This art icle was downloaded by: [ I ndiana Universit y Libraries] On: 04 Novem ber 2011, At : 12: 51 Publisher: Rout ledge I nform a Lt d Regist ered in England and Wales Regist ered Num ber: 1072954 Regist ered office: Mort im er House, 37- 41 Mort im er St reet , London W1T 3JH, UK Representation Publicat ion det ails, including inst ruct ions f or aut hors and subscript ion inf ormat ion: ht t p: / / www. t andf online. com/ loi/ rrep20 INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION: THE CASE OF SOUTH KOREA Timot hy S. Rich Available online: 04 Nov 2011 To cite this article: Timot hy S. Rich (2011): INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION: THE CASE OF SOUTH KOREA, Represent at ion, 47: 4, 457-470 To link to this article: ht t p: / / dx. doi. org/ 10. 1080/ 00344893. 2011. 611161 PLEASE SCROLL DOWN FOR ARTI CLE Full t erm s and condit ions of use: ht t p: / / www.t andfonline.com / page/ t erm s- andcondit ions This art icle m ay be used for research, t eaching, and privat e st udy purposes. Any subst ant ial or syst em at ic reproduct ion, redist ribut ion, reselling, loan, sub- licensing, syst em at ic supply, or dist ribut ion in any form t o anyone is expressly forbidden. The publisher does not give any warrant y express or im plied or m ake any represent at ion t hat t he cont ent s will be com plet e or accurat e or up t o dat e. The accuracy of any inst ruct ions, form ulae, and drug doses should be independent ly verified wit h prim ary sources. The publisher shall not be liable for any loss, act ions, claim s, proceedings, dem and, or cost s or dam ages what soever or howsoever caused arising direct ly or indirect ly in connect ion wit h or arising out of t he use of t his m at erial. INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION: THE CASE OF SOUTH KOREA Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 Timothy S. Rich How do alterations in electoral rules shape competition? This article analyses district level competition in six legislative elections in South Korea (1988 – 2008) to identify whether Duvergerian pressures are restricting the number of competitors. South Korea provides a unique opportunity to analyse variation in mixed election rules, where seats are allocated in both single-member districts (SMDs) and a nationwide list by proportional representation (PR), as the country has used three different electoral formulas over the time period. Through quantitative analysis, evidence of a gradual reduction in the number of candidates is evident, despite a history of regional voting, thus resembling Duvergerian expectations. Introduction How do alterations in electoral rules shape competition? In single-member districts one expects the reduction of parties in line with Duverger’s Law, expecting one party or candidate more than the district magnitude (M + 1). In the South Korean context however, regionalism has often been the defining trait, potentially preventing Duvegerian equilibrium from emerging. In addition, mixed member legislative systems—those which allocate seats in both single-member districts (SMDs) and by proportional representation (PR)—can have complex effects on national and district level competition, with cursory evidence from East Asia suggesting these systems behave more like pure majoritarian systems than mixed systems elsewhere. Whether the imposition of such electoral systems which appear to discourage smaller parties are actually having this effect at the district level remains poorly addressed. This article analyses district level data for all legislative elections since South Korea’s democratisation (1988– ) and in the process attempts to identify not only whether a general trend towards Duvergerian expectations is emerging but whether particular institutional or regional differences help or hinder these incentives. South Korea provides a unique opportunity to analyse six elections in total, under three slightly different but potentially salient institutional arrangements. In the first two elections after democratisation (1988 and 1992), South Korean legislative elections used a one-vote mixed system, where voters chose a district candidate and this vote not only determined district winners but was aggregated to the national level to allocate party list seats. Unlike most conceptions of a proportional party list tier, allocation was determined by the proportion of district seats won. If the party with the most district wins captured a majority of district seats, it was allotted two-thirds of the party list seats, with the remaining allocated proportionally. If the most successful party in district competition won less than half the seats, it still received half of the party list seats.1 In 1996 Representation, Vol. No. 47, No. 4, 2011 ISSN 0034-4893 print/1749-4001 online/11/040457–14 # 2011 McDougall Trust, London http://dx.doi.org/10.1080/00344893.2011.611161 458 TIMOTHY S. RICH Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 and 2000, a one-vote system endured, but the allocation of all party seats was by proportional representation. Finally for 2004 and 2008, South Korea moved to the more common two-vote style of mixed member legislative system, where voters had a separate ballot for both the district and the party list. This article will first summarise the literature relevant to district competition and how mixed legislative systems like that in Korea present challenges and potential tests to established theory. Descriptive and spatial analysis of district competition follows. Finally regression analysis provides greater leverage in understand how institutional factors shape district competition. Ultimately this article shows a general reduction in political fragmentation over time, consistent in part with the institutional incentives inherent with single-member districts (SMDs), even as historical regional voting patterns endure. Literature Review As one of the oldest and most consistent findings in political science, Duverger’s Law states that electoral systems comprised of single-member districts (SMDs) will result in twoparty competition, while proportional representation (PR) will produce multiparty systems (Duverger 1954). As only one candidate can win per district, voter and party actions ultimately restrict the number of competitors to two viable candidates. While this reduction may not be immediate, as voters, candidates and parties adapt to a new system and learn through trial and error, over time one would expect that votes would concentrate around the top two candidates as voters strategically defect from weaker candidates and smaller parties drop out (also see Cox and Shugart 1991; Riker 1982). Furthermore, with the reduction of candidates down to two, one would expect candidates to move towards the ideological centre to appeal to the median voter (Downs 1957). Counterexamples exist (e.g., Papua New Guinea), but overall Duverger’s Law has been empirically supported consistently at the district if not the aggregate national level in systems comprised solely of SMDs.2 Applying the logic of Duverger to more complex electoral institutions has a long history, but must be proceeded with caution. Previous work shows that multimember districts that are not allocated by proportional representation (e.g., the single non-transferable vote/SNTV) largely conform to Duverger’s M + 1 prediction of candidates, with greater deviation as district magnitude increases (e.g., Hsieh and Niemi 1998). Predicting the number of candidates in mixed legislative systems however is more complicated due to the coexistence of a party list tier that potentially creates incentives for a multitude of competitors in both tiers. Mixed systems—those which use two formulas to allocate seats within a single legislature—have become increasingly common in the past 20 years, with the vast majority of such systems combining SMDs with a proportional tier.3 From established Western democracies (e.g., Germany and New Zealand) to post-communist countries (e.g., Lithuania, Russia and Ukraine) to Latin America (e.g., Mexico and Venezuela), mixed systems have proliferated in virtually every region outside of North America. In East and Southeast Asia such systems have been similarly popular as the region in general has moved to more majoritarian designs compared to the global trend towards greater proportionality (Colomer 2005; Reilly 2007). In East Asia, Japan moved to a two-vote mixed system with their 1996 election, with South Korea and Taiwan later adopting remarkably similar reforms.4 The early literature in mixed systems expected that the individual tiers would act independently, similar to a natural experiment holding all other factors constant (e.g., Lancaster and Patterson 1990; Moser 1997; Reed 1999; Stratmann and Baur 2002). Increasing evidence Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION in numerous countries however suggests various forms of interdependence, usually resulting in a proliferation of parties in both tiers (Cox and Schoppa 2002; Ferrara et al. 2005; Herron and Nishikawa 2001).5 In addition, much of the literature on mixed systems, and electoral systems more generally, focus on national level competition, leaving district level competition poorly understood. Few multi-country studies of mixed systems at the district level exist (Ferrara and Herron 2005; Moser and Scheiner 2004), and those that do seldom cover multiple elections, a necessity to determine whether institutional constraints create accumulated pressures on district competition. The question thus becomes to what extent Duvergerian expectations are met in these mixed systems. Mixed systems vary in several fundamental ways. First, the type of tier linkage divides mixed systems into two broad categories. In mixed member proportional (MMP or compensatory) systems, the proportional list compensates for disproportionality in the district tier, with the overall results dependent on this list tier. The tiers are thus linked, with the list tier being more important in determining the overall distribution of seats. In mixed member majoritarian (MMM or parallel) systems, however, the two tiers are unlinked, producing potentially very disproportional results. Despite MMM being the more common variant much of the literature, at least in English, focuses on the MMP cases of Germany or New Zealand, potentially limiting our understanding of both variation within mixed systems and broader trends. Second, the number of votes may differ among mixed systems, usually with either one or two votes. In a one-vote system (e.g., Mexico), voters choose a district candidate and votes by party are aggregated nationally to allocate party list seats. Accordingly, voters who support a party but whose district candidate is not a viable contender may still vote for this candidate as a means to encourage party representation at the national level. In a two-vote system, voters have a separate vote for each tier, allowing one to split one’s vote if desired. Thus one may expect that in a two-vote system voters would follow the individual institutional incentives of the tiers, voting strategically in the district tier and voting sincerely in the party list. Parties, however, especially smaller ones, may have an incentive to continue to run district candidates regardless of their viability. This could be done to remind people to vote for the party in the party list (with evidence elsewhere suggesting that district nomination boosts votes for the party list) or as a means to gain leverage or concessions out of another party. In sum, the mixed systems appear to provide incentives for district competition to expand beyond Duvergerian expectations. The Korean Context Unlike the broader literature, evidence from mixed systems in East Asia suggests factors that may constrain political fragmentation (Reilly 2007; Reynolds et al. 2005). These factors may be culturally rooted, a resulting of the absence of Western-style social cleavages in which parties traditionally anchor themselves (Flanagan et al. 1991; McAllister 2007; Yu 2005).6 In Korea, for example, despite a more fluid party system history, new entries were largely just reformulations of earlier parties or splinters of these parties without a separate ideological identity, and therefore often disappeared after one election (Gwag 2009). Despite a growing ideological difference between the largest two parties (the Grand National Party and the Democratic Party), a large section of Korean voters continue to base support on personal characteristics. Similar to Japan, personal networks connect legislators to potential voters (Lim 2005), with supporters generally following when these legislators switch parties. Whereas long-time dominant parties endured in Japan and Taiwan for 459 Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 460 TIMOTHY S. RICH decades, in Korea parties traditionally were mere personal vehicles for dominant personalities from each region—Kim Dae Jung, Kim Young Sam and Kim Jong Pil. Even after their retirement from politics, parties remained largely based on personalities rather than deeper institutional roots. Not surprisingly, in every parliamentary election since democratisation, at least three parties have competed, all with their own regional base of support (Jaung 2000). While regionalism has pre-modern roots in the Three Kingdom Era (Hermanns 2004; Nahm 1993), for practical purposes political regionalism can be traced to authoritarian-era Korea. Presidents Park Chung Hee and Chun Doo Hwan promoted state-guided economic development in Seoul and the Yeongnam region (the home region of both leaders), while the Honam region remained largely underdeveloped. As expected Honam residents resented the discrimination (Steinberg 1998: 78). This pattern continued under the first two democratically elected presidents as well—Roh Tae Woo and Kim Young Sam—as both were also from Yeongnam. Only with the election of Kim Dae Jung in 1998 and his predecessor Roh Moo-hyun, the former from the Honam region, did such economic disparities begin to shift. Experts have argued that the saliency of regionalism is ingrained in Korean politics, with Samuel Kim (2008: 195) stating that ‘what seems to matter most in Korean politics is not what politicians do or say, but where they were born’. Browne and Kim (2003) argue that the effects of regionalism actually increased after democratisation. In 2000, for example, Kim Dae Jung’s party, the New Millennium Party (a predecessor to today’s Democratic Party) won 25 of 29 seats in his home region, winning none in Yeongnam, the stronghold of the Grand National Party (GNP). This extends to presidential elections as well, as three of the last four presidential elections have included three viable candidates with its own regional strongholds, each with at least 15% of the national vote. Even when the New Millennium Party chose Roh, a Yeongnam native in 2002, voters in the region remained overwhelmingly supportive of the GNP. Regionalism is often seen as a hindrance to the development of nationally oriented parties, with specific concerns in Korea about its effects on democracy (e.g., Cho 1998, 2000). Kim (2003) however argues that regionalism could in fact encourage district-focused competition. Although cultural and historical factors may have encouraged patterns of competition that were inconsistent with Duverger at the national level, one would still expect as unsuccessful candidates and parties realise the institutional constraints that they will gradually either bow out of competition or merge with viable entrants. Cursory evidence suggests such a move at the national level with the gradual emergence of two larger parties in Korea. Regionalism often undermines the formation of ideological-based parties, yet growing evidence from surveys and parties in Korea suggests an emerging ideological differentiation among the Democratic Party (DP) and GNP. As not all parties should be weighted equally as many non-viable entrants exist and small parties may eke out a few seats, I use the effective number of parliamentary parties, a common measure for weighting parties (Laakso and Taagepera 1979). Figure 1 shows the effective number of legislative parties (ENLP) by election.7 By 2000 the ENLP drops by nearly one effective party. The 2008 election may also be an outlier as factional splits within the GNP lead to the formation of the Pro-Park Alliance (PPA) and an increase in independents from 2004. By 2010 however PPA legislators announced their intent to rejoin the GNP. While suggesting a possible trend in Korean legislative elections at the aggregate level, this says little about competition at the district level. First, while nationally Korea may still appear above Duvergerian expectations, the theory itself would be more appropriately tested at the district level. In fact, other than Papua New Guinea, most so-called exceptions to Duverger’s Law are based on national and not district competition. Second, if district Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION FIGURE 1 Effective number of legislative parties (ENLP) by election competition is the focus, the effects of regionalism in reducing political fragmentation should result in outcomes more consistent with Duverger in these areas. Furthermore, with changes in electoral laws in Korea after every two elections, the accumulated knowledge and expectations for voters and parties may be stunted, creating greater room for deviation away from two-party (or two-candidate) competition. If a gradual progression is present, however, this would suggest that even relatively minor institutional variations can have profound effects on electoral behaviour. Variation within the category of mixed systems requires greater attention, with the case of South Korea being an appropriate means to capture some of this variation and whether competition is moving in the direction anticipated by Duverger. Generally speaking, mixed systems are presumed to encourage the proliferation of parties beyond what would be expected in a system solely comprised of SMDs. Thus, if a reduction of competitors is seen over time, this would suggest that Duvegerian pressures, even if weakened, are at play. In particular, the South Korean context allows one to see how the change from a one-vote mixed system to a two-vote system effects district competition. To my knowledge only one other mixed system ever changed from a one-vote to a two-vote system—Albania—from 1992 to 1996.8 In a one-vote system, smaller parties must run district candidates or engage in preelection coordination with like-minded parties to have any chance at seats, as the district vote at the aggregate also determines party list allocation. As such, small parties will not bow out of district competition. Similarly supporters of smaller parties will not tend to vote strategically in districts. In a two-vote mixed system, however, smaller parties have greater incentive to either focus on the list tier or, if support is regionally concentrated, focus only on these district races where they are viable. For example, the Democratic Labor Party (DLP) surprised experts when in 2004 it garnered ten seats, eight of them in the PR tier. Voters too have greater incentive to vote strategically, going with the most viable candidate closest to their preferences in districts while voting sincerely in the party list. The South Korean example provides an additional means to evaluate mixed system institution—the move from a disproportional list to a traditional proportional party list. The institutional rules for the 1988 and 1992 elections encouraged a majority government by allocating either half to two-thirds of party list seats to the party that won the most districts. This favoured broad parties, even if in reality composed of thinly connected coalitions. One may assume that 461 462 TIMOTHY S. RICH Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 under such restrictions, voters would shy away from voting for smaller parties as, even if they win a few districts, they are unlikely to receive any PR seats. Similarly potential parties, knowing there is no role as a coalitional pivot since the disproportional tier promotes majority governments, may opt not to run. With these being the first free elections as well, voters and wouldbe political hopefuls may have been hesitant to invest in smaller entities. Once a proportional tier was established in 1996, however, the incentives for small parties as well as experience in voting should have promoted additional entries. Based on the literature and basic knowledge of Korea, I plan to test the following hypotheses: H1: The move from a disproportional one-vote system (1988 –1992) to a proportional mixed one vote system (1996 – 2000) should lead to a temporary increase in vote fragmentation. H2: The move from a one-vote (1996 –2000) to two-vote mixed system (2004 –) will reduce vote fragmentation at the district level. H3: The regions of Honam and Yeongnam will have lower vote fragmentation than the rest of Korea. Analysis Although Duverger’s Law may be extended to national level elections, district level competition is the most appropriate means to evaluate the mechanical and psychological effects. Therefore I employ several means to analyse district level data from South Korea over six elections, relying on data collected by Kuniaki Nemoto (‘Korean National Assembly Election Data, 1988 – 2008′ ).9 The availability of multiple measures over six elections prevents one from overreliance on a single measure or election which may be an outlier. Furthermore, it provides for more rigorous statistical analyses. To address district competition, I focus on the effective number of district candidates, simply applying the commonly used weighting system for national party competition at this lower level. This should provide a more intuitive measure of whether district competition is conforming to Duvergerian expectations. Figure 2 shows histograms of the distribution of the effective number of candidates (ENCs) per district, showing a gradual move towards Duvergerian expectations, even while the effective number of candidates continues to hover over two. In 1988, the mean number of candidates per district was 2.87, rising as high as 3.08 in 1996 (consistent with H1) while dropping in each subsequent election. By 2008 the average number of candidates dropped to 2.43. Overall, district evidence suggests that institutional factors do have a similar restraining effect in mixed elections in Korea as a central tendency (although still above two candidates) gradual emerges. The effective number of candidates gives us some indication of district competition, but not a complete picture. For example, it does not give us any indicator as to the closeness of competition in these races. An alternative, Nagayama Triangles, provide added leverage in this regard. Nagayama (1997) noticed that graphing the vote percentage for winning candidates in single-member districts on the x-axis by the percentage received by the runner-up (which by definition cannot be more than 50%) created a triangle shaped distribution indicating patterns of competition. The top point of the triangle corresponds with competitive two-party competition where smaller parties receive little support, consistent with Duvergerian expectations in SMDs. Districts with multiparty competition and thus a larger percentage of Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION FIGURE 2 Histograms of the effective number of district candidates by election the vote going to smaller parties (in contrast to Duvergerian expectations) are found as one approaches the left corner of the triangle. Districts where two parties receive nearly all the vote, but one party dominates are located towards the right corner of the triangle. While perhaps not normatively appealing as such races are less competitive than those at the near the top peak of the triangle, such races would still be technically consistent with Duverger in that competition is effectively pared down to only two competitors. Figure 3 shows Nagayama Triangles for each National Assembly election. While a considerable number of districts remain towards the far right side consistent with one-party dominant districts in regional strongholds, in general a conversion away from multiparty competition and towards two-party competition is evident. Duvergerian expectations appear evident in district races, especially as institutional reforms have decreased rationales for voting for non-viable district candidates. Furthermore one would expect the percentage of the vote captured by the top two candidates to increase over time if Duvergerian pressures were in effect. Figure 4 shows this progression. The average percentage of the vote obtained by the top two candidates increased from 78.55% in 1988 to 86.81% 20 years later, only declining in 1996, the first year under the one-vote system with proportional tier. Similarly, the percentage of districts in which 90% or more of the vote is captured by the top two candidates increases from 27.15% in 1988 to 42.98% in 2008. The dip between 1988 and 1992 could be attributed in part to the coupling of democratic reforms and the fluidity of parties in the Korean context, while the initial move towards a one-vote system with a proportional tier in 1996 would again encourage an initial proliferation of district candidates hoping to eke out representation in the party list. Multivariate regression analysis allows for additional insight into the factors shaping district competition. First, OLS regressions broken down by year reaffirm the consistent effect of 463 Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 464 TIMOTHY S. RICH FIGURE 3 Nagayama Triangles of district competition by year FIGURE 4 Vote percentages for top two district candidates regionalism on competition (Table 1). Using the effective number of district candidates (ENCs) as the dependent variable, four independent variables were tested: controls for two regions (Honam and Yeongnam) as well as the size of the district in square meters (logged) and the number of votes in each district (logged). Across all six models, districts within Honam were negatively correlated with ENCs at the 0.001 level, resulting in a reduction of 0.41 to 1.5 ENCs. Districts within Yeongnam too were correlated and statistically significant in all but one model (1996), resulting in a reduction of ENCs between 0.27 and 0.705. Interestingly, the size of the coefficients in the regional variables generally declines over time, consistent with claims that regionalism itself is declining in saliency. Meanwhile area and the number of voters were positively correlated in all but the 1988 model, reaching statistical significance in several models but not consistently across elections. In sum, this verifies the continued impact of regional vote. That said, the variance explained in each model (based on the R2) declines greatly over the six models, from 0.5581 down to 0.1238, strongly Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 1988 Coef. ∗∗∗ Honam Yeongnam Area (log) Votes (log) Constant N R2 Notes: ∗∗∗ p , 0.001, ∗∗∗ SE 1996 Coef. ∗∗∗ SE 0.1147736 0.0924147 0.0273964 0.1539553 2.068627 -0.8243771 -0.4873555∗∗∗ 0.0261767 0.3530888∗∗ -1.434116 231 0.3145 0.1065915 0.0821173 0.0254461 0.1235872 1.722896 -1.504547 -0.1509096 0.0986955∗∗∗ 0.5212494∗∗ -4.323596 250 0.3458 0.145361 0.10873 0.029807 0.18683 2.38535 2000 Coef. SE 2004 Coef. SE 2008 Coef. SE -0.958659 -0.4120382∗∗∗ 0.0884042∗∗∗ 0.1714048 -0.7647651 225 0.2695 ∗∗ 1992 Coef. -1.275718 -0.7050237∗∗∗ -0.1839052∗∗∗ -0.0427652 7.224385∗∗∗ 221 0.5581 ∗∗∗ Honam Yeongnam Area (log) Votes (log) Constant N R2 SE p , 0.01, ∗ p , 0.05. 0.1138637 0.0821927 0.0208114 0.138972 1.708383 ∗∗∗ -0.4104786 -0.3062678∗∗∗ 0.0328178 0.0506248 1.517701 241 0.1286 0.0898352 0.0649084 0.0168139 0.1249089 1.54271 ∗∗∗ -0.4985377 -0.2706609∗∗∗ 0.0426461∗ 0.2129965 -0.5904738 242 0.1238 0.113472 0.079669 0.019559 0.158936 1.807825 INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION TABLE 1 Correlates of effective number of district candidates by year 465 Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 466 TIMOTHY S. RICH suggesting additional factors such as institutional constraints playing a major yet unmeasured influence. These regressions give us a snapshot of individual elections, but no direct indication about whether the differences in the type of mixed system employed affects competition. Thus next I run two sets of random-effects GLS regression models on the same dependent variable, pooling data from all six elections (Tables 2 and 3).10 The first controls for the effects of the imposition of reforms in 1996 and 2004 (coded Mixed 1 and Mixed 2) with the 1988 – 92 rules as the baseline. The second controls for individual elections using 1988 as the baseline. Additional controls are then added to both models, using the four variables used in the earlier election-year models. In Model 1, a decrease in the ENC is seen with each electoral reform, but only the latter is statistically significant (at the 0.001 level). In substantive terms, the ENC drops by 0.36 between the 1988 – 92 system to the present mixed system. This effect remains consistent when adding additional controls (Model 1a), all of which reach at least statistical significance at the 0.05 level or better. Substantively, the regional controls both show a greater effect that the institutional change, especially Honam which correlates with a reduction of nearly on effective candidate after controlling for other factors. Meanwhile the geographic size of the size and the number of voters in the district (both logged) were positively correlated with the ENC. Model 2 shows results generally consistent with the previous analysis. No statistically significant difference is seen between 1988 and 1992, but the first mixed election under a fully proportional tier is correlated with an increase in ENC at the 0.001 level. Every subsequent election however is negatively correlated with ENC, not only at the 0.001 level but with a larger coefficient signifying a greater substantive impact. When the same controls as in Model 1 are added (Model 2a), the same basic trend is seen. Surprisingly both controls for district size and voter population were positively correlated with the effective number of district candidates. Theoretically as the size of the potential voting population increases, the threshold for candidate viability should also increase. For example, TABLE 2 Correlates of effective number of district candidates (1988– 2008) Model 1 Mixed1 Mixed2 -0.0075823 0.444228 -0.357873∗∗∗ 0.046904 Honam Yeongnam Area(Log) Votes(Log) Constant N R2 2.83456∗∗∗ 0.037158 1410 0.0483 Notes: ∗∗∗ p , 0.001, ∗∗ p , 0.01, ∗ p , 0.05. Model 1a 0.0116932 0.0416578 -0.3439804∗∗∗ 0.0429463 -0.9328127∗∗∗ 0.0597614 -0.3928298∗∗∗ 0.0452403 0.0262525∗ 0.0120974 0.2482385∗∗∗ 0.0675994 -0.2044074 0.8698018 1410 0.2575 INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION TABLE 3 Correlates of effective number of district candidates (1988– 2008) Model 2 1992 -0.0997028 0.0542256 0.1865722∗∗∗ 0.0569775 -0.358255∗∗∗ 0.0591475 -0.3665037∗∗∗ 0.0592708 -0.5122054∗∗∗ 0.0594806 1996 2000 2004 Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 2008 Honam Yeongnam Area(Log) Votes(Log) 2.89542∗∗∗ 0.0460075 1410 0.0935 Constant N R2 Notes: ∗∗∗ Model 2a -0.1007008 0.0535933 0.2207207∗∗∗ 0.054992 -0.3396804∗∗∗ 0.0562881 -0.3673867∗∗∗ 0.0556375 -0.447709∗∗∗ 0.0579556 -0.9480683∗∗∗ 0.0592061 -0.404299∗∗∗ 0.0448756 0.0293991∗∗ 0.0120642 0.2597207∗∗∗ 0.0690028 -0.3307785∗∗∗ 0.8880615 1410 0.3085 p , 0.001, ∗∗ p , 0.01, ∗ p , 0.05. the resources necessary to be competitive in a population of 10,000 is a far lower threshold than where 100,000 are necessary. Similarly one would presume a similar constraint on would-be competitors as district size increases, both as the saliency of being a local political celebrity diminishes with distance but also the simple costs of campaigning in a larger geographical unit. Yet statistical evidence here as in most of the models in this study suggests the opposite relationship: as both area and the voting population increase, so does the effective number of district candidates. For area, the substantive effect is rather minor, with a 1% increase in district size positively correlating to less than a 0.03 increase in the ENCs across both pooled models. The voter population however appears more substantively significant, increasing the ENCs by nearly a quarter of a candidate. The underlying logic behind this relationship remains unclear, perhaps suggesting factors related to the personalization of district competition, the influence of the mixed system itself or factors truly unique to Korea, and deserves greater attention in future research. Conclusion This cursory analysis confirms that variations in institutional arrangements all commonly lumped under the category of mixed legislative systems have differential impacts on electoral competition. Regionalism may remain a strong characteristic in Korean electoral behaviour, however consistent with analysis suggesting the general decline in regionalism, this study suggests a general trend towards acting upon institutional incentives, namely the reduction of candidates at the district level and two-candidate if not two-party competition. Whether South Korean electoral politics will ever completely converge to two-party competition is 467 468 TIMOTHY S. RICH uncertain, especially as regional support for smaller parties and coordination among traditional coalitional partners allows such parties to eke out minimal representation. If evidence from Japan is any indication, a gradual move towards two-party dominant competition may take several elections under a two-vote while smaller parties endure but lose much of their influence as potential pivots. Even after five such elections in Japan, smaller parties endure while competition at both the district and national level has effectively become a two-party race. NOTES Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. I thank Prof. Aurel Croissant for his insight on these early elections. The mechanisms behind Duverger’s Law are inherently at the district level, yet are often applied to analysis of national politics, where counterexamples are common (e.g., India, Canada, Great Britain). Most of these counterexamples however are due to the concentrated regional support of an additional party, with district level competition in these areas largely consistent with Duverger. Other variants include but are not limited to multimember districts (e.g., SNTV or block vote) with a proportional tier. The literature overwhelmingly focuses on those which follow the SMD-PR combination. Interviews with officials in both Taiwan and Korea suggest that Japan was the implicit if not explicit frame of reference for such reforms. Duverger (1986) suggested that district competition could outweigh institutional incentives of the party list tier, resulting in a reduction of parties in the latter tier, however few examples of this have occurred. Rather, the literature focuses more on the party list tier contaminating district competition. Norris (2004: 10 –11) and Jou (2011: 49) also question the application and strength of leftright placements of respondents in East Asia. Independents were lumped together to calculate the ENLP. Albania also switched from MMM to MMP. I thank Daniel Bochsler for this insight. Available at: http://sites.google.com/site/knemoto1978/kuniakinemotodata. GLS models were chosen over OLS as the error term would likely be correlated over elections. This would be less of a concern in regressions restricted to one election. REFERENCES BROWNE, ERIC C. and SUNWOONG KIM . 2003. Regionalism in South Korean National Assembly elections: a votes components analysis of electoral change. Working Paper. Available at http://www.uwm. edu/≏kim/paper/Korean%20Regionalism_July%202003.pdf. CHO, KISUK . 1998. Regionalism in Korean elections and democratization: an empirical analysis. Asian Perspective 22(1): 135–56. CHO, KISUK . 2000. Jiyeokjuui Seongeowa Hapnijeok Yukwonja [Regionalism and Rational Voters]. Seoul: Nanam Press. COLOMER, JOSEP . 2005. It’s parties that choose electoral systems (or, Duverger’s Laws upside down). Political Studies 53: 1–21. COX, GARY W. and MATTHEW SHUGART . 1991. Comment on ‘Gallagher’s proportionality, disproportionality and electoral systems’. Electoral Studies 10(4): 348–52. INSTITUTIONAL VARIATION AND THE EFFECTS ON DISTRICT COMPETITION and LEONARD J. SCHOPPA . 2002. Interaction effects in mixed-member electoral systems. Comparative Political Studies 35(9): 1027–53. DOWNS, ANTHONY . 1957. An Economic Theory of Democracy. New York: Harper & Row. DUVERGER, MAURICE . 1954. Political Parties: Their Organization and Activity in the Modern State. London: Methuen & Company. DUVERGER, MAURICE . 1986. Duverger’s Law: forty years later. In Electoral Laws and Their Political Consequences, edited by Grofman Bernard and Arend Lijphart. New York: Agathon Press, pp. 69 –84. FERRARA, FEDERICO and ERIK S. HERRON . 2005. Going it alone? Strategic entry under mixed electoral rules. American Journal of Political Science 49: 16–31. FERRARA, FEDERICO, ERIK S. HERRON and MISA NISHIKAWA . 2005. Mixed Electoral Systems: Combination and its Consequences. New York: Palgrave. FLANAGAN, SCOTT, SHINSAKU KOHEI, ICHIRO MIYAKE and BRADLEY M. RICHARDSON . 1991. The Japanese Voter. New Haven, CT: Yale University Press. GWAG , JIN YOUNG . 2009. The changing alignment of Korean political parties and the instability of the political system. The Journal of the Korean Association of Party Studies 8 (1): 115– 45. Cited in Young-Rae Kim and Sang-Shin Park. 2009. The reform of electoral system and the change of party system in Korea. Paper presented at the International Conference on Elections in Taiwan, Japan, and Korea under the Mixed-Member Electoral Systems. Election Study Center at National Chengchi University, 24 May. HERMANNS, HEIKE . 2004. Local and national elections in South Korea. In Recent Elections and Political Trends in East Asia, edited by Christian Schaffer. Austria: Austrian Association of East Asian Studies, pp. 1 –39. HERRON, ERIK S. and MISA NISHIKAWA . 2001. Contamination effects and the number of parties in mixed superposition electoral systems. Electoral Studies 20:63–86. HSIEH, JOHN and RICHARD G. NIEMI . 1998. Can Duverger’s Law be extended to SNTV? The case of Taiwan’s legislative yuan elections. Electoral Studies 18: 101–16. JAUNG, HOON . 2000. Electoral politics and political parties. In Institutional Reform and Democratic Consolidation in Korea, edited by Diamond Larry and Don Chull Shin. Stanford, CA: Hoover Institution Press, pp. 43 –72. JOU, WILLY . 2011. How do citizens in East Asian democracies understand left and right? Japanese Journal of Political Science 12(1): 33–55. KIM, SAMUEL . 2008. Korea’s Democratization. New York: Cambridge University Press. KIM, WANG-BAE . 2003. Regionalism: its origins and substance with competition and exclusion. Korea Journal 43(2): 5–31. LAASKO, MARKKU and REIN TAAGEPERA . 1979. Effective number of parties: a measure with application to West Europe. Comparative Political Studies 12: 3–27. LANCASTER, THOMAS D. and DAVID PATTERSON . 1990. Comparative pork barrel politics: perceptions from the West German Bundestag. Comparative Political Studies 22(4): 458–77. LIM, SEONG-HO . 2005. Neither gridlock nor moderation, but administrative recharge: the irony of divided government in South Korea. Korean Journal of International Relations 45(5): 231–57. MCALLISTER, IAN . 2007. Social structure and party support in the East Asian democracies. Journal of East Asian Studies 7: 225–49. MOSER, ROBERT G . 1997. The impact of parliamentary electoral systems in Russia. Post-Soviet Affairs 13(3): 284–302. MOSER, ROBERT G. and ETHAN SCHEINER . 2004. Mixed electoral systems and electoral system effects: controlled comparison and cross-national analysis. Electoral Studies 23: 575–99. Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 COX, KAREN E. 469 470 TIMOTHY S. RICH Downloaded by [Indiana University Libraries] at 12:51 04 November 2011 NAGAYAMA, MASAO . 1997. Shousenkyoku no Kako to Genzai [The Past and Present of SMDs]. Paper presented at the 1997 annual conference of the Japan Political Science Association, Tokyo. NAHM, ANDREW . 1993. Korea Tradition and Transformation: A History of the Korean People. Seoul: Hollym Corporation. NORRIS, PIPPA . 2004. Electoral Engineering: Voting Rules and Political Behavior. New York: Cambridge University Press. REED, STEVEN R . 1999. Strategic voting in the 1996 Japanese general election. Comparative Political Studies 32(2): 257–70. REILLY, BENJAMIN . 2007. Democratization and electoral reform in the Asia-Pacific region: is there an ‘Asian model’ of democracy? Comparative Political Studies 40(9): 1350–71. REYNOLDS, ANDREW, BENJAMIN REILLY and ANDREW ELLIS . (Eds). 2005. Electoral System Design: The New International IDEA Handbook. Stockholm: International Institute for Democracy and Electoral Assistance. RIKER, WILLIAM H . 1982. The two-party system and Duverger’s Law: an essay on the history of political science. American Political Science Review 76: 753–66. STEINBERG, DAVID . 1998. Korea: triumph amid turmoil. Journal of Democracy 9(2): 76–90. STRATMANN, THOMAS and MARTIN BAUR . 2002. Plurality rule, proportional representation, and the German Bundestag: how incentives to pork barrel differ across electoral systems. American Journal of Political Science 46(3): 506–14. YU, CHING-HSIN . 2005. The evolving party system in Taiwan. Journal of Asian and African Studies 40: 105–23. Timothy S. Rich is a doctoral candidate in political science at Indiana University. His dissertation analyses the interaction of electoral institutions, culture, and party influences in East Asian mixed legislative systems. In 2010– 2011, he conducted field research in Taiwan, Korea, and Japan. Email: tsrich@indiana.edu