Electoral Reform and Party System Development in Japan and Taiwan: A Comparative Study
Author(s): Willy Jou
Source: Asian Survey, Vol. 49, No. 5 (September/October 2009), pp. 759-785
Published by: University of California Press
Stable URL: http://www.jstor.org/stable/10.1525/as.2009.49.5.759 .
Accessed: 15/05/2013 03:39
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
University of California Press is collaborating with JSTOR to digitize, preserve and extend access to Asian
Survey.
http://www.jstor.org
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
ELECTORAL REFORM AND PARTY
SYSTEM DEVELOPMENT IN JAPAN
AND TAIWAN
A Comparative Study
Willy Jou
Abstract
This article examines changes in patterns of party competition in Japan and
Taiwan following the introduction of mixed parallel electoral systems. National
and district level election results demonstrate the consolidation of a two-party
system in both countries. Differences in the speed and extent of this development are analyzed.
Keywords: electoral reform, parallel mixed system, effective number of parties
(ENP), disproportionality, volatility, Japan, Taiwan
Introduction
The correspondence between electoral systems and
party systems has generated an enormous amount of literature.1 While each
element may influence the other, most discussions focus on how institutional
rules influence patterns of party competition. The assumption that alterations
in the former automatically lead to changes in the latter has in part motivated
the overhauling of long-established electoral rules in Italy, New Zealand,
Willy Jou is a doctoral candidate in Political Science at the University of California, Irvine. He would like to thank the Center for Asian Studies at the University
of California, Irvine, for funding part of this research, and the anonymous reviewer for detailed comments and suggestions. Email: <[email protected]>.
1. See Arend Lijphart, Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945–1990 (Oxford: Oxford University Press, 1994); David M. Farrell, Electoral Systems:
A Comparative Introduction (Basingstoke: Palgrave, 2001); Pippa Norris, Electoral Engineering:
Voting Rules and Political Behavior (Cambridge: Cambridge University Press, 2004).
Asian Survey, Vol. 49, Issue 5, pp. 759–785, ISSN 0004-4687, electronic ISSN 1533-838X. © 2009
by The Regents of the University of California. All rights reserved. Please direct all requests for permission to photocopy or reproduce article content through the University of California Press’s Rights and
Permissions website, at http://www.ucpressjournals.com/reprintInfo.asp. DOI: AS.2009.49.5.759.
759
AS4905_02_Jou.indd 759
10/13/09 4:49:55 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
760 asian survey, vol. xlix, no. 5, september/october 2009
and Japan in the early 1990s,2 and in Italy again in 2005. These events not
only brought renewed attention to the topic of choosing electoral systems3
but also allow empirical testing of theoretical postulates on consequences
of electoral reform for party system change.
However, difficulties arise in making cross-national comparisons of how
electoral rule changes shape subsequent party system development. This is
because the reforms enumerated above, as well as others enacted during
recent decades, do not entail changes both from and to the same systems.
Thus, it is often impossible to disentangle the effects of electoral reform
from country-specific political dynamics when drawing a causal arrow from
institutional to party system change. A rare exception involves the cases of
Japan and Taiwan, both of which abandoned the time-honored single nontransferable vote (SNTV) in favor of a mixed parallel (also known as mixedmember majoritarian or mixed superposition) system.4 A comparative study
of these two countries thus permits one to distinguish features of party and
voter behavior amenable to electoral reform from those that are less readily
modified because of idiosyncratic local conditions.
In examining party system change in Japan and Taiwan, this study focuses on two related aspects: similarities in trends toward two-party competition at both national and district levels, and differences in how quickly
bipolar equilibrium is attained in each country. Pre- and post-reform election results are presented to demonstrate that while convergence toward a
two-party system is unmistakable in both Japan and Taiwan, this process
has occurred much more rapidly in the latter. Several possible explanations
for this difference are discussed, including issue cleavages, constitutional
structure (parliamentary versus presidential system), consistency of rules
for national and sub-national elections, and distinctive features contained in
the reforms enacted in each country.
This study is organized as follows: the next sections briefly describe
electoral reform in Japan and Taiwan, and review literature concerning its
2. Richard S. Katz, “Reforming the Italian Electoral Law, 1993,” in Mixed-Member Electoral Systems: The Best of Both Worlds? eds. Matthew Soberg Shugart and Martin P. Wattenberg (Oxford: Oxford University Press, 2001); Raymond V. Christensen, “Electoral Reform
in Japan: How It Was Enacted and Change It May Bring,” Asian Survey 34:7 (July 1994), pp.
589–605; Jack Vowles, “The Politics of Electoral Reform in New Zealand,” International Political Science Review 16:1 (1995), pp. 95–115.
3. Dieter Nohlen, “Changes and Choices in Electoral Systems,” in Choosing an Electoral
System: Issues and Alternatives, eds. Arend Lijphart and Bernard Grofman (New York: Praeger, 1984); Carles Boix, “Setting the Rules of the Game: The Choice of Electoral Systems in
Advanced Democracies,” American Political Science Review 93 (1999), pp. 609–24.
4. Only the House of Representatives in Japan, the more powerful chamber of the country’s bicameral Parliament, is discussed in this paper. Elections in Taiwan refer to the Legislative Yuan.
AS4905_02_Jou.indd 760
10/13/09 4:49:55 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 761
anticipated repercussions. Indicators derived from election results under
both old and new rules are then presented, including indices of fragmentation
and disproportionality. Next, constituency results are analyzed to explore
trends in party system change at the district level. Some hypotheses explaining
contrasts in patterns of party competition observed in the two countries
are probed before concluding remarks.
New Electoral Systems in Japan
and Taiwan
The SNTV approach was used to choose democratically elected legislatures
in Japan from 1947 to 1993, and in Taiwan from 1992 to 2004.5 Under this
system, each district elected two to six members in Japan, and one to 16
members in Taiwan.6 Where a district elects three or more representatives, a
party seeking a parliamentary majority must run multiple candidates, giving
rise to intra-party competition and candidate-centered campaigns. This is
because aspiring members of Parliament (MPs) unable to rely on their party
labels alone resorted to cultivating personal votes instead. Since SNTV
tends to achieve “semi-proportional” seat distributions,7 small parties could
acquire representation and had little incentive for cooperation. As for larger
parties, their seat share depended not only on the number of votes garnered
but also on the number of candidates nominated in each district and the
distribution of votes among them.8 Reed’s assessment that SNTV promoted
“issue-free elections, political corruption, and one-party dominance”9 in
Japan was equally applicable to Taiwan.
The sequence of events leading to the enactment of electoral reform in
Japan and the strategic considerations that motivated proponents of reform
are already well documented;10 Lin describes similar processes and incentives
5. SNTV was also used in Japan before World War Two, and during decades of authoritarian rule in Taiwan, though election outcomes were heavily influenced by government
intervention.
6. About one-fifth of seats were allocated under proportional representation (PR) under
the old system in Taiwan, but voters did not cast a separate ballot for district candidates and
party lists. Instead, PR seats were filled in proportion to votes cast for district candidates
nominated by each party.
7. Arend Lijphart, Democracies: Patterns of Majoritarian and Consensus Government in
Twenty-One Countries (New Haven, Conn.: Yale University Press, 1984).
8. Gary W. Cox and Emerson Niou, “Seat Bonuses under the Single Non-Transferable
Vote System: Evidence from Japan and Taiwan,” Comparative Politics 26:2 (January 1994),
pp. 221–36.
9. Steven Reed, “Japan: Haltingly Toward a Two-Party System,” in The Politics of Electoral
Systems, eds. Michael Gallagher and Paul Mitchell (Oxford: Oxford University Press, 2005).
10. Christensen, “Electoral Reform in Japan,” pp. 589–605; Sadafumi Kawato, “Strategic
Contexts of the Vote on Political Reform Bills,” Japanese Journal of Political Science 1:1 (May
AS4905_02_Jou.indd 761
10/13/09 4:49:55 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
762 asian survey, vol. xlix, no. 5, september/october 2009
in the Taiwanese case.11 Suffice it to summarize here key features of the new
system. In Japan, the legislature comprises 300 single-member districts
(SMDs) and 200 (later reduced to 180) seats allocated through PR in 11
regional blocs.12 In Taiwan, there are 73 SMDs and 34 PR seats in a single
nationwide constituency, as well as six seats reserved for indigenous
populations.13 In both countries, voters cast separate SMD and PR ballots,
and seat allocation under each component is independent of the other.
Parties in Japan must receive at least 2% of votes in each regional bloc to
gain PR seats, while the threshold in the nationwide PR district in Taiwan
is markedly higher at 5%.
Although this difference may have a bearing on the chances of small
parties to enter Parliament, in practice the 2% threshold in Japan is meaningless, because the effective threshold14 in even the bloc with the largest
number of seats exceeds this legal minimum. This means that a party obtaining slightly above 2% of PR votes would still be unable to secure a
seat. At present the average PR bloc magnitude is 16.36, which translates
into an effective threshold of 4.3%. Small parties are disadvantaged by the
use of regional blocs rather than a nationwide constituency as the unit of
PR seat distribution. In contrast, the 5% threshold in Taiwan considerably
exceeds the effective threshold of 2.1%, thus constituting a real hurdle for
small party representation.
It comes as no surprise that large and small parties adopted opposite
stances toward electoral reform. The long-ruling Liberal Democratic Party
(LDP) in Japan twice attempted to push through a first-past-the-post
(FPTP) system. Even after it was temporarily dislodged from power, the
LDP pushed to increase the percentage of SMDs under the new mixed
system, when negotiating with the new coalition government. Similarly, in
Taiwan the incumbent Nationalist Party (Kuomintang/Guomindang, KMT)
proposed a mixed system with a majority of seats allocated by FPTP, while
small parties preferred a fully proportional system. The most instructive
2000), pp. 23–51; Steven R. Reed and Ethan Scheiner, “Electoral Incentives and Policy Preferences: Mixed Motives Behind Party Defections in Japan,” British Journal of Political Science
33:3 (July 2003), pp. 469–90.
11. Jih-wen Lin, “The Politics of Reform in Japan and Taiwan,” Journal of Democracy
17:2 (April 2006), pp. 118–31.
12. With district magnitude ranging from seven to 33 seats.
13. Two three-member districts, one each for highland and lowland aborigines. These seats
are excluded from the discussion below because their allocation mechanism differs from both
SMDs and PR.
14. Effective threshold = 75%/(m + 1); m = district magnitude. See Rein Taagepera and
Matthew Soberg Shugart, Seats and Votes: The Effects and Determinants of Electoral Systems
(New Haven, Conn.: Yale University Press, 1989); Lijphart, Electoral Systems and Party
Systems.
AS4905_02_Jou.indd 762
10/13/09 4:49:56 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 763
case may be that of Komeito (Clean Government Party) in Japan: when it
was part of the New Frontier Party (NFP), a large party that encompassed
both opposition members and LDP defectors, it supported a greater emphasis on the FPTP component of the mixed system. But after the disintegration of the NFP left Komeito on its own, it began to advocate a return
to the old SNTV system and, failing this, vigorously opposed reducing the
number of seats allocated by PR.
Consequences for Patterns of Party
Competition: Some Predictions
Although Duverger’s proposition that “the simple-majority single-ballot
system favors the two party system” comes closer than most hypotheses
in the social sciences to attaining the status of a law,15 a mixed system
juxtaposing SMD and PR components mitigates the strength of this
causal relationship. The reason is that the mechanical effect Duverger
envisages would be “less punishing to marginal parties than the mechanical effect in standard SMD systems with a plurality rule.”16 Instead of
recognizing the futility of running SMD candidates under an FPTP system
and either giving up or seeking cooperation with a large party, small
parties in a mixed system retain an incentive to nominate district candidates
in order to boost their vote share in the PR component. In other words,
the existence of two tiers in the electoral system links party incentives and
strategies in one component to the other, even when seat distribution
mechanisms are unlinked.
With this “contamination effect” in mind, most scholars analyzing the
impact of electoral reform in Japan do not expect movement toward a
full-fledged two-party system. For example, Shiratori predicts that while
the SMD component would facilitate bipolar contests, the PR tier could
lead to moderate multi-party competition.17 Kohno forecasts that “Duverger’s law is not going to work and a two-party system is unlikely to
emerge,” and that the new electoral rules would instead produce coalition
15. Maurice Duverger, Political Parties: Their Organization and Activity in the Modern
State (London: Methuen, 1954).
16. Erik S. Herron and Misa Nishikawa, “Contamination Effects and the Number of Parties in Mixed-Superposition Electoral Systems,” Electoral Studies 20:1 (March 2001), pp.
63–86. See also Setsufumi Mizusaki and Hiroki Mori, “Tokuhyo Deta kara mita Heiritsusei
no Mekanizumu” [Examining the mechanism of parallel mixed system from voting data],
Senkyo Kenkyu [Japanese Journal of Electoral Studies] 13 (1998), pp. 50–59; Misa Nishikawa
and Erik S. Herron, “Mixed Electoral Rules’ Impact on Party Systems,” Electoral Studies 23:4
(December 2004), pp. 753–68.
17. Rei Shiratori, “The Politics of Electoral Reform in Japan,” International Political Science Review 16:1 (1995), pp. 79–94.
AS4905_02_Jou.indd 763
10/13/09 4:49:56 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
764 asian survey, vol. xlix, no. 5, september/october 2009
governments where small parties are needed to form parliamentary majorities.18 Regarding the number of parties, Thies believes that “the system
should continue to produce two large parties and a few small ones,”19 while
McKean and Scheiner foresee “three conservative-to-centrist parties” along
with the continued presence of the Japanese Community Party (JCP).20
Even when one narrows the discussion to SMDs alone, Cox and Schoppa
report that the effective number of electoral parties (known in political
science as ENEP) in SMDs does not decrease as Duverger’s law predicts,
but rather remains relatively stable over time. The reason, they argue, is
because “under mixed rules, Duverger’s incentives do not operate strongly
enough to move ENEP numbers below the mid-2.0 range.”21 The same logic
leads Wada to conclude that “Japan will never have a two-party system even
in the single-member district with plurality part” because small parties
continue to run in SMD races, resulting in “a terribly distorted multi-party
system.”22 In contrast, Reed identifies “a clear trend toward districts featuring bipolar competition between one representative of the governing
coalition and one representative of the opposition,” even though “movement towards a two-party system at the national level has been halting at
best” because of the main opposition Democratic Party of Japan’s (DPJ)
difficulty in presenting itself as a credible alternative government in its early
years.23
Another possible development is the perpetuation of patterns of party
competition under the previous system, notwithstanding alterations in
electoral rules. The prospect that “the new electoral system will reinforce
the unity and the electoral strength of the LDP while leaving its opposition
weak and divided”24 cannot be ruled out. This is not because party actors
are impervious to institutional changes but because such changes may
18. Masaru Kohno, “Voter Turnout and Strategic Ticket-Splitting under Japan’s New
Electoral Rules,” Asian Survey 37:5 (May 1997), pp. 429–40.
19. Michael F. Thies, “Changing How the Japanese Vote: The Promise and Pitfalls of the
1994 Electoral Reform,” in How Asia Votes, eds. John Fuh-sheng Hsieh and David Newman
(New York: Chatham House Publishers, 2002).
20. Margaret McKean and Ethan Scheiner, “Japan’s New Electoral System: La Plus Ça
Change . . . ,” Electoral Studies 19:4 (December 2000), pp. 447–77.
21. Karen E. Cox and Leonard J. Schoppa, “Interaction Effects in Mixed-Member Electoral Systems: Theory and Evidence from Germany, Japan, and Italy,” Comparative Political
Studies 35:9 (November 2002), pp. 1027–53.
22. Junichiro Wada, The Japanese Election System: Three Analytical Perspectives (London: Routledge, 1996).
23. Steven Reed, “Evaluating Political Reform in Japan: A Midterm Report,” Japanese
Journal of Political Science 3:2 (November 2002), pp. 243–63.
24. Gerald L. Curtis, The Logic of Japanese Politics: Leaders, Institutions, and the Limits
of Change (New York: Columbia University Press, 1999).
AS4905_02_Jou.indd 764
10/13/09 4:49:56 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 765
generate incentives and strategies unforeseen by, and operating contrary
to the expectations of, the designers of electoral reform. The continuation
of factions within the LDP is a clear example of organizational persistence
and adaptation in the face of changes in the rules of the game.25
Although the Taiwanese case has received less scholarly attention, one
can reasonably conjecture on the basis of the above discussion that a contamination effect would also be in evidence. At the same time, the absence
of regional PR blocs and a dual candidacy provision may mean fewer
closely contested SMDs; the higher legal threshold in Taiwan hampers
small parties to a greater extent than in Japan. Having a nationwide PR
bloc without allowance for dual candidacies may suppress the effective
number of elective parties, while a high threshold does the same for the effective number of parliamentary parties.26 These trends may in turn lead
not only to a more consolidated two-party system but even render one-party
dominance more likely. In this case, though, the identity of the dominant
party is less clear, because both the KMT and its main opponent, the
Democratic Progressive Party (DPP), have obtained popular vote majorities
before.
Moving toward a Two-Party System?
Some National Trends
Effective Number of Parties
The contours of a party system are defined by the number of parties. Instead
of merely counting how many parties obtain parliamentary representation,
however, political scientists generally take party size into account and
calculate the effective number of parties (ENP).27 The first column of
Table 1 reports this figure for all elections in Japan and Taiwan under the
new mixed system, along with a number of elections conducted under
SNTV to provide a basis for comparison. In both countries, one notes a
large decrease in ENP between the last pre-reform and the first postreform elections, affirming that a mixed system with a majority of seats
25. Gary W. Cox, Frances McCall Rosenbluth, and Michael F. Thies, “Electoral Reform
and the Fate of Factions: The Case of Japan’s Liberal Democratic Party,” British Journal of
Political Science 29:1 (January 1999), pp. 33–56; Ellis S. Krauss and Robert Pekkanen, “Explaining Party Adaptation to Electoral Reform: The Discreet Charm of the LDP?” Journal of
Japanese Studies 30:1 (Winter 2004), pp. 1–34.
26. The effective number of elective parties is calculated based on parties’ absolute vote
shares; the effective number of parliamentary parties is calculated from parties’ share of legislative seats.
27. Markku Laakso and Rein Taagepera, “The Effective Number of Parties: A Measure
with Application to Western Europe,” Comparative Political Studies 12:1 (April 1979), pp.
3–27.
AS4905_02_Jou.indd 765
10/13/09 4:49:56 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
766 table asian survey, vol. xlix, no. 5, september/october 2009
1 Effective Number of Parties and Combined Seat Shares of Two
Largest Parties
Effective Number
of Parties
(ENP)*
Japan
Pre-reform (SNTV)
1980
2.74
1983
3.24
1986
2.58
1990
2.71
1993
4.20
Post-reform (parallel mixed)
1996
2.94
2000
3.17
2003
2.60
2005
2.27
Taiwan
Pre-reform (SNTV)
1995
2.54
1998
2.51
2001
3.49
2004
3.27
Post-reform (parallel mixed)
2008
1.75
Combined Seat
Share of Two Largest
Parties (%)
(% Including
Large Parties’
Allies)
76.5
70.8
75.2
80.3
57.3
79.0
75.0
86.3
85.2
(82.4)
(82.9)
(94.2)
(91.7)
84.8
85.8
68.9
74.7
95.6
(99.1)
SOURCE: Compiled by author.
*See Note 27.
allocated by FPTP has a significant impact in reducing the number of
parliamentary parties. At the same time, it is noteworthy that whereas the
decline in legislative fragmentation in Japan only brought ENP down to
figures resembling those seen in the 1980s, in Taiwan post-reform ENP
was considerably lower than at any point during the pre-reform decade.
One should also observe that in accord with forecasts reported in the previous section, ENP in Japan shows a declining trend after initial fluctuation
(reflecting an unstable party system that saw a number of parties splitting
and merging throughout the 1990s), but never reaches as low as two. This
implies that although small parties cannot thrive as they did under the old
system, they are still able to maintain a presence in the legislature by relying on the new system’s PR component. The case of Taiwan points to the
opposite conclusion, suggesting not only that small parties have scarce
AS4905_02_Jou.indd 766
10/13/09 4:49:57 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 767
room for survival in the face of a strong majoritarian imperative but also
that instead of two-party competition, one may be observing a scenario
of one-party dominance. Explanations for these contrasting developments
will be examined below.
ENP reveals the degree of parliamentary fragmentation but not how
seats are distributed among parties. To assess whether electoral reform has
encouraged the development of a two-party system, the second column of
Table 1 displays the combined seat share of the two largest parties after
each election. This percentage increased markedly between the last SNTV
election and the first one conducted under the parallel mixed system. But
whereas similar levels of large-party strength were seen in Japan as late as
1990, in Taiwan the degree of parliamentary dominance by the two main
parties is unprecedented since democratic elections were first held in 1992.
These trends provide further evidence of evolution toward a two-party
system in both countries, but once again suggest that electoral reform has
a more drastic impact in squeezing out small parties in Taiwan.
Rather than depending solely on PR seats (including fielding hopeless
SMD candidates in order to boost the PR vote), another means of small
party adaptation involves alliance with large parties. Small parties can ensure both electoral survival and potential policy influence through these arrangements, and large parties may leave a number of SMDs open to allow
their junior partners a free run in return for extra support in most other districts. Such pacts become all the more important if a large party falls short
of a parliamentary majority and needs support from small parties to pass
legislation. This explains the LDP’s offer to form governing coalitions with
a series of small parties, most notably (and so far enduringly) with Komeito.
In the Taiwanese election of 2008, the KMT even invited small-party candidates to run under its label. When one counts not only the two largest parties but also their allies, figures in the second column of Table 1 reach even
higher proportions (shown in parentheses).
Having observed movement toward a two-party system in both Japan
and (especially) Taiwan following electoral reform, the next logical step is
to examine whether such a trend is confined to the parliamentary level or
reflects similar developments at the electoral level as well. In other words,
to what degree has electoral reform generated a psychological effect among
party elites and ordinary voters to supplement the mechanical effect translating votes into seats? Keeping in mind that the mixed parallel system is
composed of two independent parts, it makes sense to discuss the effect of
the SMD and PR components separately.
Table 2 gives the ENP at both the elective and parliamentary levels for
the same elections as shown in Table 1, with figures for single member
districts and PR blocs listed separately. Understandably, correspondence
AS4905_02_Jou.indd 767
10/13/09 4:49:57 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
768 table asian survey, vol. xlix, no. 5, september/october 2009
2 Elective and Parliamentary Effective Number of Candidates
Japan
Pre-reform (SNTV)
1980
1983
1986
1990
1993
Post-reform (parallel mixed)
1996 SMD
2000 SMD
2003 SMD
2005 SMD
1996 PR
2000 PR
2003 PR
2005 PR
Taiwan
Pre-reform (SNTV)
1995
1998
2001
2004
Post-reform (parallel mixed)
2008 SMD
2008 PR
ENP-elective
ENP-parliamentary
3.20
3.32
2.99
2.98
4.57
2.74
3.24
2.58
2.71
4.20
3.43
3.44
2.72
2.47
4.07
5.15
3.42
3.72
2.36
2.38
2.29
1.77
3.84
4.72
3.04
3.15
2.51
2.64
3.56
3.37
2.54
2.51
3.49
3.27
2.13
2.49
1.57
1.94
SOURCE: Ibid.
between the two ENP figures is much greater in the PR component than in
SMDs. The fact that ENP is always significantly higher in the PR portion
even at the elective level suggests both that many voters have been able to
distinguish between the two components of the new system from the start,
and that small parties either refrained from fielding SMD candidates or
received significantly lower vote shares in district races compared with their
party list votes.
More interestingly, when the post-reform figures shown in Table 1 are
disaggregated in Table 2, it is clear that evolution toward a two-party system in Japan is primarily a phenomenon in SMDs only, as Duverger’s law
predicts. ENP at the parliamentary level has remained relatively stable in
AS4905_02_Jou.indd 768
10/13/09 4:49:58 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 769
the first three elections under the new system, showing a decline only with
the LDP landslide of 2005. Instead, it is at the elective level where the bipolarizing logic of FPTP has exerted a strong influence, driving elective
ENP from 3.5 to 2.5. This figure partly reflects the presence of JCP candidates in nearly every district. Since the communists have decided to abandon this nomination policy in future elections,28 one can expect a further
decrease in elective ENP (though with little impact at the parliamentary
level, because the JCP has not won any district races since 1996). In contrast, parliamentary ENP has not shown a consistent decrease (in fact rising dramatically in 2000 before dropping back in the following election),
and there is little reason to expect this figure to fall much below 3.0 barring a major party realignment.
The introduction of a mixed parallel system in Taiwan also resulted in a
significant reduction in ENP under the SMD component at both elective
and parliamentary levels. Unlike Japan, the same holds true for the PR
portion. Moreover, the PR component in Taiwan yields a result much less
proportional than in Japan, at the expense of small parties. That small
parties in Taiwan face greater bipolarizing pressures can be surmised by
comparing the elective ENP in 2004 with that under the PR portion in 2008:
even where voters had an incentive to cast their ballots sincerely, they mostly
flocked to the two large parties, abandoning a number of established
minor parties and a host of new ones. Perhaps foreseeing this development,
and certainly bearing the 5% threshold in mind, one of the small parties,
the People First Party (PFP), did not even attempt to present its own list,
opting instead to negotiate with the KMT and have its candidates run
under the latter’s label.
Disproportionality
Results reported above lead appropriately to a discussion about disproportion­
a­lity, namely, the disparity between parties’ vote and seat shares. As figures
for pre-reform elections in Table 2 suggest, SNTV produced relatively
proportional results when translating votes into seats. Advocates of electoral
reform no doubt fully recognized that a mixed parallel system, especially
with a majority of seats allocated by FPTP, would increase disproportionality.
This was perceived not as a shortcoming but rather as a welcome antidote
to party fragmentation and, at least in the case of Japan, a change that
would encourage prospects of government alternation under a two-party
system (this had already taken place in Taiwan).
28. See, for example, Asahi Shimbun, September 6, 2008 (electronic edition), reporting that
the JCP will run only about 140 SMD candidates in the next general election.
AS4905_02_Jou.indd 769
10/13/09 4:49:58 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
770 table asian survey, vol. xlix, no. 5, september/october 2009
3 Disproportionality
Index of
Disproportionality*
Japan
Pre-reform (SNTV)
1980
6.57
1983
4.25
1986
7.17
1990
6.73
1993
6.35
Post-reform (parallel mixed)
1996 SMD
15.76
2000 SMD
15.57
2003 SMD
10.62
2005 SMD
22.99
1996 PR
2.63
2000 PR
2.49
2003 PR
4.01
2005 PR
4.65
Index of
Disproportionality*
Taiwan
Pre-reform (SNTV)
1995
1998
2001
2004
2.68
3.94
3.07
2.53
Post-reform (parallel mixed)
2008 SMD
23.08
2008 PR
7.30
SOURCE: Ibid.
*See Note 29.
The least-squares indices of disproportionality29 for Japan (1980–2005)
and Taiwan (1995–2008) are shown in Table 3, with separate figures listed
for the SMD and PR portions under the new system. Assuming relatively
even distribution of party support across the country, the higher the
number of closely fought SMD contests, the more proportional the outcome
is likely to be. Empirically, when the main opposition DPJ scored its best
performance against the LDP in 2003, one sees lower disproportiona­lity in
the SMD component, whereas the LDP landslide of 2005 raised dispro­
portion­ality to an unprecedented level. The same phenomenon is seen in
the high SMD disproportionality in Taiwan in 2008, when the KMT won
an overwhelming victory. In contrast to figures in previous tables, one
detects no obvious post-reform trend in Japan with regard to SMD
disproportionality. The only unmistakable change involves the significant
29. Using the formula proposed in Michael Gallagher, “Proportionality, Disproportionality, and Electoral Systems,” Electoral Studies 10:1 (March 1991), pp. 33–51.
AS4905_02_Jou.indd 770
10/13/09 4:49:58 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 771
decline in proportionality with the introduction of the mixed parallel
system in both countries.
Of greater interest are disproportionality indices in the PR segment,
where one notes a steady rise over the past several elections in Japan, although even the highest level reached so far (2005) remains below figures
obtained under SNTV. This suggests that small parties are increasingly
disadvantaged even in a component of the system designed to ensure their
continued viability. While allocation of PR seats in a single, nationwide
PR constituency would ensure greater proportionality, and hence brighter
prospects for small parties, than 11 regional blocs, this has been a fixed
feature since the introduction of the new system and cannot account for
variation in disproportionality over the past four elections. One possible
explanation is that the PR vote share of some small parties has fallen to
just below the effective threshold in a number of regional blocs.
In contrast, disproportionality in the PR segment in Taiwan considerably
exceeds both equivalent indicators in Japan and figures for SNTV elections.
This reaffirms the finding that whereas small parties in Japan are able to
maintain their presence under the new system through PR lists, those in
Taiwan have not been afforded similar opportunities. The disparity derives
from both the smaller number of PR seats and the 5% threshold imposed
in Taiwan. The more small parties there are that fail to clear this hurdle,
the greater the disproportionality. In the face of electoral oblivion, small
parties in Taiwan have a far stronger incentive than their Japanese counterparts
to seek alliance or merger with large parties. This was precisely the course
taken by the PFP, which subordinated itself under the KMT banner.
Ironically, greater proportionality would result only if other minor parties
follow suit (or exit the electoral stage).
Moving toward a Two-Party System?
District-Level Indicators
Having examined party system changes at the national level, attention now
turns to exploring patterns of party competition in individual constituencies
to further compare and contrast movement toward a two-party system in
Japan and Taiwan. This section focuses on SMDs, where the clearest evidence of such a trend is expected. One can demonstrate the emergence
and consolidation of a two-party system through the following characteristics, discussed in detail below: two main parties fielding nominees in a higher
number of districts, declining total and effective number of candidates per
district, increasing presence and competitiveness of the top two parties’
nominees in each district, and diminishing instances of minor party and independent victories in SMD contests.
AS4905_02_Jou.indd 771
10/13/09 4:49:59 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
772 table asian survey, vol. xlix, no. 5, september/october 2009
4 Number of SMD Candidates
Average
Effective
(Rural)
(Mixed)
(Urban)
Japan
1996
2000
2003
2005
4.20
4.00
3.42
3.30
2.95
2.77
2.41
2.40
2.38
2.35
2.22
2.36
2.86
2.73
2.38
2.34
3.62
3.22
2.62
2.50
Taiwan
2008
3.88
2.09
SOURCE: Ibid.
Presence of Top Two National Parties in SMDs
The average number of candidates per district for all elections conducted
under the new system is listed in the first column of Table 4. In order to
discount the impact of also-rans, the effective number of candidates is
also reported, employing the method used for ENP above. A progressive
decline in both indicators is readily observable in Japan, with a particularly sharp drop between 2000 and 2003, reflecting not only the marginalization of some small parties but also the merger between the DPJ and
Liberal Party that established the former as the only alternative governing
party to the LDP. If one disregards the perennial communist standardbearers, both the mean and the effective number of candidates per district
are app­roach­ing 2.0. Interestingly, a breakdown of SMDs by population
density30 reveals that the drop in the effective number of candidates is
largely an urban phenomenon: it has decreased by more than 1 in metropolitan districts but hardly changed at all in rural areas, where it has always
been low.
Figures shown for Taiwan in Table 4 reveal a greater disparity between
the mean and effective number of candidates than for any election in Japan,
reflecting a large number of also-ran candidates garnering minuscule vote
shares. Since 2008 was the first election in Taiwan conducted under the
mixed parallel system, small parties could not infer from past results when
estimating their level of support in each district. Even so, the bipolarizing
logic of FPTP in SMDs should have convinced most small party (and
independent) candidates that they stood little chance of winning. Once
again, with regard to the effective number of SMD candidates, movement
30. The 100 most densely populated SMDs are classified as urban; the next 100, mixed;
and the most sparsely populated 100 districts are labeled rural.
AS4905_02_Jou.indd 772
10/13/09 4:49:59 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 773
toward the anticipated equilibrium of 2.0 has proceeded at a much faster
pace in Taiwan than in Japan.
Cross-District Linkage
Reaching this equilibrium indicates bipolar competition at the district level,
but it does not necessarily mean that the same two parties are competitive
across a given country. It is possible that two parties vying for SMDs in one
part of the country only play insignificant roles in other regions. Keeping in
mind that Duverger’s mechanical and psychological effects strictly apply
only at the district level, Cox cautions that “there is nothing in the original
district-level logic (regarding either strategic voting or strategic entry) that
allows one to conclude that there will be two parties nationally.”31 Thus, the
causal arrow could be drawn not only from electoral system to patterns of
party competition but also in the reverse direction.
To examine whether the process of cross-district linkage (i.e., competition
across all districts featuring the same parties) is taking place, the first
question is whether the two leading national parties nominate candidates
across the entire country. Failure to do so may indicate weakness in organization or inability to find suitable standard bearers in certain districts, thus
permitting minor parties or independents to enter Parliament via SMDs. The
first column of Table 5 shows the number of SMDs without official nominees from the two leading parties at each election. Since large governing
parties often arrange electoral pacts with smaller parties, and refrain from
contesting some SMDs to allow the latter a clear run, figures in parentheses indicate the number of districts without any official government candidate. The weakness of the main opposition party in Japan (NFP in 1996,
DPJ thereafter) in the first few elections under the new system is striking.
The party left about one-fifth of all SMDs uncontested, although one sees
the DPJ gradually closing the gap with the LDP-led governing coalition.
This strengthening of the main opposition party into a nationally viable
contender presents evidence of increasing cross-district linkage and fulfills
a precondition for the development of a two-party system. In Taiwan, both
large parties were present in most SMDs.
Competitiveness of Top Two National Parties’
SMD Candidates
Nominating candidates does not guarantee a credible performance in
district contests, however, and one cannot speak of a nationwide two-party
31. Gary W. Cox, Making Voters Count: Strategic Coordination in the World’s Electoral
Systems (Cambridge: Cambridge University Press, 1997).
AS4905_02_Jou.indd 773
10/13/09 4:49:59 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
774 table asian survey, vol. xlix, no. 5, september/october 2009
5 Single Member Districts without Official Party Nominees
LDP
(LDP + Allies)
NFP/DPJ
Two Largest
Parties
Finished in
Top Spots*
(Including
Allies of Two
Largest
Parties)*
Japan
1996
2000
2003
2005
12
29
25
10
(6)
(3)
(5)
(1)
65
58
33
11
185
196
234
251
(193)
(211)
(234)
(259)
Taiwan
2008
KMT
4
(KMT + allies)
(1)
DPP
4
61
(64)
SOURCE: Ibid.
*Japan: out of 300 districts; Taiwan: out of 73 districts.
system if one or both main parties remain no more than a token presence in
many districts. To explore the strength of large parties in SMD races, the
last columns in Table 5 display the number of seats where the top two vote
getters were official nominees of the two leading parties nationwide.32 Taking the aforementioned electoral arrangements into account, figures in parentheses show the number of districts where the candidates representing
the governing coalition and the main opposition party finished in first and
second places (irrespective of which ended up as winner). As in column 1, a
clear trend emerges in Japan, as the percentage of SMDs featuring
competition between nominees from the top two national parties rose steadily
from 1996 to 2005. The most recent election involved a minor LDP split,
and a number of ex-LDP MPs ran and retained their seats as independents.
Because they have been readmitted to the party and will seek reelection
under the LDP banner, one can expect the figure in this column to increase
further in the next election.
Whereas the process of two-party system consolidation in Japan is
taking place only gradually, bipolar competition in Taiwan appears to
have been entrenched by the time the new system came into operation. As
Table 5 shows, candidates from both major parties dominated most SMD
races. This is less surprising on the side of the leading party and its allies,
but the second largest party has mounted a more serious challenge for the
32. In Japan: LDP and NFP in 1996; LDP and DPJ in 2000, 2003, 2005. In Taiwan: KMT
and DPP.
AS4905_02_Jou.indd 774
10/13/09 4:50:00 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 775
reins of power than any opposition party in Japan has been able to do so
far.33 By this measure also, one can conclude that Taiwan has advanced
further toward a consolidated two-party system than Japan.
Minor Party and Independent Candidates
According to Duverger’s law, both the mechanical and psychological effects of FPTP should lead to two-party competition in SMDs, because
small parties are squeezed by a bipolarizing logic. The aforementioned
contamination effect offers small parties a rationale to nominate SMD
candidates nonetheless, but only for the purpose of improving their PR
vote shares. This prompts Reed’s forecast about the impact of the new
electoral system in Japan that “the PR tier will not be able to rescue many,
if any, small parties from the powerful incentives toward two-candidate
competition in the single-member districts.”34 Independent candidates are
also not expected to fare well under the new system because they face the
pressure toward bipolar competition without the potential safety net of PR
lists.
Table 6 shows the number of SMDs secured by minor parties (i.e., other
than the two leading national contenders) and independents since the
implementation of the new system, with figures in parentheses indicating
the number of seats won by small parties not belonging to electoral pacts
with any large party.35 The two sets of figures follow divergent trends in
Japan: while small parties (at least those outside the governing coalition)
become less able to cling to districts, independent candidates continue to
thrive.36 In many districts won by independents, the main opposition party
was too weak to be a factor, leaving the battle to be fought between the
LDP and one unaffiliated candidate.37 In striking contrast to Japan, small
33. For the purpose of this discussion, the KMT is considered the leading party in Taiwan
despite having lost two consecutive presidential elections, because the party and its allies commanded a legislative majority.
34. Reed, “Evaluating Political Reform,” p. 260.
35. Even though the LDP was governing in coalition with the Social Democratic Party of
Japan (SDPJ) and the New Harbinger Party (NHP) at the time of the 1996 election, the LDP
ran candidates in three of the four SMDs won by the SDPJ, and one SMD where the NHP
won. These two junior partners also nominated candidates in several other districts in direct
competition with the LDP. The Taiwanese election of 2008 saw two SMDs where the KMT
and an allied party (the Non-Partisan Solidarity Association in one case, the People First
Party in the other) were vying for the top spot, leaving the DPP far behind.
36. This does not necessarily contradict Duverger’s law, which only hypothesizes the emergence of two leading candidates as a consequence of FPTP without stipulating that these
contenders must come from the two largest parties nationwide.
37. Because more than half of independent victories in 2005 are attributable to former
LDP MPs running without a party label after being denied official nominations, one can
AS4905_02_Jou.indd 775
10/13/09 4:50:00 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
776 table asian survey, vol. xlix, no. 5, september/october 2009
6 SMDs Won by Small Parties and Independent Candidates
SMDs Won by
Small Parties
Japan
1996
2000
2003
2005
Taiwan
2008
(Excluding Allies
of Large Parties)
SMDs Won by
Independents
26
23
12
11
(20)
(9)
(2)
(3)
9
20
15
18
2
0
1
SOURCE: Ibid.
parties in Taiwan stood no chance of winning district races even in the
first election under the new system. This is not for lack of trying, as evidenced
by the relatively high average number of candidates per district shown in
Table 2.
Volatility
Another indicator of party system consolidation can be found in declining levels of inter-party volatility. One may see the party system in a state
of flux in the immediate aftermath of electoral reform, as political elites
and entrepreneurs try to assess the potential impact of new rules of the
game in deciding whether to enter or exit the electoral market, while voters face the disappearance of some familiar party labels and the emergence of novel choices. As the party system approaches equilibrium,
however, more voters should form stable preferences as parties no longer
surface and vanish at each election, leading to lower volatility. Note that
this holds true for the stabilization of any system regardless of the number of parties at equilibrium.
The first column of Table 7 shows volatility between each pair of elections since the last polls conducted under SNTV,38 as indicated by the
expect fewer independent wins in the next election because all but one of these MPs have been
readmitted to the party.
38. Party vote shares under the mixed parallel system refer to vote percentage in PR
blocs. This gives a more accurate measure of actual levels of party support among the electorate, since voters have greater incentive to cast sincere ballots and have a fuller range of
choices when voting for party lists (most small parties do not nominate candidates in all
SMDs).
AS4905_02_Jou.indd 776
10/13/09 4:50:00 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU table 777
7 Volatility
Index of Volatility*
Japan
1993–96
1996–2000
2000–03
2003–05
Taiwan
2004–08
Adjusted volatility
Adjusted
47.29
36.70
20.66
8.40
19.25
12.31
9.65
8.40
23.80
5.86
SOURCE: Ibid.
*See note 38. Japan: Sum of Japan Renewal Party, Komeito, Japan New Party, Democratic
Socialist Party, Socialist Democratic League in 1993 = NFP in 1996; NFP in 1996 = Komeito +
Liberal Party in 2000; DPJ + Liberal Party in 2000 = DPJ in 2003. Taiwan: KMT-People
First Party-New Party-Non Partisan Solidarity Association bloc versus DPP-Taiwan Solidarity Union bloc.
Pedersen index.39 The high values seen between most elections imply considerable fluctuations in parties’ vote shares, especially in Japan from 1993
to 1996, when nearly half of all voters switched from one party to another.
In this case, as in the two subsequent periods, one can probably attribute
the fluctuating figures as much to party system reconfiguration as weak
voter attachments. A flurry of small- and medium-sized new parties appeared before the 1993 elections, but many merged, forming the NFP, before the next polls in order to mount a credible challenge to the LDP in
SMD contests. These frequent party name changes artificially inflate volatility, since parties that contested the 1993 election separately but later
amalgamated are treated as having a 0% vote share in 1996. Taking this
into account, figures given in the second column of Table 7 adjust for cases
where a predecessor or successor party is clearly identifiable, offering a
more accurate measure of volatility.
Both columns reveal a steady decline in volatility in Japan. The still-high
figure for the 1993–96 period is attributable in part to the newly established
DPJ. Although it was founded by former members of the SDPJ and the
39. The Pedersen index is calculated by summing the absolute value of changes in each
party’s vote shares and dividing the result by two. The highest possible figure of 100 means
none of the parties running in the current election received any vote in the previous one, while
0 indicates that vote shares for all parties remain exactly the same across two elections. See
Mogens N. Pedersen, “The Dynamics of European Party Systems: Changing Patterns of
Electoral Volatility,” European Journal of Political Research 7:1 (March 1979), pp. 1–26.
AS4905_02_Jou.indd 777
10/13/09 4:50:01 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
778 asian survey, vol. xlix, no. 5, september/october 2009
NHP, the latter two still contested the 1996 polls, so the DPJ’s vote share
can only be treated as having risen from 0% to 16%. The breakup of the
NFP in 1997 and the DPJ-Liberal Party merger prior to the 2003 election
largely explain figures for the two subsequent periods. While it is beyond
the scope of this paper to trace the rise and fall of individual parties, even
the brief sketch offered here suffices to capture the frequency of entrances
to, and exits from, the electoral market in the decade following electoral
reform. Alongside falling volatility, the fact that the 2005 election was the
first time no major party mergers or splits occurred (though two tiny splinter parties were founded by a few renegade LDP politicians) testifies to
party system stabilization.
The adjusted volatility figure for Taiwan is no easier to calculate, partly
because the country does not have a parliamentary system of government,
so there are no formal coalitions formed to maintain a legislative majority.
Nevertheless, the KMT concluded electoral pacts with the PFP and the
New Party in 2008, going as far as accepting candidates from these small
parties to run under its banner. An informal agreement between the KMT
and a party called the Non-Partisan Solidarity Association was also evident, as the latter was given a free run in head-to-head confrontations with
DPP candidates in several districts. The higher figure in column one of
Table 7 is largely attributable to the change in PFP vote share from 15% in
2004 to 0% in 2008 (and a corresponding increase in KMT support) because
the PFP did not present a PR list. Treating all parties cited above as a single
bloc in 2004, and combining the DPP vote with that of the Taiwan Solidarity Union as another bloc, led to the considerably lower figure shown in the
second column.40 In the case of Taiwan, one can speak of the first column
as displaying total volatility, while the second column represents inter-bloc
volatility. The latter suggests that Taiwan has already attained party system
equilibrium.
Explaining Differences in Trends toward
a Two-Party System
Despite similar electoral rules under both the old SNTV and new parallel
mixed system, party systems in Japan and Taiwan show divergent postreform trajectories, or at least variations in the speed and extent of convergence toward bipolar competition. Several explanations are explored
below, including social cleavages, constitutional structures, inconsistencies between national and local electoral rules, and features under the new electoral
40. If one does not count the Non-Partisan Solidarity Association as a part of the KMTled bloc, adjusted volatility would be 9.0.
AS4905_02_Jou.indd 778
10/13/09 4:50:01 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 779
rules adopted by each country, to explain why Taiwan has moved faster
and farther toward a two-party system than Japan.
Parties rise into prominence or fade into oblivion not only in reaction to
electoral incentives but also in response to historically shaped societal
cleavages.41 From the early post-war period, political debate in Japan had
revolved around non-economic questions such as the role of the emperor
and the military, the form of political structure, and the choice of foreign
alliances. Richardson et al. state that “these cultural political issues, to a
great extent, defined electoral alignments and the postwar party system,”
not least because other cleavages (regional, religious, class) lacked salience.42
Economic issues were also of secondary importance in Taiwan, where the
question of national identity dominated post-democratization party politics.43
Although one may anticipate that the existence of a single principal cleavage
leads to a two-party system, the emergence of the Democratic Socialist Party
and Komeito in Japan and the New Party, Taiwan Solidarity Union, and
PFP in Taiwan casts doubt on this hypothesis.
Competition between the LDP and DPJ today is generally not organized
around “cultural” cleavages, partly because the end of the Cold War
rendered some conflicts obsolete, and also because of a degree of consensus
on important issues between these two parties. However, these issues (e.g.,
constitutional revision) may still provide a raison d’être for smaller parties
such as the SDPJ and JCP. In contrast, the national identity question in
Taiwan continues to underpin competition between the KMT and DPP.
Yet in view of the aforementioned smaller parties in Taiwan being able to
enter the political arena and attract significant vote shares during the decade
41. Seymour M. Lipset and Stein Rokkan, “Introduction,” Party Systems and Voter Alignments: Cross-National Perspectives, eds. Seymour M. Lipset and Stein Rokkan (New York:
Free Press, 1967); Robert H. Dix, “Cleavage Structures and Party Systems in Latin America,”
Comparative Politics 22:1 (October 1989), pp. 23–37.
42. Bradley M. Richardson, Scott C. Flanagan, Joji Watanuki, Ichiro Miyake, Shinsaku
Kohei, “The Japanese Voter: Comparing the Explanatory Variables in Electoral Decisions,”
in The Japanese Voter, eds. Scott C. Flanagan, Shinsaku Kohei, Ichiro Miyake, Bradley M.
Richardson, Joji Watanuki (New Haven, Conn.: Yale University Press, 1991). The authors
point out that “Japan has a more homogeneous society with less regional diversity, less pervasively felt religious differences, and a somewhat weak and fragmented labor union movement” (p. 401).
43. John Fuh-Sheng Hsieh and Emerson M. S. Niou, “Salient Issues in Taiwan’s Electoral
Politics,” Electoral Studies 15:2 (May 1996), pp. 219–35. The authors add that whereas few
voters cite national identity as most important, “some politicians within each political party
find it in their electoral interest to capitalize on this issue” (p. 235). Other scholars point to the
emergence of new salient issues, e.g., Tse-Min Lin and Baohui Zhang, “Cross-cutting Issues
and the Consolidation of Democracy in Taiwan,” Democratization 5:4 (Winter 1998), pp.
118–43.
AS4905_02_Jou.indd 779
10/13/09 4:50:01 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
780 asian survey, vol. xlix, no. 5, september/october 2009
preceding electoral reform, one cannot conclude that social cleavages
constitute the main influence on party system development. Whether these
parties represent constituencies aligned along different issue dimensions, or
simply occupy niches on both extremes of the national identity spectrum,
the fact that contours of the party system changed while cleavage patterns
remained the same demands alternative explanations.
That presidential regimes produce a smaller number of parties than
parliamentary systems has been noted both theoretically and empirically
by many scholars.44 Because the presidency is a unitary position, the same
logic behind Duverger’s mechanical and psychological effects in SMDs is
applicable, in this case at the national level. Furthermore, within presidential regimes, two-party systems are most likely when the president is
elected by plurality (rather than majority run-off), and the concurrence
of presidential and legislative elections also reduces the number of parties.45 The presidential system in Taiwan today conforms to both of these
criteria, so one may argue that differences between Japan’s and Taiwan’s
evolution toward bipolar competition are rooted in contrasting constitutional structures.
Yet while it is plausible to assert that Taiwan has a more full-fledged twoparty system than Japan because of the electoral imperative of presidential
elections, this is only a static description of the current state of affairs. A
dynamic view reveals that ENP in Taiwan, at both the elective and parliamentary levels, increased from the late 1990s to the early 2000s (see Table
2), notwithstanding conditions governing presidential contests remaining
constant from 1996 to 2004 (plurality of popular vote, non-concurrence
with legislative polls). The only significant change in 2008 was election
timing, namely, the near concurrence of presidential and legislative elections after legislative terms were extended from three to four years. But
one cannot count on a “coattail” effect, because MPs were elected before
the president.46 Even if nearly concurrent elections in Taiwan account for
the sharp drop in ENP from 2004 to 2008, emphasis on relatively stable
constitutional structures cannot explain dynamic party system shifts in Japan and during the pre-reform decade in Taiwan.
44. For example, see Alfred Stepan and Cindy Skach, “Presidentialism and Parliamentarism in Comparative Perspective,” in The Failure of Presidential Democracy, eds. Juan J. Linz
and Arturo Valenzuela (Baltimore, Md.: Johns Hopkins University Press, 1994).
45. Matthew Soberg Shugart and John M. Carey, Presidents and Assemblies: Constitutional
Design and Electoral Dynamics (Cambridge: Cambridge University Press, 1992), chs. 9–10.
46. Nearly concurrent elections are now scheduled to take place every four years. Lin
predicts that “a two-person presidential race is likely to overshadow the legislative election
and to make it hard for smaller parties to make their voices heard” (Lin, “The Politics of
Reform in Japan and Taiwan,” p. 129).
AS4905_02_Jou.indd 780
10/13/09 4:50:01 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 781
A number of recent studies on party system change in Japan focus on
inconsistencies between electoral rules for national and sub-national legislatures in explaining the continued presence of more than two major candidates at the district level. Whereas SMD candidates are elected by plurality,
SNTV is still used for prefectural and local assembly elections, resulting in
what Horiuchi and Natori label a “vertical contamination effect”: SMD
candidates often rely on local politicians for campaign mobilization (a relationship known as keiretsu), yet while FPTP creates centripetal incentives,
SNTV generates centrifugal pressure.47 Where a national (single-member)
district geographically overlaps with multiple local districts, the difference
in electoral rules affects the effective number of candidates per SMD. In
short, inconsistent electoral rules impede the development of a two-party
system by inhibiting bipolar competition at the district level through contrary incentive structures.
An examination of these inconsistent rules offers important insights
into the previously neglected association between local and national electoral dynamics in unitary states. The validity of this analysis can also be
extended to the case of Taiwan, where local factions play an important
role in national elections.48 Yet precisely because the inconsistent rules in
Japan and Taiwan are similar, they cannot explain divergences in the two
countries’ evolution toward a two-party system. The puzzle of why vertical contamination has not deterred or at least delayed the emergence of
bipolar competition in Taiwan lies outside the scope of this study; here it
suffices to stress that one must seek other explanations to account for
differences in post-reform party system development between Japan and
Taiwan.
Although the mechanisms translating votes into seats under both SNTV
and the mixed parallel system are largely similar in Japan and Taiwan, it is
important not to overlook differences in a number of provisions. Focusing
on the post-reform system, these include the design of PR blocs, the
possibility of dual candidacies, and the relative distribution of parliamentary
seats allocated by SMD and PR, in addition to different thresholds already
described. The Japanese system allocates PR seats by regional blocs,
permits candidates to run in both SMD and PR components, and reserves
a higher proportion of parliamentary seats for members elected through
PR compared with Taiwan. The impact of the first factor on facilitating a
47. Yusaku Horiuchi and Ryota Natori, “Vertical Contamination Effects on the Duvergerian Equilibrium: The Case of Japan,” typescript, Australian National University, 2007, available at <http://www.horiuchi.org/files/sines.pdf>.
48. For example, see Shelly Rigger, Politics in Taiwan: Voting for Democracy (London:
Routledge, 1999).
AS4905_02_Jou.indd 781
10/13/09 4:50:02 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
782 asian survey, vol. xlix, no. 5, september/october 2009
two-party system is ambiguous, while the latter two may account for the
more rapid decline of small parties in Taiwan.
One consequence deriving from the distinction between regional and
nationwide PR districts is that the former allow (particularly major) parties
to gain representation in areas where they are not sufficiently strong to
win SMD contests. While this may not have a direct impact on parliamentary
seat distribution, it increases the likelihood of parties establishing footholds
in electorally weak regions that may serve as bases for future expansion.
Consequently, “the PR tier may actually enhance the tendency of SMDs
to produce two-candidate competition by reducing the number of noncompetitive districts.”49 In contrast, candidates running on a nationwide
PR list probably have reason to appeal to the electorate on the basis of
policy preferences or sectoral group interests rather than cultivate areaspecific ties. As a consequence, even a major party may have both lesser
incentive and fewer opportunities to strengthen its support in regions
where it is traditionally weak.
Another important difference between the new systems adopted in
Japan and Taiwan is the possibility of dual candidacy in the former.
Moreover, candidates who are simultaneously nominated to both tiers
may be ranked equally on their parties’ PR list. The position of equally
ranked candidates on the PR list who lose their SMD races is determined
by the ratio of their SMD vote to the district winners’ vote (sekihairitsu);
those whose vote share comes closest to that of SMD winners have a better chance of gaining seats through the PR tier. Voters in Japan thus have
an incentive to cast a sincere vote not only in PR but also in the SMD
portion (if the contest is not too lopsided), because even if their preferred
candidate does not win the district race, he or she may still become an MP
thanks to a high sekihairitsu.50 In contrast, Taiwanese voters faced with a
similar scenario may be more inclined to vote tactically in SMD contests,
or not turn out at all, in the knowledge that sincere voting would serve no
utilitarian purpose.
One should note that seats allocated by PR comprise only 37.5%
(reduced from 40%) of total seats in the Japanese House of Representatives,
and approximately 30% in the Taiwanese Legislative Yuan. Because most
small parties cannot hope to win many SMD contests and depend on
PR lists for parliamentary survival, the lower the percentage of PR seats
in the composition of the legislature, the more likely small parties will find
49. Steven Reed, “Evaluating Political Reform,” p. 261.
50. Steven R. Reed and Michael F. Thies, “The Consequences of Electoral Reform in Japan,” in Mixed-Member Electoral Systems: The Best of Both Worlds? eds. Matthew Soberg
Shugart and Martin P. Wattenberg (Oxford: Oxford University Press, 2001).
AS4905_02_Jou.indd 782
10/13/09 4:50:02 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 783
themselves discriminated against by the electoral system. Furthermore, overall
legislature size also affects small parties’ chances, although in the case of
Taiwan small party failure is more attributable to the 5% threshold than to
the truncated size of Parliament (reduced from 225 to 113 seats with the
enactment of electoral reform).
Discussions and Conclusions
This study has compared trends in patterns of party competition in Japan
and Taiwan following the change from SNTV to a mixed parallel system,
and examined alternative theories explaining why Taiwan has moved faster
and farther than Japan toward bipolar competition. Tracing the evolution
in effective number of parties and disproportionality highlights macro-level
changes in the party system consequent upon electoral reform, while detailing variations in individual districts over the course of several elections
provides evidence of movement toward a two-party system. Finally, volatility figures reveal the degree of party system stabilization. The phenomena of continued decline in the ENP (particularly in the SMD component)
and increase in disproportionality together corroborate prognostications
derived from the literature. Trends toward bipolar competition and greater
cross-district linkage by both leading national parties are also largely in
line with the expectations of electoral reform advocates and academic
experts alike.
A variety of indicators all point to the emergence of a two-party system
in both Japan and Taiwan, though neither at the same speed nor to the
same extent. Four explanations are offered to account for these differences:
social cleavages, constitutional structures, inconsistencies between national
and sub-national electoral rules, and features of the new system that set
the two cases apart. Most of these hypotheses offer valid elucidations of
developments in one country or in both comparatively at a given point in
time, but perform less well in illuminating cross-national and cross-temporal
distinctions. The implementation of concurrent presidential and legislative
elections, the unavailability of a dual candidacy provision, and the smaller
legislature size (coupled with a 5% threshold) in Taiwan all function as
constraints on small party survival, and hence accelerate movement toward
two-party competition. A presidential system and persistent unidimensional
cleavage may also be contributory factors, although they cannot explain the
rise of small parties during the years preceding electoral reform.
Before drawing conclusions about the decisive, even deterministic,
function of institutional rules in dictating party system development,
however, two general and one specific reminders are in order. First, the
introduction of electoral reform takes place within particular political
AS4905_02_Jou.indd 783
10/13/09 4:50:02 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
784 asian survey, vol. xlix, no. 5, september/october 2009
contexts, with each actor (not only parties but also individual politicians)
taking its own interests into consideration when devising new institutional
rules. Not surprisingly, large parties prefer plurality rules while small parties
without geographically concentrated support favor PR. Accordingly, the
higher number of both total and effective parliamentary parties in Japan
when electoral reform was passed influenced the introduction of the new system in a more proportional direction than that devised by a less fragmented
legislature in Taiwan. In fact, electoral reform might not have passed at all
without splits within the LDP leading to greater legislative fragmentation
and power alternation. In short, not only can electoral reform influence party
system change, the causal arrow may also point in the opposite direction.
On a related note, the extent of party system change brought about by
electoral reform can be analyzed not merely by observing trends in the number of parties and their relative strength but also by examining strategies individual parties adopt in response to new rules of the game. The preceding
section mentions several examples of electoral pacts and government coalitions as parties react to the logic generated by the new system. This potentially leads to centripetal competition concomitant with the development of
a two-party system because parties now have an incentive to court the median voter rather than cultivate niches of minority support.51 Even if small
parties manage to survive under the new system, they may become dependent on one of the large parties, resulting in what may be labeled a twocamp system. Recent developments in Japan offer some evidence affirming
this development, with the LDP-Komeito coalition on one side versus DPJ
and all other small parties (bar the communists) in the opposite camp.
Lastly, having seen that the 1996 election in Japan did not necessarily
portend trends in party competition in subsequent polls, caution is necessary
when making predictions about Taiwan’s party system based on results from
a single election. At the same time, one has reason to believe that unlike
Japan, the party system in Taiwan will not be in flux; it was not as fragmented
as Japan’s when the new system was adopted, and all indicators above show
that Taiwan may have already reached the equilibrium of a full-fledged twoparty system. Just one election is sufficient to prove that small parties have
minimum space for survival under the current system, so political entre­
preneurs (including dissidents within large parties) have an incentive to
advance their careers and policy objectives through one of the two existing
parties rather than establishing new ones.
While intra-party conflict and parliamentary fragmentation were by no
means the only problems electoral reform advocates proposed to address—
other perceived deleterious corollaries of SNTV include corruption,
51. Anthony Downs, An Economic Theory of Democracy (New York: Harper, 1957).
AS4905_02_Jou.indd 784
10/13/09 4:50:02 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions
WILLY JOU 785
factionalism, and non-policy-centered campaigns—the area where one
could expect the introduction of the mixed parallel system to exert the most
direct and immediate impact involves patterns of party competition. The
development and stabilization of a nationwide two-party system in both
Japan and Taiwan in place of a multi-polar one demonstrate the crucial
influence of institutional design on the behaviors of both parties and voters.
At the same time, differences in the pace and extent of evolution toward such
an equilibrium caution against conclusions based on singular explanations.
Contextual factors such as social cleavages and the degree of legislative
fragmentation; structural factors such as constitutional design and consistency between mechanisms for selecting national and sub-national representatives; as well as details of electoral rules such as legal and effective
thresholds for party lists, the percentage of seats allocated by PR, and the
size of PR district(s) can all play a significant role in shaping party systems’
evolution following electoral reform.
AS4905_02_Jou.indd 785
10/13/09 4:50:03 PM
This content downloaded from 175.159.161.64 on Wed, 15 May 2013 03:39:19 AM
All use subject to JSTOR Terms and Conditions