Three s Company: Old and New Cleavages in Chile s Party System

Three s Company: Old and New Cleavages in Chile s Party System Patricio Navia Center for Latin American and Caribbean Studies New York University Pdn200@nyu.edu March 25, 2003 9229 words Prepared for delivery at the 2003 meeting of the Latin American Studies Association, Dallas, Texas, March 27-29, 2003. The political party system in existence in Chile before 1973 is often described as reflecting a three-way division (three-thirds), with a strong left, a pragmatic center and a conservative, but democratic, right. The polarization of the extremes and the radicalization of the center are cited as responsible for the democratic breakdown and the military coup. After a 17 year dictatorship, the new democratic period witnessed the emergence of a new cleavage that allowed for the formation of two coalitions in the party system. The old center and most of the left formed the new Concertación alliance and the old right gave way to a new conservative coalition associated with the Pinochet legacy. After 12 years of democratic life and continuous electoral success by the Concertación, there is evidence that the two-way division is weakening. The Concertación seems to be running out of steam and the conservative right has every reason to believe that it will capture an electoral majority for the first time in decades in the 2005 presidential election. The growth of the right has led some analysts and political actors to pronounce the death of the two-way division and announce the return of the three-thirds. In this paper I contrast evidence that supports the thesis that the old three-thirds division will reemerge with arguments that suggest the country will continue to observe a two-way division in its political party system. 1

Three s Company: Old and New Cleavages in Chile s Party System 1 The Chilean political system was characterized by the three-way division right, center and left that constrained the formation of political alliances and shaped the government make up from 1932 to the 1973 democratic breakdown (Angell 1993, Caviedes 1979, Faúndez 1997, Garretón and Moulián 1983, Gil 1966, Scully 1992, Sigmund 1977, Valenzuela and Valenzuela 1976, Valenzuela 1995). A significant change in the political system after 1988 was the emergence of a new two-way alignment generated in the wake of the 1988 plebiscite, when supporters and opponents of the dictatorship aligned along the Yes and No options available in the presidential plebiscite. After 1988, the new two-way division seemed to have replaced the old 3-way divide. Because the old three-way divide was broadly characterized by a similar electoral weight of each one of those political thirds, many automatically assumed that the new 2-way divide would also be reflected in elections as a two-halves divide. But rather than 2-halves, the new division resulting from the 1988 plebiscite gave majority support to those who opposed the dictatorship, those who identified with the old center and left. After the No vote obtained 56% of the preferences in the 1988 plebiscite, the center-left Concertación coalition consistently commanded majority support in all elections held between 1989 and 1997. Yet, the emergence and consolidation of two large coalitions that together commanded around 90% of the vote was the most salient characteristic of the Chilean party system during the 1990s. If Chile enjoyed a multiparty system before the dictatorship, after 1990 the country experienced the formation and consolidation of a stable two-coalition system. True, before 1973 political parties often formed different alliances and legislative coalitions to exercise control of the executive and legislative. The Popular Front (PF) was one such coalition between 1938 and 1947. But the PF was not a very stable coalition. The only party that continuously remained a part of the PF was the Radical Party (PR). The Socialist Party (PS) and Communist Party (PC) entered and exited the coalition at various times and suffered internal divisions that resulted in the formation of different socialist and communist alliances and electoral parties. Rather than underlining the formation of government coalitions after 1988, analysts should emphasize the stability of the government and opposition coalitions formed after 1988. The stability in the party make up of the two coalitions formed after 1990 has made that period unique in Chile s political history. The Concertación has survived since 1988 without any major defection from its founding members. The only party that abandoned the center-left coalition was the Humanist Party (PH) in 1993. All the other founding parties have stayed. Similarly, the conservative coalition Alianza por Chile has remained primarily comprised of the Renovación Nacional (RN) and Independent Democratic Union (UDI) parties. Although some smaller centrist and conservative parties 1 The author gratefully acknowledges support from FONDECYT (Grant #1020684, Ser competente en política ). An earlier version of this paper was presented at the Chile, el País que Viene Conference, Harvard University, February 28-March 2, 2002. I am grateful for the feedback received there. 2

have entered the coalition for electoral purposes at different times, the two parties have remained the core members of that coalition. 2 Thus, even if during 2002 there were 7 parties with parliamentary representation and 9 parties with representation in municipal councils, the defining characteristic of the Chilean party system in the 1990s has been the presence of two large coalitions. Most historical analyses of the political party system existing before 1973 attribute some, if not all, of the responsibility for the democratic breakdown to the weakening experienced by the political center (Scully 1992, Valenzuela and Valenzuela 1876, Drake 1993, Sigmund 1977). Because of the excessive ideological purity of the Christian Democratic Party (PDC) and its inability, or unwillingness, to form stable and enduring alliances with parties from the right or left, the lack of a consensus-builder centrist party is identified as the leading cause of the polarization observed in Chile during the late 1960s. After the PDC replaced the PR as the largest centrist party in the 1960s, its inability to form a moderate reformist coalition during the Frei government (1964-1970) and its militant opposition to Allende (1970-73), made the three-way political division of the country incompatible with the subsistence of a democratic order. After the breakdown, 17 years of dictatorship resulted in high costs in terms of human rights and democratic values. Most studies on Chile s transition to democracy underline the importance of the alliance between the centrist PDC and the moderate left (several PS factions) in facilitating the formation of the center-left Concertación coalition in the 1980s (Valenzuela and Scully 1997, Tironi, Agüero and Valenzuela 2001, Siavelis 2000, Montes, Mainwaring and Ortega 2000). Because the opposition was unified and had agreed on a common cause, Chile s transition to democracy could take place (Boeninger 1997, Arriagada 1997). The Concertación successfully guaranteed that Pinochet s exit from power would not lead to new political and social confrontations and new bloodshed. Yet, the Concertación success was possible because the old left and old center overcame their ideological and tactical divisions and agreed on a common platform over which the Concertación was founded and built. Grounded on solid common views on democratic governance values and putting tactical and historical differences aside, the Concertación parties successfully provided presidents Aylwin (1990-94) and Frei (1994-00) with enough political and popular support to govern effectively. The PDC-socialist alliance scored consecutive electoral victories carrying overwhelming majorities of the popular vote in the 1988 plebiscite, 1989 presidential and parliamentary elections, 1992 municipal elections, 1993 presidential and parliamentary elections, 1996 municipal elections and 1997 parliamentary elections. Even in the highly contested 1999 election, the Concertación ended up winning a third consecutive presidential election. 2 That coalition has changed names over time. In 1989 it was called Democracia y Progreso; In 1993 it was Participación y Progreso; in 1993 it was Unión por el Progreso de Chile; In 1996 and 1997 it was Unión por Chile; and since 1999 it has been named Alianza por Chile. 3

Yet, the wear and tear suffered by all governments holding power for a long time also began to erode the electoral support for the Concertación. In addition, because the center-left coalition was initially formed to put an end to the dictatorship and guarantee a peaceful transition to democracy, once democracy was consolidated in the sense that no reversion to authoritarian government was possible, despite the survival of the 1980 Constitution with its deadlock authoritarian provisions the Concertación fulfilled its foundational mission. That coalition had to find a new unifying cause to reinvent itself and remain successful in the new decade. The 1999 economic crisis and the slow economic recovery that ensued brought about new tensions within the Concertación coalition. Many of those Concertacionistas who initially opposed the economic model implemented by the Pinochet dictatorship, and embraced later by the Aylwin and Frei governments, ended up conditionally accepting the model because of its positive results. Yet, the 1999 economic recession and the recent difficult economic times have fueled their previously silenced discontent. The debate between those who advocated for a change of the economic model and those who defended it first emerged in 1997, with two documents produced by different groups within the Concertación. On the one hand, those identified with the economic model produced a document titled La fuerza de nuestras ideas. There, they proposed to continue building on the economic policies adopted by presidents Aylwin and Frei. Conversely, those who sought to adopt policies that addressed more aggressively the problem of income and wealth distribution produced a document of their own titled La gente tiene razón where they criticized the self-complacent nature of the first document. The latter group was rapidly defined as self-flagellating. The division quickly led some analysts to identify a profound rift within the Concertación. Others minimized it by pointing to different priorities: while the self-complacent sought to promote economic growth, the self-flagellating group was more immediately interested in distribution policies. The debate was never fully exhausted within the Concertación and many saw it reemerging during the 1999 Lagos presidential campaign. There was a growing perception that it was increasingly making it more difficult for the coalition to speak with a unified voice and to share a common vision. Shortly after the Lagos presidential election, there was ample talk about the upcoming Lagos 6-year term as a farewell ceremony, 3 a care-taking wrap-up government that would eventually hand power over to Joaquín Lavín, the conservative candidate. Lavín had forced Lagos into a runoff election when he obtained the highest electoral support by any conservative candidate during Chile s history of full enfranchisement. In fact, many within the Concertación even suggested that it would be good and healthy for democracy that the conservative opposition achieved a victory in the presidential election of 2005. To be sure, although alternation in power is a central component of all democracies, uncertainty about electoral results also constitutes a central element of democratic systems. By anticipating, and even celebrating, a conservative victory in the 2005 presidential election, many Concertación intellectuals 3 The expresssion ceremonia del adiós was first used by socialist analyst Antonio Cortés Terzi in a paper of limited circulation. Several newspaper interviews early in 2000 made the phrase into a commonly used by analysts and press reports. 4

were undermining the very strength of democracy they claimed to defend. Yet, the perception that the Lagos 6-year period would constitute the last Concertación government became a generally accepted belief among the political elite. Logically, the concern about a farewell ceremony soon worried the Concertación parties. Political parties seek to win power and stay in power. The farewell ceremony syndrome was correctly perceived as detrimental to the interest of the Concertación parties. Reacting to the ceremonia del adiós, the different Concertación parties began to develop competing approaches and strategies to increase their electoral chances for the 2005 presidential and parliamentary election. A good deal of political strategy design during President Lagos s tenure has focused on finding ways to make the Concertación competitive. From the debate about the most useful policies that the Lagos administration should advocate to the cabinet appointments made by the president, all the decisions on policy and politics have been interpreted as reflecting conflicting views on which is the best strategy to make the Concertación a viable electoral coalition. One thesis outlined from within the Concertación to recover an electoral majority that gained some support and was widely circulated and discussed in documents and statements suggests that the Concertación must place emphasis on rebuilding its leftist and centrist support bases that originally constituted it. Those who support that view argue that by focusing on the growth of the Center and Left poles of the Concertación, the government alliance will successfully prevent the erosion in electoral support caused by the tendency of the Alianza opposition to cater to moderate voters and the efforts by out-of-the-concertación leftist parties to attract the support of left-leaning voters. The political strategy undertaken by current PDC president Adolfo Zaldívar since he was elected in January 2002 reflects the view that the Concertación can only be strengthened by strengthening first the individual parties that comprise it. Ironically, Zaldívar strategy was formally outlined by two members of the socialist party, Senator Carlos Ominami and political scientist Alfredo Joignant more than a year earlier (Joignant and Ominami 2000). Although the Joignant and Ominami text seems to be more concerned with strengthening the Concertación, the strategy under which Adolfo Zaldívar became party president is more directly aimed at strengthening the Christian Democratic Party. Yet both strategies seek to reinvigorate the three-way political party alignment of the pre 1973. Either because that will eventually increase the Concertación s electoral strength (as argued by Joignant and Ominami) or because it will prevent the further erosion of the PDC, fortifying the two poles that gave birth to the Concertación seems to be the strategy en vogue. To be sure, Zaldívar has repeatedly stated that a strong Concertación cannot exist without a strong PDC. His opponents believe that Zaldívar cares much more about a strong PDC than about a strong Concertación, but his argument ultimately coincides with the view of the two socialists in that reinforcing the center and left components of the government coalition is what needs to be done. 5

At any rate, both views have something in common. They perceive the Concertación merely as an electoral coalition formed by two distinctly different parties representing the Center and the Left. Thus, those views reject the argument that the Concertación is the result of the rearrangement of the political party system along the Yes-No options in the 1988 plebiscite. According to this view, the Concertación does not reproduce the views of those who opposed the Pinochet dictatorship in 1988. Instead it represents the alliance between the parties of the Center and Left that happened to opposed the Pinochet dictatorship. The 1988 plebiscite was just an occasion for the PDC and the socialists to unite. It did not represent the birth of a long lasting alliance (coalition of parties or eventually the birth of a new party called Concertación) between two groups that shared a common ideology. In other words, according to that view, the 1989 election marked the moment in which two of the three thirds formed an alliance to govern the country. This approach does not believe that a two-way division replaced the old threethirds division by the 1988 plebiscite. Naturally, it is empirically impossible to determine if the political cleavage that emerged in 1988 Yes versus No can survive the passage of time or if there is latent three-thirds division waiting to reemerge. If one believes that the political party system is sticky and long lasting, then one must believe that the old three-thirds division will not go away so easily. In fact, one can argue that the three-way divide will come back as soon as general Pinochet is out of the picture. Some might even argue that because Pinochet retired from political life after he was stripped of his parliamentary immunity, the time is ripe for the three-way division to reemerge. On the other hand, one might believe that new cleavages do indeed emerge and, moreover, that the 1988 plebiscite constitutes such new cleavage. If that is the case, one might well argue that the Yes-No division that characterized much of Chilean politics in that it led to the formation of two stable political coalitions will survive Pinochet s passing and will remain as the defining cleavage of the Chilean party system. I would suggest that the above is a more theoretical than empirical debate. First, cleavages need not appear accidentally. Second, political elites have something to do with how cleavages turn out to be determinants of electoral political developments. In that sense, the likelihood that the two-way division can survive beyond Pinochet or the possibility that the three-thirds breakdown will return as the defining characteristic of Chilean politics will partially depend on the decisions made by the Chilean political elite and the leading political parties. Given that the status quo is the two-way division, the reemergence of the three-thirds divide in the Chilean party system will be subject to a conscious decision by the leading political parties to realign along the old three-thirds divide. In addition, the acquiescence of the electorate to that decision by the leading political parties will also be necessary for such a realignment to take place. Neither condition by itself is sufficient for the three-thirds to replace the two-way division. Political parties alone or the electorate alone cannot unilaterally alter the status quo. The most recent effort to end the Yes-No divide that characterized Chilean politics since 1988 and resurrect the three-thirds division was undertaken by the PDC. Conservative parties seem to be satisfied with the survival of the two-way division. After 6

the most recent parliamentary elections, UDI president Pablo Longueira celebrated the consolidation of a two-way divide and defined the two coalitions as democratic centerright and democratic center-left. The left is unclear as to the convenience of resurrecting the old three-way divide. On the one hand, the PS seems more enthusiastic about the three-thirds. The PS even attempted to generate some mutual understanding with extra- Concertación leftist parties in the 1997 and 2001 parliamentary elections. On the other hand, the other leftist Concertación party, the PPD, seems to favor the two-way divide. After all, the PPD was born when that divide was in place and a three-third division would force the PPD to define itself either as a centrist or a leftist party. The PS has not actively engaged in attempting to resurrect the three-thirds because of the immediate conflicts with the PPD and the PDC that such a move would generate. Whereas the PPD would actively oppose the effort to rearticulate the left, seeking to build a coalition with the out-of-concertación leftist parties would be vehemently opposed by the PDC. Thos, to properly assess the likelihood that the three-thirds division can successfully replace the two-way divide, we must identify if the current PDC strategy of separating itself from the leftist Concertación parties has a chance of success. Given that the only party that can attempt to resurrect the three-thirds division is the PDC, a successful replacement of the two-way divide by the old three-thirds requires that the PDC strategy be successful. In order to test whether there is electoral support for the initiative to resurrect the three-thirds division, I looked at electoral results for the 1997 and 2001 parliamentary elections. The evidence suggests that voters were more likely to support a two-way divide in 1997 than in 2001. In the most recent parliamentary elections, voters split their votes for candidates from the Concertación for one chamber and candidates from the Alianza in the other at higher rates than in 1997. In addition, the most recent electoral results show a significant electoral growth of the conservative third. The Alianza has grown to obtain well over one third of the votes in the most recent election. Although it is true that Pinochet obtained 44% of the vote in the 1988 plebiscite, the vote for conservative parties did not surpass the 40% mark in any of the elections held during the 1990s. Yet, both in the 2000 municipal election and the 2001 parliamentary election, the vote for conservative parties was higher than 40%. The growth in recent electoral support for conservative parties has been widely explained by the fall in support for the PDC. Yet, the PDC began to lose electoral support in the 1993 parliamentary elections, years before conservative parties began to gain support. In part, the fall in electoral support for the PDC went unnoticed because that party suffered significant losses in parliamentary seats only in the 2001 elections. Despite having fallen from 27% to 23% of the vote between 1993 and 1997, the PDC actually gained one seat in the Chamber of Deputies, going from 37 seats in 1993 to 38 seats in 1997. Yet, in 2001, the PDC fell again by 4% in electoral support, going from 23% to 19%, but in terms of votes, the PDC lost 14 seats, by going from 38 seats to 24 in the Chamber of Deputies. In the Senate, the results were equally dramatic. While the PDC had won 10 senate seats in 10 senatorial districts in 1997, that party only clinched 2 seats in 9 districts in 2001. That dramatic fall took place despite a moderate fall in electoral support down from 29.2% in 1997 to 22.8% in 201. 7

Table 1 shows the results for the 1993 and 2001 parliamentary elections. One can easily observe a fall in the number of valid votes (which was widely discussed in the 1997 parliamentary election when it reached an all-time law of 5,795 million votes). On the other hand, there is an evident growth in the electoral support for the UDI and a decrease in the electoral support for the PDC. The electoral advantage of the Concertación over the Alianza shrank dramatically between 1993 and 2001. Although the 1993 parliamentary elections were held concurrently with presidential elections where the popular Concertación candidate more than doubled the vote for the conservative candidate a comparison between both elections clearly shows significant fluctuations. After 8 years, there was a 7.5% variation in the vote for the two largest coalitions, with a clear growth for the right and a clear loss for the Concertación. True, depending on what we compare this variation with, we might conclude it is a small variation. Yet, if we compare it with the 1989 and 1992 results, it does represent a significant variation. Table 1. 1993 and 2001 Parliamentary Election Results PARTY 1993 % 2001 % Unión Demócrata Independiente UDI 816,104 12.11 1,538,835 25.20 Renovación Nacional RN 1,098,852 16.31 840,568 13.76 Alianza (Union) por Chile, Others 556,833 8.26 327,751 5.37 Alianza por Chile Total 2,471,789 36.68 2,707,154 44.33 Partido Demócrata Cristiano PDC 1,827,373 27.12 1,155,597 18.92 Partido Radical de Chile PR 200,837 2.98 247,576 4.05 Partido Socialdemocracia Chilena PSD 53,377 0.79 Partido Socialista de Chile PS 803,719 11.93 611,305 10.01 Partido por la Democracia PPD 798,206 11.84 777,278 12.73 Concertación Others 49,764 0.74 134,044 2.19 Concertación Total 3,733,276 55.40 2,925,800 47.91 PC 430,495 6.39 318,638 5.22 PH 96,195 1.43 69,265 1.13 Independents 7,104 0.11 86,283 1.41 Valid Votes 6,738,859 100.00 6,107,140 100.00 Source: constructed by author with data from http://www.elecciones.gov.cl Yet, to conduct a more rigorous analysis, the results must be desegregated. Rather than using regression analysis a procedure that might be confusing to many readers and that turns out to provide little additional explanatory power I conducted a simple test of electoral performance in each of the country s 341 municipalities. Since the rate of electoral participation (valid votes/registered voters) did not fall uniformly across the country, I estimated an abstention rate change for each municipality between 1993 and 2001. Assuming that the fall in electoral participation equally hurt all political parties that is, that no party suffered from higher abstention more than the rest the abstention tax is the predicted fall in votes for each party between 1993 and 2001. Thus, if the fall in electoral participation in a given municipality was 13% (70% of the registered population cast valid votes in 1993 but only 57% did in 2001), each party should have experienced a 13% drop in its total number of votes product of lower participation. 8

Because this estimate is constructed municipality by municipality, the abstention tax varies widely across the cases. I estimated the abstention tax for all the municipalities in the country, counting as a single municipality those that were split into two communes after 1993, 4 except in two cases. 5 Table 2 shows the estimated drop in electoral support for each party in the 2001 parliamentary election, based on the drop in electoral participation experienced between 1993 and 2001. As shown in Table 2, even thought the after abstention tax predicted fall in votes for the PDC is 216 thousand, its observed fall was 630 thousand. That is, there are some 413 thousand votes that cannot be explained by a drop in electoral participation (again, assuming that the fall in electoral participation equally hurts all parties). The rest of the Concertación experienced a slight improvement between 1993 and 2001 after discounting the abstention tax. This is particularly relevant because the PDC had candidates in 48 districts in 1993 and 54 districts in 2001. Thus, the other Concertación parties improved their performance despite having fewer candidates. Yet, the improvement in the other Concertación parties does not mean an improvement for the entire left. If we add together the vote for the PS, PPD, PC and PH, the Left lost 261 thousand votes in 8 years. Assuming that the growth in abstention and null/blank votes is distributed proportionally to all parties based on their 1993 number of votes, the left experienced a net loss of 72 thousand votes. In a universe of more than 8 million registered voters, that drop is small but not trivial. The big winner are the parties of the Right, Alianza por Chile, that improved its vote after taking into account the abstention tax by over 430 thousand votes (7% of valid votes in 2001). Within the right, UDI improved significantly more than RN. In fact, while the UDI observed a net increase, after the abstention tax, of more than 780 thousand votes, the rest of the Alianza experienced a drop of 358 thousand votes. True, the UDI ran candidates in 29 districts in 1993 and 54 in 2001. A good portion of the larger vote totals gathered by UDI resulted from having more candidates in Chile s 60 Chamber of Deputies districts. In that sense, we can safely speak of a new alignment within the Alianza. The UDI has grown at the expense of independent conservatives and RN. Yet, it is also true that overall, the Alianza increased its vote between 1993 and 2001, after discounting the abstention tax, in more than 432 thousand votes. 4 Viña del Mar and Concón were grouped under Viña del Mar; Chillán and Chillán Viejo were grouped under Chillán; Concepción, Chiguayante and San Pedro de la Paz were grouped under Concepción; Temuco and Padre Las Casas were grouped under Temuco. 5 Padre Hurtado (Región Metropolitana) and San Rafael (VII Región) were made up by taking some territories away from several different municipalities. I simply omitted those two municipalities for the 2001 election. For that reason, my totals are different than those officially published in http://www.elecciones.gov.cl/ 9

Table 2. 1993 2001 Electoral Gains After Abstention Tax by Parties/Coalitions Net Vote Loss 1993-2001 Estimated Vote Loss 1993-2001 Party / Coalition 1993- Deputies 2001- Deputies Difference (Est Net) PDC 1.827.373 1.197.912 629.461 147.513 481.948 Other Concertación 1.905.903 1.814.986 90.917 149.560-58.643 Concertación 3.733.276 3.012.898 720.378 297.072 423.306 UDI 816.104 1.534.847-718.743 70.989-789.732 Other Alianza 1.655.685 1.166.624 489.061 136.962 352.099 Alianza 2.471.789 2.701.471-229.682 207.951-437.633 Concertación + Alianza 6.205.065 5.714.369 490.696 505.023 14.327 Left 2.432.593 2.171.112 261.481 191.729 69.752 Total Valid Votes 6.738.859 6.091.776* 647.083 647.083 0 Source: http://www.elecciones.gov.cl/ The 2001 total is lower than the official 6,107,140 because I omitted the 11565 votes of Padre Hurtado (district 31) and the 3799 votes of San Rafael (district 38) since those municipalities were created after 1993 by putting together pieces of different municipalities. If we follow the logic of the three-thirds, the growth in electoral support for the right represents a monumental challenge to the monolithic control of the political center exercised by the PDC during the 1990s. After the Lagos election, when the Concertación seemed to have moved, however slightly, toward the left, the PDC chose to actively seek to recover the political center. If the 1999 presidential election confronted the left (Ricardo Lagos) with the right (Joaquín Lavín), not offering a credible centrist candidate, the electoral results of the 2000 municipal elections and 2001 parliamentary elections seemed to indicate that the electorate was not too thrilled about supporting centrist candidates. The PPD-PS-PRSD parties within the Concertación improved their share of seats in parliament, although they slightly improved their share of the vote. The UDI increased its share of seats and votes. Yet, the PDC fell in terms of seats and votes. In part, the fall in electoral support for the PDC has something to do with the incentives of the electoral system. The so-called binominal system, a 2-seat proportional representation (PR) arrangement, has built-in incentives that lead to polarization rather than to convergence towards the median voter (Magar, Rosemblum and Samuels 1998). Figure 1 shows the centrifugal incentives of a PR 2-seat arrangement using the same d Hondt mathematical formulae utilized before 1973 as compared to the centripetal incentives of a single member district with runoff arrangement. Whereas under a 2-seat PR, a candidate can secure a seat with 33.4% (33 1/3% + 1, strictly speaking), under a single member district with runoff, 50% + 1 is required to secure a seat. As with all PR arrangements, the incentives to polarization or at least the fact that lower electoral thresholds allow minority parties to achieve representation were entrenched in the electoral system adopted by the Pinochet dictatorship. Understood as an insurance mechanism, PR wit district magnitude of 2 allows you to secure 50% of the seats with slightly more than 1/3 of the votes. 10

Figure 1. Centrifugal incentives of the binomial system + Number of voters 33.4% threshold for a senate or chamber of deputies seat 50%+1 threshold for the presidency 33.4% threshold for a senate or chamber of deputies seat When presidential elections are held concurrently with parliamentary elections, the centrifugal effect of the electoral system is undermined by the centripetal effect of the presidential election. Although candidates for parliament need only to obtain 1/3 of the vote to secure 50% of the seats, presidential candidates need to secure an absolute majority to win the election. Yet, when parliamentary elections are held not concurrently with presidential elections, as in 1997 and 2001, the centrifugal effect is not present. Although those structural incentives generated by the electoral rules were already in place in 1989, 1993 and 1997, the PDC successfully captured the support of a significant share of the electorate to become the party with the first plurality of votes and seats. Yet, for different reasons, the electorate seems to have fallen out of love with the PDC in the most recent elections and thus the political center has weakened. There are no apparent reasons to expect that such trend will be reversed in the near future. In fact, for the 2005 parliamentary elections, regardless of its electoral support, the PDC can only lose its share of seats in the Senate, since it will have 10 seats up for re-election in the 10 senatorial districts that will be up for election. Moreover, there is evidence that the weakening of the centrist PDC extends beyond mere electoral considerations. The difficulties experienced by that party in recent years seems to be indicative of deeper troubles within the PDC. Its embrace of the Zaldívar doctrine that advocates for stressing the party differences with the Lagos government ultimately seeks to resurrect the threethirds thesis. Yet, if the electorate does not go along and instead chooses to confirm the two-way division, the effort to strengthen the center will fail. Unfortunately, the electoral evidence to support either the view that voters favor the two-way division or prefer a return to the old three-thirds divide is inconclusive. There is evidence based on the 1997 electoral results that the electorate aligns along the two-way divide. But there is also some evidence based primarily in the 2001 results that the electorate might be prone to return to a three-thirds division. In the 2001 parliamentary elections, voters from 9 senatorial districts went to the polls to elect senators and Chamber of Deputies members concurrently. 6 The rest of Chile s senatorial districts only had Chamber of Deputies elections. In 1997, the other 10 senatorial districts had concurrently held senate and Chamber of Deputies elections. It is often argued that sophisticated voters go beyond party identification and vote or candidates of certain parties for given offices and candidates of other parties for other 6 Those districts were Tarapaca (I), Atacama (III), Valparaíso-Coast (V), Valparaíso-Interior (V), Maule- North (VII), Maule-South (VII), Araucanía-North (IX), Araucanía South (IX) and Aysén (XI). 11

offices. Thus, divided government results from lower levels of party identification on the part of voters and concurrently from higher levels of sophistication among voters (Alesina and Rosenthal 1995). Simply put, more sophisticated voters are more likely to split their tickets and vote for a candidate of a party for the presidency and a candidate of a different party for the legislature. Some have claimed that voter sophistication is a result of greater social and economic development, while other highlight that higher voter sophistication tends to occur in countries with more established democracies. Traditionally, polling data has been used to test these propositions (Achen and Shively 1995). Survey data shows that split-ticket voting rates in America are higher than in previous decades (Burden and Kimball 1998). Some researches have argued that ticket splitting is intentionally done to produce certain mix of policies or to create checks among parties (Alesina and Rosenthal 1995, Fiorina 1996). Other accounts include as explanatory variables the decreasing ideological differences between political parties (Born 1994, Soss and Canon 1995), competitiveness of congressional elections (Jacobson 1997), and incentives offered by certain ballot mechanisms (Beck 1997, Cox 1996). But the widely accepted view is that ticket splitting results from shrinking party identification by voters and the growth of personalized campaigns. People vote for candidates rather than for parties (Cain, Ferejohn and Fiorina, 1987). Naturally, in parliamentary systems and in presidential systems with closed-lists, it is more difficult to move away from party vote to personalized vote. Yet, Chile is a country with a presidential system but with open list provisions. Voters actually vote for candidates rather than for parties. True, they might choose to vote for candidates that represent a given party, but they still need to cast a vote for an individual party rather than for a party slate. Recent survey data in Chile has also indicated a reduction in the level of political party identification on the part of voters, 7 and recent publications have underlined the shrinking party loyalty among Chilean voters. 8 Others have argued that a consolidation of personalized political candidacies and a weakening of party identification on the part of candidates are also taking place (Tironi 1999, Halpern 2002). Thus, the phenomena observed in the U.S. and other industrialized nations seem to have reached Chile as well. By studying ticket splitting, we can determine if voters align along the two-way divide or support a three-thirds divide. If voters split their tickets within the two large coalitions but not across coalitions, there is evidence to support the argument that a twoway divide exists. However, if voters split their votes within and across coalitions, we can argue that perhaps a significant number of centrist voters alternate between supporting Concertación and Alianza centrist candidates. Thus, there more inter-coalition ticket splitting exist, the more likely voters are to prefer the three-third divide. Conversely, the more intra-coalition ticket splitting exist, the more likely is that voters identify with the Yes-No cleavage created in 1988 and indistinctly choose from any party as long as they belong to either the Concertación or the Alianza coalition. 7 See López and Martínez (1999), Centro de Estudios Públicos (1990), Centro de Estudios de la Realidad Contemporánea (1993, 1994), Hinzpeter and Lehmann (1999a, 1999b), Lehmann and Hinzpeter (2001) and most recently the Centro de Estudios Públicos poll databank, http://www.cepchile.cl/cgi-dms/procesa.pl?plantilla=/base.html&contenido=categoria&id_cat=443 8 See Agüero, Tironi, Valenzuela and Sunkel (1998) and Tironi, Agüero and Valenzuela (2001). 12

Intra-Coalition Ticket Splitting in 1997 In the 1997 parliamentary elections, the aggregate figures provide little evidence of ticket-splitting going on in either direction. While the Concertación obtained 49.8% of the vote in the Senate election (10 senatorial districts), the government coalition obtained 50.5% in the Chamber of Deputies election (120 districts, the entire country). The RN- UDI coalition obtained 36.3 and 36.6% respectively. However, if one looks at each one of the 10 senatorial districts by party, there is clear evidence of the existence of a limited, but significant, amount of intra-coalition ticket splitting going on. As shown in Chart 1, the vote for the PDC varied considerably across senatorial districts in 1997. The PDC consistently got more votes for its senate candidates than for its candidates for the Chamber of Deputies. True, the PDC had candidates for the senate in each of the 10 senatorial districts up for election, but it abstained in 5 of the 60 Chamber of Deputies districts. Regardless, even in regions where the party had Chamber of Deputies candidates in all districts, the senatorial candidates performed consistently better than the candidates for the lower house. Chart 1. Vote for PDC Senate and Chamber of Deputies Candidates by Senatorial District, 1997 350,000 300,000 Senate PDC Deputies-PDC 250,000 Votes 200,000 150,000 100,000 50,000 0 II IV RM-1 RM-2 VI VIII-1 VIII-2 X-1 X-2 XII Senatorial District Conversely, the other Concertación parties did better in Chamber of Deputies elections than in the senatorial election. As shown in Chart 2, the combined vote for Chamber of Deputies candidates from the PPD/PS/PRSD was consistently better than that of the senate candidates from those parties. Again, in 5 Chamber of Deputies districts, the two Concertación candidates belonged to the PPD/PS/PRSD, but even in those regions where the two Concertación candidates were equally split between the PPD/PS/PRSD and the PDC, the Chamber candidates of the leftwing Concertación parties did better than the Senate candidates from those parties. 13

350,000 300,000 Chart 2. Vote for PS-PPD-PRSD Senate and Chamber of Deputies Candidates by Senatorial District, 1997 Senate PS-PPD-PRSD Deputies PS-PPD-PRSD Votes 250,000 200,000 150,000 100,000 50,000 0 II IV RM-1 RM-2 VI VIII-1 VIII-2 X-1 X-2 XII Senatorial District The most dramatic difference occurred in RM-1 (Santiago s Western Senatorial District). The combined vote of the PPD/PS/PRSD candidates for the Chamber of Deputies was about 60% higher than the votes obtained by the PS senatorial candidate. The presence of PC s popular president, Gladys Marín, as a senatorial candidate might have led many PPD/PS/PRSD sympathizers to cast a vote for the communist senatorial candidate. After all, the likelihood of having both Concertación s candidates elected are minimal and so are the chances of having the opposition get twice as many votes as the Concertación. Thus, because it would have had no effect either way, many voters might have chosen to cast a vote for the communist candidate to make a statement, knowing that the Concertación and Alianza would have still end up getting one senate seat each. Table 3. Santiago Western Senatorial District Vote, 1997 PC (PS-PPD) candidates Total Concertac RN-UDI PH Chile 2000 Total Deputies 119.977 294,784 554.411 362.340 39.773 21.280 1.097.781 Senators 174.780 177,965 487.335 382.286 26.794 42.771 1.113.966 Difference (Sen Dep) -54.803 116,819 67.076-19.946 12.979-21.491-16.185 Source: calculated with data from http://www.elecciones.gov.cl Altogether, as shown in Chart 3, the Concertación vote for the Chamber and Senate elections ended up balancing itself out in most senatorial districts. That would lead to think that Concertación voters are voting indistinctly for PDC, PPD, PS and PRSD more and more, as long as they represent the Concertación. This is somewhat contradictory with what Valenzuela and Scully (1993), Montes, Mainwaring and Ortega (2000) and Siavelis (1997) have argued. They have pointed out to the continuities of the Chilean party system, historically divided among three-thirds (right, center and left). The evidence from W=Santiago Western Senatorial District might in fact point to the 14

persistence of the three-thirds. Many voters who cast votes for PS-PPD-PRSD candidates, also voted for the Communist Party senatorial candidate. Still, because throughout the country there seems to have been little inter-coalition ticket splitting going but there was intra-coalition ticket splitting within the Concertación, the evidence seems to indicate that the old three-thirds split of the Chilean party system might be giving way to a new 2-way divide (Concertación vs. Alianza). 600,000 500,000 Chart 3. Concertacion Senate and Chamber of Deputies Candidates by Senatorial District, 1997 Senate Concertacion Deputies Concertacion 400,000 Votos 300,000 200,000 100,000 0 II IV RM-1 RM-2 VI VIII-1 VIII-2 X-1 X-2 XII Senatorial District Yes-No Cleavage in the 1999 Presidential Election The 1999 presidential election results have also been used as evidence of the consolidation of a Concertación vote (rather than a specific PDC, PS, PPD or PRSD vote) among the electorate. If the 1989 and 1993 presidential elections were sufficient proof that leftwing Concertación voters were ready to throw their support behind the centrist Christian Democratic candidate, the 1999 election showed that centrist PDC sympathizers could vote for a leftwing Concertación presidential candidate. Thus, we could very much be witnessing the consolidation of a two-way split in the Chilean electorate that has slowly begun to replace the old three-way division. The poor electoral support obtained by self-proclaimed centrist Arturo Frei in the 1999 presidential election shows that faced with that admittedly mediocre alternative, voters preferred to align along a two-way divide than to reproduce the old three-way split. It might very well be that a considerable number of centrist voters might have chosen to support Joaquín Lavín rather than Ricardo Lagos. Lavín s strong 47.5% showing, higher than that obtained by Pinochet in the 1988 plebiscite, fuels the speculation that many centrist voters deserted the Concertación. However, it is also true that the leftwing candidate obtained a larger share of the vote than any other leftist candidate before in Chilean history. In the run-off election, Lagos surpassed the 1971 15

Municipal election vote obtained the Popular Unity government. Thus, just as a considerable number of centrist voters must have opted to vote for Lavín in 1999, an even larger number of those who have traditionally voted centrist opted to support Lagos in the presidential contest. Chart 4 shows that the Concertación vote has remained stable over the past years and that even in 1999, the Concertación vote did not depart significantly from what was observed in previous elections. Although there is no conclusive evidence to suggest that the three-thirds division is now gone, we can safely argue that the 1988 Yes-No cleavage remained strong throughout the 1990s, including the 1999 presidential election. Chart 4. Concertacion Vote (Lagos, 1999 First Round, 2000 Municipal and 1997 Parlamentary Elections) by Chamber of Deputies District 130000 110000 Conc-97 Concertacion-2000 Lagos 90000 Votes 70000 50000 30000 10000 1 5 9 13 17 22 23 27 33 37 42 41 49 53 57 District 2001 Inter-Coalition Ticket Splitting Just as the 1997 parliamentary election strengthened the view that there is a Concertación and Alianza vote, the 1999 confirmed the suspicion that centrist pro-pdc voters would overwhelmingly support the Concertación candidate even if he/she was from the PS/PPD. In 1997, ticket splitting might have occurred within the Concertación and within the Alianza parties but not across political coalitions. Yet, the 2001 parliamentary election results call into question that belief and fuel the argument that Chilean voters do indeed cross party and coalition lines to exercise their ticket-splitting option and vote for Chamber of Deputies candidates from one coalition and Senate candidates from a different coalition. 16

As shown in Chart 5, the PDC again performed better with senatorial candidates than with Chamber of Deputies candidates in most regions. With the notable exceptions of northern Chile (Regions I, Arica, and III, Atacama), the PDC always obtained more votes for the Senate than for the Chamber of Deputies. Again, in a few regions (most notably VII-South), there were several Chamber of Deputies districts with no PDC candidate, thus the lower vote totals. But in general, the PDC again did better in the upper house elections than in the lower house contest. 90000 80000 70000 60000 50000 40000 30000 20000 10000 0 Chart 5. Vote for PDC Senate and Chamber of Deputies Candidates by Senatorial District, 2001 PDC-Deputies PDC-Senate I III V Interior V Costa VII Norte VII Sur IX Norte IX Sur XI 160000 140000 120000 Chart 6. Vote for PS-PPD-PRSD Senate and Chamber of Deputies Candidates by Senatorial District, 2001 PPD-PS-PRSD-Deputies PPD-PS-PRSD-Senate 100000 80000 60000 40000 20000 0 I III V Interior V Costa VII Norte VII Sur IX Norte IX Sur XI Although one would expect the opposite for the PPD/PS/PRSD vote, the candidates from those parties for the Senate also often did better than the Chamber of Deputies candidates from those parties. In 6 of the 9 senatorial districts where elections were held, the Senate candidate from the PPD/PS/PRSD obtained more votes than the combined votes for the Chamber of Deputies candidates from those parties. 17

How can it be that both the PDC and the PPD/PS/PRSD do better with their Senate candidates than with the Chamber of Deputies candidates? The answer can be clearly inferred from Chart 7. In general, the Concertación obtained more votes in the Senate races than in the Chamber of Deputies races. In six of the 9 senatorial districts up for election, the vote for the Concertación Senate candidates was higher than that for the Concertación Chamber of Deputies candidates. 200000 160000 Chart 7. Vote for Concertacion Senate and Chamber of Deputies Candidates by Senatorial District, 2001 Conc-Deputies Conc-Senate 120000 80000 40000 0 I III V Interior V Costa VII Norte VII Sur IX Norte IX Sur XI This evidence of ticket splitting calls into question the preliminary conclusion from the 1997 results. Voters do split their ticket for candidates from different parties within a coalition, but they also split their tickets among candidates from parties from different coalitions. The thesis advanced by Tironi (1999) and Halpern (2002) finds support from aggregate data from the most recent election. Chileans are splitting their tickets within and across political coalitions. Partially, the reason for the phenomenon observed in 2001 has to do with the electoral strategies devised by the different political coalitions. While the Concertación continued to fill 2 candidates in every senatorial and Chamber of Deputies district, the Alianza negotiated a slate that presented only one Alianza senatorial candidate in 7 of the 9 districts up for election. 9 That transformed the senate election into a de facto closed-list for Alianza voters. Rather than having the option to select among two Alianza senatorial candidates, those who preferred to vote for the conservative opposition were forced to accept the candidate selected by the party elite. Because the campaign reach of one senatorial candidate will necessarily be more limited than that of two candidates, the total vote for the coalition will likely go down when one candidate runs in a 2-seat district against coalitions filing 2 candidates. That was the case in the 2001 parliamentary elections in Chile. One of the unintended effects of that strategic choice made by the 9 Those districts were Atacama, Valparaíso-Coast, Va lparaíso-interior, Maule-North, Maule-South, Araucanía-Nort and Aysén. Although there were two candidates in the Alianza ticket in Valparaíso- Interior, Maule-North and Manuel-South, the Alianza made it clear that there was only one privileged candidate in the coalition, and Alianza voters successfully coordinated behind that candidate to prevent intra-coalition competition. 18