Electoral reform and the costs of personal support in Japan.

Carlson, Matthew M.

How does the choice of electoral rules affect politicians' incentives to campaign on the basis of personalized support? This article examines to what extent the adoption of new electoral and campaign finance rules affects the incentive of politicians in Japan's Liberal Democratic Party to rely on personal support organizations called koenkai. The core of the analysis utilizes newly collected campaign finance data. The empirical analyses confirm a considerable weakening in the number of koenkai across systems as well as a decreased need for politicians to spend money in the proportional representation tier. These results highlight the importance of previous organizational legacies as well as the efforts of political actors to mitigate the effects of rule change on their election and reelection prospects.

KEYWORDS: Japan, LDP, campaign finance, electoral reform, campaigns, personal vote, incumbency, rational choice, institutionalism, mixed-member electoral systems

**********

Mixed-member electoral systems have been adopted in Latin America (e.g., Venezuela, Mexico, Bolivia), in postcommunist countries (e.g., Russia, Croatia, Hungary, Lithuania), and in the established democracies of New Zealand, Italy, and Japan. These cases of reform have inspired widespread interest in the challenges of institutional design and the political consequences of mixed electoral systems. In this article, I offer a detailed examination of a significant informal organization associated with Japanese election campaigns: the koenkai, which refers to "the use of a mass-membership organization with the function of organizing large numbers of the general electorate" on behalf of a particular politician. (1) Few scholars have focused on the fate of koenkai in Japanese elections, which is lamentable for both theoretical and practical reasons. On a theoretical level, a greater comprehension of the koenkai is relevant for comprehending the effects of mixed rules on the salient relationships between constituents and their members in a democratic system. On a practical level, an increasing number of citizens elect their representatives under mixed rules, and many other countries are considering electoral reform. The choice of electoral rules warrants critical attention, particularly as it may have significant effects on a country's party and political systems.

The choice of electoral rules has been shown to affect party systems in various ways, such as shaping the number of political parties, (2) influencing how parties and candidates place themselves on an ideological spectrum, (3) or determining the extent that politicians find it necessary to cultivate a personal vote. (4) With emphasis on the case of Japan, I examine how electoral rules shape incentives to cultivate a personal vote through the use of personal support groups as a main vote-gathering strategy for defeating competitors. Electoral systems that encourage large parties to nominate multiple candidates in each constituency are argued to generate strong incentives for candidates to rely upon a personal vote, such as the use of support groups, in order to distinguish themselves from other intraparty and interparty competitors. (5) Yet precisely what happens to the personal vote and support groups when an electoral system is changed? A study of the evolution of koenkai in Japan affords an excellent opportunity to consider how politicians and informal organizations adapt when a new mixed-member electoral system is introduced.

I examine the evolution and incentives shaping the use of koenkai in Japan with newly collected data on the number of registered koenkai and annual expenditure data for Liberal Democratic Party (LDP) politicians. Because many of the previous studies have relied on interview and case studies, the use of this data is valuable in considering additional consequences of Japan's new electoral system, including the impact of campaign finance reforms. Moreover, the data is available for a nationwide sample and for multiple years. In this article, I show how the incentives to create koenkai have changed by comparing figures for the last three elections held in the prereform system with the first three elections held in the postreform system. I further consider to what extent politicians' incentives to maintain koenkai in the new system are reflected in the amounts of money they spend on an annual basis. I argue that the use of koenkai in Japan's new electoral system will decrease in importance as many of the original incentives fostering their creation cease to exist. I draw upon combined insights from rational choice institutionalism as well as historical institutionalism in order to elaborate on the role played by political actors. Rational choice institutionalism assumes that political actors react to institutional rules and adopt strategies that increase their likelihood of gaining office. (6) In contrast, historical institutionalism emphasizes the importance of path dependency to explain why politicians cling stubbornly to campaign strategies developed in the previous electoral system. (7) Following Pauline Luong, I adopt an integrative approach that utilizes rational choice institutionalism and historical institutionalism's respective insights. (8)

The article proceeds as follows. I begin by briefly discussing Japan's prereform system and the general literature on mixed-member electoral systems. I then turn to discuss insights derived from rational choice institutionalism and historical institutionalism. Accordingly, I examine the use of koenkai by LDP politicians in Japan's old and new electoral systems. Finally, I discuss the implications of my findings for the prospects of institutional change in Japan and in other countries.

Campaign Strategies Under Japan's SNTV System

Few studies on Japan's prereform electoral system fail to mention the ubiquitous role of personal support groups, the clientelistic ties between politicians and constituents, and the high costs of campaigning. From 1947 to 1993, Japan used a single nontransferable vote (SNTV) electoral system, which is claimed to be one of the electoral systems most likely to encourage a personal vote. (9) With a typical range of three to five seats available in each multimember district, larger parties like the LDP had little recourse but to nominate more than one candidate in each district if they desired to capture a majority of seats in Japan's more powerful lower house.

With individual LDP members facing both intraparty and interparty varieties of competition, they found it useful to rely upon koenkai to cope with uncertainty and help ensure their election and reelection. The use of koenkai incurred considerably high financial costs to individual politicians and contributed to a system where nearly four times more money per capita is spent on politics than in Germany, the United States, and the United Kingdom. (10) With scant emphasis on policy and issue debates in election campaigns, politicians maintained koenkai by spending exorbitant amounts of time and money holding regular meetings with supporters, attending funerals and weddings, and organizing bus trips to hot-spring resorts. Politics became extremely localized, which left little time for politicians to focus on broader national issues.

The premise that politicians have incentives to cultivate personal support rests on a common thread in the literature on legislative behavior: politicians are motivated by their desire to seek reelection. (11) In single-member district systems used in such cases as the US Congress, it is argued that candidates can pursue both personal and party vote-gathering strategies. In the United States, the personal vote is used to refer "to that portion of a candidate's electoral support which originates in his or her personal qualities, qualifications, activities and record." (12) Support for candidates on the basis of partisan affiliation, fixed voter characteristics such as class and religion, and reactions to the state of the economy are not part of the personal vote.

The concept of the personal vote is clearly applicable to Japan prior to and after the adoption of the new electoral system. In the SNTV system, voters faced two voting decisions. First, they had to decide which party to support. Second, they had to decide which of the same-party candidates to vote for in the multimember districts. Basing his conclusions on the analysis of survey data, Ichiro Miyake shows that as the number of candidates running from the same party increases, the more the effects of candidate components and the personal campaigning efforts of individual candidates matter. (13) His research also indicates that the candidate components in Japanese voting decisions are further salient in Japan's new mixed system. In a second study, Miyake shows that candidate components exert strong effects when voters evaluate districts that feature an incumbent competing against a newcomer. (14)

In Japan's prereform and postreform electoral systems, the use of koenkai is argued to be the most important mechanism used by Japanese politicians to cultivate a personal reputation and a personal vote. Scholars describe the use of koenkai and the variety of constituency services they provide in the following terms: "There is often an air of disguised bribery in many of these tours; constituents pay only part of the expenses involved and the machines pay the balance." (15) The extensive use of the personal vote fostered a strong and embedded system of patron-client ties, which many argue is difficult to alter. Many political reformers and scholars of Japanese politics expect Japan's new electoral system will reduce corruption and the expensive cost of elections and promote more policy and issue debates between parties. In this article, I evaluate to what extent many politicians in the LDP are willing or unwilling to abandon their personal vote-gathering strategies mastered under the old regime.

Election Campaigns Under Japan's Mixed-Member System

In 1994, Japanese politicians adopted a mixed-member majoritarian (MMM) electoral system for the lower house that currently combines 300 single-member districts (SMD) with 180 proportional representation (PR) seats. In addition, stricter campaign finance laws and a public subsidy system for political parties were introduced. Many scholars debate to what extent these changes will affect the campaign strategies of Japanese politicians. (16) In this section, I briefly discuss some of the broader literature on mixed-member electoral systems before turning to some of the specific details of the Japanese case.

Mixed-member electoral systems are suggested to offer the "best of both worlds"--both "majoritarianism and PR" and "both personalized geographic representation and party representation." (17) However, the political consequences of mixed systems are far from uniform given their differences in such areas as the specific rules governing the ballot or vote component, how the PR and SMD tiers are linked, or the percentage of seats allocated to the PR tier. In compensatory systems like New Zealand and Germany, the number of seats parties can win in the PR tier is determined in part by the number of votes cast or seats won in the SMD tier. In parallel systems such as Russia, Mexico, and Japan, the number of seats and votes won by parties in the PR tier are not adjusted based on their performance in the SMD tier. As there is a wide configuration of mixed systems in use, scholars are not in complete agreement as to whether such systems do indeed offer the best of both worlds.

On analyzing aggregate election and survey data, many scholars have reached a consensus that mixed systems are not simply the sum of their component parts, but rather are systems with incentives and consequences quite different from pure PR or SMD systems in isolation. (18) Scholars frequently use the term contamination effects when referring to mixed systems in order to describe the phenomena when the incentives of one tier affect the behavior of political actors and organizations in the other tier. One of the main findings in the literature is that the parties' placement of candidates in the SMD tier can significantly shape their electoral performance in the PR tier. Additionally, scholars have examined the effects of mixed rules on other dimensions of the political system such as strategic voting, the number of political parties, and election campaigns. (19) However, few studies have detailed the effects of mixed electoral rules on the specific campaign strategies adopted by politicians and candidates.

In Japan as well as in other mixed systems, the rules that govern the PR and SMD tiers are of no minor consequence for election campaigns as they define the specific avenues by which parties and politicians gain representation. First, candidates can compete purely in the single-member districts, which they can win outright by capturing the most votes. Second, the formal rules of the mixed system allow parties' candidates to be listed simultaneously in the SMD and PR tiers in what is called "dual candidacy." It thus becomes possible to lose the SMD race yet be "resurrected" because of a high ranking on one of the party's PR lists. (20) Finally, parties may list candidates purely in the PR tier.

The specific avenues available for winning a seat may encourage the greater or lesser use of koenkai in Japan's new system. Candidates competing purely under PR do not face the immediate pressures of interparty competition in the single-member district, which may require candidates to develop strong personal ties to individuals and groups in a single-member district in order to capture as many votes as possible. Additionally, party leaders may opt to nominate candidates without strong support networks to run on the party lists. On the other hand, the use of dual candidacies by parties may "perpetuate the personal vote and personal support organizations" of individual candidates because of the primacy accorded to the single-member district. (21) This article examines politicians' incentives to maintain their support networks with emphasis on the different incentives and capabilities they face in their efforts to win a seat in Japan's new system.

The Expected Impact of Electoral Rule Change in Japan

Many scholars anticipate that Japan's new mixed system will weaken politicians' incentives to create and maintain personal support groups. However, the explanations offered for why support groups will weaken are attributed to different rule changes, which make it difficult to ascertain the underlying causal mechanisms. Moreover, many of the claims in the literature have not been subject to an empirical test. In this section, I review some of the major claims in the literature concerning the expected impact of Japan's new electoral system on the campaign strategies of politicians, with a focus on some of the major changes associated with the reform. I first discuss several of the institutional changes claimed to be responsible for the expected impact.

One of the main expectations in the literature is that the reform should decrease the importance of koenkai in the new system because the possibility of intraparty competition has been eliminated at the election stage. If LDP politicians no longer face competition from other members in their own party, and instead can compete against the policies articulated by other parties' candidates, they will be less likely to rely upon personalized support. Indeed, some claim that the end of intraparty competition will render koenkai "unnecessary" and allow parties the opportunity to organize local party branches. (22)

Another claim in the literature is that stricter campaign finance laws and a new subsidy system for political parties may weaken the influence of koenkai in the new system. Under the old campaign finance regime, the disclosure limits were set at [yen] 1 million. At present, however, the names of all donors who contribute more than [yen] 50,000 must be disclosed. Likewise, other changes were introduced with the goal of curbing the potentially corrupt ties that exist between koenkai and their financial supporters: corporations are no longer allowed to make direct contributions to koenkai, and individual donors cannot contribute more than [yen] 1.5 million per year.

In tandem with the new campaign finance laws, reformers introduced a public subsidy system for political parties to mitigate the expensive cost of election campaigns. Officials calculate the public subsidy to political parties at [yen] 250 for each person in Japan, an expense that totaled over [yen] 31 billion in 2006. (23) To receive the subsidy, candidates are required to establish a local party branch in their district. Along with the end of intraparty competition, the new incentive for parties and candidates to create local party branches could arguably weaken the use of koenkai in the new system.

The formal changes introduced by the new system, including the end of intraparty competition and the new campaign finance regulations and subsidy system for parties, are cited as major reasons why the use of koenkai in Japan is likely to weaken. However, these explanations by themselves are incomplete without factoring in the incentives and capabilities of political actors. In their discussion of the likely effects of the new electoral laws, some argue the rules would almost be "certain to end up favoring conservative politicians who came up under old party elites in the LDP." (24) In a similarly made argument, others suggest there is little reason why incumbents would be likely to abandon their support groups, further noting that their use might actually increase. (25)

To date, there have been a few studies on the impact of Japan's electoral reform on the role of support groups. (26) In the most current study, Ellis Krauss and Robert Pekkanen rely on interviews with Japanese politicians to evaluate the continued and new vote mobilization functions of koenkai as a result of electoral reform. (27) They argue that politicians continue to find koenkai useful as a tool for catering to a broad and diverse range of interests in the new single-member constituencies. Krauss and Pekkanen claim that politicians "make 'embedded choices'--decisions made under constraints and circumstances in part created by the extant system and influenced by the temptation to reshape existing organizations to new purposes rather than create new ones from whole cloth." (28) However, their arguments are not based on a nationwide sample and do not evaluate other impacts of reform on koenkai beyond their continued and new vote mobilization functions.

In this article, I utilize a newly collected data source for considering the impact of electoral rules and electoral rule change on support groups: the campaign finance disclosure reports of Japanese politicians. As previous studies have relied largely on interviews and case studies to study the impact of reform on the campaign strategies and vote mobilization efforts of Japanese politicians, the use of these data is valuable in revealing additional consequences of the 1994 reforms, including revisions to the old campaign finance regime. At the same time, the campaign finance data allow us to evaluate the impact of reform using a nationwide sample of both winning and losing candidates for a total of six elections.

Japanese politicians are required not only to register all of their political organizations, including koenkai, with election authorities, but also to report how much money they raise and spend throughout the year. Using the campaign finance reports, I make use of two indicators--the number of koenkai registered to each politician and the amounts of money related to income or expenditure--to make comparisons across Japan's new and old electoral systems and to analyze politicians' incentives to use koenkai in the new system. I make a theoretical distinction between incumbents and new candidates, which I then apply to an analysis of the LDP. This distinction necessarily highlights how differently situated actors face dissimilar incentives and capabilities. I then offer empirical evidence that highlights the fate and evolution of koenkai from 1986 to 2003.

The Role of Political Actors

The comparative literature on institutional change often frames assumptions about political actors using insights derived from rational choice institutionalism and/or historical institutionalism. In rational choice institutionalism, it is assumed that political actors deliberately design institutions based on the short- and long-term preferences they may have or based on the strategies they expect other self-interested actors will pursue. In the historical institutionalism approach, individual agency is assigned a lesser role because structure and history are deemed to be the most significant determinant in creating or modifying institutions. In recent studies, scholars have combined the insights from both approaches to explain institutional change in such settings as former Soviet Central Asian republics.

Luong, in her study of institutional design in Kazakhstan, Kyrgystan, and Uzbekistan, integrates assumptions from both approaches to explain the establishment of electoral systems. By emphasizing the potential areas of convergence between the perspectives, she seeks to overcome the limitations of focusing on only one approach. Using this integrative approach, her study concludes that "the adoption of new institutions during a transition does not necessarily indicate either a fundamental break with the past or its continuation, but rather a change through continuity." (29)

Building on Luong and Krauss and Pekkanen, I argue that koenkai remain as a residual legacy of Japan's previous system although some of the original rationale for their existence has disappeared. (30) Koenkai are additionally useful in the current electoral system as they help alleviate the negative impact of the rule change on the election and reelection prospects of politicians. In contrast to the previous literature, my approach evaluates the evolution and current use of koenkai in Japan by examining in detail how differently situated actors face different incentives and capabilities in their quest to secure victory. My argument consequently highlights an important theoretical distinction between incumbents and new candidates.

In debates over the consequences of Japan's electoral system change, a frequent claim is that incumbents are unlikely to alter their campaign strategies perfected under the old regime, including their extensive use of koenkai. (31) Instead, incumbents are expected to convert whatever resource advantages or personal experience they formerly had in order to win reelection under the new rules. Following the insights of the historical institutionalism approach, it is assumed that the weight of previous structures and traditions bear heavily upon the campaign strategies of incumbents. From another perspective, however, the koenkai could also be seen not as a "weight" or a function of "tradition," but as an existing tool that incumbents have at their disposal. When compared with incumbents, new candidates may be disadvantaged in trying to utilize koenkai given the different incentives and capabilities they face.

Because differently situated political actors will face different propensities to use koenkai, it is likely that more senior members within the same party organization may be the least likely to abandon the use of koenkai and the personal vote. Moreover, increased seniority and incumbency should translate into considerable fundraising advantages, particularly as incumbents can rely on their name recognition and extensive national and local connections. In contrast to incumbents, candidates who have not previously won election to Japan's lower house may place less emphasis on the use of personal support groups. Not only might the influence of previous structural and historical context not be as great, but also new candidates may be disadvantaged given the variety of constraints they may face. New candidates, for example, are likely to differ from incumbents in terms of their ability to accrue financial resources, particularly because many lack sufficient name recognition or extensive connections at the local or national level. Consequently, they may have much less money to devote to personal support groups and a personal vote.

The personal characteristics of candidates are further important to party leaders when selecting which candidates should be allowed to compete on behalf of their party organization. The dual use of SMD and PR seats in Japan requires party leaders to strategize over which candidates ultimately receive the party nomination. Thus far under the new system, party leaders in the LDP have largely nominated incumbents to occupy the highest positions on the party lists. As a result, the incentives for these incumbents to use koenkai may be reduced but are unlikely to completely disappear.

Politicians" Incentives to Create and Maintain Support Groups

In this article, I examine two incentives politicians face in their use of koenkai. The first incentive is whether and to what extent the reforms affect politicians' creation of support groups. Politicians must formally establish their support groups with election authorities, where they are required to provide a physical address and the name of the manager in charge. Politicians can also dissolve koenkai by submitting the necessary paperwork, which is sometimes done when an organization will no longer be used. Thus, I use the total number of support groups registered to LDP members as a measure of politicians' incentives to create koenkai.

The second type of incentive is whether the reforms affect politicians' maintenance of these groups. Once politicians create support groups, they must calculate how many resources to allocate to the organizations' activities and upkeep. Politicians with a large membership base may need to allocate more financial resources and time in comparison to politicians with smaller and newly established koenkai. This article consequently relies on the year-round financial costs of koenkai reported by LDP politicians as a measure reflecting politicians' incentives to maintain them. However, as data are unavailable prior to 1996, the analysis of this data is limited to the new electoral system.

The incentive to create koenkai factors into the first of two main questions posed in this article: To what extent did the electoral reforms affect the number of support groups registered to LDP politicians? By determining the average number of registered support groups for LDP candidates in both systems, I am able to consider the extent of change or continuity in their use. I also calculate the average amount of income raised by LDP politicians for three SNTV elections and for three elections under the new rules. This is necessary to determine whether politicians are utilizing fewer groups but still raising and spending more money.

The second question focuses on the incentives to maintain koenkai: Since the reforms, to what extent do the personal characteristics of politicians (e.g., incumbent status) and the new rules governing the SMD and PR tiers affect the use of politicians' support groups? I use the amounts of reported koenkai expenditures to examine whether there are salient differences between incumbents and newcomers as well as between pure PR, pure SMD, and dual-listed candidates in the new system.

Comparing Across Japan's Old and New Systems

In order to consider the effects of the 1994 reforms, I examine in this section the average number of support groups for LDP incumbents and newcomers who competed in the prereform and postreform systems. Annual reports from Japan's old campaign finance system have not been studied extensively because they have been difficult to collect and analyze. Fortunately, a campaign finance book that was published on a yearly basis in Japan prior to 1996 identifies all political organizations related to individual Japanese politicians as well as the amount of reported income for each organization. The original source for this information is the yearly campaign finance reports politicians submitted to both the former Ministry of Home Affairs and the electoral commission in each prefecture.

By using this source book, it is possible to count the total number of political organizations linked to individual LDP politicians and candidates for the last three elections (1986, 1990, and 1993) held under SNTV. (32) Moreover, it is possible to approximate a general estimate of how much money politicians raised across all of their organizations. However, it is necessary to state three caveats regarding the use of total income for Japan's old electoral system. First, the campaign finance regulations during that time encouraged politicians to register multiple political support organizations in order to evade quantitative restrictions on the amount of money they may raise. (33) Consequently, many of the support organizations may have been little more than funding bodies used to collect money. The new regulations, however, allow politicians to raise unlimited funds through their local party branch and main fund agent--which should curb some of the original incentives to register large numbers of groups. In addition, politicians must disclose any contributions over [Yen] 50,000 from a previous ceiling of [Yen] 1 million.

Second, the figures reported in the book for Japan's old system are imperfect, as the income cannot be adjusted for any double-counting of financial transfers that may occur between the organizations of the same politician. Because politicians are allowed to transfer money freely between their organizations, the total income for all of their groups may be actually higher than the officially declared levels. Finally, a third caveat is the likely possibility that politicians in Japan and in other countries understate their actual expenditures. (34) These caveats suggest some caution when using this source of data to compare across Japan's two different electoral systems and campaign finance regimes.

To derive similar indicators for the first three elections held under Japan's new system, I rely upon the same source material in order to count the number of support organizations affiliated with individual candidates as well as the amounts of their annual income. (35) There is one difference in the counting procedure followed for determining the number of support groups. I do not include the newly created party branches, which were created and envisioned as an alternative organization to the koenkai. (36) However, it is necessary to include the party branch and fund agent in calculating total income for each politician. The main difference with the figures based on the SNTV system is that I have corrected for any double-counting that may occur if politicians transfer money between groups, which generally reduced total income by a small margin for most LDP members.

Table 1 displays the average number of registered koenkai and total income for LDP incumbents, newcomers, and all LDP candidates for a total of six lower house elections between 1986 and 2003. The top half of the table reports the average number of groups and income for three election years under SNTV: 1986, 1990, and 1993. (37) The means for this period, indicated in italic in the middle of the table, show that LDP members averaged 13.9 support groups and reported approximately [Yen] 209.4 ($1.9 million) in yearly income. An examination of the figures for the two candidate types in each election year further reveals the greater prowess of incumbents over new candidates. Indeed, a means comparison test (results not shown) uncovered statistically significant differences in the means for both the number of support groups and income for incumbents and new candidates with only one exception in year 1986. (38)

The lower half of Table 1 shows the average number of groups and income for LDP politicians that competed in 1996, 2000, and 2003 under the new rules. (39) As indicated at the bottom of the table, the average number of groups is 5.8 and [Yen] 102.2 ($964,000) for total income. If we compare these figures with the previous system, recognizing some of the caveats noted above, both the number of groups and yearly income have been reduced to less than half from their levels under the old system. Examining the case of incumbents, for example, shows a reduction in the total number of groups from 9.2 in 1996 to 5.5 in 2003, with a subsequent reduction in yearly income from [Yen] 133.6 ($1.3 million) to [Yen] 117.1 ($1.1 million). When compared to incumbents, new candidates registered considerably fewer groups and reported less income. (40)

The general trends captured in Table 1 support insights derived from rational choice and historical institutionalism that institutions such as the koenkai are modified and adapted in a process of change through continuity. The use of the koenkai continues in the new system despite the reduction in the number of registered groups and politicians' total income. According to the rational choice institutionalism approach, politicians' campaign strategies are based on the short- and long-term preferences they may have or are based on the strategies they expect other self-interested actors will pursue. To this extent, the use of koenkai as well as considerable sums of money appear useful to the election-oriented goals of LDP politicians.

In contrast, historical institutionalism would emphasize the embedded nature of Japan's previous electoral system and campaign finance regime on the present use of koenkai. Consequently, one could attribute the power asymmetries between incumbents and newcomers in the new system as both a continuation and change from the previous system.

However, the continued use and evolution of koenkai implies that they may be less of a "weight" or a function of "tradition" than this perspective might suggest. To the extent that koenkai continue to be valuable, they can also be viewed as an existing tool that political actors have at their disposal. Incumbents still use koenkai, whereas new candidates find it more difficult to form them. One possible reason is related to the different capabilities and incentives incumbents and new candidates face.

The LDP's Use of Support Groups in the New System

Having examined the number of support groups and total income reported by politicians, I turn to a second set of questions, which focus on the power asymmetries between political actors as well as the impact of the rules of the new system: What is the impact of candidate characteristics such as incumbent status upon the use of koenkai? How do the rules governing the SMD and PR tiers affect politicians' use of support groups? To answer these questions, I rely on the total amount of koenkai-based expenditure as an indicator of LDP maintenance costs for the year.

A multivariate test helps gauge whether there is a marked difference between pure PR candidates and those who compete as dual-listed candidates. In both models, I include a dummy variable for pure PR and pure SMD candidates to help ascertain whether they utilize fewer groups than dual-listed candidates (the excluded category). I also include two measures that tap important candidate characteristics: incumbent status and the number of lower house terms politicians were elected to previously. Incumbents and more senior LDP members are expected to report a higher level of koenkai expenditures than newcomers.

The models are tested separately for the first three elections held under the new rules. This allows us to study to what extent the results are fairly stable or change across time. Some scholars argue that it may take several elections before the anticipated changes in campaigning, policymaking, and political parties can be seen. (41) To the extent that the first elections under the new rules are marked by transitional change and uncertainty, the results for each election year may or may not reflect a consistent pattern.

The results of the model are specified in Table 2, which presents the effects of candidate characteristics and the other variables detailed above on the koenkai-based expenditures of LDP politicians. As captured here, the variable for pure PR candidacy is statistically significant and is correlated negatively with koenkai expenditures in two of the three election years. This means that pure PR candidates spent [Yen] 16.8 million ($155,600) less than dual-listed members in 1996 and much less than this amount in 2003. This finding would support the argument that pure PR candidacies require less maintenance on koenkai than LDP members who are dual-listed. This result is significant when controlling for incumbent status, which is necessary to account for LDP nomination strategies to grant incumbents preference in pure PR positions at the top of the party lists. However, there is no statistical difference between the expenditures of pure SMD candidates and dual-listed candidates.

For the remaining variables, the measures for incumbent status and number of previous terms are statistically significant and positively correlated with the level of koenkai expenditures, with the exception of incumbents in 2003. LDP incumbents spent a sum of approximately [Yen] 7.32 to [Yen] 17 million ($63,000 to $157,000) more than newcomers. For each additional term of office, LDP candidates spent approximately [Yen] 1.40 to g2.20 million ($13,000 to $19,000) more on their support groups. Finally, the results reported in Table 3 suggest that the effects of independent variables are not consistently uniform across the three elections, which may highlight the unsettled electoral environment faced by politicians in the new electoral system.

The analysis above examines only koenkai-based expenditures, although politicians also spend money throughout the year through their local party branch and fund agent. Politicians in the LDP designate an official party branch in the SMD or PR tier in order to receive and spend the newly created subsidy for political parties that was passed in 1994. Rules based on British practice allow them also to designate a fund agent to receive and spend funds. (42) To better discern how koenkai-based expenditures compare with party branch and fund agent expenditures of LDP members, I have calculated in Table 3 the percentage of koenkai, party branch, and fund agent spending to total spending. If koenkai-based spending is indeed decreasing at the margins, have the patterns of spending shifted more to the politicians' local party branches?

The percentages of expenditures for koenkai, fund agent, and party branch are reported first for pure PR candidates. In the 1996 election, the largest percentage of expenditures for pure PR candidates was spent on koenkai (38%) followed by the fund agent (36%) and party branch (26%). By 2003, the percentage devoted to koenkai dropped to 23 percent. Likewise, fund agent spending declined, whereas party branch expenses increased from 26 to 47 percent. In the case of pure PR candidates, the trend thus far suggests a decline in koenkai-based expenses and the increasing financial importance of the local party branch organizations. (43)

Listed below the pure PR candidates are the dual-listed politicians. The majority of LDP members are in this category. From 1996 to 2003, the percentages of koenkai to total spending have declined from 27 to 20 percent. At the same time, the percentage of fund agent spending in 2003 (51%) is considerably higher than the percentages in 1996 and 2000 (34% and 32%). In contrast, party branch spending in 2003 (29%) fell considerably below the percentages for the other years (39% and 46%). Although there are only three data points thus far to examine, the initial evidence for dual-listed members in the LDP indicates that the percentage of koenkai spending to total spending dropped to 20 percent by 2003. The decline is further marked by increased spending through the fund agent and party branch.

The decline in koenkai spending for dual-listed candidates may be partially related to changes in the campaign finance laws. Currently, politicians are allowed to accept contributions from corporations and other organizations only through their fund agent or party branch. However, politicians are allowed to transfer funds from their fund agent and party branch to spend on their koenkai. To the extent that the new campaign finance regulations help centralize the ebb and flow of money through the fund agent and party branch, this may explain the decline of koenkai expenses to total expenses.

Finally, the figures for pure SMD members are included just after dual-listed members in Table 3. The percentage of koenkai to total expenditure drops from 31 percent in 1996 to the mid-twenties by 2003. In contrast, the percentage of fund agent to total spending increases from 27 percent to the mid-thirties, whereas party branch spending is relatively constant. In addition, an examination of the total amount of expenditures in the far right column suggests that pure SMD candidates in the LDP spent more on average than pure PR and dual-listed candidates. However, the multivariate analysis in the previous section uncovered no statistical difference between pure SMD and dual-listed candidates in terms of koenkai expenditures. (44)

The financial costs incurred by LDP members in Japan suggest that the dynamics of the new system still encourage strong incentives to maintain a personal vote and support groups. In Table 3, the survival and continued use of koenkai persists despite the newly created party branches and the stricter campaign finance laws. This confirms claims and interview data in the previous literature that politicians are "influenced by the temptation to reshape existing organizations to new purposes rather than create new ones from whole cloth." (45) Japan's reforms appear to have impacted not only the continued and new vote mobilization functions of koenkai, but also their number and financial strength--a point that I consider further in the next section.

Conclusion

The widespread adoption of mixed electoral systems has generated considerable interest in understanding the political consequences of such systems. Focusing on the case of Japan, I examine in this article politicians' incentives to use koenkai in the context of the 1994 reforms. Using insights derived from historical and rational choice institutionalism, I highlight in my analysis some of the resilience of the koenkai as a vote-gathering strategy for the LDP, which it continues to use to win elections. A primary reason why the koenkai have not disappeared in the new system is because of their utility for mobilizing votes. Despite the efforts of many political reformers to reduce the personal vote with new legislation, the koenkai do not appear at any risk of becoming extinct in the near future.

Most LDP politicians continue to rely on koenkai-centered campaigns, which have organizational utility in gathering votes in competition with other parties' candidates. The bulk of LDP candidates are now dual-listed in the SMD and PR tier, which means to win an SMD seat, they must capture a larger average of votes in competition than was often necessary in the SNTV system. The incentive to campaign on the basis of a personal vote becomes intensified, because if they lose their SMD race, it may be possible to be resurrected to the PR tier. (46) For the LDP, the shift away from a personal vote appears to be complicated by the fact that koenkai still represent an effective and winning strategy for collecting votes.

The impact of reform on the LDP, however, may be stronger or weaker depending on whether politicians are incumbents or new candidates. In contrast to new candidates, incumbents were expected to raise the most funds and to cling stubbornly to their reliance on a personal vote and support groups. These claims were partially supported by the use of campaign finance data presented in Table 1. Incumbents clearly dominated new candidates both in terms of the number of koenkai and in their ability to accrue funds, which is likely related to the name recognition and extensive local and national connections they possess. The evidence also suggests that both incumbents and new candidates in the LDP maintained and adapted koenkai in the context of Japan's electoral shift. The average number of groups dropped from 13.9 to 5.8 while income declined from [Yen] 209.4 ($1.9 million) to [Yen] 102.2 million ($964,000), although some caution is required given the limitations of campaign finance data discussed earlier.

An alternative argument for the evidence presented in Table 1 is simply that changes in the number and financial strength of koenkai do not reflect the impact of new electoral rules--but rather are related to factors that are partially or largely exogenous to the electoral system, such as campaign finance regulations or changes to Japan's political economy. In Table 1, for example, the figures suggest that politicians' ability to raise funds has clearly diminished in Japan's new electoral system--an observation that could be attributed to multiple causes. The introduction of proportional representation, for example, seems to curb politicians' incentives to spend money on koenkai. At the same time, the results could be the product of stricter campaign finance regulations or changes to the supply side, such as the introduction of a public subsidy system or the weakened position of donors to contribute to the LDP machine. (47)

The debate over causation is not inconsequential, particularly as some could suggest that political reformers in Japan may only have needed to tighten Japan's campaign finance system in the hope of decreasing the personal vote and role of money politics. Indeed, if reformers had not altered the existing campaign finance regulations or passed the public subsidy for political parties, perhaps the average number of koenkai would have remained the same despite the rule change. It is hoped that future research can better disentangle the effects of specific rule changes from each other.

Related to the task of sorting out the impact of campaign finance regulations is an additional question that requires consideration. If the koenkai have decreased in number or financial strength, what organizational practices have taken their place for politicians' vote-gathering strategies? One possibility is that politicians in the LDP are resorting to greater media-based strategies, including the effective use of campaign posters, literature, and the Internet. Politicians may also rely more on volunteer groups, lesser-paid staff, or the local party branch structure in their constituency. The alternative argument previously suggested could link the emergence of new organizational practices more specifically to the changes in the ebb and flow of money rather than to the effects of new electoral rules. (48)

In the case of the local party branches, the LDP headquarters distributed an average of [Yen] 19.2 million ($181,000) to the main party branches in the SMD tier in the year 2000. (49) Interestingly, in ninety-two of the 300 single-member districts, LDP members transferred an average of [Yen] 18.1 million from the party branch they control to their koenkai, effectively blurring the distinction between the individual politician and the political party. For this reason, the new subsidy system is "reducing the formerly sharp distinction between koenkai and party organization. But it has not changed the reality that election campaigns are controlled by individual politicians operating within a party franchise." (50) Clearly, new organizational structures combining parts of the old and new systems continue to evolve in election campaigns for Japan's lower house--both in response to newly created incentives (e.g., electoral and campaign finance rules) and the weight of structure and history on political parties.

By using new data to measure the evolution of koenkai in Japan, the results of this study substantiate the claims and interview data of Krauss and Pekkanen that koenkai have been maintained and continued despite the electoral system change and stricter campaign finance regulations. The evidence presented in this article suggests a decline in the average number of koenkai and in their financial strength. Moreover, the impact of reform appears to differ for incumbents and new candidates within the LDP.

Japan's electoral reform additionally suggests the challenges in analyzing the political consequences that follow when electoral rules are changed. In political systems with a dominant political party like the LDP, the effects of new rules on campaigning are shaped by the powerful organizational legacies of the past as well as the efforts of political actors to mitigate the effects of the rule change on their election and reelection prospects. The introduction of a mixed-member electoral system--as well as changes in campaign finance laws--can substantially alter the preexisting equilibrium by introducing a new set of incentives for politicians and parties to win a seat and to campaign on the basis of a personal vote. In this context, old campaign strategies are, not surprisingly, remolded and dressed in a new robe.

Notes

I am grateful to Ikuo Kabashima, Gabriella Montinola, Steven Vogel, Ellis Krauss, Robert Pekkanen, Stephan Haggard, and three anonymous reviewers for their suggestions on earlier drafts. I also wish to acknowledge research funding from the Asian Studies Program at the University of Vermont and from the Japan Program at the Norwegian University of Science and Technology.

(1.) Gerald Curtis, Election Campaigning, Japanese Style (New York: Columbia University Press, 1971), p. 128.

(2.) See Maurice Duverger, Political Parties, Their Organization and Activity in the Modern State (New York: John Wiley, 1954); William Riker, "The Two-party System and Duverger's Law: An Essay on the History of Political Science," American Political Science Review 76, no. 4 (1982): 753-766; and Gary Cox, Making Votes Count: Strategic Coordination in the World's Electoral Systems (New York: Cambridge University Press, 1997).

(3.) See Ian Budge, David Robertson, and Derek Hearl, eds., Ideology, Strategy and Party Change: Spatial Analysis of Post-war Election Programmes in 19 Democracies (Cambridge: Cambridge University Press, 1987); Matthew Gabel and John D. Huber, "Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions from Party Manifestos Data," American Journal of Political Science 44, no. 1 (2000): 94-103; and Michael Laver, ed., Estimating the Policy Positions of Political Actors (London: Routledge, 2001).

(4.) See John Carey and Matthew Shugart, "Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas," Electoral Studies 14, no. 4 (1995): 417-439; and Bruce Cain, John Ferejohn, and Morris Fiorina, The Personal Vote: Constituency Service and Electoral Independence (Cambridge: Harvard University Press, 1987).

(5.) Carey and Shugart, "Incentives to Cultivate a Personal Vote," pp. 417-418.

(6.) See, for example, Kenneth Shepsle, "Institutional Equilibrium and Equilibrium Institutions." In Herbert Weisberg, ed., Political Science: The Science of Politics (New York: Agathon, 1986), pp. 51-81; and George Tsebelis, Nested Games: Rational Choice in Comparative Politics (Berkeley: University of California Press, 1990).

(7.) See, for example, Theda Skocpol, Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United States (Cambridge: Harvard University Press, 1992); and Sven Steinmo, Taxation and Democracy: Swedish, British, and American Approaches to Financing the Modern State (New Haven: Yale University Press, 1993).

(8.) Pauline Luong, "After the Break-up: Institutional Design in Transitional States," Comparative Political Studies 33, no. 5 (2000): 563-592.

(9.) Carey and Shugart, "Incentives to Cultivate a Personal Vote," p. 429.

(10.) See Chalmers Johnson, Japan: Who Governs? The Rise of the Developmental State (New York: Norton, 1996); and Hiroshi Mizuguchi, "Political Reform: Much Ado About Nothing?" Japan Quarterly 40, no. 3 (1993): 246-257.

(11.) See, for example, Douglas Rae, The Political Consequences of Electoral Laws (New Haven: Yale University Press, 1971); and David Mayhew, Congress: The Electoral Connection (New Haven: Yale University Press, 1974).

(12.) Cain, Ferejohn, and Fiorina, The Personal Vote, p. 9.

(13.) Ichiro Miyake, "Candidate Evaluation and Voting Choice Under the Japanese Electoral System." In Bernard Grofman, Sung-Chull Lee, Edwin Winkler, and Brian Woodall, eds., Elections in Japan, Korea, and Taiwan Under the Single Nontransferable Vote: The Comparative Study of an Embedded Institution (Ann Arbor: University of Michigan Press, 1999), pp. 153-180.

(14.) Ichiro Miyake, "Seito tohyo to kohosha kojin tohyo no baransu: Shosenkyokusei ikoki no tohyo kettei" [The balance between the party and personal vote: Determinants of the voting decision under Japan's new electoral system], Leviathan 25 (1999): 7-31.

(15.) Haruhiro Fukui and Shigeko Fukai, "Campaigning for the Japanese Diet." In Grofman, Lee, Winkler, and Woodall, Elections in Japan, Korea, and Taiwan Under the Single Nontransferable Vote, p. 127.

(16.) See, for example, Steven Reed and Michael Thies, "The Consequences of Electoral Reform in Japan." In Matthew Shugart and Martin Wattenberg, eds., Mixed-Member Electoral Systems: The Best of Both Worlds? (New York: Oxford University Press, 2001), pp. 380-403; Otake Hideo, "Overview." In Otake Hideo, ed., How Electoral Reform Boomeranged (New York: Japan Center for International Exchange, 1998), pp. vi-xxxi; Ray Christensen, "The Effects of Electoral Reforms on Campaign Practices in Japan: Putting New Wine into Old Bottles," Asian Survey 38, no. 10 (1998): 986-1004; and Steven Reed, ed., Japanese Electoral Politics: Creating a New Party System (London: Routledge Curzon, 2003).

(17.) Matthew Shugart and Martin Wattenberg, "Mixed-Member Electoral Systems: A Definition and Typology." In Shugart and Wattenberg, Mixed-Member Electoral Systems: The Best of Both Worlds? p. 2.

(18.) Erik Herron and Misa Nishikawa, "Contamination Effects and the Number of Parties in Mixed-Superposition Electoral Systems," Electoral Studies 20, no. 1 (2001): 63-86.

(19.) See, for example, Thomas Gschwend, Ron Johnston, and Charles Pattie, "Split-Ticket Patterns in Mixed-Member Proportional Election Systems: Estimates and Analyses of Their Spatial Variation at the German Federal Election, 1998," British Journal of Political Science 33, no. 1 (2003): 109-127; Steven Reed, "Strategic Voting in the 1996 Japanese General Election," Comparative Political Studies 32, no. 2 (1999): 257-270; Karen Cox and Len Schoppa, "Interaction Effects in Mixed-Member Electoral Systems: Theory and Evidence from Germany, Japan, and Italy," Comparative Political Studies 35, no. 9 (2002): 1027-1053; and Matthew Carlson, "Electoral Reform and the Evolution of Informal Norms in Japan," Asian Survey 46, no. 3 (2006).

(20.) Alternatively, because parties may rank their candidates at the same PR list positions, some candidates are selected through the use of a "best-loser" provision, which is calculated on the basis of their performance in the SMD tier by taking the number of votes the candidate receives divided by votes won by the first-place finisher.

(21.) Margaret McKean and Ethan Scheiner, "Japan's New Electoral System: La Plus Ca Change.... " Electoral Studies 19, no. 4 (2000): 447-448.

(22.) Reed and Thies, "The Consequences of Electoral Reform in Japan," p. 395.

(23.) Information on amounts from Ministry of Public Management, Home Affairs, Posts and Telecommunications' homepage, at www.mha.go.jp (accessed May 1, 2006).

(24.) Harold Kerbo and John McKinstry, Who Rules Japan? The Inner Circles of Economic and Political Power (Westport, CT: Praeger, 1995), p. 98.

(25.) See, for example, T. J. Pempel, Regime Shift: Comparative Dynamics of the Japanese Political Economy (Ithaca: Cornell University Press, 1998); and Gerald Curtis, The Logic of Japanese Politics: Leaders, Institutions, and the Limits of Change (New York: Columbia University Press, 1999).

(26.) For two of the first studies, see, for example, Otake Hideo, "How a Diet Member's Koenkai Adapts to Social and Political Changes." In Hideo, How Electoral Reform Boomeranged, pp. 1-32; and Masahiro Yamada, "Nukaga Fukushiro: Climbing the Ladder to Influence." In Hideo, How Electoral Reform Boomeranged, pp. 33-58.

(27.) Ellis Krauss and Robert Pekkanen, "Explaining Party Adaptation to Electoral Reform: The Discreet Charm of the LDP?" Journal of Japanese Studies 30, no. 1 (2004): 1-34.

(28.) Ibid., p. 28.

(29.) Luong, "After the Break-Up," p. 589.

(30.) For example, I concur with Krauss and Pekkanen's claims that there is a path dependence to organizational development and that politicians are likely "to adapt familiar forms to new rational purposes in the new system" (Krauss and Pekkanen, "Explaining Party Adaptation to Electoral Reform," p. 28).

(31.) See, for example, Hideo, "How a Diet Member's Koenkai Adapts to Social and Political Changes"; Pempel, Regime Shift; and Curtis, The Logic of Japanese Politics.

(32.) Seiji shikin zensho [Complete book of political funds] (Tokyo: Nihon Kokusei Chosakai, various years).

(33.) Politicians only needed to disclose contributions over [yen]1 million. By creating multiple funding groups, they could instruct donors to contribute up to the maximum limit for each group without disclosure. For a discussion of the old campaign finance loopholes in Japanese, see Michisada Hirose, Seiji to kane [Politics and money] (Tokyo: Iwanami Press, 1989).

(34.) Gary Cox and Michael Thies, "How Much Does Money Matter? 'Buying' Votes in Japan, 1967-1990," Comparative Political Studies 33, no. 1 (2000): 44.

(35.) The figures for 1996, 2000, and 2003 were gathered from Kanpo [public registrar] (Tokyo: Ministry of Finance Printing Press) and the Koho [prefectural gazette].

(36.) Reed and Thies, "The Consequences of Electoral Reform in Japan," p. 395.

(37.) The Japanese campaign finance book did not report complete figures for the year 1993. I opted to estimate the missing data for specific organizations and total income using 1992 figures. For this reason, the 1993 figures reported here are likely to be slightly lower than if actual complete data had been used.

(38.) The means comparison tests used the t-test command in Stata 9. In 1986, the means of income for incumbents and newcomers are statistically different from each other at any level greater than 7.3 percent. However, the test did not find a statistically significant difference for the number of support groups. In 1990, the means for the number of groups and income are different at any level greater than 0.0 percent. In 1993, the means for the number of groups are different at any level exceeding 0.3 percent and income at 0.7 percent.

(39.) Included in this figure are the fund agents.

(40.) A means comparison test showed that the means for the number of groups and income for incumbents and new candidates are statistically different at any level greater than 0.0 percent for 1996, 2000, and 2003.

(41.) See, for example, Gary Cox and Emerson Niou, "Seat Bonuses Under the Single Non-transferable Vote System: Evidence from Japan and Taiwan," Comparative Politics 26, no. 2 (1994): 221-236; and Peter Cowhey and Matthew McCubbins, "Conclusion." In Peter Cowhey and Matthew McCubbins, eds., Structure and Policy in Japan and the United States (New York: Cambridge University Press, 1995), pp. 253-260.

(42.) The fund agent was originally limited to a maximum annual contribution of [yen] 500,000 from corporations and [yen] 1.5 million from individuals. In 1999, the law was revised to prohibit contributions from corporations (and labor unions). In contrast, there are no restrictions on donations from corporations, organizations, or individuals to the party branch. These regulations would seem to favor the creation and use of local party branches at the expense of the koenkai and fund agent.

(43.) This trend, of course, is complicated by contamination effects from the mixture of PR and SMD rules. Many pure PR candidates are incumbents who are unlikely to disband their koenkai as they may anticipate returning to the SMD tier in a future election.

(44.) The nonsignificant result may be related to the smaller number of pure SMD candidates in the LDP, which may have made it more difficult for statistical significance to be achieved.

(45.) Krauss and Pekkanen, "Explaining Party Adaptation to Electoral Reform," p. 28.

(46.) This is possible through the parties' use of a best-loser provision, which is calculated by taking the number of votes the candidate receives divided by votes won by the first-place finisher.

(47.) Indeed, if politicians have less funds to devote to personalized support, they may have opted to focus on a fewer number of koenkai in their constituency--evoting their attention to koenkai subunits that have enough vote-gathering potential to justify their continued use.

(48.) It is also possible to argue that the effects of campaign finance laws and electoral rules are interactive and to demonstrate their interactive effects across a variety of electoral systems.

(49.) Matthew Carlson, "New Rules, Old Politics: Electoral Laws and Campaign Strategies in Japan" (Ph.D. diss., University of California, Davis, 2003), p. 84.

(50.) Curtis, The Logic of Japanese Politics: Leaders, Institutions, and the Limits of Change, p. 166.

Matthew Carlson is assistant professor of political science at the University of Vermont. He specializes in Asian politics, public opinion, electoral systems, and human rights.

Table 1 Average Income and Number of Registered Support Groups
for LDP Members

Election Year Incumbents

1986
Number of groups 14.0
Income ([yen] in millions) 255.0 ($2.4 million)
N 193

1990
Number of groups 15.7
Income ([yen] in millions) 240.1 ($2.3 million)
N 249

1993
Number of groups 16.0
Income ([yen] in millions) 204.2 ($1.9 million)
N 175

1986-1993
 (mean # of groups) 15.2
 (mean income) 233.1 ($2.2 million)

1996
Number of groups 9.2
Income ([yen] in millions) 133.6 ($1.3 million)
N 202

2000
Number of groups 7.1
Income ([yen] in millions) 129.9 ($1.2 million)
N 249

2003
Number of groups 5.5
Income ([yen] in millions) 117.1 ($1.1 million)
N 215

1996-2003
 (mean # of groups) 7.3
 (mean income) 126.9 ($1.2 million)

Election Year Newcomers

1986
Number of groups 13.2
Income ([yen] in millions) 163.9 ($1.5 million)
N 62

1990
Number of groups 5.8
Income ([yen] in millions) 80.0 ($755,000)
N 53

1993
Number of groups 6.5
Income ([yen] in millions) 98.2 ($926,000)
N 44

1986-1993
 (mean # of groups) 8.5
 (mean income) 114.0 ($1.1 million)

1996
Number of groups 2.4
Income ([yen] in millions) 58.7 ($554,000)
N 147

2000
Number of groups 2.7
Income ([yen] in millions) 57.5 ($542,000)
N 86

2003
Number of groups 1.8
Income ([yen] in millions) 49.6 ($468,000)
N 117

1996-2003
 (mean # of groups) 2.3
 (mean income) 55.3 ($522,000)

Election Year All

1986
Number of groups 13.8
Income ([yen] in millions) 232.9 ($2.2 million)
N 255

1990
Number of groups 13.9
Income ([yen] in millions) 212.5 ($2.0 million)
N 302

1993
Number of groups 14.1
Income ([yen] in millions) 182.9 ($1.7 million)
N 219

1986-1993
 (mean # of groups) 13.9
 (mean income) 209.4 ($1.9 million)

1996
Number of groups 6.3
Income ([yen] in millions) 102.2 ($1.0 million)
N 349

2000
Number of groups 6.9
Income ([yen] in millions) 111.1 ($1.0 million)
N 335

2003
Number of groups 4.2
Income ([yen] in millions) 93.3 ($880,000)
N 332

1996-2003
 (mean # of groups) 5.8
 (mean income) 102.2 ($964,000)

Sources: Seiji Shikin Zensho, Koho [prefectural gazette] and Kanpo
[public registrar], various years.

Notes: Number of groups includes all fundraising bodies listed to
politicians. From 1996 to 2003, I do not include the newly created
local party branches as part of this number. All income figures are
adjusted for inflation (2000= 100). N represents the number of LDP
politicians for which data were available in the campaign finance
reports. Omitted are the cases where the number of groups could not be
determined.

Table 2 Regressions of Characteristics on Koenkai
Expenditures (LDP)

 1996 2000 2003
IndependentVariables Election Election Election

Constant 12.60 *** 10.80 *** 7.64 ***
 (2.85) (4.12) (2.75)
Pure PR candidate -16.80 ** -6.03 -7.22 **
 (4.76) (5.07) (3.94)
Pure SMD candidate 5.37 -6.64 5.51
 (6.70) (11.50) (6.51)
Incumbent status 17.00 *** 13.30 *** 7.32 **
 (4.42) (5.22) (3.48)
Number of 2.10 *** 1.40 ** 2.20 ***
previous terms (0.62) (0.71) (0.62)
N 355 336 336
Adjusted [R.sup.2] 0.17 0.05 0.11

Sources: Seiji Shikin Zensho, Koho [prefectural gazette], and Kanpo
[public registrar], various years.

Notes: Estimates are ordinary least squares coefficients with standard
errors in parentheses. Dependent variable is the total annual amount
of LDP koenkai expenditures ([yen] in millions). The robust regression
results differ little from those above. These figures are not adjusted
for inflation.

* p < .05 ** p < .01 *** p < .001.

Table 3 Types of Expenditures for LDP Members by
Candidacy Type ([yen] in millions)

 Fund Party
Year and Type Koenkai Agent Branch Total

1996 Pure PR 38% 36% 26% 38.5 ($356,000)
2000 Pure PR 30% 39% 32% 60.1 ($567,000)
2003 Pure PR 23% 30% 47% 33.5 ($289,000)
1996 Dual-listed 27% 34% 39% 100.0 ($926,000)
2000 Dual-listed 23% 32% 46% 109.5 ($1.0 million)
2003 Dual-listed 20% 51% 29% 91.2 ($786,000)
1996 Pure SMD 31% 27% 42% 142.8 ($1.3 million)
2000 Pure SMD 23% 34% 43% 118.0 ($1.1 milion)
2003 Pure SMD 24% 33% 42% 147.1 ($1.3 million)

Sources: Seiji Shikin Zensho, Koho [prefectural gazette], and Kanpo
[public registrarl, various years.

Notes: The percentages reflect the amount of spending through the
koenkai, fund agent, or party branch in relation to total spending
(combined total across these categories). Koenkai include multiple
groups for each LDP member, while fund agent and party branch are
officially set at one organization per LDP member. Due to rounding,
not all of the subcategory figures total 100 percent. These figures
are not adjusted for inflation.