Non-Median and Condorcet-loser Presidents in Latin America: A Recipe for Instability

Georgetown University From the SelectedWorks of Josep M. Colomer Summer September 1, 2006 Non-Median and Condorcet-loser Presidents in Latin America: A Recipe for Instability Josep M. Colomer Available at: https://works.bepress.com/josep_colomer/120/

APSA Conference paper, 2006 Non-Median and Condorcet-loser Presidents in Latin America: A Recipe for Instability Josep M. Colomer ABSTRACT A favorable condition for good governance is that elected presidents obtain the support of both the median voter and the median legislator. Several electoral rules are evaluated for their results in 111 presidential and 137 congressional elections in 18 Latin American countries during the current democratic periods. The frequency of median voter's or Condorcet-winner presidents appears to be higher under rules with a secondround runoff than under simple plurality rule. The victory of Condorcet-loser or the most rejected candidate is discarded under majority runoff rule. More than half of democratic presidents have not belonged to the median voter s party in the presidential or the congressional elections. Many of them have faced wide popular and political opposition and entered into inter-institutional conflict. Key words: * Plurality rule * Majority runoff * Condorcet-winner * Condorcet-loser * Median voter

INTRODUCTION With separate elections for the congress and the presidency, as in all Latin American countries, broad popular representation and further institutional cooperation can be attained when the president s party holds a decisive place in congress, allowing it to build a politically consistent legislative majority. Specifically, socially efficient presidents are those able to obtain the support of both the median voter and the median legislator, analogously to the typical case of prime ministers in parliamentary regimes. However, any presidential electoral rule always produces a single absolute winner that can be based on a minority of voters' first-preferences not including the median voter s, even with majority runoff rule, which may produce alienation or rejection among broad groups of electors. In addition, the winning president's party may not coincide with the majority party in congress. In the short term, this disparity may distort popular representation, make cooperation between the president and congress more difficult for both legislative decisions and cabinet formation, and induce any of the institutions to try to rule on its own. In order to approach the conditions for good governance under a regime with separate elections, I evaluate several electoral rules for the frequency they produce Condorcet-winner, non-condorcet-winner or Condorcet-loser winners, which can be considered relatively favorable, bad and very bad, respectively, for such an aim. Some definitions and specifications can be appropriate. The median voter is the one having the same number of voters on each side of the single-dimensional political or ideological space, such as the left-right axis. In order to form a consistent majority along the single-dimensional space, the median voter s has to be included. The median voter s position minimizes the sum of spatial distances 1

from all the other positions. It can, thus, be considered to minimize aggregate voters' dissatisfaction and be a socially efficient result. The Condorcet winner is the candidate winning under the Condorcet voting procedure, which requires being preferred to each of all the other candidates in contests by pairs. On a single-dimensional space, the median voter s candidate is the Condorcet winner because he or she will always be preferred by a consistent majority at least on one side of the spectrum against any other candidate at the other side. In contrast, the Condorcet-loser is the candidate that would be defeated by majority by each and all other candidates in pair-wise contests. The Condorcet-loser can be considered the most rejected and least socially efficient candidate. In the past, the relative performance of different electoral rules in giving the victory to the Condorcet-winner or the Condorcet-loser candidates has been examined by means of computer simulations. The results show higher efficiency of majority runoff rule than plurality rule in producing Condorcet-winners, as well as a decrease in efficiency for both rules as the number of candidates increases (see, for example, Nurmi 1987, Merrill 1988). However, these exercises are usually based on low numbers of voters and strong assumptions regarding their distribution of preferences. Recent criticism has noted that the typical assumption of a random society or impartial culture, by which each possible ordering of individual preferences among alternatives is taken as equally probable, is not neutral, but, on the contrary, it s the worst case scenario maximizing the probability of cycles and inefficient results (Regenwetter et al. 2006). This raises doubts about the empirical validity of the formal findings just mentioned, especially for large electorates with asymmetric distributions of preferences and some grouping structure. Broad empirical applications are needed in order to check 2

the validity of formal analyses for evaluating the functioning of real political institutions and advising on their establishment or replacement. One of the crucial difficulties to develop such kind of analysis is the requirement to identify political party s or candidates relative positions in the policy-ideology space. This seems particularly challenging for settings such as most Latin American countries, where parties and political leaders have been commonly labeled as personalistic, clientelistic and populist, that is, as lacking strong policy or ideological allegiances and frequently shifting positions (see, for instance, Roberts and Wibbels 1999, Payne et al. 2002). In this paper I present a new operational method to obtain relative positions of political parties in Latin American countries on the basis of mass survey polls by which political party s positions are estimated from party s voters selfrevealed preferences. This approach may be more relevant than expert estimates or politicians self-location to understand the degrees of voters satisfaction with electoral results and their rejection of electoral-rule dependent winners, which in turn may help explaining subsequent levels of instability of political regimes. Although single-country or single-election opinion polls and other analysis may foster discussion of some of the findings of this paper, it presents the most comprehensive dataset and test of the question to date, encompassing results for more than 111 presidential and 137 congressional elections in 18 Latin American countries during the current democratic periods. 3

MULTIPARTY SYSTEMS PRECED ELECTORAL SYSTEMS In most countries in Latin America, the rules for the election of president have been modified along with the development of multiparty systems. Parliamentary formulas by which the congress elected the executive president, initially inspired by the 1812 Spanish constitution, were used in a number of Latin American countries during the 19th century and, well into the 20th century, in Chile and Uruguay. However, soon after the independence of the new republics, the United States-inspired formula of majority rule in the electoral college with a second round in congress was widespread. When the disadvantages of indirect elections were to be suppressed, a consistent alternative was direct elections by majority rule with a second round in congress. But most Latin American countries, guided in many cases by reelected presidents supported by single party dominance (as well as some authoritarian rulers), eventually completely suppressed the role of congress and established direct presidential elections by plurality rule. Plurality rule can produce popular vote minority winners. Also, the winning president by plurality rule can be the Condorcet-loser. A series of political crises, often leading to violent civil confrontations and coups d état, occurred in Latin American countries as a consequence of political and social opposition to unpopular minority presidents elected by plurality rule (famous cases in the 1960s and 1970s include Haya de la Torre in Peru, Kubitschek and Goulart in Brazil, and Allende in Chile, all of them preceding military coups). Nowadays, simple plurality rule for the election of president is used in only a few countries: Honduras, Mexico, Panama, Paraguay, and Venezuela. In more recent times, multiparty constituent assemblies or multiside political negotiations favored the replacement of plurality rule for presidential elections with 4

more inclusive rules, namely majority rule with a second-round runoff between the two most voted-for candidates. This procedure is presently used in Brazil, Chile, Colombia, the Dominican Republic, El Salvador, Guatemala, Peru, and Uruguay. A variant of second-round procedures is the rule requiring some qualifiedplurality threshold short of a majority, such as 40 or 45 percent of votes, to win the election, while a second round-runoff is held only if no candidate obtains such a support. This type of rule, which is used now in Bolivia, Costa Rica, Ecuador, and Nicaragua, tends to be adopted in situations of moderate multipartism, in which the incumbent party calculates or simply feels that it can obtain larger support than any of the opposition groups but is not sure of its dominance. In relatively balanced negotiations, this proposal can also be accepted by the opposition if it considers itself unlikely to obtain an absolute majority of votes but able to prevent the incumbent party s candidate from obtaining such a threshold and force a second round. During the present democratic periods, presidential elections in Latin America have been mostly run with multiparty systems. For each country, the effective number of parties (ENP) in votes (that is, the number of parties weighted by their size) has been calculated for the election immediately previous to the electoral system change (with the exception of those electoral system changes following a long dictatorial period which destroyed the previous party system). The party systems previous to the adoption of second-round presidential electoral systems in ten countries are measured in Table 1. The average effective number of presidential candidates was 3.9. This demonstrates that the replacement of electoral college or plurality rule presidential elections with absolute-majority or qualified-plurality rules with a second round-runoff has followed, rather than preceded, the development of multiple parties. (For complete data and analysis see Colomer 2004a; for discussion, Negretto 2006a). 5

In parallel, the rules for the election of the low or single chamber of congress have typically moved from old procedures including lots and indirect elections, to relatively restrictive formulas based on small districts and plurality rule, to more permissive devices permitting the inclusion of some minorities, up to the present dominance of proportional representation. Typically, the establishment or the maintenance of plurality rule was the result of decisions made within dominant singleparty or balanced two-party systems, that is, in situations able to feed established leaders expectations of maintaining control of political competition. In contrast, proportional representation systems have been typically adopted by pluralistic constituent conventions or by political agreements among multiple political parties which expect to compete in more uncertain elections or to share power in multiparty congresses and cabinets. A panorama of the party systems prior to the adoption of proportional representation electoral systems for congress elections is offered in Table 2. As can be observed, in almost every country a pluralistic multiparty configuration already existed under the constrictions but against the grain of plurality electoral rules, at the time of introducing proportional representation. The average effective number of presidential candidates was 4.9. It is not thus the electoral system of proportional representation that created multipartism, but rather multipartism that led political leaders to introduce a new electoral system reducing the risks of their becoming absolute losers and permitting wider pluralistic representation. During the present democratic periods, the election-average effective number of parties in congressional elections is 3.9, according to calculations for 18 countries (excluding Cuba and Haiti), as presented in Table 3. For presidential elections the value is a little lower, 3.1, due to the incentives to concentrate candidacies and votes in a 6

small number of contenders provided by majoritarian rules producing a single absolute winner. But only a few countries can be considered as having two-party systems during their most recent democratic periods (operationalized as having an average ENP lower than 3): Honduras, Nicaragua, and Paraguay. Other traditional two-party systems have experienced increases in pluralism in recent elections. My emphasis on the causality line from party systems to electoral rules, instead of the opposite traditional assumption, projects high difficulties for any attempt to replace permissive rules associated to multiparty systems, such as proportional representation for congress and second-round rules for the presidency, with more restrictive rules such as simple plurality. The hypothetical reintroduction of restrictive electoral rules in the intention to reduce the number of parties may become unfeasible due precisely to the resistance offered by multiple parties entrenched in the existing institutional framework. If multipartism is difficult to combine with presidentialism, as has been postulated, it may be the case that it s not multipartism but presidentialism that should be reformed. But this would be another subject. Let s focus now on the performance of presidential electoral rules. NON-MEDIAN PRESIDENTS In comparison with simple plurality rule, majority rule with a second round runoff can be less bad for selecting the median voter s or Condorcet-winner candidate and is superior at preventing the selection of the Condorcet-loser or most rejected candidate. Let us address these two aspects in turn. 7

On the one hand, to the extent that simple plurality rule is associated to twoparty systems while majority-runoff and qualified-plurality rules with second-round are associated to multiparty systems, the former should produce median voter s candidates with higher probability. In any election with only two candidates, the winner by plurality is the winner by majority and the Condorcet-winner candidate. On the other hand, with more than two parties the probability of the median voter's candidate winning is double under majority rule than under plurality rule. This is due to the fact that if the median voter's candidate is present at the second round, he or she can be expected to win by a majority against any other rival. In a two-candidate contest, like the typical second round, the median voter's candidate can attract all the voters on 'his or her side', which are half of the total, plus some of those in between or undecided about the two candidates. Thus, the probability of the median voter's candidate winning is equal to the probability of being one of the two most voted-for candidates in the first round, which, a priori, that is, independently on the number of candidates, is double than the probability of being the single most voted-for candidate in the first round. The two elements associated to each of the rules likely different numbers of candidates and different likelihoods of a median voter's winner when more than two candidates run somehow counteract each other, but they may have different degrees of influence on the electoral result. Empirically, the frequency of majority and median voter's winners seems to be a little higher under rules with a second round-runoff. This may be the consequence of a relatively high number of multiparty elections under plurality rule, that is, of the growing development of multiparty systems identified above. 8

In order to obtain these results, I examined all 111 presidential elections in 18 Latin American countries during the present democratic periods. A survey of the electoral rules during these periods is presented in Table 4. In order to know whether the winning candidate has the support of the median voter, only relative positions from the other candidates to the winner need be identified. Let s assume that presidential candidates are consistent with political party s positions on the left-right ideological dimension. This may be a controversial assumption, as suggested, but it should be revised on the basis of more accurate data regarding political party s positions, not on the basis of prejudice. In the present exercise, for each political party its position on the left-right dimension is measured by the average self-placement of voters who declare their preference for that party. This, as mentioned, can be relevant to identify voter s satisfaction and reaction to electoral results in favor of a political party. Average ideological locations on a simplified five-position scale (left, center-left, center, centerright, and right) for a total of 74 political parties in 18 Latin American countries are presented in Table 5 (see more details in Colomer and Escatel 2005). I assume that the electoral winner contains the median voter's support if the other candidates on each side of the winner gather less than 50 per cent of votes. Even uncertain or shifting relative positions among parties may permit such a calculation. In the particular case that the winner has obtained an absolute majority of votes, he surely has the median voter's support. Under the assumption of a single-dimensional space, the median voter s candidate is the Condorcet winner. I identify the Condorcet-loser as an extreme party facing a majority of voters supporting parties which are located at distant positions from it and relatively close among themselves. The results of the analysis, as contained in Table 6, can be summarized the following way. In total, only 70 percent of presidential elections in 18 Latin American 9

countries during the present democratic periods have given the victory to the median voter s candidate (in 78 out of 111 elections). In only a few countries all the presidents have been supported by the median voter: Argentina, Chile and Honduras. Median voter s candidates have won the presidency in about two-thirds of elections by plurality rule and three-fourths of elections by second-round rules. More specifically, in plurality rule presidential elections, the winner has been the median voter s candidate in 65 percent of the cases (in 25 out of 39 elections), while with majority or qualified plurality rules with a second round the median voter s candidate has won in 73 percent of the cases (in 51 of 70 elections). Qualified-plurality rules should produce lower proportions of median voter s winners than majority rule (and less as the lower the threshold required). In our survey, however, the record of qualified plurality-rule appears to be better than that of majority rule, 78 percent versus 70 percent (in 18 of 23 elections and 33 out of 47, respectively), due to the heavy weight of a single country with a very long period of stable democracy and durable bipartism, Costa Rica (with 14 elections by 40 percent rule). The most important advantage of majority runoff rule regarding plurality rule is that it can prevent the victory of the Condorcet-loser or the 'worst' possible candidate. The winner in the second round cannot be the least preferred candidate for an electoral majority because he or she will be defeated, if not earlier, at least at the second round. The winner by majority runoff will never be the Condorcet-loser --that is, the candidate that would be defeated in pair-wise contests by each of all other candidates. In the worst of the cases, the elected president at the second round will be considered a lesser evil by many of the voters. In contrast, the winner in a contest by plurality rule may be the Condorcet-loser, as may have been the case during present democratic periods of, among others, 10

Balaguer in the Dominican Republic and Caldera in Venezuela, just to mention a few with simple plurality rule, as well as Gutiérrez in Ecuador and Ortega in Nicaragua with qualified plurality rules. (More details in Table 6). A comparison with electoral results in the United States by the college system mostly based on plurality rule can be in point. For the recent period 1948-2004 it can be estimated that 80 percent of the United States presidential elections have given the victory to the median voter s or Condorcet-winner candidate (in 12 out of 15 elections, the exceptions being Kennedy in 1960, by a very tiny margin as discussed in Colomer 2001: 104-6, Clinton in 1992 and Bush in 2000). The US relatively good record corresponds, of course, to the prevalence of two-party competition in most elections. However, in a previous 19 th century period with more volatile party systems (which might be comparable to more recent periods in Latin America), electoral results were dramatically inefficient. In my estimate, only 38 percent of the US presidents elected during the period 1844-1892 were median voter s or Condorcet-winner candidates (in only 5 of 13 elections, due to the role of third parties like Liberty or abolitionists, Free Soil, Constitutional Union, Greenback and Populist). At least one president was the Condorcet-loser candidate (the republican Rutherford Hayes, who was elected by the college against a democratic rival having obtained an absolute majority of popular votes). This convulse era culminated in the Civil War. Although third and additional parties have regularly ran in further periods too, their influence on the electoral result has diminished. For the whole period 1828-2004 it can be estimated that 73 percent of the US presidents have been the median voter s or Condorcet-winner candidate (in 33 out of 45 elections, including 28 who obtained an absolute majority of popular votes) a proportion not very different from that in Latin America in the most recent period. 11

DIVIDED GOVERNMENTS Even in some presidential elections in which the median voter s candidate has been elected, the president s party may not be the median voter s party in congressional elections. This disparity in results may appear with multiple parties even in concurrent elections due to the incentives provided by majoritarian presidential rules to concentrate support on a lower number of candidates. Specifically, some median voter s presidential candidates can be located at relatively high distances from the median voter s preference, which may be closer to some congressional party not having run for the presidency. In total, in less than half of the elections in 18 Latin American countries during the present democratic periods has the president s party been the median voter s party in both the presidential and congressional elections (in 48 percent of cases, or in 65 out of 137 elections). In no country all the presidents have enjoyed the support of both the median voter and the median legislator. (More details in Table 6). The elected presidents in Latin America during the present democratic periods can be classified the following way. First, median voter's presidents include those candidates who obtained a majority of votes in the first round, like the populist Evo Morales in Bolivia, christian-democrats Patricio Aylwin and Eduardo Frei in Chile, the conservative Alvaro Uribe in Colombia, Leonel Fernández in the Dominican Republic, or the leftist Tabaré Vázquez in Uruguay. Second, presidents with the support of the median voter are also those who did not obtain an absolute majority support in popular votes in the first round, but were located on an advantageous position around the center of the left-right political 12

spectrum able to attract a majority on any of the two sides against the other or at least prevent the formation of a alternative compact majority, such as justicialist Carlos Menem in Argentina, nationalists Víctor Paz Estensoro and Gonzalo Sánchez de Lozada (in his first term) in Bolivia, liberals Virgilio Barco, César Gaviria and Ernesto Samper in Colombia, populist Alan García in Peru (in his first term), and colorados Julio M. Sanguinetti and Jorge L. Batlle in Uruguay. Third, other presidential candidates obtained the support of the median voter in the presidential race thanks to the support of a multiparty coalition formed in the first round, but they did not belong to the median voter's party in congressional elections. Cases include the social-democrat Fernando Henrique Cardoso and leftist Lula da Silva (in his second term) in Brazil, both having enjoyed the support of the median voter s Democratic Movement (PMDB) which didn t present its own candidate for president, and the socialist Ricardo Lagos and Michelle Bachelet in Chile, who also enjoyed the electoral support of the median voter's christian-democrats within the Concertación broad electoral coalition. Finally, a number of elected presidents have not been the median voter s candidate in the first round of the presidential election nor the median voter s party in the congressional election. Generally holding extreme positions on any of the two sides of the spectrum, they include leftist Hernán Siles-Zuazo and Jaime Paz Zamora, as well as former rightist dictator Hugo Bánzer in Bolivia, eccentric Fernando Collor de Mello and leftist Lula Da Silva (in his first term) in Brazil, former rightist dictator Joaquín Balaguer in the Dominican Republic, several Ecuadorian presidents including leftpopulist Lucio Gutiérrez (also former golpista ) and Rafael Correa in Ecuador, conservatives Vicente Fox and Felipe Calderón in Mexico, leftist Daniel Ortega in Nicaragua (in his most recent election), authoritarian Alberto Fujimori in Peru (in his 13

last election), and conservative Rafael Caldera in Venezuela. Once in office, some of them tried to build broader post-electoral coalitions in congress, which included presidential rivals' parties. But most of them faced wide popular and political opposition and entered into interinstitutional conflict. (For surveys of institutional conflicts, see Colomer 2004b, Valenzuela 2004, Negretto 2006b). CONCLUSION Several electoral rules have been evaluated on the basis of the results of 111 presidential and 137 congressional elections in 18 Latin American countries during the current democratic periods. The empirical analysis confirms in qualitative terms some findings of previous formal exercises based on computer simulations, but it also shows that in real electorates with asymmetric and grouping preferences the degrees of social efficiency of different electoral rules tends to be higher than in more inefficiency-prone formal assumptions. The discussion sketched in the previous pages also permits an assessment of the consequences of institutional changes introduced in many Latin American countries in recent times. Specifically, the applied analysis presented in this paper gives support to the following statements: Plurality rule in presidential elections tends to produce non-median voter s or non-condorcet winner candidates with higher frequency than majority or qualified-plurality rules, although the latter do not guarantee Condorcet-winners either. In the analysis here presented, plurality rule gave the victory to median 14

voter s or Condorcet-winner candidates in about two-thirds of the cases, while second-round rules did it in about three fourths. Plurality rule can produce broadly rejected, Condorcet-loser presidents, while this result is discarded with majority runoff rule. The replacement of plurality rule with second-round rules for presidential elections in most countries of Latin America during the last few decades has increased the proportions of median voter s and Condorcet-winner presidents and decreased that of Condorcet-loser ones in comparison with past periods in which plurality rule systems prevailed. This has reduced popular rejection of elected presidents and somehow might favor more acceptance and durability of democratic regimes. However, with separate elections for presidency and congress, even if the president is elected with the support of the median voter, the president s party may not be the median voter s party in congress. The relative high frequency of presidents not enjoying the support of the median voter or the median legislator can contribute to explain that, in spite of the relative high duration of present democracies, there are high levels of inter-institutional conflict and political instability in the region. 15

Table 1. Presidential electoral system change ==================================================== Country Law year Previous election. Year Type ENP -------------------------------------------------------------------------------------- Argentina 1972 1965 Assembly 4.9 Argentina 1994 1994 Constituent Assembly 3.0 Brazil 1986 1982 Assembly 2.7 Colombia 1991 1990 Constituent Assembly 2.2 Costa Rica 1936 1932 President 2.8 Dominican R. 1995 1994 Assembly 2.8 Ecuador 1978 1978 President 4.8 Guatemala 1985 1984 Constituent Assembly 7.7 Peru 1979 1978 Constituent Assembly 4.8 Uruguay 1996 1994 Assembly 3.3 -------------------------------------------------------------------------------------- Effective Number of Parties (ENP) average 3.9 -------------------------------------------------------------------------------------- Note: Electoral system changes are from college or plurality rule to majority or qualified-plurality rules with second round runoff. ==================================================== 16

Table 2. Lower or single chamber electoral system change ========================================== Country Law year Previous election. Year Type ENP ---------------------------------------------------------------------- Argentina 1963 1962 Assembly 6.7 Bolivia 1952 1951 President 3.3 Brazil 1945 1945 Assembly 3.7 Chile 1925 1921 Assembly 5.6 Colombia 1931 1930 President 2.0 Costa Rica 1953 1948 Assembly 2.9 Cuba 1940 1940 Constituent 7.5 Ecuador 1978 1978 President 4.8 Guatemala 1985 1984 Constituent 7.7 Peru 1933 1931 Assembly 2.6 --------------------------------------------------------------------- Effective Number of Parties (ENP) average 4.9 --------------------------------------------------------------------- Note: Electoral system changes are from plurality rule to mixed or proportional representation systems. ENP = 1/ vi 2, where is the sum for all parties and vi is the percentage of votes of each party i. ========================================== 17

Table 3. Effective number of parties in votes ==================================================== Country Democratic Congress Presidency. period No. elections ENP No. elections ENP --------------------------------------------------------------------------------------- Argentina 1983-2003 10 3.1 5 2.7 Bolivia 1979-2002 6 5.3 6 5.1 Brazil 1982-2002 6 6.6 4 3.4 Chile 1989-2001 4 5.5 3 2.3 Colombia 1978-2002 7 2.9 7 2.6 Costa Rica 1953-2002 13 2.9 13 2.3 Dominican R. 1978-2002 7 2.6 7 2.6 Ecuador 1978-2002 8 7.5 7 5.5 El Salvador 1984-2003 7 3.5 4 2.7 Guatemala 1985-2003 5 3.6 5 4.3 Honduras 1981-2001 6 2.3 6 2.2 Mexico 1997-2003 3 3.3 2 2.7 Nicaragua 1990-2001 3 2.5 3 2.3 Panama 1994-1999 2 3.9 2 3.3 Paraguay 1993-2003 3 2.3 3 2.5 Peru 1980-2001 6 4.3 6 3.0 Uruguay 1984-1999 4 3.2 4 3.2 Venezuela 1958-2000 10 4.2 10 2.8 --------------------------------------------------------------------------------------- Effective Number of Parties (ENP) election average 3.9 election average 3.1 country average 3.8 country average 3.1 ==================================================== 18

Table 4. Electoral systems in Latin America ============================================================= 4.1. ASSEMBLY (lower or single chamber) ============================================================= Country Law year Term Seats Districts Magnitude Ballot Rule ============================================================= Argentina 1963 4 243-257 24 3-35 Closed list Proportional Bolivia 1952 5 102-130 8-9 7-28 Closed list Proportional 1996 5 130 9 5-31 Single Proportional - Brazil 1950 4 326-513 25-27 1-75 Open list Proportional - Chile 1980 4 110-120 55-60 2 Open list Proportional Colombia 1958 2 148-204 26 7 (average) Closed list 2-party proport. 1970 4 160-210 26-33 2-10 Closed list Proportional Costa Rica 1953 4 45-57 7 4-21 Closed list Proportional Dominican R.1966 4 91-150 30-33 2-31 Closed list Proportional Ecuador 1978 2 72-121 21-23 1-20 2 closed lists Proportional 1998 Open ballot Plurality 2000 Open list Proportional El Salvador 1984 3 60-84 14 3-20 Closed list Proportional Guatemala 1984 5 80-116 22 2-20 2 closed lists Proportional Honduras 1966 4 82-134 18 1-16 Closed list Proportional Mexico 1986 3 500 300/1 1/200 Single Mixed:Plurality/PR - Nicaragua 1984 6 90 9-18 1-25 Closed list Proportional Panama 1983 5 67-71 40 1-6 Single Mixed:PR/Plurality Paraguay 1993 5 80 18 1-14 Closed list Proportional Peru 1979 5 120-180 24 1-11 Open list Proportional 1995 120 1 120 Uruguay 1984 5 99 1 99 Open list Proportional Venezuela 1946 5 110-201 23 1-37 Closed list Proportional 1989 5 204-207 23-24 3-25 Double vote Proportional 1999 5 165 24 3-14 Closed list Proportional 19

4.2. PRESIDENCY ============================================================= Country Law year Term Rule ============================================================= Argentina 1983 6 College majority / 2nd round runoff by congress 1994 4+4 Qualified plurality (45% or 40%+10) / 2nd round runoff -------------------------------------------------------------------------------------------------------- Bolivia 1967 4 Majority / 2nd round runoff by congress -------------------------------------------------------------------------------------------------------- Brazil 1986 5 Majority / 2nd round runoff 1993 4 1997 4+4 -------------------------------------------------------------------------------------------------------- Chile 1989 5 Majority / 2nd round runoff Colombia 1910 4 Plurality 1991 4 Majority / 2nd round runoff Costa Rica 1936 4 Qualified plurality (40%) / 2nd round runoff Dominican R. 1962 4 Plurality 1995 4 Majority / 2nd round runoff Ecuador 1978 5 Majority / 2nd round runoff 1983 4 1998 4+4 Qualified plurality (40%+10) / 2nd round runoff El Salvador 1983 5 Majority / 2nd round runoff Guatemala 1984 5 Majority / 2nd round runoff 1994 4 Honduras 1966 4 Plurality Mexico 1917 6 Plurality Nicaragua 1984 6+6 Plurality 1995 6 Qualified plurality (45%) / 2nd round runoff 1999 6 Qualified plurality (40%) / 2nd round runoff Panama 1916 5 Plurality - Paraguay 1989 5 Plurality Peru 1978 5 Majority / 2nd round runoff 1993 5+5 2001 5 Uruguay 1899 5 Plurality 1996 5 Majority / 2nd round runoff 20

Venezuela 1946 5 Plurality 1999 6+6 Note: Only major changes during current democratic periods are included. Seats, number of districts, and magnitudes are given in ranks within the period. Source: Author's elaboration; for more details, see Colomer (2004a). ============================================================ 21

Table 5. Party systems in Latin America ============================================================== LEFT CENTER-LEFT CENTER CENTER-RIGHT RIGHT Argentina IU ARI-FrePaSo -- PJ, UCR -- - Bolivia MAS UDP, MIR NFR, MNR UCS ADN Brazil -- PT PSB PMDB, PSDB, PFL -- - Chile PC PS, PPD PDC RN, UDI -- Colombia -- -- -- PC, PL -- - Costa Rica -- -- -- PML, PAC, PLN, PUSC -- - Dominican R. -- PLD PRD -- PRSC - Ecuador -- PSP, ID, DP -- PSC, PRE, PUR -- - El Salvador FMLN -- -- PCN, PDC AReNa - Guatemala -- -- PDC, MAS PAN, FRG -- - Honduras -- -- -- PL, PN -- - Mexico -- PRD PRI, PAN -- -- Nicaragua -- FSLN PCN PLC -- - Panama -- CD, PRD APU, AD PDC,MoLiReNa,Parn -- - Paraguay -- -- PLRA, ANR -- -- Peru -- -- SP, PAP, UN, PP AP -- Uruguay -- EP/FA -- PN, PC -- - Venezuela -- -- MVR PJ, AD CoPEI - Note: Relative party positions are based on each party s voters average self-placement on a 1-10 left-right scale, and operationalized as: Left: 1-3; Center-left: 3-5; Center: 5-6; Center-right: 6-8; Right: 8-10. Source: Author's own elaboration with data from Latinobarometer, 2003. Relative positions are highly consistent with experts or politicians subjective estimates of fewer parties as collected in Huber and Inglehart (1995), Coppedge (1997), and Alcántara and Freidenberg (2001). I have included here relative locations for all parties which gained the presidency during current democratic periods. More parties and more details in Colomer and Escatel (2005). 22

Party names: AD: Alianza Democrática (Panama). AD: Acción Democrática (Venezuela). ADN: Acción Democrática Nacionalista. ANR: Asociación Nacional Republicana ( Colorados ). AP: Acción Popular. APRA: Alianza Popular Revolucionaria Americana. APU: Alianza Pueblo Unido. AReNa: Alianza Republicana Nacionalista. ARI-FrePaSo: Alianza Alternativa por una República de Iguales-Frente País Solidario. CD: Cambio Democrático. COPEI: Partido Social Cristiano. DP: Democracia Popular. EP/FA: Encuentro Progresista-Frente Amplio. FMLN: Frente Farabundo Martí para la Liberación Nacional. FRG: Frente Republicano Guatemalteco. FSLN: Frente Sandinista de Liberación Nacional. ID: Izquierda Democrática. IU: Izquierda Unida. MAS: Movimiento al Socialismo (Bolivia). MAS: Movimiento de Acción Social (Guatemala). MIR: Movimiento de la Izquierda Revolucionaria. MNR: Movimiento Nacionalista Revolucionario. MoLiReNa: Partido Movimiento Liberal Republicano Nacionalista. MVR: Movimiento V República. NFR: Nueva Fuerza Republicana. PAC: Partido Acción Ciudadana. PAN: Partido de Avanzada Nacional (Guatemala). PAN: Partido Acción Nacional (México). PAP: Partido Aprista Peruano (see APRA). Parn: Partido Arnulfista. PC: Partido Colorado (Uruguay). PC: Partido Comunista (Chile). PC: Partido Conservador (Colombia). PCN: Partido de Conciliación Nacional (El Salvador). PCN: Partido Conservador de Nicaragua (Nicaragua). PDC: Partido Demócrata Cristiano (Chile). PDC: Partido Demócrata Cristiano (Guatemala). PFL: Partido da Frente Liberal. PJ: Partido Justicialista (Argentina). PJ: Primero Justicia (Venezuela). PL: Partido Liberal. PLC: Partido Liberal Constitucionalista. PLD: Partido de la Liberación Dominicana. PLN: Partido Liberación Nacional. PLRA: Partido Liberal Radical Auténtico. PMDB: Partido do Movimiento Democrático Brasileiro. PML: Partido Movimiento Libertario. PN: Partido Nacional (Blancos). PP: Perú Posible. 23

PPD: Partido por la Democracia. PRD: Partido de la Revolución Democrática (Mexico). PRD: Partido Revolucionario Democrático (Panama). PRD: Partido Revolucionario Dominicano. PRE: Partido Roldosista Ecuatoriano. PRI: Partido Revolucionario Institucional. PRSC: Partido Reformista Social Cristiano. PS: Partido Socialista. PSB: Partido Socialista Brasileiro. PSC: Partido Social Cristiano. PSDB: Partido da Social Democracia Brasileira. PSP: Partido Sociedad Patriótica 21 de enero. PT: Partido dos Trabalhadores. PUR: Partido Unión Republicana. PUSC: Partido Unidad Social Cristiana. RN: Renovación Nacional. SP: Somos Perú. UCR: Unión Cívica Radical. UCS: Unidad Cívica Solidaridad. UDI: Unión Demócrata Independiente. UDP: Unión Democrática Popular. UN: Unidad Nacional. 24

Table 6. Democratic elections in Latin America ============================================================= Country / Electoral President President s Median voter s in election?. Election year rule party Presidential Congressional ============================================================= Argentina 1983 College Alfonsín UCR Yes Yes 1985 No 1987 No 1989 College Menem PJ Yes Yes 1991 Yes 1993 Yes 1995 QP runoff Menem PJ Yes Yes 1997 No 1999 QP runoff De la Rúa UCR Yes No 2001 No 2003 QP runoff Kirchner PJ Yes Yes 2005 Yes - Bolivia 1979 Majority runoff Siles UDP No No 1985 Majority runoff Paz E. MNR Yes Yes 1989 Majority runoff Paz Zamora MIR No No 1993 Majority runoff Sánchez MNR Yes Yes 1997 Majority runoff Bánzer ADN No No 2002 Majority runoff Sánchez MNR Yes Yes 2006 Majority runoff Morales MAS Yes Yes Brazil 1990 Majority runoff Collor PRN No No 1994 Majority runoff Cardoso PSDB No Yes 1998 Majority runoff Cardoso PSDB Yes No 2002 Majority runoff Silva PT No No 2006 Majority runoff Silva PT Yes No Chile 1989 Majority runoff Aylwin PDC Yes Yes 1993 Majority runoff Frei PDC Yes Yes 1997 Yes 1999 Majority runoff Lagos PS Yes No 2001 No 2006 Majority runoff Bachelet PS Yes No Colombia 1978 Plurality Turbay PL Yes Yes 1982 Plurality Betancur PC No No 1986 Plurality Barco PL Yes Yes 1990 Plurality Gaviria PL Yes Yes 1991 Yes 1994 Majority runoff Samper PL Yes Yes 25

1998 Majority runoff Pastrana PC No No 2002 Majority runoff Uribe PC Yes No 2006 Majority runoff Uribe PC Yes No Costa Rica 1953 QP runoff Figueres PLN Yes Yes 1958 QP runoff Echandi PUN Yes No 1962 QP runoff Orlich PLN Yes Yes 1966 QP runoff Trejos UN Yes No 1970 QP runoff Figueres PLN Yes Yes 1974 QP runoff Oduber PLN No No 1978 QP runoff Carazo PU Yes Yes 1982 QP runoff Moya PLN Yes Yes 1986 QP runoff Arias PLN Yes Yes 1990 QP runoff Calderón PUSC Yes Yes 1994 QP runoff Figueres PLN Yes Yes 1998 QP runoff Rodríguez PUSC Yes Yes 2002 QP runoff Pacheco PUSC No No 2006 QP runoff Arias PLN Yes Yes Dominican Republic 1978 Plurality Guzmán PRD Yes Yes 1982 Plurality Blanco PRD No Yes 1986 Plurality Balaguer PRSC No No 1990 Plurality Balaguer PRSC No No 1994 Plurality Balaguer PRSC No No 1996 Majority runoff Fernández PLD Yes Yes 1998 No 2000 Majority runoff Mejía PRD Yes Yes 2002 No 2006 Majority runoff Fernández PLD Yes Yes Ecuador 1979 Majority runoff Roldós CFP Yes Yes 1984 Majority runoff Febres PSC No No 1986 No 1988 Majority runoff Borja ID Yes Yes 1990 No 1992 Majority runoff Durán PUR No No 1994 No 1996 Majority runoff Bucaram PRE Yes Yes 1998 QP runoff Mahuad DP Yes Yes 2000 No 2002 QP runoff Gutiérrez PSP/MUPPNP No No 2006 QP runoff Correa PAIS/PS-FA No No -------------------------------------------------------------------------------------------------------- El Salvador 1984 Majority runoff Duarte PDC No 1985 Yes 1988 No 26

1989 Majority runoff Cristiani AReNa Yes No 1991 No 1994 Majority runoff Calderón AReNa No No 1997 No 1999 Majority runoff Flores AReNa Yes No 2000 No 2003 No 2004 Majority runoff Saca AReNa Yes No 2006 No Guatemala 1985 Majority runoff Cerezo DCG No Yes 1990 Majority runoff Serrano MAS Yes No 1994 No 1995 Majority runoff Arzú PAN Yes Yes 1999 Majority runoff Portillo FRG No Yes 2003 Majority runoff Berger MR-PP-PSN Yes Yes Honduras 1981 Plurality Suazo PLH Yes Yes 1985 Plurality Azcona PLH Yes Yes 1989 Plurality Callejas PNH Yes Yes 1993 Plurality Reina PLH Yes Yes 1997 Plurality Flores PLH Yes Yes 2001 Plurality Maduro PNH Yes No 2005 Plurality Zelaya PLH Yes Yes Mexico 1994 Plurality Zedillo PRI Yes Yes 1997 Yes 2000 Plurality Fox PAN No No 2003 No No 2006 Plurality Calderón PAN No No Nicaragua 1990 Plurality Barrios UNO Yes Yes 1996 QP runoff Alemán AL Yes Yes 2001 QP runoff Bolaños PLC Yes Yes 2006 QP runoff Ortega FSLN No No Panama 1994 Plurality Pérez Bal. APU Yes No 1999 Plurality Moscoso AD No No 2004 Plurality Torrijos PRD Yes Yes Paraguay 1993 Plurality Wasmosy ANR No No 1998 Plurality Cubas ANR Yes Yes 2003 Plurality Duarte ANR Yes Yes 27

Peru 1980 Majority runoff Belaúnde AP Yes Yes 1985 Majority runoff García PAP Yes Yes 1990 Majority runoff Fujimori C90 Yes No 1995 Majority runoff Fujimori C90 Yes Yes 2000 Majority runoff Fujimori P2000 No No 2001 Majority runoff Toledo PP Yes Yes 2006 Majority runoff García PAP Yes Yes Uruguay 1984 Plurality Sanguinetti PC Yes Yes 1989 Plurality Lacalle PN No No 1994 Plurality Sanguinetti PC Yes Yes 1999 Majority runoff Batlle PC Yes Yes 2004 Majority runoff Vázquez NE/FA Yes Yes Venezuela 1958 Plurality Betancur AD No Yes 1963 Plurality Leoni AD No Yes 1968 Plurality Caldera COPEI No No 1973 Plurality Pérez AD Yes Yes 1978 Plurality Herrera COPEI No No 1983 Plurality Lusinchi AD Yes Yes 1988 Plurality Pérez AD Yes Yes 1993 Plurality Caldera CN Yes No 1998 Plurality Chávez MVR Yes No 2000 Plurality Chávez MVR Yes No Notes: The Table includes only current democratic periods, which total 465 countryyears. Source: Author s elaboration, with party positions and names as given in Table 5 and electoral data from Nohlen (2005), Payne et al. (2002) and List of election results in Wikipedia. ============================================================ 28

REFERENCES Alcántara, Manuel, and Flavia Freidenberg eds. 2001. Partidos políticos de América Latina, 3 vols. Salamanca: Universidad de Salamanca. Colomer, Josep M. 2001. Political Institutions: Democracy and Social Choice. New York-Oxford: Oxford University Press. -- ed. 2004a. Handbook of Electoral System Choice. London-New York: Palgrave- Macmillan. -- 2004b. 'Taming the Tiger: Voting Rights and Political Instability in Latin America', Latin America Politics and Society, 46, 2: 29-58. -- 2005. Policy Making in Divided Government: A Pivotal-Actors Model with Party Discipline, Public Choice, 125, 3-4: 247-269. Colomer, Josep M., and Luis E. Escatel. 2005. La dimensión izquierda-derecha en América Latina,Desarrollo Económico, 44, 177: 123-136. Colomer, Josep M. and Gabriel L. Negretto. 2005. 'Can Presidentialism Work Like Parliamentarism?', Government and Opposition, 40, 1: 60-89.. Coppedge, Michael. 1997. 'A Classification of Latin American Political Parties'. University of Notre Dame: The Helen Kellogg Institute for International Studies, Working Paper #244. Huber, John, and Ronald Inglehart. 1995. 'Expert Interpretations of Party Space and Party Locations in 42 Societies', Party Politics, 1, 1: 73-111. Negretto, Gabriel L. 2006a. Choosing How to Choose Presidents: Parties, Military Rulers, and Presidential Elections in Latin America, Journal of Politics, 68, 2: 421-433. -- 2006b. 'Minority Presidents and Democratic Performance in Latin America', Latin America Politics and Society, 48, 3: 63-92. Nohlen, Dieter ed. 2005. Elections in the Americas. A Data Handbook, 2 vols. New York-Oxford: Oxford University Press. Nurmi, Hannu 1987. Comparing Voting Systems. Dordrecht: Kluwer. Payne, J. Mark, Daniel Zovatto G., Fernando Carrillo Flórez, and Andrés Allamand Zavala. 2002. Democracies in Development: Politics and Reform in Latin America. Washington, DC: Inter-American Development Bank. Regenwetter, Michael., Bernard Grofman, A. A. J. Marley and Ilia Tsetlin. 2006. Behavioral Social Choice: Probabilistic Models, Statistical Inferences and Applications. Cambridge-New York: Cambridge University Press. Roberts, Kenneth M., and Erik Wibbels. 1999. 'Party Systems and Electoral Volatility in Latin America: A Test of Economic, Institutional, and Structural Explanations', American Political Science Review, 93, 3: 575-591. Valenzuela, Arturo. 2004. Latin American Presidencies Interrupted, Journal of Democracy, 15, 4: 5-19. 29