Abstract
This study presents a new explanation to a puzzle regarding the reluctance of the Japan Socialist Party (JSP) to moderate its hard-liner Marxist platform. The author does so by focusing on the preferences and strategies of individual JSP members, in contrast to previous studies that treat the party as a unitary actor. The author shows that Japan’s electoral system created a unique environment in which the electoral prospects for some JSP incumbents would be seriously jeopardized if their party increased its popular support. The analysis demonstrates that the degree of security JSP politicians had in retaining their own seats was a significant determinant in their attitudes toward a proposed policy moderation plan that was considered during a critical period in the early 1960s that could have increased support for the party.
A central component of democracy is competitive elections, whereby the people choose who will govern them. Competitiveness is created through opposition parties that put forth different visions or policy proposals as alternatives to the current government’s actions. Without an opposition party to which voters can turn when they are dissatisfied with the incumbent party, the government’s responsiveness to the demands and wishes of the electorate may decline. When politicians in the ruling party do not perceive a realistic possibility of losing power, they are less motivated to sustain their popularity.
One of the major factors that can lower competitiveness in a democracy is when opposition parties have policies that are radically different from the center of the ideological spectrum in the country. If a main opposition party advocates a policy platform that is seen as extreme by a majority of citizens, dissatisfied voters who previously voted for the current incumbent party but now want to punish that party for its performance may not be able to view the major opposition party as an appealing alternative. Thus, the ideological location of a main opposition party can influence the competitiveness of the political system and the accountability of the government. Furthermore, a polarized party system in which major parties are far apart ideologically may make the entire polity unstable. Hence, the quality of democratic government depends on the relative position of parties with respect to their ideologies.
For political parties, where to locate the party’s platform along the country’s ideological spectrum is an important issue that is made more difficult when a party seeks multiple and often conflicting goals (Müller & Strom, 1999). Left-wing parties, in particular, are often caught in a dilemma between maintaining their original policy platform and moving to a more moderate position. On one hand, moderating its policies could attract more voters and increase its chance of entering a cabinet. On the other hand, policy moderation can alienate the activists of the party who usually hold stronger ideological beliefs.
Let us look at the Social Democratic Party of Germany as an illustrative example. The party had a strong Marxist platform until 1959, when it largely abandoned it. The shift succeeded in broadening its support base, and the party has been a regular member of coalition cabinets since the late 1960s. Similarly, the British Labor Party discarded a socialistic clause from its constitution in 1995, and it went on to win a landslide electoral victory 2 years later. The policy moderations undertaken by these parties greatly altered the electoral landscape in both countries. A party’s policy shift can change not only the party’s electoral fortunes but also the country’s pattern of party competition.
Reflecting its importance, there is a rich scholarly literature on the subject of party policy positions and policy change. Downs (1957) argued that, under a set of assumptions, two political parties move to the center to maximize their votes. Since then, a large number of scholars have extended or generalized Downs’s thesis by taking other factors into account while generally remaining in line with the vote-maximizing assumption. Those factors include electoral systems (Cox, 1990; Dow, 2001; Ezrow, 2008), changes in public opinion (Adams, Clark, Ezrow, & Glasgow, 2004), the location of other parties (Adams & Merrill, 2006), and the challenge of new parties (Meguid, 2005). There have also been scholarly efforts to consider party motivations other than vote maximization. Schofield, Martin, Quinn, and Whitford (1998) demonstrate that parties in proportional systems may abandon a vote-maximizing position in favor of a location that is advantageous in coalition negotiations. Parties may also take a position that is more extreme than a vote-maximizing location to appease policy-oriented activists and utilize the resources activists provide or create for the party (Aldrich, 1983; Miller & Schofield, 2003).
In this line of the literature, a typical approach has been to analyze the behavior of political parties as if they were unitary actors with their own preferences and strategies. Studies that have sought to explain party behavior by examining individual actors (e.g., Tsebelis, 1990, chap. 5) have been far fewer in number. Theories utilizing a unitary actor assumption can have a large amount of explanatory power in certain circumstances—especially where party leaders are strong. However, a party’s actions are sometimes the result of intraparty politics and are not best explained by looking at the party as a whole. As will be demonstrated, this study offers one such example.
The case I examine in this study is the Japan Socialist Party (JSP), which was the largest opposition party in Japan during the period of one-party dominance by the Liberal Democratic Party (LDP). Japan is a well-known case where one party—the LDP—stayed in power for an exceptionally long time. Even though the country has been a long-standing democracy with a strong economy, the lack of alternation in ruling parties has led people to consider Japan as a unique or “uncommon” (Pempel, 1990) case among advanced democracies in the world. From the beginning of LDP dominance in 1955, the JSP was always the largest opposition party until 1993, when the LDP lost power for the first time. The JSP retained a strong leftist platform throughout this period and, as I will discuss later, was not seen by the public as a credible alternative to the LDP. It is now widely accepted that the JSP’s failure to moderate its leftist policy position was one of the factors that made the LDP’s long tenure possible (see, e.g., Kohno, 1997). The reason why the JSP kept its electorally disadvantageous leftist position is an important question both from a theoretical point of view and from a practical interest in understanding the nature of democracy in postwar Japan.
There have been several explanations put forth for why the JSP failed to become a less radical center-left party. Perhaps the most widely accepted explanation is the organizational dependency hypothesis (e.g., Flanagan, 1984, and references cited therein). In this view, the JSP did not abandon its leftist policies because of its reliance on the leftist labor unions for support and funding. Another explanation focuses on the JSP’s disastrous experience in the late 1940s. After participating in two coalition governments with conservative parties between 1947 and 1948, the JSP suffered a devastating setback in the 1949 election, losing two thirds of the seats it gained in the previous election. Stockwin (1992) argues that this experience brought about “a steady shift in the balance of power [within the party] from the Right, which was initially dominant, toward the Left” (p. 87). Thus, the view that it lost a great deal of its support after cooperating with conservative partners led to reluctance to moderate its leftist positions (also see Tani, 1992, pp. 80-84). Yet another explanation emphasizes the role of the Japanese Communist Party (JCP). Kohno (1997) argues that it was difficult for the JSP to change its position because of the fact that it was located between the JCP on its left side and a few small centrist parties on its right side.
Although these explanations are generally plausible, they suffer from several shortcomings. First, the existing explanations tend to be abstract because they try to account for a political process lasting 30 to 40 years in its entirety. As I will demonstrate, there was a crucial period in the early 1960s during which the party underwent a severe internal conflict, and decisions made at that time subsequently restricted the behavior of the party for decades. By focusing on this crucial period, I am able to provide a more concrete account of the JSP’s ideological immobility. Second, previous approaches have neglected the individual socialist legislators as a useful level of analysis in determining the party’s policy positions. Instead, most studies have tended to treat the JSP as a coherent single entity with its own preference and strategy. As will be shown, there were instances of fierce internal conflicts within the JSP over the party platform during the period under examination, and individual legislators were the important actors during these conflicts. My analysis demonstrates the usefulness of studying the behavior of political parties as the end result of internal political struggles within the party organization.
My main argument is that Japan’s electoral system created incentives among some socialist politicians to prevent their party from becoming more popular. Strangely, there were incumbent JSP members who remained better off if the party failed to increase its support and hence resisted any proposed changes to the party’s platform. How could such an incentive structure originate?
The Policy Position of the JSP and the Structural Reform Plan
A Short History of the JSP and Its Platforms
Like many leftist parties around the world, the JSP has a history of internal debates and fights between those who embrace what they regard as the pure Marxist ideal and those who take more moderate positions. The JSP was established in 1945, soon after the end of World War II, and participated in two coalition cabinets during the early years of postwar politics (1947–1948). In 1951, the party split into the Leftist JSP and the Rightist JSP. The two reunited in 1955—the same year the LDP was established by the merger of two conservative parties. For the next four decades, the LDP and the JSP were the two largest parties in Japan, with the LDP maintaining continuous power. This durable party system came to be known in Japan as “the 1955 regime.”
Even after the reunification, the internal rivalry within the JSP between the hard left and moderates remained and in some respects intensified. In 1959, 4 years after the reunification, a rightist faction led by Suehiro Nishio left the party and established the center-left Democratic Socialist Party (DSP). There was also an attempt to moderate the party’s platform in the early 1960s (the Structural Reform plan), which, for reasons I will discuss in detail, failed after an intense intraparty conflict. Instead, the party ended up adopting a hard-liner leftist document in 1964 titled “The Road to Socialism in Japan” that would ultimately bind the behavior of the party for decades to come (Mori, 2001). Hara (2000, p. 199) notes that, except for a short period in the early 1960s, the leftists were always the dominant force within the party, which means that Eda’s failed attempt in the early 1960s was the only realistic chance to moderate the party’s platform. Although Japan would experience rapid increases in wealth and living standards under LDP rule and a market economy, the JSP continued to argue for the transformation of Japan’s capitalist economy into a socialist system. Rarely did the JSP put forth any alternative policy proposals, instead acting almost solely as an anti-LDP, antisystem party. The party continued to “advocate a ‘dictatorship of the proletariat,’ a slogan long since abandoned even by most communist parties, including Japan’s” (Pempel, 1982, p. 35).
After discarding “The Road” as their official platform in 1986, the JSP experienced some success in increasing its legislative seats by capitalizing on an unpopular LDP tax reform plan and several corruption scandals in the late 1980s. The gains, however, were short-lived. In the 1993 election in which several new parties entered the electoral competition, the JSP suffered the largest defeat, reducing the number of seats from 136 to 70. It participated in several coalition cabinets during the 1990s while becoming smaller and smaller. Currently, the party controls only 6 seats in the 480-seat House of Representatives. 1
The fact that the support for the party suddenly declined when new parties emerged suggests that voters did not consider the JSP a credible and viable alternative to the LDP. Instead, the JSP functioned as a watchdog organization monitoring the LDP government. Inoguchi (1983) and Kabashima (1988) are representative of the view that much of the support for the JSP came from so-called “buffer players” who did not want a JSP government but voted for JSP candidates in an effort to keep the LDP from becoming too strong. The votes for the JSP from those “buffer players” made the popularity of the JSP look far larger than the size of its real support base. The JSP was never seen by any sizable segment of the population as a real contender for running the government, thus ensuring the LDP’s long standing as the dominant party in Japan.
The Structural Reform Plan and Its Demise
Why was the JSP reluctant to moderate its policy position? We have to look to the one occasion in the early 1960s when the moderation of the party platform was actually on the agenda and seriously considered within the party. It was arguably this occasion when the JSP came the closest to becoming a party that could viably compete for governmental power. Saburo Eda was the main proponent of the plan to moderate the party platform in the early 1960s. Eda had belonged to the leftist wing of the JSP when the party was split into two from 1951 to 1955. He became the party secretary (the number-two position) in March 1960 and soon brought up the “Structural Reform” plan (hereinafter referred to as SR) that sought to moderate the party. This plan is often compared to the position change of the German Social Democratic Party at the Bad Godesberg Conference in 1959 (e.g., Ishikawa, 1984, p. 150; Soga, 1989). Unlike its German counterpart, however, the JSP failed to adopt the SR and instead clung to the hard-liner leftist position.
Since becoming party secretary, Eda had argued in favor of the SR on numerous occasions in party speeches and magazine articles. For example, in a speech presented at a party meeting in July 1962, he praised “America’s high living standards, the Soviet Union’s thorough social welfare, Britain’s parliamentary democracy, and Japan’s pacifist constitution” and said these things should be seen as the image of “the new socialism in the modern society, which is different from the ones in the Soviet Union and China” (Ishikawa, 1984, p. 67). Praising the United States (even for its living standards) was a bold step for a socialist politician.
Eda’s arguments (and the SR debate in general) ultimately split the party. Although some favored it, others fiercely criticized Eda for being a revisionist and accepting “the current capitalist regime” (Ishikawa, 1984, p. 67). At the 21st party congress in January 1962, there was heated debate over the SR, and a motion was adopted that prohibited the party to immediately accept the SR as the party’s principle. Later, at the 22nd party congress in November 1962, the anti-SR group intensified its attack on Eda, and a motion to censure Eda was adopted by a narrow margin. 2 Eda then resigned from the party secretary post, and the conflict within the party subsided after this event (Mori, 2001, p. 57). The JSP then adopted a hard-liner leftist document, “The Road to Socialism in Japan,” in 1964, which subsequently bound the party’s behavior for a long time to come (Mori, 2001). When Eda and his argument for policy moderation lost, the JSP reverted to a leftist-dominated party, and the chance to become a competitive party might also have vanished. Kakuei Tanaka, an LDP heavyweight and prime minister in the 1970s, once said that Eda was the most formidable man in the JSP and the LDP might be defeated if Eda became the JSP’s leader (Shiota, 1994).
Explaining the failure of Eda and the SR should be the central question in understanding the behavior of the JSP during the period of LDP dominance in Japanese postwar politics. For JSP members, whether their party maintains its Marxist platform or moderates it would of course be a sensitive issue, and their beliefs and ideologies—how earnestly they believed in Marxism—should no doubt be an important factor in determining their attitudes toward the SR. We might assume that the division over the SR mirrored closely the left–right division within the JSP. A closer examination, however, of JSP legislators at the time shows that some hard-liner leftists supported the SR whereas some from the rightist factions actually opposed it. In addition, quite a few JSP members switched factions during the period. This narrative suggests two things about why the SR was rejected. First, ideology by itself does not explain much about JSP members’ attitudes toward the SR. Second, power struggles between factions in the JSP notwithstanding, the individual legislators appear to be the important actors in this event. Obviously, more than ideology affected individual JSP members’ attitudes toward the plan to moderate the party’s position.
I argue that the single nontransferable vote (SNTV) electoral system created various incentives for JSP politicians that led to differences in their behavior. Specifically, the increase in the JSP’s popularity that the SR would have brought was expected to increase the reelection prospects of some JSP members but decrease the prospects of other members. I contend that the legislators whose reelection quests would have been jeopardized by the position change were those who ended up opposing the SR.
Why could someone’s reelection become difficult if his or her party becomes more popular? This should sound implausible, but it was made possible by the SNTV system. In the next section, I examine how the SNTV electoral system created incentives among certain JSP members to keep their party smaller and less popular.
The Incentive Structure Induced by the SNTV System
The SNTV system was used to elect the members of the House of Representatives of Japan’s National Diet from 1947 to 1993. The district magnitude, or the number of members elected from a district, varied throughout the country, but it was typically between three and five seats. Each voter casts one ballot for an individual candidate, and the top N candidates in an N-member district were elected. In the 1960 general election, which elected the legislators who were incumbents when the JSP dealt with the SR issue, there were 40 three-member districts, 39 four-member districts, and 38 five-member districts, in addition to an exceptional single-member district that covered some remote islands of Kagoshima prefecture. In total, there were 467 members of the lower house, and 145 of them belonged to the JSP.
Under the SNTV system, a candidate’s chance of winning a seat depends on many different factors. First, the size of the personal and partisan support base in the district is, of course, important as in any electoral system. Second, the number of seats in the district (district magnitude) significantly changes the difficulty of election. In a three-member district, one needs 25% of the total votes to secure a seat, whereas 20% is needed in a four-member district and 16.67% in a five-member district. Third, whether there are candidates from the same party running in the same district makes a big difference. If support for the JSP is sufficiently large enough to support two candidates running in the same district, the party may try to secure two seats instead of just one. Yet there is always the risk that running two candidates from the same party in the same district might split an insufficient support base, so that neither candidate ends up getting enough votes to secure a seat. Fourth, in a multiparty system, which parties field candidates in the district also makes a difference. If two parties that have similar positions compete with each other in a district, their candidates have to compete for the same type of voters, and thus it becomes more difficult to obtain enough votes to guarantee a seat.
The intention of Eda’s SR plan was to make the JSP a more moderate party that could attract a wider support base and compete against the ruling LDP. To be competitive against the LDP, one thing the JSP absolutely needed to do then was to increase the number of its candidates. In the 1960 election, it nominated 186 candidates, which means that even if every one of them had won, the party would have controlled only 40% of the 467-seat House of Representatives. To become a party that could defeat the LDP and win power, the JSP would have to increase the number of its candidates. 3
However, fielding more candidates would have made the JSP incumbents worse off in terms of their reelection prospects. As noted above, if two or more JSP candidates run in a district, the votes of the party supporters are split between the candidates, making a reliable victory more difficult for them. Hence, some JSP incumbents at the time were in a dilemma in which the plan that was supposed to make the party more popular would jeopardize the possibility of their reelection.
Not all JSP incumbents were in this dilemma; the extent to which the SR would lower the chance of reelection was different depending on the situation of each incumbent’s home district. For instance, there were 13 members who were elected from three-member districts where the JSP nominated two candidates. Since it is extremely unlikely that the JSP would run three candidates in a three-member district, these 13 members were safe from the possibility of having additional JSP candidates in their districts. We might expect that these members would have welcomed the SR since it would likely expand the party’s popularity and make their reelection easier.
Thus, the possible effects of the adoption of the SR on the reelection quests of the JSP incumbents depended on the district magnitude and the number of JSP candidates who ran in the previous election. There were some members elected from districts where the JSP was unlikely to increase the number of candidates, and hence the SR would not have jeopardized their reelection prospects. In contrast, there were members from some districts where the JSP ran few candidates so that an increase in the number of candidates fielded for these districts would be expected in a program to increase the party’s popularity and national standing.
Table 1 shows the number of JSP candidates in the 1960 election, classified by the district magnitude (rows) and the number of JSP candidates running in the district (columns). 4 The percentages of the JSP candidates who were elected are reported in parentheses. As would be expected, the fewer the number of candidates running in a district, the more likely those candidates were to win their respective elections. 5 In the 27 districts where the district magnitude was 4 or 5 and the JSP ran only one candidate, every JSP candidate won a seat. Since the JSP candidates could mobilize labor unions, it was not too difficult to consolidate the number of votes needed to secure a seat in a larger district. These 27 sole candidates in four- or five-member districts were enjoying the easiest campaigns of all the JSP candidates. As long as the party did not nominate second candidates in their districts, these “safe-seat” JSP politicians could presumably retain their positions with little effort until they decided to retire.
The Number of the JSP Candidates in the 1960 Election, Classified by the District Magnitude and the Number of the JSP Candidates in the District
Winning percentages are in parentheses.
For the party, it was a safe strategy to nominate only one candidate in a large district because it was almost certain that the candidate would win. Yet if the JSP was to seriously compete for power, it would have to abandon this safe strategy and start running more candidates in the large districts; and that was what would have happened if the SR had been adopted and the party’s popularity had increased.
Hence, the party’s (and perhaps party leaders’) seat-maximizing incentive collided with some incumbents’ reelection incentives, which would not typically happen under other electoral systems. Those with “safe seats” may well have resisted the change to the status quo if their foremost concern was protecting their careers as legislators rather than realizing a socialist government. In contrast, the legislators other than the “safe seaters” would generally welcome the change that might increase the party’s popularity and make their reelection easier.
Some may find my argument excessively cynical and question if the JSP legislators, many of whom were labor union activists before becoming politicians, were really concerned about their own reelections. Yet it is not entirely new to view the JSP members as reelection seekers. Reed and Bolland (1999) indeed find a “downward ratchet” mechanism in the JSP’s candidate nomination pattern. That is, “whenever the JSP runs one less candidate than in the previous election, it finds it difficult to increase the number of candidates thereafter” (Reed & Bolland, 1999, p. 214), and they argue that this tendency is caused by “the power of incumbents in the nomination process” (p. 217). I am also not arguing that the reelection incentive was the sole determinant of the JSP members’ attitudes toward the SR. The members’ ideologies, of course, were one of the most important factors. I argue that, after controlling for the effects of the ideology, whether they had safe seats that would have been lost by the adoption of the SR made a difference in their attitudes.
Hypothesis 1: The JSP legislators who were the sole JSP candidates in four- or five-member districts resisted the SR plan more than the ones who were not. 6
Along similar lines, another factor that should have made a difference in JSP incumbents’ attitudes toward the SR is whether they were competing against other opposition parties’ candidates in their home districts. In the early 1960s, there were two opposition parties other than the JSP: the JCP with three lower house seats, which was at the far left, and the DSP with 17 seats, which was located between the JSP and the LDP.
Located between two other opposition parties, the JSP was in a difficult situation in choosing its ideological position. If the party had moved closer to the center, some leftist JSP supporters might have switched their support to the JCP; but if it had moved more leftward, centrist voters would have abandoned the JSP and voted instead for the DSP (or even for the LDP because the DSP was so small then). Ramseyer and Rosenbluth (1993, p. 42) suggest that the JSP’s adherence to the leftist policy was a “niche strategy” to locate itself between the other parties. Kohno (1997) also elaborates this thesis and presents empirical evidence.
Although I consider the “niche” hypothesis a plausible explanation of the failure of the JSP to moderate its policies in general, for this particular event in the early 1960s, the level of analysis must again be shifted to the individual JSP politicians who had the power to initiate the SR or defeat it. Using the individual JSP incumbent as the unit of analysis, I argue that the level of competition with other opposition candidates in their home districts (especially with DSP candidates, for a reason to be explained shortly) was an important factor in determining their attitudes toward the adoption of the SR. Table 2 classifies JSP members according to the competition they faced from other opposition candidates in the 1960 election. The JCP ran candidates in all 118 districts, as it did in subsequent years, but won only three seats in 1960. Therefore, all JSP incumbents were in competition against a JCP candidate. As the table shows, there was little variation among JSP members in terms of facing a JCP rival. Moreover, 88% of the JCP candidates in the 1960 election received less than 5% of votes in their respective districts. Since every JSP incumbent would expect a JCP competitor, coupled with the fact that JCP candidates appeared to be almost invariably weak competitors, one would not expect the competition with the JCP to be a major discriminating factor in JSP incumbents’ positions toward the SR.
Japan Socialist Party (JSP) Incumbents and Their Competition Against Other Opposition Parties in the 1960 Election
The same cannot be said of JSP incumbents who faced DSP candidates. In all, 16 JSP members had no DSP competition, 110 faced DSP candidates who lost in their districts, and another 19 JSP members competed against DSP candidates who won a seat in the election. Because the DSP was located between the JSP and the LDP, the presence of a DSP candidate in a district should have had a large impact on the strategy of the JSP members running in the district since the adoption of the SR would have the effect of moving the JSP close to the policy space held by the DSP. Without a DSP candidate, the JSP candidate would be in a sole position to gain support from center and center-left voters if the JSP moved toward the center. However, if a DSP candidate occupies that policy space, then any move by the JSP toward the center would not produce much in the way of electoral benefit as the centrist vote would then be split between the two candidates. In sum, I argue that whether or not JSP candidates had to compete with DSP candidates should have produced a noticeable difference in the amount of electoral benefit those JSP members would have expected from the adoption of the SR. 7 Thus, I propose the following hypothesis:
Hypothesis 2: The JSP legislators who were in competition against DSP candidates in their home districts were more likely to resist the SR plan than those who lacked DSP competition.
Empirical Investigation
Identifying the Attitudes of the JSP Legislators
The two hypotheses stated above are tested using the data of individual JSP legislators in the early 1960s. The dependent variable is their attitudes toward the SR plan, a proposal to moderate the JSP’s leftist platform. In identifying their attitudes, we can obtain useful information from the factional affiliation of the JSP legislators. Table 3 shows the factional affiliations of JSP lower house members in 1961 and 1963. The sources are the 1961 and 1963 editions of Kokkai Binran (Nihon Seikei Shinbunsha, 1961, 1963). The 1961 edition was published in February 1961 and reflects the initial factional affiliations of JSP members who were elected in the November 1960 election. The 1962 edition is quite similar to the 1961 edition in that it shows the factional affiliations at the time but does not reflect the factional realignment that was taking place. The 1963 edition indicates that each of the four largest factions— Kawakami, Wada, Suzuki, and Heiwa 8 —suffered from an internal cleavage.
Factional Realignment Within the Japan Socialist Party (JSP) During the Conflict Over the Structural Reform Plan
The source for the factional affiliations is Nihon Seikei Shinbunsha (1961, 1963). See the text for the source for the attitude toward the structural reform (SR). Values are the numbers of JSP lower house members. Shaded cells indicate five or more members.
The Kawakami faction, led by Jotaro Kawakami who became the party leader in March 1961, had two subgroups in 1963: the mainstream group and the group under Eda’s influence. The Wada faction, led by Hiroo Wada, contained three subgroups: the mainstream group as well as groups that followed Eda and Seiichi Katsumata, respectively. The Suzuki faction, led by former party leader Mosaburo Suzuki, was splitting into two groups led by Eda and Kozo Sasaki (the leading critic of the SR), respectively. The Heiwa faction contained two members who were under Sasaki’s influence and 13 members of the mainstream group. As displayed in Table 3, many members were switching sides during this 2-year period with the Kuroda faction and the Leftist Group disappearing completely by the 1963 report.
Since the factional realignment between 1961 and 1963 took place concurrently with the fierce intraparty conflict within the JSP over the SR and the rivalry between Eda and Sasaki, it would be safe to assume that the factional affiliation in 1963 reflects the members’ attitudes toward the SR. In Table 3, I indicate whether each faction or group in 1963 was for or against the SR. As described earlier, Eda and Sasaki were the main figures during this bipolar conflict. Thus, there is no doubt that the members of Kawakami-Eda, Wada-Eda, and Suzuki-Eda were supportive of the SR, and those of Suzuki-Sasaki and Heiwa-Sasaki opposed it.
Journalistic evidence from this period suggests that Katsumata was one of the proponents of the SR and was supportive of Eda, 9 and hence we can classify the Wada-Katsumata group as pro-SR. Similarly, the Kawakami-main can be considered as a pro-SR group. 10 Conversely, the Heiwa faction and the Nomizo faction—both leftist—were against the SR and supported Sasaki. 11 The Wada-main group is somewhat difficult to classify. Evidence suggests that most of the members of the Wada faction supported Eda and the SR, 12 although it was also reported that Wada personally took an ambiguous or neutral stance. 13 Thus, it can be inferred that the members of the mainstream group within the Wada faction were less supportive of the SR than those who followed Eda or Katsumata but more supportive than those who strongly opposed the SR. As a result, I code the Wada-main group between the pro-SR people and the anti-SR people. The three legislators of the “centrist” grouping are also considered “in between.” Last, the nine JSP members whose factional affiliations are not listed in the 1963 Kokkai Binran are excluded from the analysis.
The coding procedure described above results in a sample of 136 JSP legislators classified according to their attitudes toward the SR: 47 are pro-SR, 25 are in between, and 64 are anti-SR. In the next subsection, the members are cross-tabulated by their attitudes toward the SR and the independent variables as a preliminary analysis.
Cross-Tabulation Analysis
Both hypotheses predict that support for the SR by JSP legislators will vary depending on the level of electoral competition each member faces. Before presenting a full analysis with control variables, this subsection shows an overview of the results using cross-tabulation tables.
Hypothesis 1 predicts that sole JSP candidates in large districts will be more likely to oppose the SR. As seen in the upper panel of Table 4, the percentage of anti-SR members among sole candidates is 62.5%, which is considerably higher than the percentage among non–sole candidates (43.8%). Also, the percentage of pro-SR members is higher among the non–sole candidates (38.4%) than among the sole candidates (16.7%). The pattern is consistent with Hypothesis 1, although the relationship does not reach statistical significance (the p value from a chi-square test is .118, the p value from a Fisher’s exact test is .109), which is not altogether surprising considering that a bivariate analysis does not control for the effects of other variables.
Cross-Tabulation Analysis
SR = structural reform; JSP = Japan Socialist Party; DSP = Democratic Socialist Party.
Hypothesis 2 predicts that JSP candidates who are competing against DSP candidates in their districts will oppose the SR. Ideally, the independent variable should measure the likelihood that a JSP incumbent will face a DSP candidate in the next election, which is not directly observable. As a proxy, I use the information from the previous election and classify JSP incumbents into three groups: those whose districts had a DSP candidate who was elected, those whose districts had a DSP candidate who lost, and those whose districts did not have a DSP candidate in the last election. This should be a reasonably good proxy for the likelihood of seeing a DSP candidate in the next election since DSP incumbents are very likely to run in the next election and districts without DSP candidates are generally places where the DSP was weak and presumably will remain so in the near future.
The findings in the lower panel of Table 4 suggest strong support for Hypothesis 2. The percentage of anti-SR members is the highest (68.4%) among those JSP members who were competing against DSP candidates who had been elected in the previous election. JSP members who were running against DSP candidates who had lost in the previous election were less likely to be against the SR (47.5%) than those members who were facing DSP incumbents. Those JSP members who were the least likely of all to be anti-SR ran in districts that did not have a DSP candidate (18.8%). The percentages of pro-SR members also show a pattern that is supportive of Hypothesis 2. Where the JSP candidate would likely face no competition from the DSP, 68.8% were pro-SR, compared to 10.5% among those who would have to compete against a DSP incumbent. The relationship in this table is statistically significant (the p value from a chi-square test is .009, the p value from a Fisher’s exact test is .008).
Maximum Likelihood Estimation
Overall, the cross-tabulation analysis of the JSP members supports my two hypotheses. In this subsection, I utilize the maximum likelihood method to perform another round of testing for the hypotheses. The dependent variable is the members’ attitudes toward the SR, which I identified earlier. It is coded 1 if pro-SR, 2 if in between, and 3 if anti-SR. The ordered logit model is used because of the fact that the dependent variable is an ordinal variable. Since our total number of cases (136) is somewhat small for such a model, the results should be interpreted as complementary to the previous cross-tabulation analysis.
To test Hypothesis 1, a dummy variable is included that takes the value of 1 if the member was a sole JSP candidate in a four- or five-member district and 0 otherwise (this variable is denoted as Sole). For Hypothesis 2, I employ two ways to operationalize the presence of the DSP. The first method is to include two dummy variables in the model: one indicating whether the member’s district had a DSP candidate who won a seat (DSP Winner) and the other whether there was a DSP candidate who lost (DSP Loser). If both variables take 0, there was no DSP candidate in the district. As a robustness check, I include a second operationalization of DSP competition measured by the share of the vote received by the DSP candidate (DSP Votes).
An important factor that needs to be controlled is the members’ ideologies since it is quite plausible that those who strongly believed in Marxism tended to oppose the SR more than the rightist (center-left) people in the party. To control for the members’ ideologies, the model includes a dummy variable that takes the value of 1 if the member belonged to one of the leftist factions in 1961 (Suzuki, Heiwa, Kuroda, Nomizo, and the Leftist Group) and 0 otherwise (Leftist). 14 Since the factional realignment that took place between 1961 and 1963 was extensive, as illustrated in Table 3, it should not be a problem to use the 1961 factional affiliation as a control variable while constructing the dependent variable using the 1963 factional affiliation.
Two more control variables are included in the model. First, I include the vote share that went to the JCP candidates (JCP Votes). 15 Second, I attempt to control for the age of the members (Age) so as to capture possible generational effects. There is almost no correlation between Age and Leftist.
Table 5 shows the result of the ordered logit analysis. The two models differ with respect to the two methods by which I operationalized the presence of the DSP. In both models, the variables of interest have the expected signs and are statistically significant, providing strong support for my hypotheses. For easier interpretation, I used the Clarify program (King, Tomz, & Wittenberg, 2000; Tomz, Wittenberg, & King, 2003) to calculate the predicted probabilities while changing the combination of the independent variables and keeping JCP Votes and Age at their respective mean values (3.0 and 52). The results are presented in Table 6. For each set of conditions, the predicted probabilities of falling into each of the three categories of the attitudes toward the SR (i.e., pro-SR, in between, and anti-SR) are calculated.
Ordered Logit Estimation of the Japan Socialist Party Members’ Attitudes Toward the Structural Reform
DSP = Democratic Socialist Party; JCP = Japanese Communist Party.
p < .05. ***p < .001.
Predicted Probabilities
Note: Numbers in parentheses show the 95% confidence intervals.
Since the results from the two models are quite similar to each other, let us focus on the results from Model 1. As expected, those legislators who had the highest probability of being against the SR (91%) were members who were leftists (Leftist = 1), who were the sole candidates from four- or five-member districts (Sole = 1), and whose districts had a DSP candidate who won a seat (DSP Winner = 1). All JSP legislators in the sample who fit all these conditions were indeed anti-SR. The probability of being anti-SR declines to 63% without any DSP competition, to 80% if Sole = 0, and to 44% if Leftist = 0. For those candidates who were not leftists, were not sole candidates, and did not compete against a DSP candidate, the predicted probability of being anti-SR is just 5% and the probability of being pro-SR is as high as 86%. There are 10 such members in the data; 9 of them are indeed pro-SR, and the other 1 is in between. The maximum likelihood estimations and resulting probabilities presented provide further evidence of very large differences in preferences toward the SR among different JSP members that conform nicely to the theory put forth to explain these differences.
Conclusion
This article puts forth a theoretical approach to JSP legislators’ preference toward the SR as a dilemma, where the dictates of individual rationality, relative to each member of the group, conflicted with the interests of those individuals at the group level. The empirical analysis showed strong support for my argument that those legislators whose reelection prospect would have been jeopardized by the party’s transformation to a center-left party were the people who opposed the SR. The SNTV electoral system appears to have trapped some legislators in a unique dilemma in which an increase in the popularity of their own party decreases the probability of their reelection. It is not difficult to think that there were members who chose to pursue personal gain (reelection) over the party’s collective gain.
If we assume that a political party is a unitary actor and behaves to maximize its votes, seats, or power, moving closer to the center of the political spectrum by adopting the SR should have been the optimal action for the JSP. Yet if we shift our focus to individual JSP politicians, it can be understood that the optimal behavior of a vote-, seat-, power-seeking party was not necessarily consistent with every member’s interest. The JSP’s nonoptimal behavior can be explained by its incumbent legislators’ purposive actions.
The JSP was thus caught in a situation in which an effort to make the parry larger was in conflict with some (or many) of its members’ individual interests. Similar situations were observed in later times. In 1990, the support for the JSP was increasing partly because of scandals and unpopular policies of the LDP and partly because of its popular new leader, Takako Doi, the first female leader of any major party in Japan. Doi tried to greatly increase the number of JSP candidates for the 1990 election but failed because of the resistance of the party’s incumbents (Reed & Bolland, 1999). The SNTV system was replaced by a system that combined single-member districts and proportional representation in the 1994 electoral law reform, and the institutional structure that entrapped the JSP in a dilemma disappeared. Fifteen years after the electoral reform, the LDP suffered a major defeat and lost power in the August 2009 election; but the election was won not by the JSP but by the Democratic Party of Japan that was founded after the electoral system reform.
The new explanation of the JSP’s ideological immobility provided by this article is important not only because the JSP’s inability to compete against the LDP was a major factor that made possible a long one-party dominance in Japan but also because it has implications for the theoretical research on political parties’ behavior. As discussed at the beginning, treating a party as a unitary actor has been a typical approach in the literature on which this study is based, and this study has succeeded in shedding new light on an important puzzle by shifting the focus from a party as a whole to individual actors. Even though this article examines only one party in a country with a unique electoral system, the idea that individual legislators may value their own reelection over their party’s success as a whole can be applied and extended in other contexts. Depending on various characteristics of electoral systems, such as district magnitude, nomination process, and the use of runoff, an electoral environment can arise in which a party’s actions increase the chance of reelection for some members while decreasing it for other members. Under such situations, intraparty politics may become more complex than in systems where the reelection prospects of legislators who belong to the same party are tied together. More case studies will be necessary to generalize the idea presented in this article, and a general theory of the influence of electoral systems on intraparty politics and party behavior may also be constructed. I believe that this study has taken a significant step toward major advancements in the research on party behavior.
Footnotes
Acknowledgements
An earlier version of this article was presented at the Midwest Political Science Association conference, Chicago, Illinois, April 7–10, 2005. The author would like to thank the participants of the panel and Hiroki Mori for their comments. The author remains responsible for all errors.
The author declared no potential conflicts of interest with respect to the authorship and/or publication of this article.
The author received no financial support for the research and/or authorship of this article.