UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Institutional reform and party systems : how direct presidential elections fragmented the 2004 Indonesian… Allen, Nathan Wallace 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2006-0356.pdf [ 6.55MB ]
JSON: 831-1.0092688.json
JSON-LD: 831-1.0092688-ld.json
RDF/XML (Pretty): 831-1.0092688-rdf.xml
RDF/JSON: 831-1.0092688-rdf.json
Turtle: 831-1.0092688-turtle.txt
N-Triples: 831-1.0092688-rdf-ntriples.txt
Original Record: 831-1.0092688-source.json
Full Text

Full Text

INSTITUTIONAL REFORM AND PARTY SYSTEMS H O W D I R E C T P R E S I D E N T I A L E L E C T I O N S F R A G M E N T E D T H E 2004 I N D O N E S I A N L E G I S L A T I V E V O T E by N A T H A N W A L L A C E A L L E N B A , Acadia University, 2005 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF M A S T E R OF A R T S in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Political Science) U N I V E R S I T Y OF B R I T I S H C O L U M B I A August 2006 © Nathan Wallace Allen 2006 Abstract It has long been noted that low district magnitudes tend to produce consolidated party systems. Between 1999 and 2004, Indonesia considerably reduced its district magnitude, yet experienced a significant increase in party fragmentation. W h y did Indonesia's party system fragment under institutional pressure to consolidate? This thesis demonstrates that district magnitude did not affect the fragmentation of the legislative vote in Indonesia. The cause of the increased party fragmentation was the introduction of direct presidential elections. Aspiring presidential candidates bypassed the established party system and introduced their own parties as presidential vehicles in the legislative election. Voters also used their ballots to express different preferences in the legislative and executive elections, supporting inclusive leaders that could guarantee national unity in the presidential election and smaller ideological or sectarian parties in the legislative contest. The introduction of presidential vehicles combined with voters' increased willingness to support non-presidential parties effectively explain the increase in party fragmentation. n Table of Contents Abstract Table of Contents List of Tables List of Figures Acknowledgements Chapter 1 Introduction Competing Hypotheses Presidentialism and Party Fragmentation Methodology 1 Chapter 2 : District Magnitude and Electoral Fragmentation 2 A Review of Electoral Reforms 2 Mechanical Effects 2 The Number of Contesting Parties 3 Magnitude and the Effective Number of Electoral Parties 3 Voter Learning 3 Magnitude and Social Cleavages 3 Evaluating the Effect o f Magnitude 4 Chapter 3 : Direct Presidential Elections and Party Electoral Fragmentation 4 A Review of Presidential Reforms in Indonesia 4 Presidentialism and the Limited Success of Coat-Tail Parties 5 Party Leadership and Voting Behavior 5 Split-Ticket Voting and the Rise of the P K S 6 Chapter 4 : Conclusion 6 Bibliography.. 7 Appendix A 7 Appendix B 8 List of Tables Table 2.1 - Magnitude Change 28 Table 2.2 - Mechanical Effects 29 Table 2.3 - Competing Parties 33 Table 2.4 - Ethnicity in Indonesia 41 Table 2.5 - Social Cleavages and Magnitude 46 Table 3.1 - 'Presidential Party' Vote Share 60 iv List of Figures Figure 2.1 - Magnitude and E N E P in 1999 35 Figure 2.2 - Magnitude and E N E P in 2004 36 Figure 2.3 - Magnitude and E N E P in Java 37 Figure 2.4 - Magnitude and E N E P in the Outer Islands 37 Figure 2.5 - 1999 Deviation and E N E P Change 39 Figure 3.1 - P D Vote by PDI-P Losses 54 Figure 3.2 - Pet of S B Y - K a l l a First Round Vote Share vs. Pet of P D Vote Share ;.. 55 Figure 3.3 - Leadership, Party ID, and Voting (Liddle & Mujani) 62 Figure 3.4 - P K S Gains and P P P - P A N Losses 64 Figure 3.5 - Rais Presidential Support by P K S Vote Share 65 v Acknowledgements First, I would like to thank Benjamin Nyblade, my thesis supervisor. Ben offered a rare blend of patience, enthusiasm, and criticism that made writing this thesis a challenging yet enjoyable endeavour. Diane Mauzy provided insightful comments and critiques that wi l l continue to influence my work for years to come. I am in debt to both for their superb academic guidance. A s well , numerous long conversations with Shane Barter about Indonesian politics were invaluable for provoking the ideas that led to this thesis. Thank you to Wi l l i am Liddle for sharing a forthcoming manuscript. Andrea Lawlor provided useful comments on all drafts of this thesis. I benefited from methodological discussions with Go Murakami. Lastly, I would like to thank Do l l , Bruce and the rest of the family for making Vancouver feel like a home away from home. Chapter 1 Introduction In 2004, Indonesia conducted its second legislative election of the post-Soeharto era. The election brought about significant changes to Indonesia's party system. The established 'B ig -5 ' parties from the 1999 election were joined by two new significant political players. Partai Golkar, the winner of the legislative plurality, managed to attract a mere 21.6% of the electoral vote, less than two-thirds of the vote garnered by the previous victor. Additionally, support for small fringe parties expanded to over 20% of the electoral vote. The 2004 elections had left Indonesia with a significantly more fragmented legislature. The increased fragmentation in Indonesia is surprising because it occurred in an institutional context in which one would expect to see a consolidation of the party system. The subdivision of provinces into multiple electoral districts in 2004 led to an impressive reduction in the average district magnitude from what it had been in 1999. Scholars have long been aware that low district magnitudes produce pressure on voters and elites to consolidate their support and efforts into a smaller number of political parties (Taagepera & Shugart 1989). Why, then, did party fragmentation in Indonesia expand under institutional pressure to contract? The study of party fragmentation is significant because it is closely tied to the prospects of democratization in Indonesia. The 2004 electoral results led some observers (Liddle & Mujani 2006) to revisit Mainwaring's (1993) warning about the 'Difficult Combination' o f presidentialism and multipartism. Indeed, soon after the election there were signs of deadlock, ideological polarization, and coalition fragility. Given multiparty presidentialism's reputation for institutional breakdown, discovering the causes of 1 electoral fragmentation in Indonesia are pertinent to the country's larger process of democratic consolidation. The thesis asks the following question: why was there a significant increase in the effective number of parties in Indonesia between 1999 and 2004? It is argued that the switch to direct presidential elections increased the fragmentation of the legislative vote. In 1999, Indonesians could only express their support for a particular party in a legislative election. The 2004 election saw the introduction of direct presidential elections, thus giving the Indonesian electorate the opportunity to vote for a preferred legislative party on one ballot and their preferred presidential candidate on another. It was this reform to the executive-legislative structure that caused the significant increase in party fragmentation. Competing Hypotheses While this thesis suggests that the introduction of direct presidential elections explains the increased party fragmentation in Indonesia, there are four competing hypotheses to consider. The first two - an anti-incumbent reaction and an anti-establishment reaction -are popular in the case-study literature. Hypothesis three and four look at district magnitude and social cleavages, the two most common ways party fragmentation is explained in the comparative literature. The thesis argues that the case-driven explanations are inadequate and incomplete while hypotheses three and four cannot properly account for the increased fragmentation in Indonesia because magnitude had no effect in this case and there was no significant change in Indonesia's social structure. 2 Competing Hypothesis 1: Anti-Incumbent Reaction It has long been noted that there is often an anti-incumbent reaction that manifests itself in a transitional country's second election. A s Samuel Huntington explains: Politically, the years after the first democratic government came to power were usually characterized by the fragmentation of the democratic coalition that had produced the transition, the decline in the effectiveness of the initial leaders of the democratic governments, and the growing realization that the advent of democracy would not, in itself, produce solutions to the major economic and social problems confronting the country. (1993, 256) In short, a rejection of the governing party and the resulting breakdown in the anti-authoritarian coalition can fragment a party system. Numerous observers of Indonesian politics have pointed out the public dissatisfaction with incumbent (Megawati Sukarnoputri) and her party (Partai Demokrasi Indonesia - Perjuangan, PDI-P). Sherlock (2004, 26) notes, "In many people's eyes, PDIP in government has become not the party o f the masses but the party of the elite." Sebastian (2004, 266) comments, "Megawati's shortcomings, coupled with the mediocre performance o f the PDI-P legislators during her term in office, were evident when the general election results were announced." Liddle & Mujani (2005,122) applied a similar logic to the presidential contest, noting, "[Presidential candidate Susilo Bambang Yudhoyono] was able to capitalize on pervasive dissatisfaction with Megawati's leadership of the country." Thus a number o f scholars saw evidence of an anti-incumbent reaction. Anti-incumbent sentiment cannot explain the extent to which the party system fragmented however. First, the anti-authoritarian reform movement was always fragmented. Even in the first elections there were three major reformist parties. In other words, the collapse o f a reformist coalition can explain little in a case where there was no 3 cohesive electoral coalition to begin with. Second, an anti-incumbent reaction does not explain why disenchanted PDI-P voters did not simply move to a pre-established party. Indeed, an analysis of early polling led some to predict that voters would flock to Golkar, the former ruling party (International Crisis Group 2003, 13). There may have been an anti-incumbent sentiment in 2004, yet this is an insufficient explanation for the increase in party fragmentation. Competing Hypothesis 2: Anti-Establishment Reaction In addition to the anti-incumbent effect, Huntington has also noted an anti-establishment reaction. This occurs when voters not only reject the incumbent, but all parties that have participated in the system during the previous term. The incentive to ride a wave of public dissatisfaction by running against the system adds new players to the political spectrum, potentially fragmenting the party system. Some scholars have pointed to the anti-establishment reaction to explain political developments in Indonesia. A s early as 2002, Paige Johnson Tan (2002, 485) remarked, "Indonesia's political parties have spent much of the goodwill bestowed on them by the population as the proper organizing principle for a post-Soeharto political system; this has led to a significant anti-party reaction." Qodari uses an anti-establishment argument to explain the legislative election results: Dissatisfied with the performance that the activists and leaders of those five major parties had turned in, the electorate told them in effect that they had had the opportunity to rule the country but had failed to use it well . Parties that . symbolized new hopes were the ones to benefits (Qodari 2005, 79). There are two fundamental problems with using the anti-establishment hypothesis to explain the increase in party fragmentation in Indonesia. First, there is no compelling reason why a protest vote against the establishment parties would spread among many 4 parties rather than be consolidated in support of one. This is especially true i f one considers the performance of presidential candidate Yudhoyono, a previously non-partisan figure who created his own party to support his political ambitions. Yudhoyono the presidential candidate did far better than the Democratic Party (Partai Demokrat, PD), his neophyte legislative party; a difference that cannot be explained simply by considering Yudhoyono's presidential alliances. Even i f one accepts that anti-establishment sentiment converged on Yudhoyono in the presidential election, this does not explain why the same anti-establishment sentiment did not converge on his party in the legislative election. Therefore, the anti-establishment hypothesis offers an incomplete explanation of increased party fragmentation. A second problem with the hypothesis is that it may be endogenous to presidential systems. Regarding anti-establishment sentiment, Samuel Huntington notes, "[The anti-establishment response] was more frequent in presidential systems, where candidates for the top office run more on an personal than a party basis; hence it tended to be more prevalent in Latin America, where it was identified with populism" (1993, 266). It may, in fact, be presidentialism itself that channels anti-incumbent sentiment into voting behavior that fragments the party system. Competing Hypothesis #3: Social Cleavages Students of party fragmentation note the rough bifurcation in the literature between those that focus on institutional causes and those that focus of sociological causes to explain the effective number o f parties (Cox & Amorim Neto 1997; Clark & Golder 2006). The sociological approach centres on the number of political issue dimensions that divide a 5 society. The more politically salient issues, the more parties form and win votes, thus causing higher party fragmentation. Operationalizing the concept of social cleavages for comparative study has proven difficult. Previous reviews of the literature identify seven primary issue dimensions on which parties divide. 1 Whether or not these issues are multidimensional can be expected to vary from country to country. Likewise, some issue dimensions overlap. Cross-national work deals with the issue of operationalization by focusing on social heterogeneity, specifically, ethnic heterogeneity. Ethnic heterogeneity is exogenously determined and there is more confidence in the statistical data on ethnicity than in other forms of social identity (Ordershook & Shvetsova 1994, 108).2 Previous research finds a strong correlation between ethnic heterogeneity and the effective number o f electoral parties (Ordershook & Shvetsova 1994; Amor im Neto & Cox 1997; Clark & Golder 2006), which suggests that social cleavages must be taken into account when considering party fragmentation. A consideration of social cleavages cannot account for the increasingly fragmented party system in Indonesia. There was no large-scale change in Indonesia's social heterogeneity between the two elections. N o new issue dimensions were introduced that clearly divided the electorate. A study of social cleavages can reveal the forces that have shaped party fragmentation but it cannot explain a significant change to the Indonesian party system between 1999 and 2004. ' These issues are: 1) socioeconomic; 2) religious; 3) cultural/ethnic; 4) urban-rural; 5) regime support; 6) foreign policy; 7) postmaterialism (Lijphart 1999, 79-87). 2 Stoll.(2005b, 15) has criticized the use of ethnic data, noting that, "lists of ethnic groups are feared by constructivists to reflect a politicized instead of a latent cleavage." Additionally, socioeconomic status and religion are more likely to be politicized in established democracies than ethnicity (Stoll 2005b, 7). Nonetheless, ethnicity is the most common proxy variable used to measure social heterogeneity. 6 Competing Hypothesis 4: District Magnitude District magnitude refers to "the number of seats filled at an election in a district" (Taagepera & Shugart 1989, 19). The number o f districts and the distribution of seats per district vary from country to country. Typically, in countries with multi-member districts, the magnitude varies from district to district. The allocation rules are the method by which votes are translated into seats. The two broad categories of allocation rules are plurality (which awards all seats in a district to the competitor with the highest number of votes) and proportional representation (which utilizes a mathematical formula to award seats based on a competitor's share of the vote). Given that proportional allocation in a district with a magnitude of one would produce the same results as a plurality formula in the same district, as well as the fact that the majority of countries that use plurality allocation rules have single-member districts, most discussions of district magnitude assume some form of proportional allocation. Maurice Duverger is credited for making the theoretical link between party fragmentation and district magnitude. The oft-cited Duverger's Law asserts that plurality systems in single-member districts tend to produce two-party competition. The flip side of this sociological law, known as Duverger's Hypothesis, states that proportional systems tend to produce multi-partyism (Riker 1982). Whereas Duverger refers to allocation rules, Reed (1990) applies Duverger's original insight to district magnitude. Reed argues that multi-member districts tend to produce competition among n + 1 parties (where ' n ' equals the district magnitude). Cox (1997, 139) asserts that M + 1 (where ' M ' equals district magnitude), "imposes an upper bound on the effective number o f 3 Several scholars (Da Silva 2006; Clark & Golder 2006) have recently pointed out that Duverger believed that a permissive electoral system would result in party-system fragmentation only if sufficient politically salient social cleavages exist. The issue of social cleavages is explored below. 7 competitors that w i l l appear in equilibrium." If voters are acting strategically the number of effective parties should not be above the upper bound. In sum, the basic logic of the institutionalist literature is that an increase in district magnitude can cause an increase in the fragmentation of the party system. Expanding on Duverger's work, scholars note two processes through which district magnitude shapes the level of party fragmentation. The first is known as the mechanical effect. The mechanical effect refers to the translation of votes to seats after votes have been cast. Since the calculation takes place after votes have been cast, the mechanical effect directly influences the level of legislative fragmentation rather than the fragmentation of the vote. High magnitude districts are able to convert votes into legislative presence in a more proportional manner than low magnitude districts; therefore it is said that higher levels o f district magnitude typically have lower levels of deviation from proportionality (Taagepera & Shugart 1989). The second process by which district magnitude shapes the level o f party fragmentation is through the psychological effect. Whereas the mechanical effect is directed at the level of party fragmentation in the legislature, the psychological effect is felt before the election and as people vote, thereby shaping the level o f electoral fragmentation. The psychological effect occurs due to the anticipation of the mechanical effect. It effects the decisions of both voters and political elites (Taagepera & Shugart 1989; Blais & Carty 1991; Benoit 2002). Political elites see little benefit in investing in campaigns in which there is little chance of converting their votes to seats. In low magnitude districts there is less chance that an entrepreneurial elite can win a political seat without the backing of an established party. Consequently, low district magnitudes 8 provide incentives for political elites to coalesce into larger electoral parties. Likewise, voters do not want to vote for parties that they do not expect to have a chance of attaining a legislative presence in their district. There is an institutionally induced pressure to choose among the competitive parties in order to avoid wasting one's vote. Thus political elites are expected to avoid districts in which their party would be a lost cause while voters are expected to act strategically by voting for competitive parties. Since deviation from proportionality diminishes as magnitude increases, and psychological effects are based on the anticipation o f potential deviation, one would expect the centrifugal impact of the psychological effect to decrease as magnitude increases. Therefore, as district magnitude increases, the vote should become increasingly fragmented. Numerous cross-national studies confirm the impact of district magnitude on the effective number of electoral and legislative parties (Taagepera & Shugart 1989; Blais & Carty 1991; Ordershook & Shvetsova 1994; Amor im Neto & Cox 1997; Benoit 2002; Clark & Golder 2006). Given the fact that few countries with multi-member districts have a uniform magnitude and that the mechanical effect is supposed to work at the district level, cross-national studies have attempted to operationalize magnitude in a number o f ways. 4 A l l the above mentioned studies return results consistent with the general theory that an increase in magnitude leads to an increase in legislative or electoral fractionalization. Sub-national investigations also confirm the effect of district magnitude. Reed's (1990) study of Japan found that competition in multimember districts tends to stabilize around n + 1 parties. B y looking at the average fractionalization of the vote ordered by 4 The independent variables have included: allocation rules (Blais & Carty 1991); 'effective magnitude' (Taagepera and Shugart 1989); average magnitude (Ordershook & Shvetsova 1994); and the log of the median magnitude (Amorim Neto & Cox 1997). 9 district magnitude, Taagepera & Shugart (1989) offer some evidence that suggests that higher magnitudes lead to a more fragmented party vote within a political system. Cox and Shugart (1996) find support for strategic voting in Colombian and Japanese districts. Similar to the cross-national research, existing sub-national investigations tend to confirm the theoretical predictions regarding the effect o f district magnitude. Though early debates on the subject of party fragmentation pitted those making the institutionalist case against those underlining the effect of social cleavages, cross-national empirical research is led by those that have posited an interactive effect between the two variables. Clark and Golder explain the modifying thesis: [Electoral institutions modify the effect o f social forces on the creation o f political parties. Social forces create more or less pressure for the multiplication of political parties and electoral laws either permit these pressures to be realized or they constrain them by discouraging the formation of new parties (Clark and Golder 2006, 681) Electoral institutions are thought to function as a 'brake' that can slow the level o f party fragmentation that would occur in a socially heterogonous country. Clark and Golder note, "Only a polity characterized by both a high degree o f social heterogeneity and a highly permissive electoral system is expected to produce a large number o f parties" Clark & Golder 2006, 683). A n interactive relationship between district magnitude and ethnic heterogeneity is confirmed by several research projects (Ordershook & Shvetsova 1994; Amor im Neto & Cox 1997; Clark & Golder 2006). Despite the evidence that has supported theoretical predictions about the effect of district magnitude, there are still several cautionary points to make about the conditions which must be met before magnitude w i l l produce the expected political response from voters and elites. Gary Cox (1997) notes the effect of magnitude relies on two 10 assumptions. First, it is assumed that voters know the preferences o f other voters. Second, it assumed that voter expectations about the electoral outcome are generated from common, publicly available sources such that all voters have similar expectations. If these conditions are not met, voters w i l l be unable to vote strategically because they would not know who to abandon and who to support. Cox ' s argument assumes a voter is strategic and, given the right information, w i l l respond to institutional incentives when casting a vote. Reed (1990), on the other hand, argues that an institutional equilibrium is reached through an evolutionary process of voter and elite learning. If a country has had only a limited number of elections voters and elites may not have the experience needed to respond to electoral incentives. Though the arguments of Reed and Cox are distinct, both outline contexts in which voters may not have the theoretically predicted response to magnitude. A n y attempt to explain changes to the Indonesian party system with reference to district magnitude fails because the Indonesian magnitude decreased between the two elections while the theory states that only an increase in district magnitude should cause an increase in electoral fragmentation. It is l ikely that neither o f Cox ' s two conditions for strategic voting were met in both the post-Soeharto elections due to the political uncertainty inherent in any transitional regime. A s well , the small number of post-Soeharto elections has provided a limited opportunity for learning by the elites and the voters. A n explanation of increased party fragmentation in Indonesia must go beyond institutional impact of district magnitude. 11 Presidentialism and Party Fragmentation Direct presidential elections caused increased fragmentation in the Indonesian party system for two reasons. First, the split-ballot allowed the electorate to express a wider variety of preferences. While the 1999 legislative campaign tied the electoral success of presidential hopefuls to the electoral success o f their parties, 2004 allowed voters to choose their preferred leader and party independently of one another. A s a result, smaller niche parties were able to expand their share of the vote. Second, the direct presidential election allowed ambitious presidential hopefuls to look beyond the established party system. This provided some incentive to create personality parties that were unconnected to any pre-established voting constituency or social organization. One of these presidential vehicles, Yudhoyono's P D , was able to trade off of the popularity of its candidate to make notable gains in the legislative election. This section explains why Indonesia's executive-legislative structure in 1999 should be considered parliamentary while the system in 2004 should be considered presidential. It then reviews the debate between those scholars who think presidentialism w i l l consolidate the number of legislative parties in a country and those who think it w i l l lead to a more fragmented party system. The section suggests that the latter 'expansionist' argument offers the most insight into the Indonesian case. In 1999, Indonesia's executive-legislative structure was a hybrid of parliamentary and presidential features. Voters selected party lists to f i l l 462 of 500 seats in a legislative assembly called the Peoples Representative Council (Dewan Perwakilan Rakyat, D P R ) . The remaining seats were reserved for the military. The D P R along with 135 indirectly elected regional representatives and 65 selected individuals representing various interest groups then convened in a 700 member People's Consultative Assembly (Majelis 12 Permusyawaratan Rakyat, M P R ) . The M P R then selected a president and vice-president to serve a fixed 5-year term. A presidential system, as defined by Shugart & Carey (1992, 2), is one in which "the process o f forming the executive is institutionally distinct from the process o f filling seats in the assembly, as both branches are popularly elected." In 1999, the forming o f the executive was not institutionally distinct from the filling of seats in the assembly. Voters had to choose a party, not a president, in a single-ballot election. The executive-legislative reforms that took place between the two elections moved Indonesia from a "quasi-parliamentary" (International Crisis Group 2003, 2) to a presidential system. There is an ongoing scholarly debate on the effects that direct presidential elections may have on the degree of fragmentation experienced in legislative elections. One group of scholars have advanced a 'compression' argument (Amorim Neto & Cox 1997; Golder 2005; Clark & Golder 2006) that claims that the direct election o f the executive wi l l reduce the number o f electoral parties in legislative elections. Another 'expansionist' school (Fillipov, Ordeshook & Shvetsova 1999; Clark & Wittrock 2005; Samuels 2002) notes the ways in which direct presidential elections can increase the fragmentation of the electoral vote. The compression argument proceeds as follows. Political elites, voters, and the national media commonly assume that the office o f the president is the most important electoral prize. Presidential candidates become the centre of attention and the symbol o f their respective parties, creating an incentive for legislative candidates to organize their campaigns around their national figurehead. Voters use information from the presidential race as a short-cut for making a decision in the legislative elections. These instances of 13 'straight-ticket' voting are believed to depress the level of fragmentation that would have occurred i f the election was not dominated by a presidential competition. There are a few caveats to the compression argument. The first involves the timing of elections. Literature on the American electoral cycle notes the division between concurrent and nonconcurrent legislative elections (Shugart 1995, 328). 5 In the former, the public votes for the executive and legislature candidates at the same time; in the latter, only the legislative candidates appear on the ballot. The 'coattail effect' is assumed to be sensitive to the temporal proximity of the two elections: in a concurrent election a popular candidate may lift her party's fortunes in the legislative contest, thus consolidating the effective number of parties, while a nonconcurrent election w i l l not produce the same pressure for straight-ticket voting, thereby opening the possibility for a more fragmented vote. In empirical studies of the coat-tail effect, some scholars have used a simple institutional dummy (Mainwaring & Shugart 1995) and others a more sophisticated measure of proximity (Shugart 1995; Amor im Neto & Cox 1997; Golder 2005; Clark & Golder 2006). A l l have found support for a time-sensitive coat-tail effect. Another aspect of presidentialism that has drawn attention is the electoral formula. It has been argued that, all things being equal, plurality elections should have a more consolidated legislative vote than majority runoff systems (Mainwaring & Shugart 1995; Jones 1999). In the case of plurality, the winner-take-all competition encourages strategic withdrawal by elites and strategic voting by the public. The two-way competition thus creates a strong coattail effect, thereby reducing the effective number of electoral parties. Majority runoff elections, on the other hand, provide elites with an 5 As Shugart notes, the American literature refers to these as 'on-year' and 'off-year' elections. 14 incentive to run in the first round and potentially use their second round endorsement as leverage in a coalition with one of the front-runners. Amorim Neto and Cox (1997) question whether the electoral formula has any distinct impact on the fragmentation of the vote. They argue that the key variable is not the formula itself, but rather the effective number of presidential candidates. A majority runoff competition that happens to produce a two way race should have the same level o f coattail effect as a plurality race with an equal number of viable contenders. While a majority runoff formula may be more likely to produce a higher number of effective presidential candidates than a plurality formula, according to Amor im Neto and Cox, it is unnecessary to consider the systems distinct. Subsequent empirical work endorses the use of presidential fragmentation as an important independent variable for the determination of the effective number of electoral parties (Mozaffar, Scarritt & Galaich 2003; Golder 2005; Clark & Golder 2006). The compression theory is challenged by a number of regional and case studies. Fil l ipov, Ordeshook & Shvetsova (1999) have argued that presidentialism may in fact increase the effective number of electoral parties. In an adaptation o f the coattail concept, the authors suggest that a desire for national visibility provides incentives for even marginally competitive elites to enter the presidential race. Party proliferation is encouraged "since forming and heading parties provides presidential aspirants with the clearest path to that office and to the [sic] national political visibility in a transitional democracy" (Fillipov, Ordeshook & Shvetsova 1999, 18). Alternatively, Clark & Wittrock (2005) suggest that increased fragmentation occurs in presidential systems due to the strength of the executive office. In countries with strong executive powers, there is 15 less incentive for the president's party to control the legislature. The indifference of the major parties leaves an opening for smaller parties and independent candidates to increase their vote in legislative elections. Though the two approaches offer different causal mechanisms, both are attempting to explain why the electoral vote is vulnerable to fragmentation in presidential systems. David Samuels argues that presidential and parliamentary systems produce different incentives that influence party behavior (Samuels 2002). Samuels suggests that the influence o f the institutions can be discerned by comparative study using Kaare Strom's model of party behavior (Strom 1990). Strom's model assumes that parties pursue three goals: votes, office, and policy. Parties must make trade-offs in the pursuit of these goals. Vote maximizing parties that seek to control government must appeal to a broad section of the population and they may have to make policy sacrifices to do so. A n office-maximizing party may pursue the benefits of office "over and above their electoral or policy value" (Strom 1990, 567). Policy-seeking parties typically have longer time horizons and a less wil l ing to trade policy influence for votes or political office. Though parties are concerned with all three goals, they typically neglect one goal for the benefit of another.6 Strom's original work centers on explaining party behavior in parliamentary systems. Samuels points out that there should be significant differences in party behavior between presidential and parliamentary regimes. The main cause o f this difference is the influence of the presidential competition. Winning a presidential competition requires a candidate to appeal to a broad spectrum of the electorate. Given the winner-take-all 6 There is, of course, a temporal aspect to a party's behavior. For example, a small party may portray itself as policy-seeking before an election only to pursue an office-seeking during coalition negotiations. One would assume that such a strategy would cost them votes in the next election. 16 nature o f the presidential race, parties have a stronger incentive to follow a vote-seeking strategy. The vote-seeking strategy of the parties with viable presidential candidates affects all o f the parties in the system. Samuels explains: [P]arties running only legislative candidates do not simply revert to a parliamentary parties' world, because the presence of the presidential election affects all parties' strategies in ways parliamentary elections do not: Their strategies in both the presidential and legislative races are affected by what the larger parties do, and they in turn can affect larger parties (Samuels 2002, 469). The vote-seeking behavior of the 'presidential' parties allows the smaller parties to run a niche policy- or office-seeking strategy either with or against one o f the major parties. Thus there are solid theoretical grounds for assuming different patterns o f party behavior in parliamentary and presidential regimes. To illustrate his theory o f presidential party behavior, Samuels looks at developments of party behavior in France and Israel, two parliamentary countries that have adopted direct executive elections.7 There are a number of important findings. First, viable presidential candidates stood above and apart from their respective parties. Policy was largely dictated by the candidate. The presidential parties acted as vehicles for the vote-seeking candidate. France has even witnessed the emergence o f parties formed for the primary purpose of supporting a presidential hopeful. 9 Second, in Israel, a country with a high district-magnitude, the smaller parties responded to the vote-seeking strategy of the presidential parties by encouraging their supporters to split their tickets. This allowed them to f i l l niche positions that were neglected by the vote-seeking presidential 7 Israel's hybrid system allowed for separate legislative and prime ministerial elections yet continued the requirement that the executive be responsible to the legislature. For matter of simplicity, however, the following discussion will simply refer to Israel as a presidential system. 8 In France, this has earned them the derogatory title of 'parti de godillots' or 'party of bootlickers' (Samuels 2002, 475). 9 Giscard's Union for French Democracy is Samuel's prominent example of the exclusively presidential party. 17 parties. Samuels' argument is similar to Clark and Wittrock's in that both argue that 'presidential' parties prioritize the executive election at the expense of the legislative election. In addition to party prioritization of the executive contest, however, Samuels' focus on the success of sectarian parties in Israel also suggests that separate executive and legislative elections may change voter preferences in ways that can increase electoral fragmentation. A s w i l l be shown, the expansionist arguments regarding the effects of direct presidential elections effectively explain increased party fragmentation in Indonesia. Methodology This section defines the key dependent variable of interest, describes how the thesis proves the effect of presidentialism on the dependent variable while showing competing hypotheses to be inadequate, and outlines the limitations of the research design. The use o f electoral fragmentation rather than legislative fragmentation to measure the dependent variable is justified on the grounds that the key independent variable of interest -presidentialism - effects a voters decisions but not the translation of votes to seats. Next, the section describes how the decisive impact of presidentialism on the electoral fragmentation is shown using a mix of thick description and statistical analysis. Following this, there is a summary of the statistical techniques used to prove that the change in district magnitude had no effect on electoral fragmentation. Last, the section explains why the research design does not allow the thesis to definitively rule out an interaction effect between presidentialism and anti-incumbent or anti-establishment sentiment. 18 Dependent Variable The standard method o f measuring party fragmentation is to calculate the effective number of parties. The effective number of parties in a political system can be attained by inversing the Herfindahl-Hirschman concentration index for either the number o f seats a party wins or the number of votes it receives (Taagepera & Shugart 1989). 1 0 The latter measurement is commonly referred to as the effective number of electoral parties (ENEP) while the former is called the effective number o f parliamentary parties (ENPP). A high number indicates a high level of fragmentation within a party system while a low number indicates a high degree of concentration. This research project is primarily interested in the E N E P . While district magnitude is expected to affect both E N E P and E N P P , the executive-legislative structure does not affect the translation of votes into seats. O f course, one of the reasons E N E P is studied stems from the fact that it has significant impact on E N P P . A comprehensive analysis o f the effects of institutional change requires that both fragmentation measures be considered; however, much of the data that the thesis examines is district level, thus making E N E P a far more revealing variable than E N P P . Unless otherwise stated, all discussion of fragmentation and concentration in this thesis w i l l refer to E N E P . A clear increase in electoral fragmentation is revealed by calculating the national E N E P for 1999 and 2004. Calculating the district level E N E P in both elections shows an increase in electoral fragmentation across the country. Chapter two describes the change 1 0 The Herfindahl-Hirschman concentration index (HH) is attained by summing the squared values of each components share of an entity. The mathematical expression for H H is: H H = S(s,) 2 where s,- is the entity proportion for component /. Therefore the expression for the effective number of parties is: E N P = 1/1(5,)2 where si is the proportion of seats or votes for party /. 19 in the dependent variable in further detail. For the purposes of this section it can simply be noted that there was a significant change in the dependent variable that strongly suggests that it was not merely caused by stochastic factors. Independent Variable A brief review of executive-legislative reform in Indonesia, as presented in Chapter 3, establishes that there was a change in the presidential variable between 1999 and 2004. To prove that this variable affected electoral fragmentation the thesis combines thick analysis with statistical tests in a research design Collier, Brady & Seawright (2004) term "nested inference." The intense focus on Indonesia allows for a sizable number of observable implications to be drawn from the theoretical argument. In doing so the thesis highlights the causal mechanisms through which the executive-legislative reform resulted in higher party system fragmentation and explains why this change cannot be explained simply by an anti-incumbent or anti-establishment sentiment in the electorate. Direct presidential elections caused increased fragmentation in two ways. The first cause was the introduction of presidential electoral vehicles. Using media reports and secondary sources, the thesis describes the formation and limited success of Yudhoyono's Democratic Party. With no connection to any major social organization or political machine, the party is best understood as a platform for Yudhoyono's presidential aspirations. Comparing the electoral vote o f Yudhoyono's presidential vehicle and Yudhoyono the presidential candidate shows that the latter had a far larger vote-share than the former, thus proving that P D had only limited success at riding the coat-tails of presidential candidate. Using district-level data from the 1999 and 2004 elections shows an increasing linear trend between P D vote share and PDI-P losses. Whi le this suggests 20 that P D may have benefited from an anti-incumbent vote, it also shows that the majority of those who abandoned the governing party did not support the party of the popular anti-incumbent presidential candidate. Though Yudhoyono the candidate was able to consolidate an anti-incumbent vote his legislative party failed at this task. The second mechanism that fragmented the party system was the growth of niche parties that was facilitated by split-ticket voting. Using secondary sources and results from the 2004 election, the thesis argues that Indonesian voters tended to support a presidential candidate ticket that was strong on the national unity issue while preferring a legislative party that could express a niche sectarian or ideological position. Existing survey data is used to confirm that Indonesian voters supported their preferred party in the legislative election and their preferred leader in the presidential election. A study of the Islamist Prosperous Justice Party (Partai Keadilan Sejahtera, P K S ) is used to demonstrate the argument that split-ticket voting increased electoral fragmentation. The P K S was able to pick up Islamist votes from the United Development Party (Partai Persatuan Permbangunan, PPP) and the National Mandate Party (Partai Amanat Nasional, P A N ) , both of whom were involved with presidential politics. This migration can be shown using district level data from the 1999 and 2004 elections: there is a clear linear trend between P P P - P A N losses and P K S gains. This was not simply an anti-establishment reaction however. P K S voters clearly supported P A N ' s presidential candidate Amien Rais, a fixture of Indonesian politics since the late Soeharto era. Plotting Rais 's district level presidential vote against P K S ' s 2004 legislative vote shows a positive linear trend, a phenomenon supported by media reports and secondary case literature. The P K S was indicative of a movement that saw voters support increasingly 21 narrow legislative parties while simultaneously supporting presidential candidates that sought to transcend major ideological and social cleavages. Having explained what evidence wi l l be presented to support the causal relationship between direct presidential elections and increased party fragmentation, the section moves to consider the evidence used to rule out the competing hypotheses. Competing Hypotheses It is suggested above that, in addition to executive-legislative reform, there are four competing hypotheses that could explain fragmentation in Indonesia. A detailed investigation on the effects of presidentialism reveals that the anti-incumbent and anti-establishment are inadequate stand-alone explanations for increased electoral fragmentation. There is no evidence that a major change in Indonesia's social cleavage structure (as popularly defined by party-system fragmentation scholars) took place between the two elections. Given that a substantial body of literature suggests that district magnitude has an important role in determining the effective number of electoral parties, this thesis focuses on proving the non-effect of this variable on the electoral vote. To demonstrate that the changes to magnitude did not impact the 2004 legislative vote the theoretical argument is broken down into four parts: 1) the role o f strategic elite behavior; 2) the direct impact on the electoral vote; 3) the role of voter learning from previous elections; 4) the 'braking' impact on the expression o f existing social cleavages. Regarding the first point, theory suggests that elites should create new parties when electoral conditions are permissive and there is a chance o f winning; on the other hand, they should withdraw parties when electoral conditions become more restrictive and there is a reduced chance of winning legislative seats. At the national level this can 22 be tested by comparing the number o f competing parties by the average district magnitude. A s well , elite learning can be tested though a comparison o f district magnitude and the number of competing parties at the district level. Theoretically, there should have been higher rates of strategic elite withdrawal in 2004 because the parties should know their regional bases of support and understand the effect of district magnitude on the chance of winning a seat. Very little evidence is found to support to the theory that party behavior was influenced by elite awareness of district magnitude. The second point can be tested using elections data from the 1999 and 2004 elections. If district magnitude effected a voter's decisions, then district level E N E P should increase with magnitude. Surprisingly, 2004 saw a significant number of districts with an E N E P higher than the 'upper bound,' the maximum effective number of parties Cox (1997) predicted would exist in a state o f electoral equilibrium. Results in neither election conform to the predicted positive linear trend between magnitude and E N E P . Reed (1990) has argued that the effect o f magnitude on voter choice is an evolutionary process. It could take a number o f elections to learn the effects of magnitude and adjust behavior accordingly. The thesis discounts any learning effect by plotting 1999 district level deviation from proportionality against the change in the E N E P between the two elections. 1 1 Reed's theory suggests that high levels of deviation in 1999 should have prompted voters to spread their voters among fewer parties in 2004. In Indonesia, however, the districts with high levels o f deviation in 1999 had the highest increase in their E N E P . 1 1 Taagepera & Shugart (1989) provide a general formula for deviation from proportionality D = ( l / 2 ) I | S i - V j | where s, is the percentage of votes and v j is the percentage of seats for the i-th party. Because the sum of all deviation must equal 0 the equation divides the sum of the absolute values. 23 Testing for a possible interactive effect between magnitude and different social cleavages on district level fragmentation demands a statistical procedure that goes beyond simply observing a linear trend between one independent variable and one dependent variable. Fortunately, the high number of electoral districts in 2004 ensures that an ordinary least squares regression analysis can be used to determine the independent effect that multiple social variables had on electoral fragmentation. Two models of district fragmentation are proposed. One tests the effect of magnitude and social variables with no interaction. The model assumes that social cleavages cause electoral fragmentation independently of district magnitude. The second model tests for an interactive relationship between social variables and magnitude. It assumes that the fragmentation caused by social variables is greater when the electoral system is more permissive. The analysis finds no evidence o f any interaction effect, which strongly suggests that magnitude did not affect the 2004 electoral vote. Limitations of the Research Design Indonesia provides a rare natural experiment where the effects of executive-legislative structure on party fragmentation can be explored. There are two primary observations on this variable: party fragmentation before and after the introduction direct presidential elections. A possible anti-system/anti-incumbent sentiment, however, is directly correlated with executive-legislative reform: it did not exist in 1999, but did exist in 2004. Though the thesis argues that neither anti-incumbent nor anti-system arguments can effectively explain the change in the dependent variable, a possible interaction effect between executive-legislative reform and one of the 'frustrated electorate' hypotheses 24 cannot be completely discounted. While the switch to direct presidential elections would likely have caused an increase in party fragmentation regardless of the presence or non-presence of a frustrated electorate, the research design does not provide sufficient evidence to support the argument that all o f the change in party fragmentation was caused by the executive-legislative reforms. The structure o f the thesis proceeds as follows. Chapter 2 describes the changes to Indonesia's district magnitude that took place between 1999 and 2004 and shows that the changes did not affect the 2004 electoral vote. Chapter 3 describes the changes to Indonesia's executive-legislative structure and explains the two ways in which these changes increased electoral fragmentation. Chapter 4 considers the significance of the findings and highlights a number of areas that require further research. . ' 2 For a discussion of small-N studies and interaction effects, see Lieberson (1991). 25 Chapter 2 : District Magnitude and Electoral Fragmentation This chapter argues that district magnitude reform between 1999 and 2004 had no impact on the effective number of electoral parties in the 2004 elections. The chapter proceeds as follows. First, electoral reforms in the post-Soeharto era are reviewed. Next, brief consideration is given to the mechanical effects of the electoral system. Third, the chapter shows that there was no clear effect of magnitude change on elite strategies by looking at the number of competing parties. Fourth, a simple plotting of magnitude and effective number of electoral parties shows no sign o f strategic voting behavior by the Indonesian electorate. Fifth, the possibility of voter learning is discounted by plotting 1999 levels of deviation and changes in the effective number o f electoral parties. Finally, the chapter presents a regression analysis that checks for a possible interactive effect between district magnitude and prominent social cleavages, thereby eliminating the possibility that magnitude acted as a 'brake' on social heterogeneity. A Review of Electoral Reforms Throughout the New Order, the ruling Golkar party used manipulated elections to help legitimate its rule in a show o f organizational strength. Each province was designated a given number of legislative seats that were divided using a list system. The fall o f the Soeharto regime and the transition to democracy opened the question of electoral reform. Noting the lack o f accountability between New Order politicians and the communities they were purported to represent, some reformers proposed the introduction of mixed system of primarily single-member districts with a reserved number o f seats to correct for proportionality (King 2000). These proposals were rejected by opposition forces, in part 26 for fear that local elections were more susceptible to manipulation by well financed party machines. The system that was put in place for the 1999 election was an awkward compromise between P R and local representation. Rather than presenting a provincial list, parties assigned candidates to represent regencies. Parties were awarded seats on the basis of the provincial vote; however, the candidates that filled these positions would be determined by the performance of the party at the regency level (King 2003, 56). There was widespread abuse of the system as parties sought to maintain control of seat assignment. Additionally, the large number of seats granted to some provinces diluted perceptions of direct accountability. The regency representation design did not perform as intended. In 2003, the D P R enacted two electoral reforms to address the perceived lack of legislative accountability. The first was the replacement of the regency-representation design with an open-list system. Theoretically, the voters were given the ability to determine the order by which party members would be allocated seats in their district. Yet international observers reported, "[T]he DPR...significantly eroded the actual impact of this innovation by adding a clause that requires a candidate to achieve a quota of votes in order to be elected under the partial open-list system" (NDI 2004b, 4). In the end only two candidates reached the quota and in both cases they would have been elected anyway. The second electoral reform involved strengthening the voter-representative link by decreasing the size of the electoral districts. The D P R provided the Indonesian 1 3 Tan (2002, 493-4) points out that both PDI-P and Golkar, the two largest parties, required pre-election letters of resignation from their candidates so as to allow the parties to direct the allocation of their seats. 27 Electoral Commission {Komisi Pemilihan Umum, K P U ) with a number of guidelines to follow in the re-districting process. First, each province was assigned a minimum of three seats. Second, no province was to be assigned fewer seats than it had in 1999. 1 4 Third, districts were assigned between three and twelve seats. Fourth, average population should vary between 325,000 and 425,000 residents per representative. Finally, the K P U was to accomplish this task while remaining sensitive to traditional patterns of local interaction. Table 2.1 presents a summary o f the effects of the K P U ' s redisricting efforts. The number of elected seats was enlarged between the two elections. The size o f the D P R itself was increased from 500 to 550. A s well , the reserved military seats were distributed among the districts. The average magnitude in 2004 was slightly less than half of what it was in 1999. Java witnessed the most significant change. Its number o f districts ballooned from 3 to 35, while its average magnitude shrunk from 46.8 to 8.7. The change in the outer islands was comparatively modest, although their total share of the elected seats fell from its previous share of just under 50% to 45% in 2004. Year # of Districts Average Magnitude Seats 1999 Total 27 17.1 462 Java 5 46.8 234 Outer Islands 22 10.4 228 2004 Total 69 8.0 550 Java 35 8.7 303 Outer Islands 34 7.3 247 This caused confusion and protest from the Malukus when the KPU interpreted that the requirement was met if the seats of new provinces and the seats of the parent province combined to equal the 1999 total. 28 Mechanical Effects Before considering the effect of district magnitude on the effective number of electoral parties it's worth looking briefly at the mechanical effect in both elections. The mechanical effect concerns the translation of votes into seats. At the national level it can be gauged through a comparison of the effective number of electoral parties and the effective number of parliamentary parties. A decrease in average magnitude would be expected to produce an increase in the difference between E N E P and E N P P . Table 2.2 presents a summary of the mechanical effect in both elections. In 1999, the electoral system was close to proportional in the translation o f vote into seats. There was only 0.34 less parliamentary than electoral parties, a modest difference. This value jumps significantly in 2004. The expansion of the E N E P outpaced the expansion o f the E N P P . Table 2.2 - Mechanical Effects Year Average Magnitude ENEP ENPP Difference 1999 17.1 5.06 4.72 0.34 2004 8.0 8.55 7.07 1.48 2004 Hypothetical* 17.6 8.55 7.85 0.71 *The secession of East Timor causes the change in average magnitude Sources: 1999 results compiled by IFES. 2004 data available online: http://www.kpu.go.id/suara/hasilsuara_dpr_sah.php. A l l data available upon request. Only part of the increase in disproportionality can be explained by redistricting. Even the higher magnitude of the 1999 system would have had difficulty translating such a fragmented vote. Given that there is an interactive relationship between district magnitude and the choice of a voter, there is no way to make an exact prediction o f the 2004 vote i f there had not been any changes made to the electoral districts. Given that the 1999 electoral system was more permissive, the 2004 E N E P likely would have been higher i f redistricting had not occurred. Nonetheless, reconstructing the 2004 vote according to the 1999 districts gives a rough idea of the change in the mechanical effect. 29 The hypothetical 2004 results show that, even i f there had been no redisricting, there would have been a marked increase in disproportionality. The reduced average magnitude in 2004 played an important role in slowing the growth of the E N P P . The Number of Contesting Parties One way to measure the psychological effect of an electoral system is to count the number o f competing parties. Elites should be more wil l ing to start political parties in permissive systems because there is a reasonable possibility o f attaining seats in the legislature. Blais and Carty (1991) find that more parties compete in P R systems than S M D systems. Likewise, Taagepera and Shugart (1989) find that, in Finland, the more permissive the district the more parties w i l l compete. Three hypotheses regarding the number o f competing parties in the 1999 and 2004 elections in Indonesia can be generated from this body of case studies and cross-national work. First, the decrease in the national average magnitude should have led to fewer parties entering the electoral contest due to the decreased permissiveness of the electoral system, and thus the decreased probability of winning a seat. Thus the following hypothesis is proposed: Hypothesis 1: Between 1999 and 2004 there should be a decrease in the number of parties contesting the national legislative elections The immediate evidence supports Hypothesis 1. There were 24 competing parties in 2004, half as many as there were in 1999. However, other important factors played a role in reducing the number o f competing parties in Indonesia. First was a tightening of the ballot access rules. Party registration laws were lax in 1999, necessitating merely fifty signatures and a review by the Ministry of Justice to ensure that the political ideology did not contradict Pancasila, the official state ideology. Ballot-access required parties to prove that they had organizations in one-third of all provinces and one-half of all 30 regencies in these provinces. For the 2004 election, the Ministry o f Justice and Human Rights only registered parties that could prove they had organizations in half the regencies in half the provinces, a more difficult demand than the previous ballot access rules. To make it on the ballot, parties had to have organizations in two-thirds of all provinces and two-thirds of all regencies within the provinces. 1 5 Thus the 2004 elections required a greater organizational capacity on the part of competing parties. A second factor that decreased the number o f competing parties was the increased time allotted to the K P U in 2004 to verify that parties had met registration requirements. In 1999, the laws that dictated ballot access rules and set the timing of the election for June 7 were not passed until late January. Party registration was to be completed by late March. The short registration period and the enormity of the task created difficulties in verifying the organizational claims of the parties. Nurcholish Madjid, the head of the verification team, did not have the resources to do a thorough verification before the deadline. A s Kev in O'Rourke (2002, 214) explains, "Nurcholish's team therefore opted for the least controversial alternative: allowing a larger than expected number of parties to pass." In contrast, the 2004 legislative elections, which were held on Apr i l 5, had a registration process that began in July of 2003 and ran through to the end of December of that year. The increased amount of time and organizational experience undoubtedly led to a more thorough screening of the parties. It is l ikely that the additional registration period allowed the K P U the time it needed to conduct thorough investigations and bar parties that may have made it onto the ballot in 1999. 1 5 Parties that received 2% of the seats in the DPR or 3% of the seats in regional legislatures spread over half the regencies in half the provinces. 31 A third relevant change in the elections was the campaign finance rules. In 1999, all competing parties were guaranteed an equal share of public funds to help cover election expenses. A s K i n g (2003, 52) notes, "This probably served as an incentive for party formation, especially since the law took effect during a time of severe economic contraction." In 2004, funds were disbursed according to a party's percentage o f the 1999 vote, reducing the incentive for a half-hearted, cash inspired political campaign. Fourth, the transitional nature of the 1999 election created a mix of uncertainty and enthusiasm that l ikely caused a temporary increase in the number of aspiring parties. Two hundred political parties were formed in the period following Soeharto's departure (Tan 2002, 485). The proliferation occurred in what Bilveer Singh referred to as, "[A] background of increasing euphoria surrounding the downfall of the largely authoritarian New Order regime" (Singh 2003, 433) Those parties that made it onto the ballot soon discovered the extent of their popularity. The decline of democratic enthusiasm probably played a role in contracting the number of competitive parties. Given the combination o f factors that may have inflated the number of contesting parties in 1999, it is too difficult to determine the independent effect that the decrease in average magnitude had on the number of contesting parties in 2004. When all o f the developments between 1999 and 2004 are considered, what becomes surprising is not the decrease in the number o f parties that gained access to the ballot but the fact that over one-hundred parties registered with the Ministry o f Justice and Human Rights in hopes o f competing in 2004 (Kurniawan 2003). The second hypothesis refers to the number o f parties that contest the election at the district level. Once a party was cleared to run in the national election it was not 32 required to run a list in every district. Given that many districts experienced a decrease in magnitude, as well as the fact that the number o f small districts expanded, there should have been higher rates of strategic withdrawal in 2004. This leads to the second hypothesis: Hypothesis 2: There should be an increased percentage of parties -withdrawing from districts in 2004 Additionally, parties can be expected to withdraw from small magnitude districts in which there is a smaller likelihood o f winning a legislative seat. Thus the third hypothesis states: Hypothesis 3: The number of competing parties should increase with the district magnitude Hypothesis 2 is particularly easy to investigate. In 1999 all 48 parties competed in all 27 districts. A n y evidence o f district level withdrawal in 2004 would support the hypothesis. Indeed, 2004 did not see the same 100% participation rate by all the parties. Table 2.3 presents a summary o f district withdrawal: District Name District Magnitude Number of Competing Parties West Irian Jaya 3 21 East Nusa Tenggara 2 7 22 East Java 9 11 23 Though the evidence does support hypothesis 2, some caution is needed. The number of electoral districts more than doubled between 1999 and 2004. In total, four parties accounted for the six instances of withdrawal. This still shows a high level of commitment to run in as many districts as possible. That being said, there is some weak confirmation for hypothesis 2. 33 The evidence for hypothesis 3 is mixed. The fact that Partai Pelopor ran in all the districts in the province o f East Java except for East Java 9 suggests that the cause of the withdrawal was administrative rather than strategic. Both of the competitors that chose not to submit a candidate list for East Nusa Tenggara 2 were small Mus l im parties. 1 6 East Nusa Tenggara has one of the highest concentrations o f Christians in Indonesia. This withdrawal was likely strategic, but had little to do with district magnitude. The three instances o f withdrawal from West Irian Jaya may have been related to the low district magnitude. The district is far away from the administrative centre o f Jakarta and has a high concentration of non-Muslims, making it a potentially unappealing district in which to compete. However, all 24 parties competed in the adjacent district of Papua, which has both o f the drawbacks of West Irian Jaya, yet has a district magnitude of 10. Again, there is some evidence that weakly confirms hypothesis 3. Magnitude and the Effective Number of Electoral Parties In addition to political elites, magnitude is also said to effect voters. To avoid 'wasting' their vote, voters may choose to forgo casting their ballot for their most preferred party and instead vote for their most preferred competitive party. The larger the magnitude, the larger the likelihood that a voter's most preferred party w i l l be competitive in a district. Thus E N E P should increase with magnitude. Taagepera and Shugart (1989) have suggested a simple method o f measuring this theoretical relationship. They found the average E N E P for a given magnitude and sorted the results by magnitude. A modest increasing linear relationship was found in both 1 6 The traditionalist Partai Persatuan Nahdlatul Ummah Indonesia (PPNUI) and the modernist Partai Bintang Reformasi (PBR). 34 Venezuela and Finland (Taagepera & Shugart 1989, 119). This leads to the following hypothesis: Hypothesis 4: The ENEP in a district should increase with magnitude . Figure 2.1 and 2.2 show E N E P plotted by each district in 1999 and 2004 respectively. In addition to E N E P , the figures also display Cox ' s (1997) 'upper bound' (Magnitude + 1). This shows the maximum number of effective parties that one would expect to see in a state of equilibrium. Figure 2.1 - Magnitude and ENEP in 1999 Figure 2.1 - Magnitude and E N E P in 1999 y = -0.0122X + 4.4599 R 2 = 0.0028 24 20 16 Q . U l u_ 12 8 4 0 •Mil 111 jjijj ll Sill i i i i i illl • 4-i l l • • 0 . 5 10 i 15 20 25 30 Magnitude Three high magnitude districts are excluded from the graph. The values are as follows. West Java: Magnitude 82, ENEP 5.06; Central Java: Magnitude 60, ENEP 4.03; East Java: Magnitude 68, ENEP 3^ 83. 35 Figure 2.2 - Magnitude and ENEP in 2004 z 8 -UJ o> Q. LU 0) 12 14 10 0 -I 2 0 2 4 6 8 10 12 14 Magni tude There appears to be no^relationship between magnitude and E N E P in 1999. Save for one district, Bengkulu, all are below the upper bound. Surprisingly, 2004 shows a . negative relationship. There were 15 districts, over 20% of the total, with an E N E P above the upper bound. This strongly suggests that magnitude prompted no strategic voting. Figures 2.3 and 2.4 present the same procedure sorted by region. In Java, there is some weak evidence to support hypothesis 4. Only one West Java district exceeded the upper bound. There is no apparent relationship between magnitude and E N E P in the outer islands. Over 40% of the districts fell above the upper bound. The results indicate that there was no uniform psychological effect produced by the district magnitude. The reduced district magnitude did not create an immediate pressure to consolidate the vote. Therefore hypothesis 4 can safely be rejected. 36 Figure 2.3 - Magnitude and ENEP in Java 14.00 12.00 10.00 8.00 CD E 6.00 V > < 4.00 2.00 0.00 Figure 2.3 - Magnitude and ENEP in Java y = 0 . 2 6 8 7 X + 3.8142 R2 = 0.1601 HflBBMHHHMSNSf • B M l N M i NRlill • • i * ? n i l 1—* — I • # • lljiplllillilil •m 10 12 Magnitude 14 Figure 2.4 - Magnitude and ENEP in the Outer Islands Figure 2.4 - Magnitude and ENEP in the Outer Islands y = -0.0241X + 8.1822 R2 = 0.0008 O-LU 14.00 12.00 10.00 8.00 2 6.00 > < 4.00 2.00 0.00 S H I IBHI ffplttlli»l j i i l i i i i i i •pill o ^ ^ ^ ^ • ••' i§§ll| lill "••liiliif iil!i|*IIl^ f^cs » , A - - , , - - v : ' r i " V ^ • - -i • B I B ill 6 8 Magnitude 10 12 14 37 Voter Learning Though there does not appear to be any evidence to support an immediate psychological effect, there is still a possibility that it could start producing consolidating pressure in the near future. A s Reed (1990) and others have pointed out, voters may need time to learn the mechanical effects of the electoral system. After learning which parties are and are not competitive in a district voters are expected to act strategically in the future. One way o f studying the learning effect is to consider the impact of deviation from proportionality in the previous election. Deviation indicates a high number of 'wasted' votes. A t the district level, high deviation is almost inevitable at low district magnitudes. A low E N E P would be expected in these districts however. High levels of deviation indicate that there should be incentive to reduce the E N E P . If there was a lagged learning effect in Indonesia, then those districts that experienced high levels of deviation in 1999 would be expected to see more consolidation (or less expansion) of E N E P in 2004. A positive finding would also indicate that the 2004 reductions in magnitude should have a psychological effect on the upcoming 2009 elections. W e thus arrive at the following hypothesis: Hypothesis 5: The higher the level of deviation in 1999, the larger the decrease (or the smaller the increase) in ENEP in 2004. Deviation is calculated at the 1999 district level . 1 7 To determine the change in the E N E P the 1999 vote was reconstructed according the 2004 districts. 1 8 This 1999 E N E P was then subtracted from the 2004 E N E P . The difference provides a measure o f the level of expansion or consolidation of the E N E P . According to the hypothesis, the difference 1 7 As noted above, the 1999 districts corresponded with provincial boundaries. 1 8 There are some minor discrepancies in the calculations. See Appendix A for discussion. 38 should be smaller in those provinces that experienced high levels of provincial deviation in 1999. Figure 2.5 - 1999 Deviation and ENEP Change Figure 2.5 -1999 Deviation and ENEP Change „_ „„„ y = 12.188x + 1.4203 8 j~ -R2 = 0.2856 7 6 -o • lange (2( 5 -SI 4 - -• + ^ — • o CL 111 3 Mm " . v > ^ £ | j p ~ " | z LLI 2 -\ 1 -HH 0 -V ' llPSiSIIIllili 1 1 t 1 , 1 , 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Provincial Deviation (1999) Figure 2.5 plots the E N E P change against the 1999 deviation. There is no evidence that a high level o f deviation in 1999 caused a consolidation of the E N E P . In fact, those districts that were in a province with high 1999 deviation had a greater level of E N E P expansion in 2004. There are clearly factors beyond magnitude that are compelling high levels o f electoral fragmentation. Magnitude and Social Cleavages The first section of this chapter considered magnitude as a stand alone factor. Recent research on the E N E P , however, has focused on making the connection between magnitude and social cleavages. Parties reflect divisions within a society. Institutions can provide different levels o f pressure to contract the electoral competition. Yet studying magnitude in isolation may not accurately capture the effect it has within a country or 39 district. This section introduces a number o f Indonesian social cleavages that may interact with magnitude to determine the E N E P . The social cleavage receives the most attention from institutionalists is ethnicity. The findings support the argument that the expansionary effect of ethnic fragmentation on E N E P is higher in permissive electoral systems. Though most of the research thus far is cross-national the theoretical argument can easily be transported to the sub-national level, where the mechanical effects of district magnitude are more definitively witnessed by voters. Indonesia is a country with a significant level of ethnic fragmentation. A t 42% o f the total population, there is a strong plurality of Javanese. However, the vast majority of groups have under 5% of the national population (see table 2.4). Politically, there is a salient division between the ethnic Javanese and the rest. Though ethnicity is a sensitive subject, to the point that the Indonesian State refused to publicly collect ethnicity data for the first 55 years o f its independence, it remains a factor that shapes electoral behavior in important ways. Suryadinata, Ar i f in , and Ananta have done work connecting ethnicity and regional expressions of party support, for example the high level of support enjoyed by the PPP among the Madurese in East Java (Suryadinata, Ar i f in , & Ananta 2003, 181). K i n g (2003), noting the low 1999 E N E P in Bal i and South Sulawesi, suggested an ethnic coat-tail effect that led regionally concentrated ethnic groups to offer solid support for leaders connected or sympathetic to their ethnic groups. 1 9 Thus there is research to suggest that ethnicity may affect E N E P . 1 9 Megawati, head of PDI-P, had a Balinese grandmother. Golkar's Habibie was of Buginese and Makasaree descent. The PDI-P received strong majority support in Bali, as did Golkar in South Sulawesi, thereby lowering the ENEP (King 2003, 149). 40 Table 2.4 - Ethnicity in Indonesia Ethnic Group Percentage Javanese 41.71% Sundanese 15.41% Malay 3.45% Madurese 3.37% Batak 3.02% Minangkabau 2.72% Betawi 2.51% Buginese 2.49% Bantenese 2.05% Banjarese 1.74% The independent variable used to test the effect of ethnicity on E N E P is the provincial level effective number o f ethnic groups ( E N E G ) . L ike E N E P , the E N E G is calculated using the Herfindahl-Hirschman concentration index. 2 0 This measure has been standard in much of the previous cross-national work (Ordershook & Shvetsova 1994; Amor im Neto & Cox 1997). Ethnicity data are from the results o f the 2000 national census. It was taken from provincial demographic statistics in Suryadinata, Ar i f in , & Ananta's Indonesia's Population (2003). Assuming that electoral districts within a province share a uniform level of ethnic diversity is less methodologically hazardous than constructing a national average magnitude and a national effective number of ethnic groups, as is common practice in cross-national studies. Though definitive claims cannot be made using only provincial level data, it should be accurate enough to discern trends in the interaction of ethnic fragmentation and electoral fragmentation. Theoretically, a higher E N E G should lead to a higher E N E P . The equation for ENEG is ENEG = l/Xfe) 2 where 57 is the population share of ethnic group / in a given province. 2 1 The construction of ENEG likely underestimates the ENEG. This underestimation should be greater in highly fragmented provinces. See Appendix B for explanation. 41 One politically salient social cleavage commonly referenced by scholars of Indonesia is the concept of aliran. Aliran is an anthropological term that references both class and religious observance. At its most basic, it divides the population according to religious devotion. The abangan are assumed to have comparatively casual, syncretic religious beliefs and a more secular outlook. Santri refers to those with a devout Islamic orientation. These social groups are broken down further. The secular oriented administrative class is labeled priyayi. A s well , a distinction is made between traditionalist and modernist santri, the former of which are said to be more influenced by Java's Hindu-Buddhist heritage. Political scientists have used these broad social groupings to explain Indonesian political behavior. Though the anthropological use of the word focuses on socio-cultural practices, political scientists have used aliran in reference to political parties and the social organizations that are linked to them in formal and informal networks (Lanti 2004). Some scholars have argued that the each of the four dominant parties in the 1955 election represented one of the four major "streams," or aliran (King 2003, 34). Recent work has provided evidence that there was some regional continuity in aliran voting between 1955 and 1999 (King 2003). There is ample reason to believe that aliran is a social-cleavage that impacts the E N E P . Operationalizing aliran in a quantitative study is a complicated task. Using recent party votes is an inappropriate measure because, in this case, both the dependent and independent variables would be constructed using the same data. Organizational 2 2 As Leo Suryadinata (2002) points out, notable Indonesianists such as Herbert Feith argued that there were issue dimensions in the 1955 election not captured by the aliran interpretation. 2 3 Of course, most of the party names have been changed and one, the PKI, is still officially banned by the state. 42 membership wi l l only provide a measurement for the two santri streams. King ' s findings suggest that the 1955 election data could be used as a rough measure. 2 4 Calculating the E N E P based on region offers one possible proxy for a social heterogeneity variable based on aliran. There are two reasons to be cautious with the results, however. First, it is unreasonable to expect that the regional class structures and levels of devotion have remained static for fifty years. Second, Indonesia has experienced in-migration that could have shifted the previous balance. Having noted these limitations, previous theoretical and empirical research indicates that is worth exploring the fractionalization o f the 1955 vote as a proxy for social heterogeneity. The results of the 1955 election are taken from Kev in Evans' The History of Political Parties & General Elections in Indonesia (2003). The measure is based on the vote share o f the four primary parties ( P N U , Masyumi, PNI , PKI) whose base of support is commonly linked to distinct aliran. Each party's provincial vote share is divided by the total provincial vote share for the four aliran parties. Fragmentation values are then calculated using the standard equation for the effective number of parties. The value provides a measure o f social diversity within a province. A third social variable that may cause social fragmentation is modernist Islam. A s mentioned, modernism refers to one of the two religious (santri) streams. Islam, in general, is a non-centralized faith. Explains Bakhtiar Effendy: Islam.. .is one religion but its expressions and interpretations are many. The fact that Islam recognizes no religious priesthood that could provide a unified legal 2 4 The 1955 elections were conducted using a PR system. There were 15 electoral districts. A common divisor of one seat for every 300,000 citizens was used in all districts. When combined with large electoral districts, the uniform divisor ensured very low deviation from proportionality at the national level. As one observer of Indonesian electoral history has pointed out, "To all intents and purposes, the system applied in 1955 amounted to proportional representation within one district consisting of the entire country." (Van Marie, 1974). It is reasonable to assume that there was an absence of institutional pressure on the decision of the voter in 1955. 43 product that binds all Musl ims is a major source o f weakness in any attempt to institutionalize Islamic teachings in a single legal entity (2003, 219) Though Effendy was referring to the difficult task of reaching a consensus on the issue of Islamic law, Islam's susceptibility to fragmentation could just as easily be applied to 25 party politics. In Indonesia, modernism is particularly vulnerable to division. Unlike the rival traditionalist religious stream, modernism de-emphasizes Javanese traditions that value order and consensus. While modernism was largely united in 1955, there is reason to believe that the significant increase in literacy since that point has reinforced the non-hierarchical nature of the religious stream. 2 6 One obvious (and available) proxy for Islamic modernism is the strength of the 1955 Masyumi vote. Unfortunately the data has also been used to construct the aliran proxy, thus the two are related. K i n g , Keohane, and Verba (1994, 173) have warned against using such a variable, arguing, "In general, we should not control for an explanatory variable that is in part a consequence of our key causal variable." Though there is a correlation, the Masyumi vote explains only half o f the variation in the aliran 27 fragmentation measure. It thus presents itself as a possibly useful proxy variable for Islamic modernism. Two additional control variables are included. The first is an 'Outer Island' dummy. Suryadinata observes (2002, 5), "There is competition, i f not conflict, between Java and the Outer Islands." In terms of salient political issues, the regional divide is asymmetric. Resentment against the centre is easier to mobilize than resentment against 2 5 For a related discussion of Muslin political fragmentation in Europe see Warner and Wenner (2002). 2 6 For example King (2003) found a negative correlation between illiteracy and support for modernist leaning political parties in 1999. 2 7 There is a diversity of possible values on the two variables. East Java, for example, has a relatively low modernist score and a high level of aliran fragmentation while East Nusa Tenggara is low on both values. Likewise, Lampung is scored relatively high in both while Riau has only a high modernist score. 44 the periphery. Since we would expect regional identities and grievances to be more important off of Java, there should be a positive correlation between the Outer Island dummy and the E N E P . The second control adds a dummy variable for the presence of separatist conflict. Beyond the question of whether a free vote is possible in a province in the midst of a battle with government forces, there should be unique electoral patterns in separatist areas. In a country going through the process o f a free nationwide vote, large-scale rigging of elections is an undesirable option for a new democracy. Banning a popular separatist movement from taking part in the elections is within the legal boundaries however. Despite this, the addition of a separatist issue dimension should increase the E N E P . We thus arrive at the following hypotheses: Hypothesis 6: Higher levels of ethnic diversity will lead to a higher ENEP Hypothesis 7: A tradition of aliran diversity will lead to a higher ENEP Hypothesis 8: A tradition of modernism will lead to a higher ENEP The above hypotheses are simply additive. What we are interested in is their interaction with district magnitude. We therefore add three more hypotheses: Hypothesis 9: Higher levels of ethnic diversity will lead to a higher ENEP only when the district magnitude is sufficiently large Hypothesis 10: A tradition of aliran diversity will lead to a higher ENEP only when the district magnitude is sufficiently large Hypothesis 11: A tradition of modernism will lead to a higher ENEP only when the district magnitude is sufficiently large Hypotheses 6-8 can be tested using the following additive model: E N E P = (30 + plln(Magnitude) + [32ENEG + (33Aliran + (34 Modernism + P50uter_Island + p6Seperatism + e Hypotheses 9-11 can be tested using the following multiplicative model: 45 E N E P = PO + piln(Magnitude) + p 2 E N E G + p3Aliran + p4 Modernism + P50uter_Island + p6Seperatism + p7ENEGXln(Magnitude) + p8AliranX ln(Magnitude) + P9ModernismX ln(Magnitude) + e Table 2.5 displays the results of both models: Table 2.5 - Social Cleavages and Magnitude Social Cleavages and the 2004 E N E P -Additive Model Estimate Variable (std. err.) ln_Magnitude ENEG Aliran Modernism Outer Island Separatism Const 0.023 (0.563) 0.141** (0.063) 1.682*** (0.386) 8.048*** (1.622) 1.631*** (0.592) 2.075* (1.243) -1.371 (1.794) Social Cleavages and the 2004 E N E P Interactive Model . Estimate Variable (std. err.) InMagnitude ENEG Aliran Modernism Outer Island Separatism ENEGXlnMagnitude AliranXlnMagnitude ModernismXInJVIagnitude Const -0.846 (4.446) 0.386 (0.253) 0.111 (2.453) 9.526 (11.174) 2.089*** (0.633) 2.171* (1.256) -0.160 (0.138) 0.806 (1.192) -1.003 (5.511) -0.262 (9.126) N 67 N 67 _ _ _ I 0-54 R2 | 0.57 Papua and West Irian Jaya are dropped from the equation because they were not under Indonesian sovereignty in 1955 and thus did not participate in the elections. The high ENEP and the existence of separatist sentiment in both provinces suggest that separatism may be a statistically stronger variable than is reported by the regression results. ***p<01 **p<0.05 *p<.10 In the additive model, all three measures of social diversity are significant at the .05 level. A legacy of modernism has a strong effect on E N E P . Keeping all values at the mean, we would expect a district that polled 30% for Masyumi in 1955 to have 1.62 more effective parties than a district in which Mayumi gathered 10% of the 1955 vote. The effect o f ethnic fragmentation is relatively weak. A district with ten effective ethnic groups would only have 1.27 more electoral parties than a district dominated by one ethnic group. A s predicted, the aliran variable is positive. The correlation coefficient 1.68 suggests that stream voting was less consolidated than it was in 1955. 46 The interactive model finds no evidence that magnitude played any role in consolidating the electoral vote. None of the interactions between magnitude and the social variables approached statistical significance. The interactions with E N E G and modernism produced a negative correlation coefficient, the opposite of what we would theoretically expect to see. The regressions offer support for hypothesis 6, 7, and 8 while refuting hypothesis 9, 10, and 11. There is evidence that ethnic, social, and religious cleavages effected voters' decisions but no evidence that magnitude played any role in these decisions. Evaluating the Effect of Magnitude This chapter finds no evidence that magnitude affected the 2004 electoral vote. A s well , there is no evidence that institutional learning from the 1999 election affected the electoral vote. The level of social diversity within a district did affect the level of electoral fragmentation; however, this affect was independent of magnitude. Caution is required when drawing any conclusions about magnitude from the Indonesian case. It is l ikely that there was high voter uncertainty regarding both public preferences and expectations of party performance in both the 1999 and 2004 elections. I f this was the case then Indonesia did not fulfill the Cox ' s (1997) two conditions for strategic voting. More data about voter knowledge are needed to reach a conclusion on this point however. 47 Chapter 3 : Direct Presidential Elections and Party Electoral Fragmentation This chapter explains how the adoption of direct presidential elections fragmented the Indonesian electoral vote. Two distinct mechanisms are examined. The first considers the role of presidential vehicles, those legislative parties created to support the presidential aspirations of popular political figures. One presidential vehicle in particular, Yudhoyono's Partai Demokrat, was moderately successful at riding Yudhoyono's coat-tails, yet failed to achieve the widespread support of its presidential candidate. The second mechanism considered is the role of split-ticket voting. The separation of legislative and presidential elections allowed Indonesians to vote for their preferred leader in the presidential contest while supporting a minor party in the legislative contest. The relative decline of the 'presidential' parties in the 2004 legislative election contributed to increased electoral fragmentation. The split-ticket phenomenon is illustrated through a brief study of the P K S , a sectarian party that benefited from the adoption of direct presidential elections. A Review of Presidential Reforms in Indonesia Throughout the New Order, presidential elections were acclamatory procedures in which a rigged legislative body would overwhelmingly offer its support for five more years of rule by Soeharto. When Soeharto was forced to step down from the presidency in M a y of 1998, Vice-President B .J . Habibie inherited the executive office. According to the Constitution, Habibie could serve out the remainder of Soeharto's term, thereby granting him just less than five years of rule. Habibie, however, recognized that the process that led to his ascension was not perceived as legitimate by an increasingly assertive public. 48 He called for new legislative elections to take place in June of 1999 followed by an indirect presidential selection in October. To ensure that the new elections would be perceived as legitimate, Habibie drafted a number of new political laws. 2 8 The body that elects the president, the M P R , would be significantly reduced, from one thousand members to seven hundred. Most o f the reduction would be at the expense of unelected delegates. Al so , the five hundred member D P R saw its reserved military seats nearly cut in half, from 15% to 8% of the legislature. Thus 66% of the M P R would consist of directly elected parliamentarians. 2 9 The presidential selection was set to be determined by the strength of the elected factions in the D P R . Before the 1999 election there was uncertainty regarding the relative strength of the electoral parties. There was some expectation that the strong parties would be those closely associated with either a large social organization or pre-established party machines from the New Order; thus PDI-P, P A N , P K B , Golkar, and PPP were perceived as frontrunners.3 0 How the electorate would be divided among these parties was largely unknown. The uncertainty contributed to a situation in which the leading candidates viewed the parliamentary elections as a leadership contest. Despite an informal alliance with Megawati, the PDI-P leader, Abdurrahman Wahid predicted that his party ( P K B ) would receive the most seats and went on to list some possible members of his cabinet (Mclntyre 2005, 205). Likewise, P A N leader and presidential hopeful Amien Rais 2 8 For a full description of these laws see Dwight King's (2000) "The 1999 Electoral Reforms in Indonesia: Debate, Design and Implementation." 2 9 As mentioned in Chapter 3, the system retained some representation for indirectly elected regional representatives and appointed members of civil society groups. 3 0 Golkar and PPP were both New Order parties while P A N and P K B were closely associated with Muhammadiyah and Nahdlatul Ulama respectively. Though officially a new party, PDI-P inherited many party activists from the New Order PDI, of which Megawati had been Chairwoman before Soeharto had her deposed in 1996. 49 claimed on the eve of the election that the party with the most votes had the claim to the presidency (Mclntyre 2005, 207-8). When the results showed that PDI-P had the largest share o f the votes and seats, Megawati assumed that she was entitled to the presidency, a view that was initially conceded by incumbent president and Golkar leader B . J . Habibie (Mclntyre 2005, 208). A combination of personal aloofness, poor political maneuvering, and Islamic puritanism created a legislative backlash against the prospects of a Megawati presidency. A n Amien Rais-led coalition of Modernist parties known as the Central A x i s vowed to support an alternative candidate.3 1 Following the vote, Habibie, the parliamentary runner-up, was dogged by financial scandal and his association with Soeharto, subjected to criticism over his handling of the East Timor crisis, and facing wavering party support. His weak political position influenced his decision not to challenge Megawati 's presidential bid in the M P R . The Central A x i s and substantial segments o f Golkar threw their support behind Wahid, giving him the support he needed to win the presidential vote. Despite the vocal anger of PDI-P activists, Megawati resigned herself to the outcome and accepted the Vice-Presidential position as a consolation prize. Since Wahid's party could only claim 11% of the seats in the D P R he initially formed a large coalition that included both Golkar and PDI-P. The cooperative effort soon broke down. Wahid, plagued by financial scandal, largely shut out the two main parties in an August 2000 cabinet shuffle. Following a presidential censure in Apr i l 2001, Wahid attempted to declare dissolve the legislature and declare a state of emergency. The Rais admitted that his party's modest vote share (just over 7%) prevented him from offering his name as a possible candidate. 50 M P R forced him from office when both the Army and the Supreme Court refused to go along with his unconstitutional maneuvers. Megawati assumed the presidency. With the ascension of Megawati there was general agreement that the system of executive selection heeded reform. The institutional process that granted the presidency to the leader o f a relatively minor party with no national mandate had set the stage for a constitutional crisis. There was broad agreement that the powers o f the legislature and executive branches needed to be separate and more clearly defined. A s well , most political factions agreed that there needed to be a separate electoral mechanism that would ensure a national mandate for the president. Between 2001 and 2003, a number of laws were passed defining the nature of executive-legislative relations and the method of selection for both branches. It was decided that the electorate would be given the opportunity to directly elect a joint presidential and vice-presidential team. To ensure that the elections would not be dominated by Javanese voters, the reforms mandated that a presidential ticket would have to win 50% or more of the popular vote and at least 20% o f the vote in half of the provinces. If no team reached the threshold there would be a winner-take-all run-off between the two tickets with the highest number o f votes in the first round. Whi le these reforms guaranteed the executive would have a clear national mandate, they did not prevent the possibility that a minor party could win the presidency. The larger parties therefore proposed that only parties with 20% o f the vote in legislative elections would be granted the right to nominate a presidential ticket. The smaller parties objected and a compromise position was reached that required a nominating party or coalition to have garnered 5% of the legislative vote or 3% of the seats in the D P R . These 51 ballot access rules necessitated staggered legislative and presidential elections so as to allow an accurate count o f the legislative vote and ample time for a presidential 32 campaign. Presidentialism and the Limited Success of Coat-Tail Parties The first way in which presidentialism fragmented the party systems was the rise of presidential vehicles. Party-system fragmentation literature from Eastern Europe highlights the role of parties created as vehicles for a candidate's presidential bid (Fillipov, Ordeshook & Shvetsova 1999). These parties are disconnected from the established party machines and serve primarily to support a presidential bid by increasing the party leader's profile. The presidential vehicles are able to draw some measure of support in legislative elections due to their close association with a popular personality, thus the parties 'ride on the coat-tails' o f their presidential candidate. Expanding on the lessons from Eastern Europe leads to the following hypothesis: Hypothesis 1: A switch from a parliamentary to a presidential system should increase the vote-share of parties that serve as leadership vehicles Looking at Indonesia, one can see signs of the presidential vehicle even in 1999. PDI-P, P A N , and P K B were all closely associated with the presidential aspirations o f Megawati, Rais, and Wahid respectively. Despite the dominating presence of leaders, all three parties had a foundation based in either a major social organization or a pre-existing party machine. Leadership may have been a major issue in 1999; nonetheless, to be a serious candidate required a credible party with access to a pre-existing organizational structure. For a full description of the presidential laws see the "Indonesia Backgrounder: A Guide to the 2004 Elections" (International Crisis Group, 2003). 52 The executive-legislative reforms changed the political opportunities for those seeking the presidency. A s long as a presidential team met the relatively low ballot access requirements (5% o f votes or 3% of seats in the legislative elections) they could enter the executive race without much concern regarding their party's presence in the legislature. This is not to say that all presidential hopefuls believed that creating a presidential vehicle was in their best interest. Mus l im intellectual Nurcholish Madjid and General Wiranto both entered the Golkar leadership race despite their status as party outsiders. Other presidential candidates took a different strategy, adding new parties and altering the political landscape. The most important presidential vehicle to gain legislative representation in 2004 was Yudhoyono's Democratic Party. Founded in 2002, Yudhoyono kept an official distance from the party during his tenure as a minister in Megawati 's cabinet. Early party membership was dominated by intellectuals and retired military figures that were committed in their support of Yudhoyono's future candidacy. The party itself was careful not to advance political positions that could alienate particular societal groups. Ponti Pandean, the party's deputy secretary-general described the party as "an alternative for people from all walks of life" (Jakarta Post 2004). Following the legislative elections, Leonard Sebastian (2004, 265) echoed the view of most analysts when he remarked, "The P D ' s strong performance can be attributed solely to the popularity o f Susilo Bambang Yudhoyono." It can be safely concluded that the P D was a presidential vehicle that reflected the vote-seeking political positions o f presidential candidate Yudhoyono. That a new party was able to win 7.5% of the legislative vote with no connections to any major 53 social organizations or pre-existing structure provides strong evidence in support of hypothesis 1. The P D , however, had only limited success in the legislative election. Analysts and pollsters have both noted that the P D ' s rise came largely at the expense of PDI-P (LP3ES; Sebastian 2004; Qodari 2005). Figure 3.1 plots the 2004 P D vote against PDI-P losses between 1999 and 2004. Figure 3.1 - PD Vote by PDI-P Losses Figure 3.1 - PD Vote by PDUP Losses y = o 2894x + 2 7401 R 2 = 0.2677 25 lllll||§||||p^ 1-o 20 -o -•-» 15 - — ; O PD Vote by 10 -5 \-4~. • PD Vote by • • » • • ~W • • • • • • u Q. 0 . • 1 1 1 1 1 • 1 1 0 5 10 . /" 15 20 25 30 35 Pet of PDI-P Vote Share Loss by District (1999-2004) There is a positive correlation between the strength of the P D vote and the extent o f PDI-P losses, thus supporting the argument that disaffected nationalists migrated to Yudhoyono's party. P D ' s 7.5% performance is slightly less than half the total share of voters that abandoned the PDI-P. Golkar, whose national vote fell less than 1%, most likely picked up some of the PDI-P losses while hemorrhaging modernist voters. It was former PDI-P voters that helped boost the vote share of the minor parties. This does 54 suggest a limitation in PD 's ability to attract support despite the favorable conditions created by an anti-PDI-P sentiment. Yudhoyono's widespread popularity makes the unconsolidated anti-incumbent vote even more curious. Figure 3.2 plots both the P D vote share in the legislative election and the Yudhoyono vote share by district first round of the presidential election. Figure 3.2 - Pet of SBY-Kalla First Round Vote Share vs. Pet of PD Vote Share Figure 3.2 - Pet of SBY-Kalla Vote Share vs. Pet of PD Vote Share ^ P D Vote Share (by District) . SBY-Kal la Vote Share (by Province) 0 20 40 60 80 Electoral District (2004) Source: KPU. Available Online: http://www.kpu.go.id'hasil_piIpres/suara_sah-l .php Yudhoyono the presidential candidate was far more effective at attracting voters than his presidential vehicle. Part of Yudhoyono's performance was due to political alliances and campaign idiosyncrasies. Yudhoyono's running mate, Jusuf Kal la , attracted some Golkar voters. P B B , a minor party with 2% of the vote, officially endorsed Yudhoyono. A s well , Wahid, the N U presidential candidate, was barred from participation for health reasons. Approximately 28% of N U voters supported Yudhoyono in the first round (LP3ES 2004). 3 3 The presidential vote is broken down by province. Due to data limitations, the graph assumes a uniform distribution within provinces. Result available online: http://www.kpu.go.id/hasil_pilpres/suara sah-1.php « 50 * mm ^ p , • _ " * * 1. i-\ it < • * 55 Beyond the official alliances, polls also show that Yudhoyono did particularly wel l among those who voted for the minor parties. Again, this raises the question of why Yudhoyono's political vehicle did not pick up these votes in the legislative elections. There are several reasons why Yudhoyono, the candidate, may have succeeded where his party failed. One has to do with the timing of the campaigns. In a Jakarta Post article, Qodari (2004) argued that P D ' s vote share would have been much larger i f the campaign period had been longer than three weeks. Yudhoyono's limited opportunity to campaign for his party may have caused a reduction in the coat-tail effect. This answer works well in explaining why the P D did not cut deeper into the vote share of the established parties; however, a significant number of voters abandoned the ' B i g - 5 ' only to pass their votes on to a party with far less national visibility than Yudhoyono's P D . Qodari's argument also raises the issue of how Yudhoyono was pursuing his political goals in a staggered campaign. The legislative election was a test of strength for the presidential aspirant; yet, providing he could pass the threshold or cobble together an electoral coalition, it would not be the defining contest in his presidential run. It is difficult to say whether Yudhoyono would have created a presidential vehicle had the institution not been changed, but assuming he had, the one-ballot nature of the contest likely would have prompted him to lead a more extensive and well-coordinated campaign for his party. A second reason the P D failed to attract more voters had to do with the vote-seeking nature o f the party and the limits of the coat-tail effect. Given P D ' s ambiguous policy positions, a vote for the party was, in effect, a vote in support of Yudhoyono and 56 his future ability to govern. A s one P D voter noted in a focus group before the presidential election: It's a pity, the governmental cabinet they made: Kabinet Gotong Royong. That makes things worse. That's why I prefer S B Y as a president with all the cabinet from Partai Demokrat - one package. (National Democratic Institute 2004a). In this case the voter is expressing frustration with Megawati's tendency to create broad, inclusive cabinets. There is, however, only limited appeal for a government dominated by one party. Additionally, not all voters were as concerned with the presidential contest in the legislative race. Wi th the expectation that Yudhoyono would make it on to the presidential ballot, future Yudhoyono voters could support either an office or policy seeking party in the legislative contest. Thus, part of P D ' s failure to consolidate disaffected voters was related to political limitations of presidential vehicles. Party Leadership and Voting Behavior The coat-tail effect has been referenced by scholars on both sides of the consolidation/fragmentation debate. It asserts that voters use information from the presidential race to make their decision about which legislative party to support. The underlining logic implies that the issues that matter to voters in the presidential race are the same issues that w i l l matter to them in the legislative elections. The preceding study of Yudhoyono's presidential vehicle suggests that there are limitations to the coat-tail approach. Samuels and Shugart argue that presidential systems allow voters to express a more complex set of preferences then they can in parliamentary systems. The authors note, "[0]ne of the benefits o f presidentialism is that voters may believe that members of the executive and legislative branches have mandates for different things" (Samuels & 57 Shugart 2003, 40). If voters believe that the two branches o f government are responsible for distinct issue areas then they are apt to make two distinct electoral decisions. The intense spotlight on a candidate (and possibly her running mate) means that presidential elections, when compared to legislative elections, focus more on personality traits and general leadership issues. Beyond the question of leadership qualities, however, there may be particular issue areas that voters want their presidential candidate to be competent in. For example, the adoption of direct executive elections in Israel changed the way voters conceived their electoral choices. Whereas parliamentary elections had been dominated by security issues, when given the opportunity to cast both a legislative and executive vote, the Israeli electorate used the latter to reflect their preferences on security issues and the former as a means of expressing a narrower, sectarian identity (Hazan & Rahat 2000, 1322). A s a result, the vote share given to sectarian parties soared while legislative parties that had staked out extreme positions on the security issue suffered notable losses. In short, a switch from a parliamentary to a presidential system can change how the electorate views the mandate of each branch of government in ways that affect voting behavior. Electoral fragmentation in Indonesia increased, in part, due to the shift that took place in the way the public viewed legislative elections following the introduction of the direct presidential ballot. The presidential election became the arena through which the electorate expressed its leadership preferences. Leadership contests in Indonesia are not just about personality traits though; they are also one o f the central ways Indonesians express preferences about national unity. Indonesia is an ethnically and religiously diverse country spread over a vast archipelago. Communal violence and separatism are 58 reoccurring problems. In order to maintain peace and stability, Indonesian voters expect their head of state to be able to transcend the major social divides. A n important task that Indonesian leadership candidates face is convincing the electorate that they have a cross-cutting appeal. Leadership candidates may have sectarian or regional bases of support; however, they need to appear inclusive both to ensure a broad range of support and to guarantee that their electoral base does not consider a different, less polarizing leadership figure. In 1999, the legislative election was a clash o f party machines; nonetheless, it was also the voters' opportunity to cast a leadership vote. Not wanting to offend any part of the electorate, the presidential hopefuls made broad appeals and generally avoided contentious policy issues. In the run-up to the 1999 elections Megawati was granted the power to shape PDI-P as she saw fit, which included running a populist campaign based on her ability to speak for the nation (Mclntyre 2005, 196-7). Rais 's P A N and Wahid's P K B cast themselves as inclusive, catch-all parties and tried to prove their point by recruiting secular-minded nationalists and religious minority candidates. Though less leadership oriented than PDI-P, P A N , or P K B , Golkar and Habibie portrayed the party as a unifying organization committed to vague, developmentalist goals. The only major party leader to not make a cross-cutting appeal was the PPP 's Hamzah Haz, who was generally not viewed as a credible presidential candidate (International Crisis Group 2001, 3 ) . M In the 2004 presidential election, personality and national unity dominated the stage. The candidates raced to present balanced presidential tickets. Yudhoyono, an 3 4 Megawati's secular-oriented PDI-P would later support Hamzah Haz in the MPR vice-presidential election due to the fact that he was politically non-threatening and because his Islamic credentials brought religious balance to Megawati's presidency (International Crisis Group 2001, 3). 59 ethnic Javanese, selected as his vice-presidential candidate Jusuf Kal la , an ethnic Buginese from the outer island of Sulawesi. Megawati and Wiranto, both of whom are secular oriented, chose high-ranking religious activists as their respective running mates. The Islamic modernist Rais chose a running mate with secular-nationalist credentials. After some speculation that he would be Megawati 's vice-presidential candidate (Wahyuni & Hajramurni 2004), Hamzah Haz entered the presidential race himself, choosing a nationalist military figure in the hopes of wooing more secular-oriented voters. 3 5 The 2004 legislative elections were less dominated by leadership and national unity than both the 1999 legislative election and the 2004 presidential election. Voters knew the election would not be the definitive leadership contest and thus expressed a wide array of political preferences. The electorate was more wi l l ing to support small parties making niche sectarian or ideological appeals. A s a result, the inclusive, vote-seeking parties with credible presidential candidates suffered. Table 3.1 shows the legislative vote share for the inclusive presidential parties in 1999 and 2004: Table 3.1 - 'Presidential Party' Vote Share Party 1999 2004 PDI-P 33.73% 18.53% Golkar 22.43% 21.58% P K B * 12.60% 10.57% P A N 7.11% 6.44% PD 7.45% Total 75.88% 64.56% 3 5 Hamzah Haz was the only presidential candidate from an explicitly sectarian (Muslim) political party. He garnered 3% of the first round presidential vote thus making him the only candidate to do worse than his party did in the legislative elections (PPP gathered 8.2% of the 2004 legislative vote.) 60 * Abdurrahman Wahid, PKB's presidential candidate, was barred from running in the presidential election after the legislative vote. A s a whole, the parties seriously contending the presidential election received less of the legislative vote share in 2004 than they did in 1999. This would suggest that Indonesian voters tended to use the election to emphasize issues beyond national unity and the control of the executive. The proclivity toward split-ticket voting, or supporting one party in legislative elections and another party's candidate in presidential elections, resulted in increased electoral fragmentation. One way to test the split-ticket hypothesis is to look at the effect of leadership concerns on legislative and presidential vote choices. If the split-ticket hypothesis is correct, the electorate should have been less likely to vote for the party o f their preferred leader in the 2004 legislative elections as compared to the 1999 legislative elections. This leads to hypothesis 2: Hypothesis 2: The likelihood that a voter will support the legislative party of their most preferred leader will decrease following the separation of legislative and executive elections. In a study o f effect o f leadership on voting behavior in Indonesia, Liddle and Mujani (forthcoming) argue that party leader preferences have been the strongest variable to effect Indonesian voters' electoral decisions. The authors note, "Indonesian voters are strongly attached to national party leaders, an attachment that appears to be a principal reason for voting for political party or president." Using survey data, the authors conducted bivariate analysis of leader preference and vote choice in the legislative elections in 1999 and 2004, as well as the first and second rounds of presidential voting in 2004. They conducted a similar analysis using party preference as the independent 61 variable. Lambda coefficients were used to indicate the strength and significance of the relationships in each election. Figure 3.3 presents a graphical re-creation o f their results: Figure 3.3 - Leadership, Party ID, and Voting (Liddle & Mujani) Figure 3.3 -Leadership, Party ID, and Voting (Liddle & Mujani) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 • Leader Preference (Lambda coefficient) • Party Preference (Lambda coefficient) 1999 Legislative 2004 Legislative 2004 Presidential 2004 Presidential (1st Round) (2nd Round) Liddle and Mujani's results show that a large drop in leader-driven voting occurred in the 2004 legislative elections. The authors note that this was largely due to the cross-party appeal of Yudhoyono. Likewise, the first round of the presidential election exhibited strong leadership-based voting and weaker party-based voting. This pattern leads the authors to the hypothesize that, "Attractive leadership can trump strong party loyalties, perhaps especially in systems characterized by separate legislative and presidential elections." The results o f Liddle & Mujani 's study would appear to support hypothesis 2. Split-Ticket Voting and the Rise of the PKS A s mentioned above, the separation of the legislative and presidential ballots in 2004 led voters to support legislative parties that emphasized issues beyond leadership and As Liddle and Mujani explain: "The lamda (k) coefficient ranges from 0 to 1.0. Zero indicates no association and 1.0 a perfect association." 62 national unity. Some voters used their legislative vote to support niche sectarian parties. The most successful of the sectarian parties was the P K S . Their 6% increase in the vote . share has commonly been credited to their stance against systemic corruption (Liddle & Mujani 2005; Qodari 2005). While P K S may have had a credible anti-corruption message, observers have failed to notice the way in which institutional reforms gave a boost to P K S ' s electoral strategy. The success of the P K S provides a convenient illustration of the way executive-legislative reform affected voting behavior. Formed by modernist students and intellectuals, P K S ran in the 1999 elections as the P K , picking up 1.36% of the legislative vote. Unlike P A N , the officially secular party with close ties to Muhammadiyah, 3 7 the P K S did not shy away from explicitly Islamist positions. Though the party's 2004 campaign emphasized the application o f Islamic values to practical problems rather than the application of Islamic law, the P K S did not stray far from its Islamist roots and its committed activists continued to view party goals in terms o f Islamic propagation (Soekanto 2004b). In the 2004 legislative elections, the P K S gathered 7.34% of the vote, making them a significant player in Indonesian party politics. P K S is one of the few parties to eschew political office and embrace the role of political opposition. It is the only 'B ig -7 ' party in 2004 to put forward a candidate for neither the presidency nor the vice-presidency. This put the P K S in a position to chip away at P A N and PPP. Despite a split that saw the departure of some prominent liberals, P A N remained mindful of the upcoming presidential race and thus maintained an officially secular position. PPP was an exclusively Mus l im party. However, its Islamist credentials were compromised by its involvement in presidential politics. Hamzah Haz 3 7 Muhammadiyah is the primary social organization for modernist muslims. 63 had been Megawati 's vice-president. There was every reason to believe the party would trade its Islamist policy positions for political office in the future. Thus presidential politics made P A N and PPP vulnerable to Islamist challengers. Figure 3.4 plots the district level sum of PPP and P A N losses against the increase the increase in P K S vote share. Figure 3.4 - PKS Gains and PPP-PAN Losses PKS Gains by PPP-PAN Losses (2004) y = 0.271x + 4.0701 R 2 = 0.3389 -5 -I , - --20 -10 0 10 20 30 40 PPP-PAN Losses The increasing slope displays evidence that some modernist voters whom previously had voted PPP or P A N shifted their support to the less compromising P K S in 2004. Presidential politicking by PPP and P A N created some space for P K S growth. This is not to say, however, that P K S had no interest in the presidential race. Most activists wanted to see Megawati defeated. A s well , many P K S activists were involved in Muhammadiyah, the modernist organization that officially endorsed Amien Rais 's candidacy (though not P A N ) . With strong Islamic credentials and a clean reputation, Rais 64 was able to secure the official endorsement of P K S . Figure 3.5 shows a positive correlation between P K S support (by district) and Amien Rais 's share of the presidential vote (by province). Figure 3.5 - Rais Presidential Support by PKS Vote Share Rais Presidential Support by PKS Vote Share (By District) y = 0.952x + 8.1826 R 2 = 0.2352 60 Si Vol 50 .2 ent 40 '35 a) Pr 30 sz I*— 20 o o 0_ 10 m "to <TS DC 0 "''fit-. . _ A M R BIllilBIii ~'f'' ^ ' " '' " .;" _____ — • - : — * I B P T 1 t l l i l M » « S i l l l l l tllllliililil 10 15 PKS Vote Share 20 25 The above results are consistent with a Jakarta Post report that approximately 57% of P K S voters supported Rais while 24% supported Yudhoyono (Taufriqurrahman 2004). The high level of support offered to Rais shows that P K S voters preferred to have a reputable, modernist candidate as president. The internal debate regarding Wiranto and the sizable minority of votes offered to Yudhoyono demonstrate an anti-Megawati undercurrent within the party. Despite its non-participation in the presidential contest, P K S members had a clear interest in the race. In the previous single-ballot system there Though not before an internal debate in which some suggested endorsing former General Wiranto, the Golkar candidate who was seen as having a better chance of defeating Megawati (Soekanto 2004a; • Soekanto 2004b). 65 would have been increased pressure to cast a vote for a party with a presidential candidate that could represent the leadership preferences o f P K S members. It is l ikely that the double-ballot in 2004 complemented P K S ' s political strategy and contributed to the party's surge, fragmenting the party system in the process. In sum, the adoption o f direct presidential elections increased party fragmentation through two mechanisms. The first was the introduction o f presidential vehicles. Partai Demokrat was able to ride Yudhoyono's coat-tails; however, only a limited number of voters responded to personality appeals in the legislative election. The second mechanism propelling fragmentation was split-ticket voting. Voters selected their preferred leader in the presidential election while tending to prefer a more narrow appeal in the legislative election, thus bolstering the vote share of small parties. 66 Chapter 4 : Conclusion Executive-legislative structures shape elite strategies and the political expectations of voters. In Indonesia, reform of the executive-legislative structure resulted in unexpected consequences for the party system. Yudhoyono saw that he could bypass the established parties and still win the presidency. Voters, on the other hand, found that they could forgo the inclusive 'presidential' parties in the legislative election in favour of a more narrow electoral choice. In what would appear to be a contradictory pattern o f political behavior, Indonesians increased their support for both unrooted presidential vehicles and small niche parties. Despite the successful efforts of the K P U to reduce the number of competing parties, the 2004 legislative elections left Indonesia with a significantly more fragmented party system. The findings of this thesis have a number o f implications for comparative scholars. Most o f the work on presidentialism and party systems thus far has focused on the coat-tail effect. This thesis supports previous work that has suggested the coat-tail effect can fragment the vote which, in and of itself, does not necessarily contradict more refined versions o f the 'compression' argument that have emphasized that the consolidating effects of presidentialism depend on the effective number o f presidential candidates. What has been missing from the coat-tail debate has been knowledge and understanding of the breadth o f this effect. Increasing numbers of presidential candidates may in fact increase party fragmentation, a point this thesis agrees with; however, i f a large portion of the electorate views the executive and legislative branch as having two distinct mandates then the present articulation of the coat-tail effect is only telling one part of the story. What the literature needs is increased appreciation for the conditions in 67 which we would see a wide coat-tail effect - in which voters use cues from the presidential race to make their legislative choices - and a narrow coat-tail effect - in which only a limited number o f voters are compelled to cast a legislative ballot for the party o f their preferred leader. Research in this area would also contribute to the broader knowledge regarding performance and representation in presidential systems. The two Indonesian elections provide a snapshot o f the effect of district magnitude in a transitional country. It is difficult to reach firm conclusions about the effect of magnitude from two elections. The thesis does suggest, however, that voters in transitional countries are not sensitive to changes in district magnitude. Whether this is due to a lack of information or a lack o f opportunity for institutional learning is unknown. Cross-national research that considers the district-level effect of magnitude in transitional countries is needed. What conditions prevent or promote a trend toward equilibrium voting patterns? Is there any relation between predictable district level competition and democratic consolidation? The examination of magnitude and electoral fragmentation presented in this thesis is one of the first attempts to apply methods from the cross-national party-system fragmentation literature to a district level analysis. It is possible that the transitional nature of Indonesia makes the country a particularly weak test-case for some of the theories that have developed regarding the interaction of social heterogeneity and district magnitude. Thus this thesis echoes Stoll 's (2005b) call for more detailed case studies to explore the application of cross-national electoral fragmentation models at the district level. 68 This thesis is the second attempt to account for cross-district variation in Indonesian electoral fragmentation and the first to include magnitude and ethnic heterogeneity as key variables. 3 9 The thesis has suggested some causes of electoral fragmentation (the influence of Islamic modernism; ethnic heterogeneity); however, much work is left in this area. The models presented in Chapter 2 did not attempt to control for dispersed as opposed to concentrated ethnic groups. There is reason to believe that a dispersed ethnic group may feel more socially vulnerable than a concentrated ethnic group, thus making them more likely to vote for a large, unifying political party. A s well , the operationalization of several variables can be improved. A district-level tally of Muhammadiyah members would certainly be a refinement of the 'modernism' variable. A s well , previous cross-national research has suggested that ethno-linguistic fragmentation may be a better operationalization of ethnic heterogeneity than a basic calculation of the effective number of ethnic groups. The models presented in Chapter 2 did not fully take into account religious heterogeneity. Thus, there is still much research to be done in this area. The central argument of the thesis has been that the adoption of presidentialism caused increased electoral fragmentation. Future Indonesian election results provide one way o f testing the validity of this claim. The rules of the presidential game are scheduled to change before the 2009 elections. Presidential ballot access w i l l be restricted to parties or coalitions with at least 15% of the legislative vote. This higher threshold should reduce the incentive for presidential hopefuls to start their own legislative parties. Some o f the core claims o f the thesis, such as the electorate's preference for a national unity presidential ticket and its comparative willingness to cast a niche legislative vote, can still 3 9 King (2003) used an eclectic mix of variables that ranged from illiteracy to economic inequality. 69 be evaluated using data gathered from future elections. A t the very least this thesis suggests that a full understanding of Indonesian party politics requires some reference to the countries unique executive-legislative structure. 70 Bibliography Amor im Neto, Octavio & Gary Cox. 1997. "Electoral Institutions, Cleavage Structures, and the Number o f Parties." American Political Science Review 41 (January): 149-74. Benoit, Kenneth. 2002. "The Endogeneity Problem in Electoral Studies: A Critical Re-examination of Duverger's Mechanical Effect." Electoral Studies 21 (March): 35-46. Blais, Andre & R. K . Carty. 1991. "The Psychological Impact of Electoral Laws: Measuring Duverger's Elusive Factor." British Journal of Political Science 21 (January): 79-93. Clark, Terry D . & J i l l N . Wittrock. 2005. "Presidentialism ad the Effect of Electoral Law on Postcommunist Systems: Regime Type Matters." Comparative Political Studies 38 (March): 171-88. Clark, Wi l l i am Roberts and Matt Golder. 2006. "Rehabilitating Duverger's Theory: Testing the Mechanical and Strategic Modifying Effects o f Electoral Laws." Comparative Political Studies 39 (August): 679-708. Collier, David, Henry E . Brady, and Jason Seawright. 2004. "Toward an Alternative V i e w of Methodology: Sources of Leverage in Causal Inference." In Henry E . Brady and David Collier (eds.) Rethinking Social Inquiry: Diverse Tools, Shared Standards. Lanham, M D : Rowman & Littlefield. Cox, Gary. 1997. Making Votes Count: Strategic Coordination in the World's Electoral Systems. Cambridge: Cambridge University Press. Cox, Gary W . & Matthew Soberg Shugart. 1996. "Strategic Vot ing Under Proportional Representation." Journal of Law, Economics & Organization 12 (October): 299-324. Croissant, Aurel . 2002. "Electoral Politics in Southeast and East As ia : A Comparative Perspective." In Aurel Croissant (ed) Electoral politics in Southeast & East Asia. Singapore: Friedrich-Ebert-Stiftung. D a Silva, Vi rg i l io Afonso. 2006. "Duverger's Laws: Between Social and Institutional Determinism." European Journal of Political Research 45 (January): 31-41. Effendy, Bahtiar. 2003. Islam and the State in Indonesia. Athens, O H : Ohio University Press. 71 Emmerson, Donald K . 2004. "Indonesia's Approaching Elections: A Year of Voting Dangerously?" Journal of Democracy 15 (Januaru): 94-108. Evans, Kev in Raymond. 2003. The History of Political Parties & General Elections in Indonesia. Jakarta: Arise Consultancies. Feith, Herbert. 1962. The Decline of Constitutional Democracy in Indonesia. Ithaca: Cornell University Press. Filippov, Mikha i l G ; Peter C. Ordeshook & Olga V . Shvetsova. 1999. "Party Fragmentation and Presidential Elections in Post-Communist Democracies." Constitutional Political Economy 10 (March): 3-26. Fukuyama, Francis, Bjorn Dressel & Boo-Seung Chung. 2005. "Challenge and Change in East Asia : Facing the Perils of Presidentialism." Journal of Democracy 16 (April): 102-16. Golder, Matt. 2005. "Presidential Coattails and Legislative Fragmentation." American Journal of Political Science 50 (January), 34-48. Golder, Sona Nadenichek. 2006. "Pre-Electoral Coalition Formation in Parliamentary Democracies." British Journal of Political Science 36 (April) : 193-212. Hazan, Reuven Y . & Gideon Rahat. 2000. "Representation, Electoral Reform, and Democracy: Theoretical and Empirical Lessons From the 1996 Elections in Israel." Comparative Political Studies 33 (December): 1310-36. Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman, O K : University o f Oklahoma Press. ( IFES. 2003. Some Questions About the Electoral System for the 2004 Indonesian General Election Answered. Available Online: <http://www.ifes.org/files/indonesia/political-laws/Q&A_Elec_Sys_2004.pdf>. Jakarta: IFES. Accessed: August 6, 2006. IFES. 2003b. Overview of Legal Framework for 2004 General Elections in Indonesia. Available Online: <http://www.ifes.org/files/indonesia/political-laws/Election_Law_Summary.pdf> Jakarta: IFES. Accessed: August 6, 2006. IFES. 1999. Allocation of Seats to Political Parties for the People's Representative Assembly (DPR): Implementation of the Proportional Representation System in Awarding National DPR Seats Based on Results of June 1999 Elections. Available Online: <http://www.ifes.org/publication/763267629bcc0295cf27ba0e73f49cl3/Allocatio 72 n_of_DPR_seats_to_Political_Parties_in_1999.pdf>. Jakarta: IFES. Accessed: August 6, 2006. International Crisis Group. 2003. Indonesia Backgrounder: A Guide to the 2004 Elections. Available Online: <http://www.crisisgroup.org/library/documents/asia/071_indonesia_guide_2004_ elections.pdf>. Jakarta: International Crisis Group. Accessed: August 6, 2006. International Crisis Group. 2001a. The Megawati Presidency. Available Online: <http://www.crisisgroup.org/library/documents/report_archive/A400412_100920 01 .pdf>. Jakarta: International Crisis Group. Accessed: August 6, 2006. International Crisis Group. 2001b. Indonesia's Presidential Crisis. Available Online: <http://www.crisisgroup.org/library/documents/report_archive/A400243_210220 Ol .pdfX Jakarta: International Crisis Group. Accessed: August 6, 2006. "Security comes before recovery: Democratic party." 2004. Jakarta Post. Jones, Mark P. 1999. "Electoral Laws and the Effective Number of Candidates in Presidential Elections." The Journal of Politics 61 (February): 171-84. King , Dwight Y . 2003. Half-Hearted Reform: Electoral Institutions and the Struggle for Democracy in Indonesia. Westport: Praeger. K ing , Dwight Y . 2000. "The 1999 Electoral Reforms in Indonesia: Debate, Design, and Implementation." Southeast Asian Journal of Social Science 2: 89-110. King , Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton: Princeton University Press. Kurniawan, Moch . N . 2003. "50 Parties make it to final K P U list." Jakarta Post, October 10. Lanti, Irman G . 2004. "Outlook on the Indonesian Parliamentary Election 2004." Unpublished Manuscript. Available Online: http://www.ntu.edu.sg/idss/publications/W orkingPapers/WP61 .PDF Liddle, Wi l l i am R. 2005. "Year One of the Yudhoyono-Kalla Duumvirate." Bulletin of Indonesian Economic Studies 41 (December) 325-340. Liddle, Wi l l i an R. & Saiful Mujani. Forthcoming. "Leadership, Party and Religion: Explaining Voting Behavior in Indonesia." Comparative Political Studies. Liddle, Wi l l i an R. & Saiful Mujani. 2006. "Indonesia in 2005: A New Multiparty Presidential Democracy." Asian Survey 46 (January/February): 132-9. 73 Liddle, Wil l ian R. & Saiful Mujani. 2005. "Indonesia in 2004" Asian Survey 45 (January/February): 119-26. Lieberson, Stanley. 1991 "Small N's and B i g Conclusions: A n Examination of the Reasoning in Comparative Case Studies Based on a Small Number of Cases." Social Forces 70 (December): 307-320. Lijphart, Arend. 1999. Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries. New Haven: Yale University Press. L P 3 E S . 2004. PDI-P Voters Switch to Democrat Party, PPP and PAN Voters Move to PKS. Available Online: <http://ndijurdil.ndi.org/pemilu04/download/press-release-april-7-e.pdf>. Jakarta: L P 3 E S . Accessed August 6, 2004. Mainwaring, Scott. 1993. "Presidentialism, Multipartism and Democracy: The Difficult Combination." In Comparative Political Studies 26 (July): 198-228. Mainwaring, Scott & Matthew Soberg Shugart 1997. "Presidentialism and Democracy in Latin America: Rethinking the Terms of the Debate." In Mainwaring and Shugart (eds) Presidentialism and Democracy in Latin America. Cambridge: Cambridge University Press. Mclntyre, Angus. 2005. The Indonesian Presidency: The Shift from Personal toward Constitutional Rule. Lanham, M D : Rowman & Littlefield. Monroe, Burt L . & Amanda G . Rose. 2002. "Electoral Systems and Unimagined Consequences: Partisan Effects of Districted Proportional Representation." American Journal of Political Science 46 (January): 67-89. Mujani, Saiful & R. Wil l ian Liddle. 2004. "Indonesia's Approaching Elections: Politics, Islam, and Public Opinion." Journal of Democracy 15 (January): 109-23. Mozaffar, Shaheen; James R. Scarritt & Glen Galaich. 2003. "Electoral Institutions, Ethnopolitical Cleavages, and Party Systems in Africa's Emerging Democracies." American Political Science Review 97 (August): 379-390. National Democratic Institute for International Affairs. 2004a. The People's Voice: Presidential Politics and Voter Perspectives in Indonesia. Available Online: <http://www.accessdemocracy.org/library/1729_id_focusgfoups.pdf>. Jakarta: National Democratic Institute for International Affairs. Accessed: August 6, 2006. National Democratic Institute for International Affairs. 2004b. Advancing Democracy in Indonesia: The Second Democratic Legislative Elections Since the Transition. Available Online: 74 <http://www.accessdemocracy.org/library/1728_id_legelections_063004.pdf>. Jakarta: National Democratic Institute for International Affairs. Accessed: August 6, 2006. Ordeshook, Peter C. & Olga V . Shvetsova. 1994. "Ethnic Heterogeneity, District Magnitude, and the Number o f Parties:" American Journal of Political Science 38 (February): 100-23. O'Rourke, Kevin . 2002. Reformasi: The Struggle for Power in Post-Soeharto Indonesia. Crows Nest, N S W : Al len & Unwin . Qodari, Muhammed. 2005. "Challenge and Change in East As ia : Indonesia's Quest for Accountable Governance." Journal of Democracy 16 (April): 73-87. Qodari, Muhammed. 2004. "Explaining reasons behind Democratic Party's Popularity." Jakarta Post Ap r i l 12. Reed, Steven R. 1990. "Structure and Behaviour: Extending Duverger's Law to the Japanese Case." British Journal of Political Science 20 (July): 335-56. Riker, Wi l l i am H . 1982. "The Two-Party System and Duverger's Law: A n Essay on the History o f Political Science." American Political Science Review 76 (December): 753-66. Samuels, David. 2002. "Presidentialized Parties: The Seperation of Powers and Party Organization and Behavior." Comparative Political Studies 35 (May): 461-83. Samuels, David. 2000. "Concurrent Elections, Discordant-Results: Presidentialism, Federalism, and Governance in Braz i l . " Comparative Politics 33 (October): 1-20. Sebastion, Leonard C. 2004. "The Paradox of Indonesian Democracy." Contemporary Southeast Asia 26 (August) 256-79. Sherlock, Stephen. 2004. The 2004 Indonesian Elections: How the System Works and What the Parties Stand For. Available Online: <http://www.cdi.anu.edu.au/CDIwebsite_1998-2004/indonesia/indonesia_downloads/Sherlock_Election%202004.pdf>. Canberra: Centre for Democratic Institutions. Accessed: August 6, 2006. Shugart, Matthew Soberg. 1995. "The Electoral and Institutional Sources of Divided Presidential Government." American Political Science Review 89 (June): 327-43. Shugart, Matthew Soberg and John M . Carey. 1992. Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press. 75 Shugart, Matthew Soberg; Melody El l is Valdini and Kati Suominen. 2005. "Looking for Locals: Voter Information Demands and Personal Vote-Earning Attributes of Legislators under Proportional Representation." American Journal of Political Science 49 (April): 437-49. Singh, Bilveer. 2003. "The 2004 Presidential Elections in Indonesia: Much Ado about Nothing?" Contemporary Southeast Asia 25 (December): 431-48. Situngkir, Hokky. 2004. "The Political Robustness in Indonesia." Unpublished Manuscript. Available Online: <http://cogprints.org/3625/01/robustness%5Fof%5Fpolitics%5Find.pdf>. Accessed August 6, 2006. Soekanto, Santi W E . 2004a. "Part 1 of 2: What's taking the P K S so long to make up its mind?" Jakarta Post June 21. Soekanto, Santi W E . 2004b. "Part 2 of 2: What's taking the P K S so long to make up its mind?" Jakarta Post June 22. Stoll, Heather. 2005a. "What's On the Political Agenda? Cleavage Salience and Dimensionality in Comparative Perspective." Unpublished Manuscript. Available Online: <http://www.polsci.ucsb.edu/faculty/hstoll/research/agenda.pdf>. Accessed: August 6, 2006. Stoll , Heather. 2005b. "Social Cleavages and the Number o f Parties: How the Measures you Choose Affect the Answers Y o u Get." Unpublished Manuscript. Available Online: <http://www.polsci.ucsb.edu/faculty/hstoll/research/cleavages.pdf>. Accessed: August 6, 2006. Strom, Kaare. 1990. " A Behavioral Theory of Competitive Political Parties." American Journal of Political Science 34 (May): 565-598. Sulisfyo, Hermawan. 2002. "Electoral Politics in Indonesia: A Hard Way to Democracy." In Aurel Croissant (ed) Electoral politics in Southeast & East Asia. Singapore: Friedrich-Ebert-Stiftung. Suryadinata, Leo. 2002. Elections and Politics in Indonesia. Singapore: Institute o f Southeast Asian Studies. Suryadinata, Leo; Ev i Nurvidya Arif in and Aris Ananta. 2004. Indonesian Electoral Behaviour: A Statistical Perspective. Singapore: Institute of Southeast Asian Studies. 76 Suryadinata, Leo; E v i Nurvidya Ar i f in and Aris Ananta. 2003. Indonesia's Population: Ethnicity and Religion in a Changing Political Landscape. Singapore: Institute o f Southeast Asian Studies. Taagepera, Rein. 1999. "The Number o f Parties as a Function of Heterogeneity and Electoral System." Comparative Political Science 32 (August): 531-548. Taagepera, Rein and Matthew Soberg Shugart. Seats and Votes: The Effects and Determinants of Electoral Systems. New-Haven: Yale University Press. Tan, Paige Johnson. 2004. "Party Rooting, Political Operators, and Instability in Indonesia: A Consideration of Party System Institutionalization in a Communally Charged Society." A Paper Presented to the Southern Political Science Association, New Orleans, Louisiana January 10. Available Online: <http://people.uncw.edu/tanp/SPSA2004.html>. Accessed: August 6, 2006. The As ia Foundation. 2003. Indonesia: A Report on Public Opinion and the 2004 Elections. Available Online: <http://www.asiafoundation.org/pdf/elections_survey_indo_03.pdf>. San Francisco: The As ia Foundation. Accessed: August 6, 2006. Tan, Paige Johnson. 2002. "Anti-Party Reaction in Indonesia: Causes and Implications." Contemporary Southeast Asia 24 (December): 484-508. Taufriqurrahman, M . 2004. "Many Muslims ignore party lines." Jakarta Post July 8. van Marie, A . 1974. "Indonesian Electoral Geography under the Orla and Orba." In Oey Hong Lee (ed.) Indonesia after the 1971 Elections. London: Oxford University Press. Vatter, Adrian. 20.03. "Legislative Party Fragmentation in Swiss Cantons: A Function of Cleavage Structures or Electoral Institutions?" Party Politics 9 (July): 445-61. Wagner, Steven. 1999. "Survey of the Indonesian Electorate Following the June 1999. Elections." Available Online: <http://www.ifes.org/publication/7ec63656c009fl 916259b30540e7820e/Indonesi a_Public_Opinion_2000.pdf>. Washington: IFES. Accessed: August 6, 2006. Wahyuni, Sri and And i Hajramurni. "2004. Susilo names Jusuf Kal la as running mate." Jakarta Post A p r i l 18. Wanandi, Jusuf. 2004. "The Indonesian General Elections 2004." Asia-Pacific Review 11 (November): 115-31. 77 Warner, Carolyn M . and Manfred W . Wenner. 2002. "Organizing Islam for Politics in Western Europe." Harvard W C F I A Project on Religion, Political Economy and Society, Working Paper Number #18. 78 Appendix A In 1999, each province consisted of a number of regencies. A policy of decentralization saw the creation of many new regencies between 1999 and 2004. In most cases the 2004 electoral districts closely corresponded to the one or more broad regencies that existed in 1999 (for which 1999 elections data is available). In a small number of cases the original 1999 regency was split between two 2004 electoral districts. The table below provides a list o f the 1999 regencies and the 2004 electoral district they were matched with for the purposes of this analysis. 2004 Electoral District 1999 Regency 2004 Electoral Distr ict 1999 Regency Aceh 1 Aceh Barat West Java 1 Bandung, Kota Aceh Besar West Java 2 Bandung Aceh Selatan West Java 3 Cianjur Aceh Sungkil Sukabumi Aceh Tenggara Sukabumi, Kota Banda Aceh West Java 4 Bogor Pidie Bogor, Kota Sabang West Java 5 Ciamis Simeulue Kuningan Aceh 2 Aceh Tengah Aceh Timur West Java 6 Karawang Purwakarta Aceh Utara West Java 7 Cirebon South Sumatra 1 Musi Banyu Asin Musi Rawas Cirebon, Kota Indramayu Palembang West Java 8 Majalenga Pangkal Pinang Subang South Sumatra 2 Lahat Sumedang Muara Enim West Java 9 Bekasi Ogan Komaring Mir Bekasi, Kota Ogan Komaring Ulu Depok, Kota Bangka Belitung Bangka Belitung West Java 10 Garut Tasikmalaya Bengkulu Bengkulu Bengkulu Selatan Bengkulu Utara Rejang Lebong Bantenl Cilegon, Kota Lebak Pandeglang Serang Jambi Batang Hari Jambi Banten 2 Tangerang Tangerang, Kota Kerinci Central Java 1 Kendal Muara Bungo Tebo Salatiga, Kota Sarolangun Bangko Semarang Tanjung Jabung Semarang, Kota 1 Riau Bengkalis Dumai Indragiri Hilir Central Java 2 Demak Jepara Kudus Indragiri Hulu Central Java 3 Blora 79 Kampar Pekanbaru Grobogan Pati Rem bang Riau Islands Batam Kepulauan Riau Central Java 4 Karanganyar Sragen Wonogiri West Sumatra 1 Padang Panjang, Kota Padang, Kota Pesisi Selatan Sawalunto/Sijunjung Sawalunto/Sijunjung, Kota Solok Solok, Kota Tanah Datar Central Java 5 Boyolali Klaten Sukoharjo Surakarta, Kota Central Java 6 Magelang Magelang, Kota Purworejo Temanggung Wonosobo West Sumatra 2 Agam Bukittinggi, Kota Limapuluh Kota Padang Pariaman Pasaman Payakumbah, Kota Central Java 7 Banjarnegara Kebumen Purbalingga North Sumatra 1 Deli Serdang Medan Tebing Tinggi Central Java 8 Banyumas Cilacap Central Java 9 Brebes Tegal Tegal, Kota North Sumatra 2 Labuhan Batu Mandailing Natal Nias Sibolga Tapanuli Selatan Tapanuli Tengah Tapanuli Utara Toba Samosir Central Java 10 Batang Pekalongan Pekalongan, Kota Pemalang East Java 1 Sidoarjo Surabaya, Kota North Sumatra 3 Asahan Binjai Dairi Karo Langkat Pematang Siantar Simalungun Tanjung Balai East Java 2 Pasuruan Pasuruan, Kota Probolinggo Probolinggo, Kota East Java 3 Banyuwangi Bondowoso Situbondo East Java 4 Lumajang Jember Lampung 1 Bandar Lampung Lampung Barat Lampung Selatan Tenggamus East Java 5 Malang Malang, Kota East Java 6 Blitar Blitar, Kota Kediri Kediri, Kota Tulungagung Lampung 2 Lampung Tengah Lampung Timur Lampung Utara Metro Way Kanan Tulang Bawang East Java 7 Magetan Ngawi Pacitan Ponogoro West Kalimantan Bengkayang Kapuas Hulu 80 J Ketapang Trenggalek Pontianak East Java 8 Jombang Pontianak, Kota Madiun Sambas Madiun, Kota Sanggau Mojokerto Sintang Mojokerto, Kota Central Kalimantan Barito Selatan Nganjuk Barito Utara East Java 9 Bojonegoro Kapuas Gresik Kotawaringin Barat Lamongan Kotawaringin Timur Tuban Palangakaraya East Java 10 Bangkalan East Kalimantan Balikpapan Berau Bulongan Pamekasan Sampang Sumenep Kutai Jakarta 1 Jakarta Timur Pasir Jakarta Utara Samarinda Jakarta Pusat Tarakan Jakarta 2 Jakarta Barat South Kalimantan Banjar Jakarta Selatan Banjar Baru, Kota Yogyakarta Bantul Banjarmasin Gunungkidul Barito Kuala Kulonprogo Hulu Sungai Selatan Sleman Hulu Sungai Tengah Bali Badung Hulu Sungai Utara Bangli Kotabaru Buleleng Tabalong Denpasar Tanahlaut Gianyar Tapin Jembrana South Sulawesi 1 Bantaeng Bone Bulukumba Karang Asem Klungkung Tabanan Gowa West Nusa Tenggara Bima Jeneponto Dompu Maros Lombok Barat Selayar Lombok Tengah Sinjai Lombok Timur Takalar Mataram Ujung Pandang Sumbawa South Sulawesi 2 Barru Enrekang Luwu Majene Mamuji Pangkajene Kepulauan East Nusa Tenggaral Alor Ende Flores Timur Manggarai Ngada Sikka Pare-pare East Nusa Tenggara 2 Belu Pinrang Kupang Polewali Mamasa Kupang, Kota Sidenreng Rapang Sumba Barat 81 Soppeng Tanatoraja Wajo . Sumba Timur Timor Tengah Selatan Timor Tengah Utara Central Sulawesi Banggai Buol Toli-toli Donggala Palu . • Poso Maluku Ambon, Kota Maluku Tengah Maluku Tenggara North Maluku Halmahera Tengah Maluku Utara North Sulawesi Bitung Bolaang Mangondow Manado Minahasa Sangihe Talaud Papua Biak Namfor Jayapura Jayapura, Kota Jayawijaya Merauke Mimika Nabire Paniai Puncak Jaya Yapen Waropen Gorontolo Gorontalo Gorontalo, Kota Southeast Sulawesi Buton Kendari Kendari, Kota Kolaka Muna West Irian Jaya Fak-Fak Manokwari Sorong 82 Appendix B Suryadinata, Ar i f in , and Ananta's Indonesia's Population (2003) provides provincial level ethnicity data. The data are expressed in terms of an ethnic groups percentage of the provincial population. Only percentages are given for the most prominent ethnic groups in the province. The total of the prominent ethnic groups does not always equal 100%. This means that there is often a large share of the population that is unaccounted for. For the purposes o f this thesis, only the ethnic groups over 5% of the provincial population were added to the equation with the specific value given in Indonesia's Population. The 'other' group was dealt with according to the following method. First, the total share of the 'other' category was divided by .05. The quotient was rounded up to the nearest whole number. It was assumed that this was the number of ethnic groups in the 'other' category. The 'other' share was divided by this hypothetical number of ethnic groups. This provided the population share for each hypothetical ethnic group. The effective number of ethnic groups was then calculated. In reality, many ethnic groups in the 'other' category fall far below 5% of the provincial population. It is thus l ikely that the measure underestimates the amount of ethnic heterogeneity in a province. 83 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items