¹ Party defection/discipline and cohesion/dissension inside a party are different concepts. However, as we will explore in detail later, throughout this article, we assume that ‘dissidence against the party leadership’ can contain a continuum of possible behaviour, ranging from the extreme of defection (‘exit’), through public disagreement (‘voice’), to going along with the party's wishes (‘loyalty’). On this point, see Hirschman, Albert O., Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States (Cambridge, Mass.: Harvard University Press, 1970)Google Scholar. For a political model using this idea, see Kato, Junko, ‘When the Party Breaks Up: Exit and Voice among Japanese Legislators’, American Political Science Review, 92 (1998), 857–70Google Scholar. We acknowledge the help of an anonymous reviewer in clarifying this point.

² Bowler, Shaun, Farrell, David M. and Katz, Richard S., ‘Party Cohesion, Party Discipline, and Parliaments’, in Farrell, David M., Bowler, Shaun and Katz, Richard S., eds, Party Discipline and Parliamentary Government (Columbus: Ohio State University Press, 1999), pp. 3–22Google Scholar.

³ See, for example, Morgenstern, Scott and Nacif, Benito, eds, Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2002)CrossRefGoogle Scholar.

⁴ On electoral rules, see Carey, John M. and Shugart, Matthew S., ‘Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas’, Electoral Studies, 14 (1995), 417–39Google Scholar; Hix, Simon, ‘Electoral Institutions and Legislative Behavior: Explaining Voting-Defection in the European Parliament’, World Politics, 56 (2004), 194–223CrossRefGoogle Scholar. On particularism, see Ames, Barry, The Deadlock of Democracy in Brazil (Ann Arbor: University of Michigan Press, 2001)Google Scholar; Desposato, Scott W., ‘Parties for Rent? Ambition, Ideology, and Party Switching in Brazil's Chamber of Deputies’, American Journal of Political Science, 50 (2006), 62–80CrossRefGoogle Scholar; Samuels, David J., ‘Incentives to Cultivate a Party Vote in Candidate-Centric Electoral Systems: Evidence from Brazil’, Comparative Political Studies, 32 (1999), 487–518CrossRefGoogle Scholar. On diversity inside a coalition, see Gaines, Brian J. and Garrett, Geoffrey, ‘The Calculus of Dissent: Party Discipline in the British Labour Government, 1974–1979’, Political Behaviour, 15 (1993), 113–35Google Scholar; Tsebelis, George, Veto Players (Princeton, N.J.: Princeton University Press, 2002)Google Scholar.

⁵ Cox, Gary W. and Rosenbluth, Frances, ‘Anatomy of a Split: The Liberal Democrats of Japan’, Electoral Studies, 14 (1995), 355–76CrossRefGoogle Scholar; Kato, ‘When the Party Breaks Up’; Reed, Steven R. and Scheiner, Ethan, ‘Electoral Incentives and Policy Preferences: Mixed Motives Behind Party Defections in Japan’, British Journal of Political Science, 33 (2003), 469–90CrossRefGoogle Scholar.

⁶ Cox, Gary W., Rosenbluth, Frances McCall and Thies, Michael F., ‘Electoral Reform and the Fate of Factions: The Case of Japan's Liberal Democratic Party’, British Journal of Political Science, 29 (1999), 33–56Google Scholar; Krauss, Ellis S. and Pekkanen, Robert, ‘Explaining Party Adaptation to Electoral Reform: The Discreet Charm of the LDP?’ Journal of Japanese Studies, 30 (2004), 1–34Google Scholar; Scheiner, Ethan, Democracy Without Competition in Japan: Opposition Failure in a One-Party Dominant State (Cambridge: Cambridge University Press, 2005)CrossRefGoogle Scholar; Scheiner, Ethan, ‘Pipelines of Pork: Japanese Politics and a Model of Local Opposition Party Failure’, Comparative Political Studies, 38 (2005), 799–823Google Scholar.

⁷ See Huber, John D., ‘The Vote of Confidence in Parliamentary Democracies’, American Political Science Review, 90 (1996), 269–82CrossRefGoogle Scholar; Diermeier, Daniel and Feddersen, Timothy J., ‘Cohesion in Legislatures and the Vote of Confidence Procedure’, American Political Science Review, 92 (1998), 611–21Google Scholar; Desposato, ‘Parties for Rent?’

⁸ Hirschman, Exit, Voice, and Loyalty; Kato, ‘When the Party Breaks Up’.

⁹ See Kam, Christopher, ‘Demotion and Dissent in the Canadian Liberal Party’, British Journal of Political Science, 36 (2006), 561–74Google Scholar, on trade between party-line voting and posts.

¹⁰ Note that the existing literature on defection, based on a strong assumption that a dissident is a 100 per cent resolute defector who takes the risk of defection behaviour, only looks at dichotomous results (defection or not) and neglects the possibility that dissension can take various forms.

¹¹ The literature on personal electoral connections is large. See, for example, Cain, Bruce E., Ferejohn, John A. and Fiorina, Morris P., ‘The Constituency Service Basis of the Personal Vote for U.S. Representatives and British Members of Parliament’, American Political Science Review, 78 (1984), 110–25Google Scholar; Carey and Shugart, ‘Incentives to Cultivate a Personal Vote’; Cox, Gary W. and McCubbins, Mathew D., ‘The Institutional Determinants of Economic Policy Outcomes’, in Haggard, Stephan and McCubbins, Mathew D., eds, Presidents, Parliaments, and Policy (Cambridge: Cambridge University Press, 2001), pp. 21–63Google Scholar; Crisp, Brian F. and Desposato, Scott W., ‘Constituency Building in Multimember Districts: Collusion or Conflict?’ Journal of Politics, 66 (2004), 136–56Google Scholar; Crisp, Brian F., Escobar-Lemmon, Maria C., Jones, Bradford S., Jones, Mark P. and Taylor-Robinson, Michelle M., ‘Vote-Seeking Incentives and Legislative Representation in Six Presidential Democracies’, Journal of Politics, 66 (2004), 823–46Google Scholar; Ingall, Rachael E. and Crisp, Brian F., ‘Determinants of Home Style: The Many Incentives for Going Home in Colombia’, Legislative Studies Quarterly, 26 (2001), 487–512CrossRefGoogle Scholar.

¹² Ames, The Deadlock of Democracy in Brazil; Desposato, ‘Parties for Rent?’; Samuels, ‘Incentives to Cultivate a Party Vote in Candidate-Centric Electoral Systems’.

¹³ Weingast, Barry R., Shepsle, Kenneth A. and Johnsen, Christopher, ‘The Political Economy of Benefits and Costs: A Neoclassical Approach to Distributive Politics’, Journal of Political Economy, 89 (1981), 642–64CrossRefGoogle Scholar.

¹⁴ Ames, The Deadlock of Democracy in Brazil; Axelrod, Robert, Conflict of Interest (Chicago: Markham, 1970)Google Scholar; Tsebelis, Veto Players.

¹⁵ Gaines and Garrett, ‘The Calculus of Dissent’.

¹⁶ Jones, Mark P., ‘Explaining the High Level of Party Discipline in the Argentine Congress’, in Morgenstern, Scott and Nacif, Benito, eds, Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2002), pp. 147–84.Google Scholar

¹⁷ On bilateral contracts and the notion of mutual hostage-taking, see Williamson, Oliver E., ‘Credible Commitments: Using Hostages to Support Exchange’, American Economic Review, 73 (1983), 519–40Google Scholar.

¹⁸ Aoki, Masahiko, ‘Toward an Economic Model of the Japanese Firm’, Journal of Economic Literature, 28 (1990), 1–27Google Scholar.

¹⁹ Coker, David C. and Crain, W. Mark, ‘Legislative Committees as Loyalty-Generating Institutions’, Public Choice, 81 (1994), 195–221CrossRefGoogle Scholar; Crook, Sara Brandes and Hibbing, John R., ‘Congressional Reform and Party Discipline: The Effects of Changes in the Seniority System on Party Loyalty in the US House of Representatives’, British Journal of Political Science, 15 (1985), 207–26CrossRefGoogle Scholar.

²⁰ Neto, Octavio Amorim and Santos, Fabiano, ‘The Executive Connection: Presidentially Defined Factions and Party Discipline in Brazil’, Party Politics, 7 (2001), 213–34CrossRefGoogle Scholar; Michael Laver and Kenneth A. Shepsle, ‘How Political Parties Emerged from the Primeval Slime: Party Cohesion, Party Discipline, and the Formation of Governments’, in Shaun Bowler, David M. Farrell and Richard S. Katz, eds, Party Discipline and Parliamentary Government, pp. 23–48.

²¹ Kam, ‘Demotion and Dissent in the Canadian Liberal Party’.

²² Those members who assume ‘pivotal positions’ in a legislature are more likely to change their voting positions. See Krehbiel, Keith, Pivotal Politics: A Theory of U.S. Lawmaking (Chicago: The University of Chicago Press, 1998)CrossRefGoogle Scholar.

²³ Strøm, Kaare, ‘A Behavioral Theory of Competitive Political Parties’, American Journal of Political Science, 34 (1990), 565–98CrossRefGoogle Scholar.

²⁴ Cox and Rosenbluth, ‘Anatomy of a Split’; Kato, ‘When the Party Breaks Up’. The insignificance of electoral variables is even more puzzling because they look only at the extreme case of dissension–defection in 1993, when personal vote should have mattered most.

²⁵ Reed and Scheiner, ‘Electoral Incentives and Policy Preferences’.

²⁶ One exception is made by Masahiko Tatebayashi, Giin Kōdō no Seiji Keizaigaku: Jimintō Shihai no Seido Bunseki [The Logic of Legislators’ Activities: Institutional Analysis of LDP Dominance in Japan], (Tokyo: Yūhikaku, 2004 [in Japanese]), who implies that there was a trade-off between the constituency servicing and policy specialization strategies in pre-1994 Japan. He reveals the negative correlation between politicians’ region-based vote-dividing strategy and the policy specialization likelihood in the same district.

²⁷ For example, Carey and Shugart, ‘Incentives to Cultivate a Personal Vote’; Cox and McCubbins, ‘The Institutional Determinants of Economic Policy Outcomes’.

²⁸ Extreme cases only actually happened before 2005 in 1976 and 1993. Only the 1993 event resulted in the fall of the LDP from the government. In 1960, 1974, 1979–80 and 2000, the dissidents in the LDP had opportunities to show dissension, but in these cases they were just persuaded and/or they managed only to show their grievances at best.

²⁹ Therefore, we would add to the functional ‘party aggregation’ argument that exogenous shocks to a system would eventually cause local politicians whose re-election and policy incentives would not otherwise coincide with the leadership to co-operate, because a ‘provincialized’ structure does not require party aggregation at the national level. See Chhibber, Pradeep K. and Kollman, Ken, The Formation of National Party Systems: Federalism and Party Competition in Canada, Great Britain, India, and the United States (Princeton, N.J.: Princeton University Press, 2004)Google Scholar. We acknowledge the help of an anonymous reviewer in pointing out the relevance of this literature.

³⁰ Flanagan, Scott C., ‘Mechanisms of Social Network Influence in Japanese Voting Behavior’, in Flanagan, Scott C., Kohei, Shinsaku, Miyake, Ichiro, Richardson, Bradley M. and Watanuki, Joji, The Japanese Voter (New Haven, Conn.: Yale University Press,1991), pp. 143–97Google Scholar; Ramseyer, J. Mark and Rosenbluth, Frances McCall, Japan's Political Marketplace (Cambridge, Mass.: Harvard University Press, 1993), pp. 23–8Google Scholar.

³¹ Takashi Inoguchi and Tomoaki Iwai, Zoku Giin no Kenkyū [The Study of Policy Tribes] (Tokyo: Nihon Keizai Shimbun Sha, 1987 [in Japanese]); Ramseyer and Rosenbluth, Japan's Political Marketplace, pp. 28–34.

³² Ramseyer and Rosenbluth, Japan's Political Marketplace, pp. 50–7.

³³ Cox, Gary W. and Rosenbluth, Frances, ‘The Electoral Fortunes of Legislative Factions in Japan’, American Political Science Review, 87 (1993), 577–89CrossRefGoogle Scholar; Cox, Gary W. and Rosenbluth, Frances, ‘Factional Competition for the Party Endorsement: The Case of Japan's Liberal Democratic Party’, British Journal of Political Science, 26 (1996), 259–69CrossRefGoogle Scholar.

³⁴ Krauss and Pekkanen, ‘Explaining Party Adaptation to Electoral Reform’.

³⁵ Seisaburō Satō and Tetsuhisa Matsuzaki, Jimintō Seiken [The LDP Administration] (Tokyo: Chūō Kōron, 1986 [in Japanese]).

³⁶ Sadafumi Kawato, ‘Jimintō ni Okeru Yakushoku Jinji no Seidoka‘[The Institutionalization of Rules Governing Promotion in the LDP], Hōgaku, 59 (1996), 933–57 [in Japanese]; Sadafumi Kawato, ‘Sinioriti Rūru to Habatsu: Jimintō ni Okeru Jinji Haibun no Henka’ [The Development of Seniority and Interfactional Balancing Rules in the LDP], Revaiasan, Extra Issue (1996), 111–45 [in Japanese]; Satō and Matsuzaki, Jimintō Seiken.

³⁷ Christensen, Raymond V., ‘Electoral Reform in Japan:HowIt Was Enacted and Changes It May Bring’, Asian Survey, 34 (1994), 589–605CrossRefGoogle Scholar; Curtis, Gerald L., The Logic of Japanese Politics: Leaders, Institutions, and the Limits of Change (New York: Columbia University Press, 1999), pp. 137–70Google Scholar; Reed, Steven R. and Thies, Michael F., ‘The Causes of Electoral Reform in Japan’, in Shugart, Matthew Soberg and Wattenberg, Martin P., eds, Mixed-Member Electoral Systems: The Best of Both Worlds? (Oxford: Oxford University Press, 2001), pp. 152–72Google Scholar.

³⁸ They are summarized in Masato Shimizu, Kantei Shudō [The Cabinet Office Initiative] (Tokyo: Nihon Keizai Shimbun Sha, 2005 [in Japanese]); Harukata Takenaka, Shushō Shihai [The Rule of Prime Minister] (Tokyo: Chūō Kōron Sha, 2006 [in Japanese]). Also see George-Mulgan, Aurelia, ‘Japan's “Un-Westminster” System: Impediments to Reform in a Crisis Economy’, Government and Opposition, 38 (2003), 73–91CrossRefGoogle Scholar; and her ‘Japan's Political Leadership Deficit’, Australian Journal of Political Science, 35 (2000), 183–202, for the weakness of prime ministers in Japan.

³⁹ Theoretically SMD systems do not necessarily completely eliminate the need for candidates to mobilize a personal vote. See Cain et al., ‘The Constituency Service Basis of the Personal Vote for U.S. Representatives and British Members of Parliament’. There is also ample empirical evidence that kōenkai survived: see Krauss and Pekkanen, ‘Explaining Party Adaptation to Electoral Reform’; Ichiro Miyake, Seitō Shiji no Kōzō [The Structure of Partisan Support] (Tokyo: Bokutakusha, 1998 [in Japanese]); Masaki Taniguchi, Gendai Nihon no Senkyo Seiji: Senkyo Seido Kaikaku wo Kenshōsuru [Electoral Politics in Contemporary Japan: Analyzing the Electoral Reform] (Tokyo: Tokyo Daigaku Shuppankai, 2004 [in Japanese]). The PR portion of the mixed-member system only encouraged PR candidates to act as local candidates; except for pure PR members, losers in an SMD can be saved as so-called ‘zombies’ by the PR portion, where rank is determined by their vote margin from the SMD winner. For more elegant descriptions on this complicated system, see McKean, Margaret and Scheiner, Ethan, ‘Japan's New Electoral System: La plus ça change …’, Electoral Studies, 19 (2000), 447–77CrossRefGoogle Scholar; Pekkanen, Robert, Nyblade, Benjamin and Krauss, Ellis S., ‘Electoral Incentives in Mixed Member Systems: Party, Posts, and Zombie Politicians in Japan’, American Political Science Review, 100 (2006), 183–93Google Scholar.

⁴⁰ See Scheiner, Democracy Without Competition in Japan, on how locally-oriented legislators maintained their clientelistic networks through the centralized fiscal structure of Japan, even after 1993–94. A rather journalistic account is available in Iwai, Tomoaki, ‘Shin “Zoku Giin no Kenkyū” ‘ [The New ‘Study of Policy Tribes’], Ekonomisuto, 80 (2002), 33–7Google Scholar [in Japanese]. Also see Mishima, Ko, ‘The Changing Relationship between Japan's LDP and the Bureaucracy: Hashimoto's Administrative Reform Effort and Its Politics’, Asian Survey, 38 (1998), 968–85CrossRefGoogle Scholar, for how zoku giin interrupted and ultimately halted Hashimoto's administrative reform initiative. On the continuous importance of factions, see Cox, Rosenbluth and Thies, ‘Electoral Reform and the Fate of Factions’; Krauss and Pekkanen, ‘Explaining Party Adaptation to Electoral Reform’.

⁴¹ The largest opposition party in 1996 was the New Frontier Party, which gained 156 seats, or 31.2 per cent of the total lower-house seats, but the party was soon split. The Democratic Party of Japan emerged as a new force, but it also failed to gain a substantial portion in the 2000 election: 127 seats, or 26.5 per cent.

⁴² In fact, individual members were ideologically similar to the LDP. See Miura, Mari, Lee, Kap-Yun and Weiner, Robert J., ‘Who Are the DPJ? Policy Positioning and Recruitment Strategy’, Asian Perspective, 29 (2005), 49–77Google Scholar.

⁴³ The average number of terms served by LDP members in the cabinet was radically decreased from 7.41 (the 1997 Hashimoto cabinet) and 7.45 (the 2000 Obuchi cabinet) to 5.18 (the 2002 Koizumi cabinet). Four of the important party posts were distributed to Koizumi's favourites, including Shinzō Abe, who became Secretary-General when he had experienced only three terms in the lower house, although traditionally only the most senior members could assume this powerful post.

⁴⁴ The Hashimoto faction, the biggest faction, did not get any of the big four posts in the LDP until 2003 with Koizumi as prime minister, although the faction had always secured one or more of the big four posts between the late 1970s and 2001.

⁴⁵ Ellis S. Krauss and Benjamin Nyblade, ‘ “Presidentialization” in Japan? The Prime Minister, Media and Elections in Japan’, British Journal of Political Science, 35 (2005), 357–68.

⁴⁶ Koizumi even strategically used the mass media to label the antagonistic force inside his own party as the ‘resistance force’ (teikō seiryoku) obstructing the reform. Many feared being labelled so, for electoral reasons.

⁴⁷ Maclachlan, Patricia L., ‘Post Office Politics in Modern Japan: The Postmasters, Iron Triangles, and the Limits of Reform’, Journal of Japanese Studies, 30 (2004), 281–313Google Scholar.

⁴⁸ Koizumi continued to imply that he would dissolve the lower house (Sankei Shimbun, 2 April 2005, p. 1; Sankei Shimbun, 17 May 2005, p. 1), and two-thirds of LDP members believed that he would surely dissolve it (Sankei Shimbun, 4 June 2005, p. 1).

⁴⁹ Koizumi dissolved the House of Representatives (HOR) even though the bill passed it and could not pass the HOC, because Japanese prime ministers can only dissolve the HOR. The rationale for this move was to offer the general voter a yes-or-no vote on the bill and Koizumi.

⁵⁰ For example, Mainichi Shimbun, 9 August 2005, p. 2.

⁵¹ As the vote approached, the leadership implied that it would give no endorsement in the next election (Mainichi Shimbun, 27 June 2005, p. 2; Sankei Shimbun, 2 July 2005, p. 4). We know that at least some possible rebels received specific threats from the LDP leadership: some changed their minds and voted for the bill because they knew expulsion or other types of sanction would be a huge loss (Yomiuri Shimbun, 2 July 2005, p. 2; Yomiuri Shimbun, 5 July 2005, p. 3). One defector even revealed in an interview that the party leadership had told him on the very morning of 5 July that the party would not give him an endorsement and would support the party's candidate in his district (Tokyo Shimbun, 10 July 2005, p. 3). Since these were reported in the media, it is plausible to believe that all or at least many potential rebels were aware of these threats, and thus the possible consequences of their actions.

⁵² Asahi Shimbun, 6 July 2005, p. 1. Technically, the LDP had 250 Lower House members in total in July 2005, but two could not be present because one was bed-ridden (Seisaburō Nakamura) and the other was in a foreign country (Toshitsugu Saitō). We have excluded them from the analysis.

⁵³ Asahi Shimbun, 6 July 2005, p. 4.

⁵⁴ Although not directly relevant to our arguments, readers may be interested in more recent developments involving the rebels. In December 2006, a dispute emerged about whether the former rebels should be allowed to return to the party. As the House of Councillors July 2007 election approached, some leaders in the LDP, especially those in the HOC, started to demand that the LDP should recall ex-LDP members to increase the LDP's electoral strength (Nihon Keizai Shimbun, 29 May 2006, p. 2). The former ‘assassins’ did not like the idea, as they had challenged the rebels in the election and knew there would be endorsement conflicts in the next HOR election. Also, public opinion was generally critical of the return. Abe Shinzō, Koizumi's successor as party president and prime minister, and the new party leadership were ambivalent about the issue (Nihon Keizai Shimbun, 24 October 2006, p. 2). Despite this internal division, the LDP finally decided to let those who submitted themselves to strict requirements return to the party. The conditions included writing an essay repenting of anti-party misdeeds, stating that they supported postal privatization and promising not to oppose the party in the future.

Some interpret the party's action as an indication of its return to its pre-Koizumi days. Indeed, Prime Minister Abe's popularity among the public has dropped as a result; but it should be noted that previously when conservative independents had challenged LDP candidates and won, they had afterwards been admitted to the party unconditionally. The stringent conditions for re-admission this time can also be seen as a sign of the more centralized party leadership after Koizumi (Nihon Keizai Shimbun, 11 November 2006, p. 2). Eleven of the twelve former rebels who had applied for re-instatement submitted to the requirements and were allowed to return to the party on 4 December 2006.

⁵⁵ Krauss and Pekkanen's Japan Legislative Organization Database [J-LOD]. It includes all LDP Diet Members in 1980–2006, with their constituency characteristics, personal backgrounds, election results, and all their party, legislative and executive posts from 1986 to 2006. The primary source is Seiji Handobukku (Tokyo: Seiji Kōhō Sentā, various years) and Seikan Yōran (Tokyo: Seisaku Jihōsha, various years).

⁵⁶ The 1994 electoral reform and the resultant redistricting created the co-ordination problem. The problem was easily solved if some candidates had few regional overlaps in the old district, while it was hard if candidates’ kōenkai were regionally overlapped. In some of the latter cases, two candidates agreed that one would run in an SMD and the other in PR in one election and vice versa in the next.

⁵⁷ On this point, see Greene, William H., Econometric Analysis, 5th edn (Upper Saddle River, N.J.: Prentice Hall, 2003), pp. 736–40Google Scholar.

⁵⁸ Therefore, for an SMD winner, Strength represents the margin between him/her and the best loser. For a ‘zombie’ saved by PR, Strength becomes negative, meaning the distance between him/her and the winner.

⁵⁹ In fact, the big three were the Hashimoto, Mori and Horiuchi factions as of June 2005, but we do not include the Mori faction in the big three as it was the mainstream – Koizumi was from the Mori faction.

⁶⁰ Scheiner, Democracy Without Competition in Japan; Scheiner, ‘Pipelines of Pork’.

⁶¹ Technically, DIDs are those areas that satisfy two conditions: first, they should have a population density of more than 4,000 people per square kilometre; and, secondly, they should have a population of more than 5,000 people. The DID ratio is now a commonly used urbanization index, for example as used by Scheiner, Democracy Without Competition in Japan. The data is available at Taku Sugawara's website (<http://freett.com/sugawara_taku/data/2003did.html>). We thank Taku Sugawara for making this data public.

⁶² We thank Naoko Taniguchi for sharing this data.

⁶³ Michael Tomz, Jason Wittenberg and Gary King, CLARIFY: Software for Interpreting and Presenting Statistical Results, Version 2.1 (Stanford University, University of Wisconsin and Harvard University, 2003). Available at <http://gking.harvard.edu/>.

⁶⁴ On these developments, see Estévez-Abe, Margarita, ‘Japan's Shift toward a Westminster System: A Structural Analysis of the 2005 Lower House Election and Its Aftermath’, Asian Survey, 46 (2006), 632–51CrossRefGoogle Scholar.