaign was characterized by lack of unity and the political parties' general state of disarr-ay. Many people, particularly in rural areas which were not benefiting economically, probably had doubts about several aspects of Fujimori's policy, and there was little risk in voting "no". A good example of one of these aspects, which was a major issue in the opposition's campaign and in the municipal elections as well, was widespread popular opposition to a govern- ment proposal to decentralize education.38 The common perception was that the benefits cf the reform would be concentrated in resource-rich Lima, at the expense of schools in poor and remote municipalities. Regardless of the reasons for Fujimori's losses outside Lima, the govermment seems to have taken the message seriously and responded with major increases in discretionary public expenditures outside the capital, and in particular in some departments where the "no" vote had been particularly high. The strategy had mixed results. In some cases expenditure increases correlate with increased support for Fujimori in 1995; in others they do not. In 1995 Fujimori won the presidential elections in the first round, with 63.7% olf the vote, well above the 51 % required to avoid a second round. Perez de Cuellar came in second with 22.4% of the votes. APRA received only 4.1% of the vote, the United Left 1.92%, and AP (Popular Action) 1.64%. Parties were clearly discredited.39 Perez, meanwhile, who had spent most of his life working abroad as a diplomat, seemed as out of touch with Peruvian realities and as unlikely a candidate as Vargas Llosa had been. Fujimori won over 50% of votes in all depart- ments but one: Loreto.40 There is not a single department in which his support was below that 3 8 The proposal was modelled on a similar scheme in Chile. There is a wide debate over the effects that the Chilean reforms had on overall performance indicators, with some evidence suggesting marginally positive effects. There is far more agreemert that the reforms resulted in a marked detrioration in quality and performance in schools in disadvantaged areas. 39 This discrediting was exacerbated by the new electoral law, which spedified that each party automatically lost official recognition if it failed to reach 5% of the vote and which created a single national district. Michael Coppedge, in a comparative study in five Latin American countries, has found that average district magnitude has significant effects on party fragmentatior. These effects are secondary, however, those of the underlying patterns of politicization in society. For detail on the effects of district magnitutde, see Coppedge (forthcoming), and for the electoral law in general, see Crabtree (1995). 40 It is interesting to note that Fujimori also failed to win this department during the 1990 election,with only 44% of the votes, and also lost in the 1993 referendum, with 54.5% voting "no". Perez received about one third of the votes in Loreto. Loreto seems to be an outlier: voting patterns were not determined by terrorism (a rate of 0.5/1000 in 1988 when the national average was 14) or 16 of the first 1990 round. In the south, where Perez was expected to do best and where anti- Fujimori sentiment was strongest in 1993, Fujimori attained very high levels of support: 67% in Cuzco, 63% in Puno, 66% in Apurimac, and 70% in Ayacucho. For the most part, these are departments with high poverty and terrorism rates, and where Fujimori had high levels of support in 1990. Nationwide, the only departments where Fujimori's support was below 60% were Arequipa (57%), La Libertad (59%), Lambayeque (59%), Loreto (48%), Tumbes (57%) and Callao (55%). Of these, Arequipa has traditionally voted for independents, and La Libertad is the heart of APRA support (even there, APRA received less than 15 % of the votes). None of these are in the sierra, and all have average income or above (Figure 1). There are departments in which the 1995 voting result was below that of the 1990 second round. The gap is particularly high in Piura, where Fujimori obtained 86.7% of the votes during the second round in 1990, compared with 65.7% during the first round of 1995. This also occurred in Lima, although with a less dramatic drop in levels of support. Again, there was probably at least some element of protest vote, as there was a widespread perception that Fujimori's contender, Perez de Cuellar, had very little chance of winning. In Lima, the results may have been a response to the notable cutback in the levels of municipal transfers relative to those to municipalities in the rest of the country.41 This may also explain the drop in support for Fujimori in neighboring Callao, which is often considered part of Lima. A caveat when comparing results between previous elections and 1995 is that the 1993 Constitution changed the basis for calculating vote percentages, shifting from calculations based on total vote cast to those based only on valid votes, excluding null and blank votes. In this case, for comparative purposes, we have adjusted the 1990 percentages along these lines; otherwise they would be disproportionately low. Another change in 1995 was that the presidential candidates' picture or party symbol appeared on the ballot rather than candidates' names. While party symbols had been used before, pictures had not, and there may have been some effect between the "known" quality of the presidential persona and the substantially higher levels of support for Fujimori in departments with lower levels of literacy than in 1993.42 It is likely that several other factors, including the different nature of the two elections and public expenditure trends, have a more important role in explaining the latter outcome. Finally, the results of the November 1995 municipal elections confirmed two trends that were evident in the 1993 municipal elections: firstly, the surge in support for independents, and secondly, the inability of Cambio 90 to operate as a national level front. Independents took all department capitals with the exception of five, two of which were won by AP, one by APRA, one by Frenatraca, and one by Obras. Cambio was unable to capture any of the major cities. by ethnicity (native Quechua speakers are 1.7%). One plausible explanation is the continued strong presesnce of the AP party, at least through the municipal elections in 1993. 41 This cutback was part and parcel of an overall reallocation of expenditures to areas outside Lima, but also a clear attempt on the part of Fujimori to undermine the highly popular mayor of Lima, Ricardo Belmont, who was a political rival and a contender in the 1995 presidential elections. 42 See Palmer (1995). 17 Even in Lima, Fujimori's hand-picked candidate Jaime Yoshivama, former president of the Chamber of Deputies. lost to the independent Alberto Andrade by a slight margin (52.7% to 47.3%).2 While some voters cited Yoshiyama's better position to access government resources as the reason the) planned to vote for him, the majority opted for an independent with no ties to Fujimori. suggesting that support for the President is far from unconditional. and that at the local level the trend in favor of independents is at least as strong as support for Fujimori. Public Expenditure Patterns: 1990-1995 Foncodes We concentrate on three kinds of public expenditures, all of which are h1ighly discretionary: the social fund, Foncodes; municipal transfers; and the school building fund, INFES. Municipal funds operate as a transfer from the central budoet. Foncodes and INFES, meanwhile, are channeled through thle Ministry of the Presidency, which was revived by Fujimori largely as a means to coordinate and control a number of social programs and to channel resources to the regional governments. The Ministry increased its presence and relative importance as discretionary expenditures increased with the improved fiscal situation after 1992, and by 1995 was allocated 22.6% of the total government budget.45 The most visible of the expenditures, Foncodes, a social fund designed to mitigate the social costs of adjustment and modelled on similar programs in other countries, was set up in August 1991.46 Foncodes got off to a slow start due to lack of initial government interest in the program. However, in mid-1992, due to international pressure and to an improvement in the fiscal situation, the government increased available funds and appointed a private sector manager rather than a political loyalist as director. Within the year the program was funding projects in every departnent in the country. There does not seem to be any obvious relationship between Foncodes expenditures in 1991-94 and support for Fujimori in 1995. This is not surprising and is in keeping with the usual political effects of demand-based social funds: as they require some sort of community participation, local governments and/or NGO's often can claim as much credit for the program. as can the central government. Thus positive program results do not necessarily translate into 43 Embassy of Peru. Washington, D.C., November 1995; and El ComeciQ, 13 and 15 November 1995. 44 This conclusion is based on a series of interviews that one of the authors had in the pre-electoral period with political observers in Lima, October 1995, and from press coverage of the municipal elections that time. 4 5 Of the Ministry's budget, the majority goes to finance social expenditures by the regional governments; the remainder is divided between Foncodes, INFES, and several smaller programs, with Foncodes clearly being the largest of these. See footnote #15. 4 6 For detail on these programs, as well as on the setting up of Foncodes, see Graham (1994). 18 direct political/electoral support for the central government.47 However, while a general or overall trend is not discernible, there are significant differences in levels of Foncodes expenditures - and their political effects - across departments. On average over the four years. per capita Foncodes disbursement was US$12. In Lima. Callao. Lambayeque, and Tacna, per capita expenditure was $3-6. At the other end of the spectrum, per capita expenditure reached $52 in Madre de Dios. Fujimori support in some departments where Foncodes presence was modest was still high. For instance in Tacna, where per capita expenditures were among the lowest in the country ($6), Fujimori obtained 67% of the votes. In Ucayali and Huanuco. where Fujimori obtained some of his best scores, per capita disbursement was below average ($11), although still higher than in Lima and Callao. Differences in the projects funded by these disbursements, however, could have effects that aggregate figures cannot capture. Some projects might have more visibility and directly benefit certain groups, while others have more diffuse effects."8 In addition, the impact of relatively low but still unprecedented expenditures in remote and relatively neglected departments is probably much greater than a higher level of expenditure somewhere like Lima, where there are a host of competing government and NGO programs. The specific role of Foncodes is therefore difficult to disaggregate. It is likely that where support for Fujimori was already high, an increase in Foncodes expenditures made little difference in electoral outcomes. However, in some cases where support for Fujimori was low in the referendum, a significant increase in levels of expenditure may well have made a difference, such as in Cuzco and Puno, where both expenditures and electoral support increased markedly after the 1993 referendum. There were also departments such as Ancash, Apurimac, and Pasco, where Fujimori lost in 1993 - although not by as large a margin as in Cuzco and Puno - which also received significant increases in Foncodes funding, and electoral support also increased markedly in 1995. [See Table 2] In departments such as these, which are among the poorest in Peru and have been traditionally neglected by government programs, one could plausibly argue that the expenditures did have some impact on people's electoral behaviour, tilting it in favor of Fujimori, and, indirectly, in favor of a continuation of the government's economic program. 47 See Graham (1994). 4 8 There are four categories of Foncodes expenditures: social assitance, social infrastructure, economic infrastructure, and production support. Departments such as Tacna, Tumbes, and Moquequa did not benefit from any depanment-specific social assistance programs. In Ucayali, neither social assistance nor economic infrastructure accounted for a large portion of expenditures. 19 Table 2 Changes in Support for Fujimori and Increases in Foncodes Expenditures by Department' Per-Capita 91- 92 Per-Capita 93 Per-Capira 94 Per-Capita 93-94 Fotnc-Exp3 Budgetedf Budgeted Buidgeted Buldgeted Department Votes/9dJ Yes Votes/93 Votes/95 % Increase FONCODES FONCONDES FONCONI)ES FONCONDES Amazonas 62.3 34 60.5 24.3 0.971 18.345 5.269 23.614 Apurimac 83.5 41.5 65.9 18.5 1.068 7.052 12.739 19.791 Arequipa 56.1 47.3 57.4 0.0 0 0 2.02 2.02 Ayacucho 81 39.9 70.7 4.5 6.794 16.741 13.521 30.262 Cajamarca 67.8 40.5 68.6 9.5 1.97 8.647 10.07 18.717 Callao 58.3 51.4 55.1 0.0 n/a n/a 1.829 1.829 Cuzco 78.5 32.3 67.5 10.4 2.275 12.963 10.727 23.69 1iuancavelica 84 32.5 64 8.4 2.648 11.518 10.615 22.133 Huanuco 64.1 43.8 77.3 13.3 1.117 7.731 7.152 14.883 Ica 64.4 49.4 60.6 7.8 1.463 8.28 1.684 11.427 Junin 67.4 53.2 77.2 6.7 2.51 9.543 4.882 16.935 La Libertad 75.1 42.4 58.1 7.1 1.727 6.894 3.06 12.281 Lambayeque 63.8 44.4 59.5 29.4 0.334 3.887 5.933 9.82 Lima 65.4 57.1 63.5 2.1 5.196 4.175 1.537 10(.908 Loreto 44 39.1 47.9 20.1 0.741 9.56 5.334 14.894 Mad/de/Dios 71 45.7 64.2 190.7 0.204 37.061 1.844 38.905 Moquegua 75.5 38.8 60.6 222.0 0.038 5.012 3.386 8.436 Pasco 61.5 42.4 64.8 61.5 0.285 8.87 8.f48 17.518 Piura 86.7 44 65.7 13.0 1.16 8.274 6.832 15.106 Puno 58.4 17.2 63.1 9.4 1.967 9.471 9.057 18.528 San Martin 41.5 43.4 65.9 22.8 0.592 10.276 3.242 13.518 Tanca 86.7 42.7 67 107.7 0.067 3.781 3.366 7.214 Tumbes 58.4 42.4 57.2 271.8 0.043 10.097 1.591 11.688 Ucayali 41.5 59.1 79.5 219.3 0.043 8.003 1.425 9.428 ' This version of the table is based on budgeted rather than disbursed expenditures. 2 Totals are for the second 1990 round. All tallies are in percentages. 3 Percentage increases in 1993/94 over 1991/92. a The total amoun t spent rather than budgeted appears in Table 3. We have used the budgeted amount for each year here as they reflect the governmilen t's intcntiorls better than disbursed amounts, which also reflected difficulties in project implementation 20 When expenditure patterns are broken down over shorter time periods, it is easier to determine the extent to which allocations were politically driven, as well as what kind of impact they had. Taking the 1993 referendum as a turning point, there are sianificant changes between budgeted expenditures across departments between 1991 and 1994. These are reflected in changing levels of average per capita expenditure by department relative to the national average. Departments in which a big increase took place in 1994 may indicate a government attempt to influence voting. The departments with the biggest changes in the relative size of the 1994 budget are (in decreasing order of importance): Lambayeque, Cajamarca, Madre de Dios, Tumbes, Huanuco, and Piura. A common feature of all these departments, with the exception of Huanuco, is that a majorit voted against Fujimori in the 1993 referendum.49 Other departments where Fujimori lost the referendum, such as Apurimac, Ayacucho, Cuzco, Pasco, and Puno, also received relatively large if not the largest increases in Foncodes expenditures, and support for Fujimnori increased markedly in 1995 compared to 1993. We posit that the distribution of Foncodes expenditures in 1994 was clearl influenced by the outcome of the 1993 referendum.50 It is telling that the director of Foncodes, Arturo Woodman, resigned in early 1994, apparently due to conflicts with the minister of the presidency over the political manipulation of projects.5 Even if Foncodes expenditure allocation was influenced by electoral outcomes in the post 1993 period, the results may not have been negative from either a political or poverty reduction standpoint. These departrnents in general had been relatively neglected by government expenditure in the past, and had high levels of poverty. Government programs have traditionally been concentrated in Lima, with the poorest regions of the country relatively neglected if they were reached at all. If these departnents expressed disapproval of government performance in 1993, and the government responded with an attempt to increase services to the poor, it is difficult to argue that this is an entirely negative trend. It is telling to compare the expenditure allocations of the major anti-poverty program of the APRA government, the PAIT employment program, and those of Foncodes. While the vast majority of PAIT funds in both relative and per capita terms went to Lima, per capita Foncodes expenditures were much higher in most departments outside the capital.52 [See Table 3] When seen in this light, one could argue that 4 9 Huanuco is one of the poorest departments in Peru, and therefore expenditure increases there may have been justified on techinical rather than political terms, particularly as Huanuco's average per capita levels remained below the national average for the 1991-94 period. so This hypothesis is supported by a general econometric analysis as well as by the individual department breakdown. See Cheikh Kane, 'Peru: Political Economy of Public Expenditures'. Memo. The World Bank, Washington, D.C., April 1994. 51 Roberts (1995), p. 105. 5 2 For details on PAIT, see C, Graham, "The APRA Government and the Urban Poor: The PAIT Programme in Lima's Pueblos Jovenes', Journal of Latin American Studies, Vol.21, Part 1, February 1991. Government (and donor) feeding programs, for example, are highly concentrated in Lima, even after the 1990 shock, when feeding programs played an important role in protecting the living standards of the poor and were indeed the only programs in place to cushion the effects of the shock. The proportion of families that used food programs was high only in Lima, where 47% of families in the lowest income strata participated. In the highest strata of Lima families in teh sample (the wealthiest 20% of families were excluded), almost 9% used a food program, a percetnage that exceeded that of even the lowest socioeconomic levels in other cities! See Hilary Creed Kanashiro 21 these trends made the electoral process and the government more responsive to broader popular demands than it was under a party-and clientelist based system. Table 3 A Comparison of PAIT and Foncodes 60 2500 -i 2000 U)I D *t9 40 C., - 1500 *PT ~~~~ 20 ~ ~ ~ ~ ~ ~ ~ ~ ~ -1000 -, ~~~~~500 0 0 Cal Lim Moq Lor PIu Uca Turn Apu Jun Cuz Pun Anc Mad Lam Tan Lib Pas Caj Hco Are Ica San Ama Aya Hua Departnients The increase in the "retrospective" element of voting may have made the government more responsive to voting trends, with traditionally marginalized regions outside the capital benefiting. The correlation of expenditures with support for the government is less clear. Increased levels of expenditure did not necessarily translate into "blind" support for Fujimori, for the reasons noted above. Fujimori won in 1995 in all departments with the exception of Loreto. In those departments with the highest levels of Foncodes expenditure, his level of support increased in comparison to the 1990 second presidential round in three: Cajamarca, Huanuco, and Puno, but decreased in four: Cuzco, Lambayeque, Madre de Dios, and Piura. [Table 2] Fujimori won many departrnents in the 1995 elections which he lost in the 1993 referendum. Yet in the 1993 elections voters were not taking "risks" by voting for alternatives, rather they were expressing opinions about the government's performance. In 1995, voters had a choice, and the vast majority voted for Fujimori. Still, the increase in Foncodes expenditures may indeed have played a role in influencing if not determining the 1995 outcomes. High levels of Foncodes expenditure are associated with significant increases in support for Fujimori in 1995 compared to and Guillermo Lopez de Romana, 'The Effects of Economic Crisis on Feeding Patterns and Infantile Malnutrition in Urban Populations in Peru', Mimeo, Instituto de Investigacion Nutricional, Lima, 1991. 22 1993, particularly in the departments where support for Fujimori was the lowest in 1993: Puno (from 17 to 63%), Cuzco (from 32 to 67 %), Huancavelica (from 32 to 64%). Amazonas (34 to 60.5%), and Ayacucho (from 39 to 70%), increases in support ranging from 77 to 266%. Yet when one compares levels of support for Fujimnori in 1990 - where there was also choice - rather than 1993. and 1995, then the impact of Foncodes expenditures is much less clear: high levels of expenditure correlate with increased levels of support for Fujimori in some departments. but seem to have had much less of an impact in others. [Table 2] The government fared very poorly in the nationwide municipal elections held in January 1993. Cambio 90 did not win a single important provincial capital, while independents won 31 % of the total vote. Accion Popular 21%, APRA and the United Left 14%. In October of that same year, Fujimori won the Constitutional Referendum by a close margin, but fared poorly in most departments outside Lima. In December 1993, through Decree 776, the government made a major revision to the tax law, changing the manner in which funds were distributed to the municipalities. Municipal Funds The municipalities have three major sources of revenue: local taxes, contributions and service charges, and a 2% value-added tax that feeds the national fund of municipal compensation. The sharing mechanism adopted in the new law divides the pool of resources among provincial municipalities on the basis of population and infant mortality. Within these provincial municipalities, with the exception of Lima, the resources are divided among the district municipalities on the basis of population, with the rural population being given twice the weight of the urban population. Within the provincial municipality of Lima the distribution across district municipalities is done on the basis of housing quality indicators and illiteracy. The result of this new rule was a drastic cut in the funds allocated to Lima, both in relative and absolute terms. While in 1991-92 Lima accounted for about 45% of municipal transfers, this share dropped to 17% in 1994. [See Table 4] Table 4: Distribution oir the Municipal Fund by Department (1991-94) (in percentage) I Ratio of Department to National GDPi' Head Departments 199] Share 1992 Share 1994 Share Cihange 91-94 1992 Amazonas - 1.32 1.04 1.61 +20.85 66.88 Ancash 2.83 ) 3.69 2.83 0.00 50.00 Apurimac 1.25 1.60 2.32 +85.46 26.62 Arequipa 4.88 1.05 3.58 - 26.69 125.97 Avacucho 1.72 2.93 3.81 + 121.39 29.87 Cajamarca 2.75 4.69 6.01 t]18.18 40.26 Cuzco 7.26 3.42 6.65 -8.48 53.90 Huiancavelica 0.87 1.41 2.89 +232.51 52.60 Huanuco 12.49 1.89 3.03 -75.68 51.95 Ica 1.09 0.67 1.97 +79.64 94.81 Junin 1.84 2.78 4.90 +165.82 80.52 La Libertad 8.88 2.19 4.47 -49.55 100.00 Lambayeque 1.03 1.48 3.82 +270.93 ]109.09 Lima 36.52 53.95 17.42 -52.30 134.42* Callao 1.42 2.51 1.83 +28,87 NA Loreto 1.06 3.17 3.88 +263.17 196.10 Madre De Dios 0.12 0.10 0.62 +384.51 148.70 Moquequa 1.03 0.33 0.57 -44.80 370.78 Pasco 4.85 0.75 1.55 -67.97 106.49 Puira 2.58 3.70 7.29 +182.37 94.81 Puno 2.21 4.25 6.50 +194.59 4 1.56 San Martin 0.90 1.40 2.15 +139.00 63.64 Tacna 0.22 0.09 1.01 +343.48 137.66 Tumbes 0.22 0.28 0.58 +157.00 92.86 Ucayali 0.53 0.54 8.60 +1509.30 86.36 Total 100 100 100 Note: GDP for Lima is includes Callao District municipalities as a grouip benefited from the new sharing mechanism vis-a-vis provincial municipalities. Big district municipalities (over 100,000) inhabitants increased their share from 4.3% in 1993 to 14.24% in 1994. The corresponding increase for small district municipalities was from 6.43 to 8.15%.[See Table 5] By contrast, the combined share of small and big provincial municipalities fell from 89 to 77%. In light of these trends it is relevant to note the extent to which size was a factor in the referendum: the NO dominated in all provinces with less than 50,000 voters, with 45-48% voting no, versus 37-38% voting yes. In contrast, in provinces with more than 1 million voters, the yes vote was, on average, 57% and the no 38%. The correlation between size and provinces was so evident that the vote was tied for medium provinces. [See Table 6] 24 Table 5: Distribution of the Municipal Fund By Tvpe of Municipality and Size (1993-94) Provincial Districtal All Big Small Big Small 1. 1993 (Million) A. Current Revenues 59.75 14.80 8.66 2.28 85.48 B. Municipal Fund 11.11 2.90 | 0.68 1.01 15.70 11. 1 994 (Million) A. Current Revenues 55.37 18.02 11.13 3.45 87.98 B. Municipal Fund 37.17 8.27 8.34 4.77 58.55 1. 1993 (Shares) A. Current Revenues 69.89 17.30 10.13 2.66 100 B. Municipal Fund 70.78 18.46 4.31 6.43 100 II. 1994 (Shares) A. Current Revenues 62.94 20.48 12.65 3.92 100 B. Municipal Fund 63.48 14.11 14.25 8.15 100 Source: Minitrv of Finance Note: A small municipality is defined as having a population below 100,000 habitants Table 6: Referendum Results and Province Size Voters Number of Yes No Blank Not Valid Total Provinces > 1000.000 1 57.19 38.17 1.24 3.40 100 100,001-1000,000 10 47.73 46.09 1.79 4.39 100 50,001-100,000 13 44.79 45.38 2.70 7.14 100 20,001-50,000 44 38.51 47.84 4.04 9.61 100 < 20,000 120 36.94 44.86 5.82 12.38 100 Total 188 47.54 43.46 2.64 6.35 100 Source: JNE Apovo One clear objective of changing the municipal fund allocation seems to have been to reduce the financial power of one of Fujimori's main rivals, Ricardo Belmont, the mayor of Lima. As the section on INFES demonstrates, Fujimori made up for the drop in municipal resources to Lima by spending on school infrastructure, an expenditure for which he rather than the mayor could take credit. His strategy was only partially successful, however. While Belmont did not run in the November 1995 municipal elections, they were won by the independent Alberto Andrade, who defeated Jaime Yoshiyama, Fujimori's hand-picked candidate, by a slight margin. Fujimori then attempted to recover political momentum and reduce Andrade's relative power by appointing Yoshiyama as the new Minister of the Presidency, a move which suggested that discretionary social spending would continue to form part of the government's political strategy. Andrade's request for a re-evaluation of the municipal fund allocation in order to provide more resources for Lima, meanwhile, was very publicly dismissed by Fujimori, by 25 Yoshiyama, by the Finance Minister, and by several members of Cambio 90 in congress. all of whom cited the benefits of the new allocation for smaller and poorer municipalities.s' A secondary objective in changFing the law was to channel more resources to poor departments. and to smaller municipalities wxithin them. One possible reason for this may have been to reduce the remaining power of parties in those departments, or at least to counter-balance the power of parties in the provincial capitals by building support in the smaller cities. This also may have been a response to Fujimori's low levels of support in smaller sized municipalities in 1993. The shift towards smaller municipalities is also in keeping with Fujimori's focus on direct contact with the population and his lack of trust of institutions. Such an approach is more likely to be effective in smaller and more remote communities where formal political organization is likely to be weak, than in larger, more urbanized and politically organized areas. In general departments whose share declined the most have a per capita income above the national average. In Lima, where per capita income is about 134% higher than the national average, the share of the municipal fund fell by 52%. Sirnilarly, Moquegua, which has a per capita income over three times higher than the national average, had a decline of its share of 45%. An exception to this trend is one relatively poor department, Huanuco, which suffered a severe decline in funds. Per capita income is 52% of the national average and its infant mortality of 87.4/1000 is the sixth highest of Peru's departments, yet its share fell by 76% (by far the biggest loss). Very few Cambio candidates contested the municipal elections in Huanuco, and all but one of them lost. The only province where a Fujimori candidate won was Puerto Inca, with 53% of the vote, while in provinces such as Maranon, Fujimori's Cambio 90/Nueva Mayoria won less than 6% of the votes versus 48% for the PPC. An independent took the provincial capital. The most cynical interpretation is that the government saw the departnent as irrelevant and a "lost" cause politically. A more generous interpretation would still find it difficult to explain the decrease of funds to such a poor department. In contrast to Huanuco is Ucayali, which benefited the most from the modified sharing mechanism. Its share of municipal funds increased from 0.53% in 1991 to 8.6% in 1994 (1509%). This department is also relatively poor: per capita income is about 86% of the national average. As in Huanuco, Fujimori did not fare particularly well in the 1993 municipal elections: AP took the department capital, Pucallpa, with 20.1 % of the vote while the Cambio cand:idate received only 1.8%. AP also took two other provincial capitals, and an independent took the third. Cambio candidates did not even contest most of the district municipalities, and most of these were taken by independents. In contrast to Huanuco, however, in Ucayali the traditional parties (AP in particular) rather than independents had a relatively strong presence. It is difficult to gauge the political effects of municipal fund reallocations by comparing these two cases. Firstly, while their municipal shares contrast, both of them received relatively high levels of Foncodes expenditure. Fujimori lost the 1993 referendum in Huanuco, receiving only 43.8% of the vote; in 1995 he received 77.3%, a 76% increase, making Huanuco one of 53 ElComercio, 13 and 15 November 1995. 26 the ten departments where support for Fujimori increased the most from 1993 to 1995. In contrast, Fujimori received 59.1% in Ucavali in 1993 and 79.5% in 1995, a much smaller increase (34.5%) than in Huanuco. One explanation is that initial support for Funjimori was alreadv high in Ucavali and therefore expenditure increases mattered less. Another is that municipal fund increases do not necessarily generate support for the central government granting them. A number of rich departments also received a sharp increase in expenditures, suggesting that there were political objectives in the fund re-allocation. The "outlier" departments. which had both high levels of per capita income and high increases in municipal expenditures were Lambayeque, Loreto, Madre de Dios, and Tacna. [See Table 4]. An example of why these departments are considered "outliers" is Madre de Dios, which has a per capita income 149% above the national average and an infant mortality (67/1000) among the lowest in all departments.54 Despite this, the department had the second highest increase in municipal funds after Ucayali. Interestingly, Madre de Dios also had the highest amount of Foncodes funds per capita in 1993-94. In all these "outliers', with the exception of Tacna (the only place where Fujimori's front won a department capital in 1993), the department capitals were taken by opposition parties rather than independents (as was the capital of Ucayali), and in all cases with a relatively high margin. [See Table 7] AP won in Lambayeque with 56.1% of the vote and in Loreto with 40%, and APRA won in Madre de Dios with 24.2%. (In the five other departments where parties won capitals, the allocation of funds was more appropriate to relative income levels; most of them were poor and merited a rather large increase. 55) Also, in all of these "outlier" departments, including Ucayali, parties took most of the provincial municipalities. Parties fared reasonably well in the district municipalities, although independents tended to dominate, particularly in the smallest ones. s4This rate seems low as there are, without a doubt, pockets of extreme poverty in Madre de Dios. The low rate may in part reflect under-reporting of infant deaths among the very poor. Among other reasons, there is a charge for officially registering the dead for burial, an additional financial burden that most poor families wish to avoid. ss Parties also took the district capitals in Huancavelica (PPC), which had a high increase but a very low income levels; Ica (APRA). which had slightly below average income levels and some fund increase; Junin (AP), which is poor and received a high increase; La Libertad (APRA). which is relatively wealthy and had a decrease; and Piura (APRA), which is quite poor and received a high increase in funds. 27 Table 7: Municipal Elections 1993 Department Capitals Department Capital Winner Support for Fujimori Candidates ____ I __ ,_ _ (%) Amazonas Chachapoya Ind. 27.9 26.7 Ancash Huaraz Ind. 24.0 -- Apurimac Abancay Ind. 28.5 7.1 Arequipa Arequipa Ind. 43.6 -- Ayacucho Ayacucho n/a -- Cajamarca Cajamarca Ind. 31.3 2 .7 Cuzco Cuzco Ind. 58.9 16.4 Huancavelica Huancavelica PPC 21.9 -- Huanuco Huanuco Ind. 57.0 Ica Ica APRA 32.8 Junin Huancayo AP 32.9 23.1 La Libertad Trujillo APRA 63.1 Lambayeque Chiclayo AP 56.1 Lima Lima Ind. **44.3 Lima Callao PPL 49.7 Loreto Iquitos AP 40.0 Madre de Dios Pto Maldonado APRA 24.2 6.6 Moquegua Moquega Frenatraca 37.3 Pasco Pasco Ind. 23.6 10.1 Piura Piura APRA 25.8 2.6 Puno Puno Ind. 18.6 San Martin Moyobamba Ind. 32.2 6 Tacna Tacna C90 46.6 Tumbes Tumbes Ind. 45.6 20.7 Ucayali Pucallpa AP 20.1 1.8 * Cambio 90/Nueva Mayoria: the front in support of Fujimori. ** The independent front behind Ricardo Belmont, Obras, took Lima. Sources: Peruvian Embassy, Washington, Press coverage, Lima 28 Table 8: Voting 1993-1995 Highest Percentage Increases bv Department Department Votes!93 Votes/95 % Increase Puno 17.2 63.1 266.9 Cuz 32.3 67.5 109.0 Huanca 32.5 64.0 96.9 Ay 39.9 70.7 77.2 Am 34.0 60.5 77.9 Huanuco 43.8 77.3 76.5 Caj 40.5 68.6 69.4 Ap - 41.5 65.9 58.8 Moq 38.8 60.6 56.2 Pas 42.4 64.8 52.8 San 43.4 65.9 51.8 Tan 42.7 67.0 56.9 Piu 44.0 65.7 49.3 Jun 53.2 77.2 45.1 Mad 45.7 64.2 40.5 LaLib 42.4 58.1 37.0 Ucay 59.1 79.5 34.5 Tumb 42.4 57.2 34.9 Lamb 44.4 59.5 34.0 Lor* 39.1 47.9 22.5 Ar 47.3 57.4 21.4 Ica 49.4 60.6 22.7 Lima 57.1 63.5 11.2 Cal 51.4 55.1 7.2 * Loreto was the one department where Fujimori failed to attain a majority in 1995. One plausible hypothesis for the disproportionate fund increases to these "outlier" departments is that Fujimori felt more of a political threat from the parties, which still had some level of local organizational capacity, than from independents. Fujimori's front, Cambio 90- Nueva Mayoria, had difficulty organizing on a national level and fielding credible local government candidates in the municipal elections.56 Thus in response to the electoral results, Fujimori may have attempted to build support in the smaller district municipalities to counter the power of the parties in the provincial capitals. The municipal fund re-allocation, without a doubt, redistributed resources from large to small municipalities. While it is far from clear that these reallocations had significant effects on the results, it is relevant that parties fared less well in department capitals in the 1995 municipal elections, winning only in five and with independents 56 See Crabtree (1995). 29 taking the rest.57 It did not benefit Cambio, which was unable to win any department capitals. including Lima where Fujimori' s hand-picked candidate contested the elections (discussed above). While it is plausible that Fujimori had political objectives in the redistribution of the municipal funds in these "outlier" cases, they were secondary to the objective of reducing Lima's access to resources. This helps to explain the diffuse political effects of the fund reallocations. None of the "outliers" was in the top ten departrnents where support for Fujimori increased the most in 1995; indeed. one of them was the one department where he lost in 1995: Loreto. [See Table 8] Nor did Cambio candidates benefit in November 1995. Parties were able to maintain some of their traditional support in some provincial cities. Also, it is not clear whether voters associate local government issues and performance with national level politics, suggesting the difficulties of influencing voter behaviour through such expenditures. The effects of increasing munipal funds on electoral outcomes at the national level are probably as diffuse as thcse of expenditures made through demand based social funds like Foncodes. Regardless of the electoral outcome, the fund re-allocation seems to have had positive effects in getting resources to poor and previously neglected municipalities. INFES Another form of discretionary expenditure, INFES, which is the school infrastructure fund managed by the Ministry of the Presidency, was also clearly influenced by the government's political objectives after 1993. Public opinion polls taken in late 1994 cited school infrastructure as one of the primary achievements of the government after the defeat of Sendero. While 36.3% of those polled cited school infrastructure as the primary achievement, only 10.8% cited control of inflation and 9.5% the stabilization of the economy.58 The government was, no doubt, aware that this was a potentially lucrative expenditure in political terms, and indeed may have tried to reverse the negative public image it created for itself with the decentralization of eduction proposal in 1993. A major campaign promise of Fujimori's was to innaugurate a school per day from the onset of 1995 until the elections. In 1990 INFES expenditures were (in 1994 $U.S.) approximately $32,000 and in 1991 they were $194,650. These expenditures increased to $29.8 million in 1993, $112 million in 1994, and $192 million for the first five months of 1995. There was also a change in. the structure of expenditures. In 1990 investment (ie. school construction) accounted for 82%7 of total expenditures. This share increased to over 99% in 1994-95. Lima and Callao received 61 % of the total schools in 1994 and 51 % during the first three months of 1995. To some extent this counters the trend that is observed earlier, and in particular 57 Most of the capitals where panies won in 1996 were traditional strongholds: Arequipa (Frenatraca), Chiclayo (AP), an,i Trujillo (APRA), AP was re-elected in Huancayo, meanwhile, and Obras won in Piura. s8 El Comercio, 9 September 1994. 30 for the municipal fund. When the number of schools is divided by the population. Lima still has the highest rate, with 42 new schools per million inhabitants (1994-95). [See Table 9] However, other departments such as Huancavelica and Amazonas stand as gainers with rates of 35-36 new schools per million inhabitants. What is particularly striking is that the number of schools relative to population cannot be explained by differences in the illiteracy rate. Except for Huancavelica, many of the high illiteracy rate departments benefited disproportionately from the INFES program. For instance Apurimac, despite a high illiteracy rate of 37% (the highest among all departments), received only 3 schools in 1994-95 or 7 schools per million inhabitants. In contrast, a department such as Ica, with the lowest illiteracy rate (1.9%) and which is also relatively wealthy compared to many departments in the sierra received 20 schools or 34 schools per million inhabitants (the third highest rate). Clearly the INFES program was not guided by a need to reverse differences in illiteracy across departments. To the extent that INFES expendi- tures had any direct effects on voting behaviour, it is interesting to note that those departments that benefited the most from INFES expenditures, Amazonas, Ayacucho, Cuzco, Huancavelica, Moquegua, and Pasco, comprise six of the ten departments where support for Fujimori increased the most from 1993 to 1995. [See Table 8] Table 9: Distribution of INFES Schools (1994-March 1995) Number of Schools Schools Per Million Population Department 1994 1995 Total 1994 1995 Total Illiteracy Rate March March Amazonas 4 9 13 11.05 24.86 35.91 17.00% Ancash 15 1 16 15.09 1.01 16.10 19.00% Apurimac 3 0 3 7.48 0.00 7.48 36.60% Arequipa 8 3 11 8.36 3.13 11.49 5.20% Ayacucho 6 6 12 11.74 11.74 23.48 34.40% Cajamarca 7 6 13 5.31 4.55 9.86 25.30% Callao 0 0 0 0 0 0 1.70% Cuzco 13 9 22 12.00 8.31 20.31 21.50% Huancavelica 7 7 14 17.37 17.37 34.74 30.20% Huanuco 9 0 9 12.95 0.00 12.95 21.50% Ica 18 2 20 30.46 3.38 33.84 1.90% Junin 6 6 12 5.41 5.41 10.81 12.00% La Libertad 10 10 20 7.62 7.62 15.24 10.00% Lambayeque 6 2 8 6.17 2.06 8.22 7.50% Lima and Callao 231 79 310 31.73 10.85 42.58 2.20% Loreto 1 2 3 1.32 2.64 3.96 8.40% Madre de Dios I 0 1 13.51 0.00 13.51 3.00% Moquegua 3 1 4 22.56 7.52 30.08 5.20% Pasco 1 3 4 4.17 12.50 16.67 13.30% Piura 7 4 11 4.89 2.80 7.69 11.60% Puno 4 0 4 3.57 0.00 3.57 21.50% San Martin 2 4 6 3.34 6.69 10.03 9.50% Tanca 1 0 1 4.33 0.00 4.33 3.60% Tumbes 0 0 0 0.00 0.00 0.00 2.60% Ucayali 2 0 2 5.73 0.00 5.73 6.40% Total 365 154 519 15.81 6.67 22.48 10.70% Note: Using the 1994 population for the ratios 31 UWhen electoral concerns are considered. then it should come as less of a surprise that Lima received high levels of expenditures. Particularly as both Foncodes and the municipal funds were concentrated outside Lima, Fujimori may have wanted to use at least one form of "highly visible discretionary expenditure to guarantee continued political support in Lima, and INFES was a tool which was outside the realm of influence of the highly popular mayor of Lima, Ricardo Belmont. INFES expenditures probably had some social positive impact, yet they also were relatively haphazard. as, for example, school buildings will hardly do much good in areas where teachers and school supplies are not available. Nor, as is mentioned above. were they allocated according to illiteracy rates. In the case of municipal funds, while there are some major outliers that are likely to have political explanations, in general allocations seem more closely correlated to income levels and size of municipality than in the case of INFES. The same is true for Foncodes expenditures, which had a reasonably close correlation with poverty levels. Even though the overall effects of these expenditures were on balance progressive, there was sufficient deviation, such as INFES in Lima and the skewed distribution of some Foncodes and municipal fund expenditures, to raise genuine concerns about the governnent's objectives. The Abstention Factor The degree of participation is an important factor in the dynamics this paper attempts to explore, and abstention had an important impact on elections in Peru. The premise is that voting played a key role in the pattern of public expenditure, a pattern which then had some impact on the political sustainability of the reform program. The evolution of abstention during that period is relevant: one would expect that departments that were less active in the political arena were also less successful in receiving assistance through a demand-based program like Foncodes, and possibly even other goverment programs. In addition, one would also expect abstention to be high in the high terrorist rate regions where Fujimori had strong support in 1990. The three departments of the highest terrorism rates display very high abstention rates in 1990 and 1995 despite a decrease in terrorism. In Ayacucho absenteeism was 41% and 57% in the first and second round of 1990 elections; in 1995 it was still 45%. In Huancavelica, absenteeism exceeded 40% in 1990 and 1995. [See Table 10] Nationwide abstention has risen steadily since 1985, continuing to grow after the defeat of Sendero in 1992. Abstention rates were 8.8% in 1985, 22.8% in the first round of 1990, and 31% in 1995. In the vast majority of departments (17 out of 25), absention was higher in 1995 than in 1990. Even after the "defeat" of Sendero Luminoso, continued high abstention rates may reflect related demographic trends. For example, it is likely that there are a number of voters who are still registered in their home departments, migrated elsewhere to escape violence, but did not change their registration as they planned to eventually return. Yet the trends still suggest that terrorism is not the sole explanatory variable. A plausible explanation is that absten,tion increased with the increase in terrorism and simultaneous collapse of the state, and recent govenmment efforts to re-direct public expenditure are not sufficient to address a decade-long trend. Indeed, the most significant increase in abstention took place in Madre de Dios, where it 32 rose from 20% in the 1990 second round to 47% in 1995. Madre de Dios had the lowest incidence of terrorism in the country; it received the highest per capita Foncodes disbursement, at $52 over three times higher than the national averaae, and it had a very high increase in municipal funds despite relativelv high average per capita income. Another possible explanation for the high levels of abstention is the remote nature of the department. and the absence of strong political or cultural link-ages with the rest of the country. While terrorism was certainlv a factor in some departments, these results suggyest that other variables, such as civic community and political culture, may also be involved. Table 10: Abstention Rates in Presidential Elections 1985 1990 1995 Round 1 Round 2 Amazonas 8.70 37.8 36.5 32 Ancash 6.50 27.1 25.8 31.7 Apurimac 17.90 27.9 27.8 33.8 Arequipa 7.60 16.7 11.9 33.3 Ayacucho 17.10 11.3 57 45.1 Cajamarca 12.50 27.4 30 22.6 Callao 6.00 13.7 15.2 30.5 Cuzco 12.90 24.4 24.2 39.2 Huancavelica 21.70 50 44.3 41 Huanuco 13.60 60 44.5 31.7 Ica 5.10 14.4 13.7 24.8 Junin 9.90 49.4 38.3 36.2 La Libertad 6.50 18 19 25.1 Lambayeque 7.70 15.8 16.5 24.9 Lima 11.20 17.6 17.5 30.1 Loreto 12.80 27 24.2 39 Madre de Dios 16.60 9.6 19.8 47.4 Moquegua 7.20 37.4 13.8 31.1 Pasco 13.10 22.9 27.8 36.7 Piura 8.50 21.6 14.7 29.3 Puno 9.00 31.2 21.6 33.7 San Martin 14.50 7.1 28.7 34 Tanca 6.30 9.2 57 25.8 Tumbes 4.40 29.6 6.4 27.1 Ucayali 14.10 53.7 22.8 32.8 Extranjero 25.70 54.7 58.5 NA Total 8.800 22.8 20.1 31 In this light, it is relevant to consider Robert Putnam's work on social capital. Putnam finds that the level of "civicness" in communities was a much more important factor in determining differences in performance among decentralized regions than was physical or human capital. 59 The degree of "civicness" in different Peruvian departments mav well have an important impact on electoral behaviour. While an attempt to measure "civicness" is beyond the scope of this article. and probably quite difficult in rural Peru, one can still posit that it may have played some role. This could explain the low level of "responsiveness" to significant increases in public expenditures and continued high rates of abstention in Madre de Dios. It is also relevant as Foncodes expenditures are. at least in theory, allocated according to priorities set by the recipient communities themselves. The extent to which this was not the case in Madre de Dios, where expenditure levels were disproportionately higher than in other departments, mayl, also help to explain the lack of reponse to expenditure increases. Indeed it is difficult to imagin.e that these high levels actually reflect real levels of autonomous demand for projects in Madre de Dios. This also supports the thesis that expenditres which are allocated according to political criteria are less likely to have positive effects than are those which are allocated in response to autonomous demand or to genuine needs. The high abstention rates may also reflect the extent to which alternatives to Fujimori in 1995 - the parties, Perez de Cuellar - were not credible ones, limiting interest in voting. Most worrisome, the trends suggest that despite high economic growth rates and the re-orientation of public expenditures, a significant part of the population remains marginalized, both in economic and political terms. Conclusion This article began with the proposition that the economnic and political changes in the past five years in Peru have had important effects on voting patterns and on government responses to those patterns. Severe economic crisis, the deterioration of state capacity, and the discrediting of the traditional party system resulted im a pattern of voting that resembles the "retrospective voting model", a model in which voters reward or punish incumbents based on their performance in government, rather than voting according to partisan affiliation or for other reasons, such as personality of the particular candidates. We posit that Peruvian voters evaluated government performance according to at least three criteria: economic trends, progress in fighting terrorism, and public expenditures and transfers. Accepting the dominant importance of the first two, we specifically explored the role of the latter. We found that discretionary public expenditures very clearly responded to voting trends, and in particular, expenditures were re-directed to areas where Fujimori lost the :1993 referendum. In general this resulted in a substantial redistribution of public resources from the capital city, Lima, to remote rural -regions of the country previously neglected by Peruvian ss Robert Putnam, Makino Democracs Work: Civic Traditions in Italy (Princeton: Princeton University Press, 1993). 34 governments.60 However, there were also several instances of expenditures being allocated in a manner that had little to do with relative poverty levels or social indicators, with benefits going disproportionately to less needv groups. The effects of these expenditure trends on voting patterns are less straightforward than the effects of voting on expenditures. In the case of Foncodes, if one compares votes in 1993 and 1995, there clearly was an increase in support for Fujirnori in many areas where expenditure was increased, particularly those departments where Fujimori's support in the referendum was the lowest. Yet these may not be comparable elections, as there was no choice of candidates in 1993. The 1990 elections are more comparable, and then trends are more mixed. In some regions with high expenditures support increased vis-a-vis 1990, in others it actually decreased. In the case of municipal funds, in the departments where expenditures increased the most, between 1990 and 1995 support for Fujimori increased in some and decreased in others. It increased in virtually all of the high municipal expenditure departments if one compares 1993 and 1995, but again that reflects a nationwide trend and less than comparable election results. Of the highest INFES recipients, meanwhile, there was actually a decrease in support across the board between 1990 and 1995. At the same time, the six highest INFES recipients comprise the majority in the top ten departments where support for Fujimori increased the most between 1993 and 1995. [Table 8] There are several explanations for this which in part stem from the nature of the expenditures themselves. Firstly, it is not clear how directly an increase in funds to local governments affect national level voting behaviour. Secondly, public expenditures distributed via demand-based social funds such as Foncodes rarely have direct effects on electoral outcomes. As beneficiaries must organize and contribute in order to obtain projects, the central government is not always credited for the benefits. This in and of itself is a positive element which may not have been the government's intent when making the expenditures: people in remote communities obtain experience in organizing and soliciting the state for services, but they do not feel required to vote accordingly.6" The logical next question is whether these changes in voting patterns are permanent, what kind of effects they will have on Peruvian politics over the longterm. This depends on whether the govemrnment continues to respond to voting patterns - and how the population evaluates the government - in the next round of elections.62 Go One exception may have been the first government of Fernando Belaunde Terry (1 963-68), which spent a large proportion of public funds extending education to rural areas of the country, although these were not allocated in response to electoral results. In addition, critics argue that much of the expenditure, which was on higher education, did not benefit the poor. 6i At least in its early years, Foncodes' relations with NGO's and local organizations were fare from trouble-free, due in part because the program was not operating in a truly demand-based manner. Several changes in management since then, as well as the reach and record of the program since then suggest that its performance on this front has improved, although it still remains overly centralized. For detail on the early years, see Graham (1994), Chapter 4. 62 The Novermber 12, 1995 municipal elections, which reflected a continuing and increasing tendency to vote for indepen- dents, were too close to the April 1995 elections to reflect any major public expenditure shifts. 35 Also irnportant in terms of the permanent effects of these trends is whether they contributed to the political sustainabilitv of reform. The results suggest that they at least influenced voter behaviour in some depar-tments, indirectly increasing voter support for continuation of the reform program. This raises the question whether building support for reformn at times requires "inefficient" public expenditures. New schools such as those built by INFES are a positive benefit. but the money could have been better spent on a comprehensive effort to improve basic education. The latter kind of expenditure may be more optimal from a public sector and poverty reduction standpoint, but tends to be less visible, takes a long time to yield positive results, and therefore has less political impact.63 As in many instances reforms are reversed by popular protest or by elections prior to their completion. it is no surprise that governments try to deliver more visible and immediate results. If some discretionary expenditure contributes to increased public good will towards the government and influences electoral behaviour in favor of reform at a critical political moment, then it indirectly contributes to growth and poverty reduction in the longterm. And if the discretionary public expenditure is directed at the poor, then it may have positive equity effects as well. In Peru, overall trends were progressive. Yet the discretionary nature of the expenditures increases the tempation for governments to use these funds for political purposes, and in Peru there were a number of "outlier" cases. There is little guarantee that the outcome would be progressive in other contexts, highlighting the risk of replicating the experience. Discretionary expenditures had three kinds of effects in Peru. The first was the short-term welfare benefits the expenditures yielded. The second effect was providing political space for the completion of reforms and the initiation of growth. Thirdly, there may also have been an additional and longer term political benefit: through the social fund previously marginalized groups obtained experience in organizing, in cooperating with local institutions, and in making more effective demands on the government. And the extension of resources to cash-strapped local governments in poor and remote areas may have provided a boost to local government, at least in the cases where resources were not poorly utilized. How lasting these effects are can only be determined as the results of future elections become available. In closing, it is important to note that a great deal remains on both democracy and economic reform agendas in Peru. The erosion of parties may have had important catalytic affects on Peruvian electoral behaviour. Yet parties play an important representative function in all democracies." In addition, there is also some evidence that countries with strong and stable 63 The same funds allocated to reform of mainstream institutions, such as the Education and Ilealth ministries, might be opitmal in public expenditure terms, but would probably take years to demonstrate positive or visible results. Recent literature has made distinctions between the design of adjustment programs along three distinct lines: political optimality, economic optimality, and equity optimality. and suggests that there are significant trade-offs between the three. [See Christian Morrisson, Stephan Haggard, and Jean-Dominique Lafay, The Political Feasibility of Adiustment in Developing Countries (Paris: OECD, 1995] A slightly different view is that properly designed safety net programs can get around some of these trade-offs and contribute to both the political sustainability of reform and to poverty reduction. [See Graham (1994)] 64 See. for example, Maurice Duverger, Political Parties; Their Organization and Activity in the Modern State (London: Methuen, 1967); and Giovanni Sartori, Parties and Party Sytems (Cambridge: Cambridge University Press. 1976). 36 party systems have an advantage in stabilizina inflation and implementing economic reforms.65 The development of a viable party system remains on the agenda for Peru. There is also a critical need to develop democracy from a civil society point of view. The high levels of abstention, while in part attributable to terrorism, also indicate low levels of "civicness". public awareness. and experience with democratic politics. The high levels of support for Fujimori's closing of Congress and the Judiciary, for example, not only reflect the weakness of those institutions, but also low levels of civic understanding of their importance. In the short-term, the Fujimori "revolution" may indeed have had some very important - if unintended - effects on democracy in Peru. In the longer term. consolidating the democratic process will require substantial insitutional development and civic education. A great deal of progress has been made in economic reform, not only in the arena of macroeconomic policy, but in some important institutional areas, such as tax reform and reform of the social security system. But important reforms remain on the agenda, in particular those which would have longrun equity enhancing effects, such as of the institutions that deliver basic social services. Progress on this front is also key to the development of civic education. While discretionary public expenditures and/or safety net measures can have welfare enhancing effects in the short term, any positive longer run effects hinge on a macroeconomic framnework which can generate sustained growth, and on institutions which deliver basic health and education services. In addition, discretionary public expenditures, precisely because they are such, can be reversed or re-directed at any time at the whim of particular political leaders. Achieving a distribution and allocation of public expenditure which is not discretionary and/or determined by shifts in voting patterns, but rather by efficiency and equity criteria, will be critical to the sustainability of reform in Peru, for both political and economic reasons. A final consideration is the extent to which the analysis of public expenditures and voting in Peru is relevant to other countries in the process of reform. At the least the results indicate a number of important themes worthy of further exploration. The first of these is the role of public expenditures in sustaining reform. In the case of Peru, the expenditures were not optimal ones from a poverty reduction standpoint. Yet the expenditures did contribute, directly or indirectly, to electoral support for Fujimori in some areas in particular where support had fallen in 1993, suggesting that such expenditures may help sustain reform at critical political moments. Yet when expenditures were allocated in an opportunistic manner rather than according to need or to autonomous demand, the results suggest that they were far less effective at generating political support and may even have had negative effects. Case in point are the examples of Loreto, where Fujimori still lost in 1995 despite a disproportionate increase in municipal funds, and Madre de Dios, which received extraordinarily high levels of Foncodes expenditures and high increases in municipal funds, but support for Fujimori was still lower in 1995 than it was in 1990. A second theme that emerges is that of the role of party systems. There is fairly broad agreement on the importance of parties. Yet their performance varies in less developed democra- 6s See Haggard and Webb (1994). The exception to this is cases of extreme crisis - e.g. periods of hyperinflation - when even highly polarized party systems like Bolivia and Poland have succeeded in stabilizing and implementing major reforms. 37 cies. Where systems are elitist and clientelistic. as in the case of Peru. or highly partisan and semi-authoritarian, as in the cases of Mexico and Senegal, for example. they may actually limit democracv. In the case of Peru. deep socio-economic crisis and inadequate government response resulted in a major discrediting of the party system, and in a surge of independent candidates backed by broad fronts rather than by established parties. The short-term effects of this se-m to have been-the increase in "retrospective" rather than partisan-oriented voting patterns, and a consequent re-evaluation of the weight of votes outside Lima. The govemrnent's response wvas to re-direct public expenditure from the capital to poor and remote regions on an unprecedented scale. While this does not discount the importance of parties in the longterm, it does suggest that a reduction in the role of clientelistic parties, whether triggered by economic crises or by internal party reforms, can change the manner in which governments respond to voter behaviour, anad by doing so, indirectly build a broader base of support for economic reform. Finally, the results of this analysis suggest the need for a re-evaluation- of the util:ity of discretionary public expenditures during reform. They suggest that such expenditures, if directed to poor and previously marginalized groups in a demand-based manner, can have a positive impact on sustaining reform, and perhaps even encourage independent political behaviour. This is particularly relevant for countries where social sector institutions are very inefficient or underdeveloped, and it will take a long time before reforming them yields results. Yet whatever positive lessons emerge, there is also a risk involved in relying on discretionary expenditures. When expenditures are allocated in an opportunistic political manner rather than according to genuine need or autonomous demand, they are unlikely to have positive political effects. This article is merely a starting point in exploring these themes, and further research could contribute to both the academic literature, as well as provide useful insights for policymakers. 38 Table 1.1 1985 Presidential Elections DEPARTMENTS Emitted ABSENT APRA IZQUIERDA CODE AC'CION OTI IWIIRS VOTES VOTERS UNIDA POPULAR Amezonas 91.30 8.70 56.400 17.800 5.500 18.300 2.000 Ancasn 93.50 6.50 64.200 2'.b00 4.800 8.800 I .R00 Apunimac 82.10 17.90 50.200 33.800 4.500 7.000 4.600 Arequipa 92.40 7.60 33.200 38.500 17.400 6.600 4.300 Ayacucho 82.90 17.10 49.600 31.500 4.900 6.700 7.300 Cajamarca 87.50 12.50 61.800 17.700 2.800 16.300 1.400 Caliao 94.00 6.00 60.900 18.000 15.100 4.200 1.800 Cuzco 87.10 12.90 44.100 36.800 5.800 6.400 7.400 Iluancavelica 78.30 21.70 41.300 43.100 6.000 5.800 3.800 Huanuco 86.40 13.60 61.400 24.900 8.500 5.000 (.100 Ica 94.90 5.10 55.600 23.300 11.400 7.700 2.000 Junin 90.10 9.90 43.400 39.900 8.900 5.500 3.200 Lalibertad 93.50 6.50 77.800 10.900 6.100 4.000 1.100 Lambayequa 92.30 7.70 64.300 20.000 7.300 7.200 1.100 Lima 88.80 11.20 53.950 22.450 14.650 7.000 2.050 Loreto 87.20 12.80 48.100 21.000 6.700 26.300 1.700 MadredeDios 83.40 16.60 45.800 17.100 4.500 14.500 3.000 Monquegua 92.80 7.20 48.400 30.200 11.200 7.800 4.100 Pasco 86.90 13.10 52.800 30.600 5.400 8.500 2.70(0 Piura 91.50 8.50 55.900 25.400 6.900 10.300 1.500 Puno 91.00 9.00 40.500 30.300 3.400 8.500 17.300 San Martin 85.50 14.50 36.300 14.500 3.700 24.200 1.400 Tanca 93.70 6.30 32.900 33.000 17.900 8.500 9.700 Tumbes 95.60 4.40 49.800 22.900 9.000 16.500 1.600 Ucayall 85.60 14.10 64.400 15.800 7.000 10.400 2.300 Etranjero 74 25.700 52.60 11.40 28.80 4.50 2.70 Total 91.200 8.800 52.10 26.70 11.90 7.20 2.10 39 Table 1.2 FIRST ROUNI) 1990 PRESIDENTIAL ELECTIONS Ballot Distribution Distribution of Casted Votes Valid Votes Emitte(d Absent Valid(s Nulos Viciados FREDEMO Camihio90 Al'RA '11 '1S Olher Blancos Amezonas 62.2 54.7 74.9 11 14.1 42.3 2.7 34.4 15.4 3.3 1.2 Ancasn 72.9 27.1 78 10.4 13.5 20.4 27.3 36.1 9.4 5.1 1.7 Apunimac 72.1 27.9 61.7 18.9 19.5 13.9 25 12.2 39.4 9.3 6.3 Arequipa 83.3 13.7 50.5 4.5 5 36.9 35 15.2 7.5 3.5 2 Ayacucho 88.7 41.3 63.1 16.4 20.5 21.4 25.2 12.5 12.9 4 22.1 Cajamarca 72.6 27.2 75.4 13.2 13.3 27.5 5.6 45.5 10.2 5.1 3.1 Caliao 88.3 13.7 99.6 3.7 3.7 33.7 40.8 17.1 4.3 3 6 Cuzco 76.6 24.4 77.5 10 12.6 15.9 37.6 15.5 19.6 4.3 4.7 Huancavelica 50 40.4 63.7 15.4 17.5 26.5 37.4 5.3 It) 5.4 7.2 Huanuco 40 50.1 70.1 13.1 18.8 33.9 24.4 17.7 10.7 6 1.5 Ica 85.6 14.4 87.4 5.1 7.5 30.6 25.5 27.9 9.4 4.7 3.5 Junin 50.6 49.4 80.1 6.2 11.7 29.5 52.5 8.8 5.2 2.7 1.3 Lalibertad 82 18 65 7.1 7.9 20.4 9.8 40.7 4 9.8 1.4 Lambayequa 84.2 15.8 66.5 6.5 7 30.7 12 43.6 6 6.! 2.1 Lima 82.5 17.6 84.6 7.5 7.9 30.4 33.6 22.7 8.8 4.4 1.3 Loreto 73 27 65.6 8.2 5.4 53 10.5 21.1 ().9 2.6 2.9 Madre de Dios 90.4 28.4 83.4 5.3 11.3 26.7 12.7 25.1 25.5 2.2 4.8 Monquegua 62.6 9.6 67.4 6.1 6.5 20.5 31.6 32.3 8.3 2.7 5.5 Pasco 77.1 37.2 74.5 11.3 14.4 25.4 48.4 11.2 8 4.1 0 Piura 78.4 22.8 63.7 7.7 8.8 30.9 18.2 31.9 12.3 7.8 17.4 Puno 68.8 21.6 71.9 16 11.2 9 30.2 17.5 12.5 4.6 2.2 San Martin 92.9 31.4 73.2 13.2 13.5 43.1 8.4 30.8 14.9 1.6 2 Tanca 90.8 7.1 80.9 4 5.1 23 51.3 16.2 3.3 2.2 0.4 Tumbes 70.4 9.4 88.7 7.4 5.9 40.5 10.1 30. I 8.5 7 5.9 Utcayali 46.3 28.6 82.1 7.9 9.9 44.1 25.1 15.5 6.8 2,6 2.9 Etranjero 46.3 54.7 83.5 5.3 11.1 54.7 15.8 8.8 4.2 3.8 0 Total 77.4 22.8 144.4 17.3 15.0 132.7 29.1 22.6 5.2 4.8 2.0 40 Table 1.3 SECOND ROUND 1990 PRESI)ENTIAL ELECTIONS Ballot Distribution Distribution of Emitted Ballots Dist ribution of Votos DEPARTME NTS Emitted Absent Valids Nulos Viciados Fred emo Cambio 90 Blandos LLosa Fujinmori Amezonas 63.5 36.5 87.1 10.5 2.4 37.7 62.3 Ancasn 74.2 25.8 83.8 13.6 2.6 23.1 76.9 Apuniimac 72.2 27.8 74.4 4.6 21 16.5 83.5 Arequipa 88.1 11.9 93.5 5.4 1 43.9 56.1 Ayacucho 43 57 72.9 21.5 5.6 19 81 Cajamarca 70 30 80.6 16 3.4 32.2 67.8 Caliao 84.8 15.2 95.4 3.9 0.7 41.7 58.3 Cuzco 75.8 24.2 84 12.1 3.9 21.5 7895 Huancavelica 55.7 44.3 77 18 5 16 84 Huanuco 55.5 44.5 81.8 14.6 3.6 35.9 64.1 Ica 86.3 13.7 94.2 4.8 1 35.6 64.4 Junin 61.7 38.3 89.1 9.3 1.6 32.6 67.4 Lalibertad 81 19 89.5 8.7 1.8 24.9 75.1 Lambayequa 83.5 16.5 91.5 7 1.5 36.2 63.8 Lima 82.5 17.5 90.1 8.2 1.7 34.6 65.4 Loreto 75.8 24.2 92.8 6.1 1.1 56 44 Madre de Dios 80.2 19.8 92.8 1.3 5.9 29 71 Monquegua 86.2 13.8 93.8 1.2 5 24.5 75.5 Pasco 72.2 27.8 87.9 5.1 7 38.5 61.5 Piura 85.3 14.7 89.1 9 1.9 13.3 86.7 Puno 78.4 21.6 82.7 14.6 2.7 41.6 58.4 San Martin 71.3 28.7 89.4 0 10.6 58.5 41.5 Tanca 43 57 93.2 1.6 5.2 13.3 96.7 Tumbes 93.6 6.4 94.1 1.2 4.7 41.6 58.4 Ucayall 77.2 22.8 91.3 1.5 7.2 58.5 41.5 Etranijero 34.5 65.5 93.2 5.2 1.6 41.2 58.8 Total 79.9 20.1 90.5 7.8 1 .7 37.5 62.5 41 Table 1.4 1995 Presidential Elections DEPARTMENTS (Cambio 90 APRA VPP UPP Obras A. Fujimori A. Mayst T M. Cabani A. Toledo J. Perez C. R.D R. Belmon 'Otros Blanco Nul1c/V.C Ausent Amezonas 60.5 0.8 5.5 2 20.1 10.9 0.1 1.1 13.9 8.9 32 Ancasn 62.8 0.5 7.8 10.8 15.4 0.8 0.4 1.5 9.7 10.4 31.7 Apunimac 65.9 1.7 3.1 2.5 22.4 1.5 0.5 2.4 13.9 9.4 33.8 Arequipa 57.4 0.6 1.6 3.6 34.4 0.8 0.2 .4 5.8 4.4 33.3 Ayacucho 70.7 0.3 0.7 7.2 16.2 0.8 0.9 2.6 4.3 12.7 45.1 Cajamarca 68.6 2 5 1.5 16.1 1.3 0.1 5.4 12.6 6.9 22.6 Caliao 55.1 0.4 3.8 2.5 30.5 0.7 5.8 1.1 6.5 4.4 30.5 Cuzco 67.5 2.3 2 2.1 22.6 1.2 0.8 1.5 13.6 8.1 39.2 lluancavelica 64 1.1 0.9 10.8 20.4 0.6 0.8 1.4 23 1i 41 Huanuco 77.3 0.5 2.9 5 9.5 1.9 0.3 2.6 10.6 10.1 31.7 Ica 60.6 0.5 5.3 2 24.6 3.8 1.1 2 1 7.2 5 24.8 Junin 77.2 0.3 1 3.4 15.5 0.9 0.6 1.1 7.8 2.7 36.2 Lalibertad 58.1 0.2 14.4 3.9 20.7 0.6 0.2 1.3 6 3.3 25.1 Lambayequa 59.5 0.4 8.4 2.8 22.9 4.5 0.3 1.2 6.5 5.8 24.9 Lima 63.5 0.3 2.6 3 23.1 0.9 5.3 1.3 5.7 3.2 30.1 Loreto 47.9 0.6 5.5 2.6 33.5 8.8 0.5 0.6 7.3 4.8 39 Madre de Dios 64.2 0.9 8.4 2.5 26.7 6.1 0.2 1 9 7.2 47.4 Monquegua 60.6 0.5 2.9 3.3 30.4 0.8 0.3 1.2 6.4 2.5 31.I Pasco 64.8 0.8 1.8 7.1 22.6 1.1 0.5 2.3 10.7 4.5 36.7 Piura 65.7 0.6 5.2 1.5 25.1 1.2 0.2 0.5 9.2 5.2 29.3 Puno 63.1 1.1 2.3 4.6 22 I (.4 4.9 12.8 14.9 33.7 San Martin 65.9 0.3 4.6 1.8 13.7 12.6 0.2 0.9 10.4 6 34 Tanca 67 0.2 1.6 3.8 25.2 0.6 0.6 1 6.6 2.5 25.8 Tumbes 57.2 0.4 7.7 2.1 28.9 3 0.2 0.4 6.4 4.9 27.1 Ucayall 79.5 0.3 2 2 12.6 1.6 (.8 1.2 6.1 3.9 32.8 Etranjero NA NA NA NA NA NA N N N Total 63.7 0.6 4.2 3.4 22.4 1.7 2.4 1.6 8 5.4 31