Abstract
Why and when does party headquarters grant autonomy to its local branches, and what trade-offs ensue when providing such discretion? This article tackles these broader questions by focusing on a largely under-investigated area: the impact of directly elected governors and mayors on intra-party dynamics. To facilitate analysis, the article borrows from the ‘delegation’ and ‘franchise’ models, which seek to explain how party organizations adapt to multi-level electoral environments. Sources of intra-party conflict over gubernatorial campaigns derive from theories of party behaviour under presidential systems. Empirical evidence is provided by examining how Japan’s two major parties have dealt with gubernatorial campaigns in recent years. Select cases and data demonstrate how local units are driven by differing incentives from the national leadership, the consequent intra-party conflicts and the limited success of party headquarters in steering its local units. The article discusses the extent to which the ‘franchise’ or ‘delegation’ models successfully capture the way in which the two Japanese parties have engaged in gubernatorial campaigns. To illustrate generalizability beyond Japan, the article ends with a brief comparative counterpoint: intra-party conflict over the selection of the Greater London Authority’s mayoral candidate in 2000 and 2004.
Introduction
In the limited but growing literature on multi-level intra-party dynamics, scholars have attempted to measure the autonomy local party branches possess from the central party leadership in terms of various local party activities. These include the selection of candidates and leaders, formation of policy, electoral campaign strategies and governing strategies (Thorlakson, 2009: 162–163; Van Biezen and Hopkin, 2006: 26–33). Most of this party organizational literature focuses on European states which do not have a tradition of powerful directly elected mayors or governors at local level. As such, little of the comparative literature on intra-party relations has considered the effect of local chief executives on the dynamics between party headquarters and local branches.
For states such as Japan, South Korea, Indonesia, Brazil and other nations recently adopting the system of directly elected local chief executives, gubernatorial (and mayoral) electoral campaigns are key elements of local party activity and autonomy. In these states, how the national party leadership chooses to control or delegate tasks related to these governors is an under-studied, but critical, variable in understanding party organization.
In Japan, local branches of both the Liberal Democratic Party (LDP) and the Democratic Party of Japan (DPJ) appear to exercise considerable discretion over which candidates to endorse and support for local chief executive elections. Attempts at top-down imposition of gubernatorial candidates often trigger resistance from local organizations. These local branches at times back their own preferred candidates instead of those officially endorsed by the party executive. It is not uncommon in Japan for party branches to endorse gubernatorial candidates who are also being endorsed by opposition parties at national level. At times, these local organizations may refuse to nominate and endorse a candidate from their own party despite instructions from their leadership to do otherwise.
The article attempts to explain such behaviour by exploring two questions: First, what divergences in structural incentives between party headquarters and local branches lead to conflicts over gubernatorial campaigns? Second, how does party leadership deal with such conflicts in terms of the discretion it provides and the sanctions it applies to local branches?
To facilitate analysis, the article borrows from the ‘delegation’ and ‘franchise’ party models (Carty, 2004; Van Houten, 2009), which seek to explain how party organizations adapt to multi-level electoral environments. After a discussion of the two models, the article provides hypotheses as to why a party leadership may provide discretion to local branches over local chief executive contests. It then explores the potential conflicts arising between the two as a result of structurally divergent incentives as predicted by theories of party behaviour under presidential systems. A brief discussion next explains the structural factors which shape the relative importance of gubernatorial campaign for a party organization.
Empirical evidence is provided by examining how Japan’s two major parties have dealt with gubernatorial campaigns in recent years. Select cases and data demonstrate how local units are driven by differing incentives from the national leadership, the consequent intra-party conflicts and the limited success of party headquarters in steering their local units. This is followed by a discussion of whether the two intra-party models – ‘franchise’ and ‘delegation’ – successfully capture the way in which the two Japanese parties have dealt with gubernatorial campaigns. To illustrate generalizability beyond Japan, the article provides a brief comparative counterpoint looking at intra-party conflicts in the selection of the Greater London Authority’s mayoral candidate in 2000 and 2004.
Delegation or franchise
Two recent articles (Carty, 2004; Van Houten, 2009) have proposed models describing how parties operate in multi-level systems focusing on the need of the national party leadership to allow local party branches to adapt to local conditions. One of the models stylizes the party at the centre as granting autonomy to local branches in the manner of a ‘delegative’ principal–agent relationship (Van Houten, 2009). The other model views different arms of the party organization forming mutually interdependent contracts in the form of a ‘franchise system’ (Carty, 2004: 10).
These two models differ in assumptions of where ultimate power lies within a party. The ‘delegation’ model is a top-down model in which the national party leadership is principal and local party branches are agents (though in some cases this relationship appears reversed, such as for Swiss and Canadian parties). The model assumes that party leadership retains ultimate authority to overrule decisions taken by its local branches on delegated matters (Van Houten, 2009: 148). The party leadership as principal possesses various control mechanisms which it may employ to keep its local branch agents in line.
The ‘franchise’ perspective, in contrast, does not assume that ultimate power lies in any branch of the party organization. Constructed as an analogy of business franchises, different organizational levels (headquarters or local units) in the party franchise can possess greater or lesser autonomy over particular activities. Party headquarters and each of its territorial units strike an ‘organizational bargain’ which allows each arm to act independently, mindful of the mutual interest of the party as a whole (Carty, 2004: 13–14). The model thus captures relations in party organizations that are mutually autonomous and share power broadly, yet are bound by a common party image and brand (Carty, 2004: 11–12). The central insight of the franchise model is that different arms of the party organization join together flexibly on an equal standing for mutual benefits, rather than party headquarters having ultimate authority to sanction or veto decisions taken by other arms, as with the delegation model.
Both models, however, recognize that the need to provide local autonomy stems from the modern party’s ‘stratarchic imperative’ (Carty, 2004: 21). This demands that local branches of state-wide parties become adequately ‘localized’, so that the party can maximize electoral success and legislative influence throughout varied electoral districts and across the different levels of government in which it competes. Party headquarters must therefore provide discretion to local branches for a range of party activities including: candidate selection, campaign strategies for local and national elections in the district, as well as local policy formation. In states such as Japan with directly elected local chief executives, contesting and securing high-profile, influential gubernatorial (and mayoral) posts is potentially a vital concern for both national and local party organizations. As such it is worth examining this particular local party process in terms of the two intra-party models.
Sources of intra-party conflict over gubernatorial campaigns
How much discretion?
Faced with a gubernatorial campaign, the party at the centre faces a ‘delegative’ choice: it could either provide discretion over these decisions to local branches or it could seek to control the process. Both options could lead to intra-party conflict.
In the first option, the national party leadership could allow local units a free hand on various aspects of the gubernatorial campaign. There are clear incentives to permit such local discretion. First, the party branch may be more capable of discovering attractive candidates and campaigning for them because it is more knowledgeable of local preferences and issues. Second, the party branch may be more motivated to campaign for a candidate it has selected, rather than a candidate parachuted from the centre. The locally chosen candidate may therefore have a better chance of electoral success than a centrally imposed one as it will secure greater support from the party on the ground (local politicians, members and supporters).
Yet there are clear risks for such a permissive arrangement. A central concern is that the candidate (and/or campaign strategy) selected by the local party branch is not the one preferred by the national party leadership. The party leadership may be opposed to a locally selected candidate because she may be seen to diverge from the party in terms of its ideology, policy platform or image. Or it may simply be a candidate deemed unlikely to win. There is also an additional risk that the local party branch may not back a candidate at all, and choose to sit out the gubernatorial/mayoral contest altogether.
To avoid such risks, the party at the centre may attempt to control the process entirely with little input or support from its local branches. This would mean parachuting candidates into the locality and conducting the electoral campaign from party headquarters. In this case, the gubernatorial candidate and electoral campaign would be in full accord with the preferences of the party leadership. Naturally, there are considerable risks arising from such top-down gubernatorial campaigns. The local party branch may be reluctant to accept candidates imposed from the centre, perhaps as the local branch already has a preferred candidate or local campaigning strategy. The local branch may perceive of the imposed candidate as less likely to win than their own candidate. Local branches and their members may respond to imposed candidates by supporting another candidate, sabotaging the campaign or even defecting to another party. Likewise, local voters may view candidates parachuted in from the centre as neither legitimate nor suitable and may withdraw their support for the party at local, and perhaps even national, level.
‘Local presidents’ and parties
Why, after being granted discretion, would local units select a candidate and pursue a gubernatorial campaign strategy at odds with the national party leadership? After all, aren’t both arms of the party largely united by ideology, policy goals and the desire of expanding party support and ultimately capturing votes and offices locally as well as nationally? Closer inspection reveals potential structural sources of conflict between party headquarters and local branches when faced with gubernatorial/mayoral campaigns.
For the national party leadership, contesting local elections (particularly high-profile local chief executive elections) is a way of expanding and maintaining the party base by mobilizing existing supporters and recruiting new ones during the campaign. These expanded electoral forces could be used for securing votes for the party in other elections – especially national level elections. Victories in local chief executive elections could generate positive press coverage and ‘momentum’ for national level elections. Party-backed chief executives could potentially campaign and support candidates for national level elections in their region in the future. Securing partisan control of local executive posts could also be instrumental in national parties seeking to pursue national projects in regions that require the agreement of the local chief executive. Local party legislators back local chief executive candidates for similar benefits: future support from the chief executive in their own legislative elections and local policy goals. Yet, in both these areas, local legislators have a much greater stake in the outcome of their gubernatorial contest than the national party leadership. 1
The literature on party behaviour under presidential systems highlights this dynamic. In such a separation of power systems parties must seek to capture the executive branch, because the president controls (to differing degrees, depending on the presidential system) access to things which the party seeks – influence over policy and distribution of offices. For parties in presidential systems, ‘winning control of the executive branch directly offers the only guaranteed path to office and policy pay-offs’ (Samuels and Shugart, 2010: 38). The president’s ability to deliver on the party’s policy or budget preferences also affects the re/election chances of individual party legislators. Hence parties in presidential systems act and organize themselves primarily around the goal of controlling the executive branch.
In local governments with local presidents, particularly strong ones, local party legislators will be primarily concerned with their relationship to the chief executive. Ideally, local partisan legislators would prefer to control the executive branch directly by having their own partisan candidate in the post. At the least, they will seek to ensure the electoral victory of a chief executive sympathetic to their policy preferences. Such a desire to maintain access to executive power in the local presidential system may over-ride partisan affiliations and considerations.
This is where the preferences of national and local branches towards their local chief executive candidate could crucially diverge. In order to secure policy access and influence the executive post-election, local legislators may choose to bandwagon with other parties rather than fight a partisan contest and fall into the opposition. They may opt to sit out a partisan contest perceived as a tough contest in the hope of not being treated as opposition post-election. Local party organizations have less appetite to back a partisan candidate and risk falling into opposition locally than the national party leadership. The costs of losing a partisan contest are far greater for the local branch than for the national party leadership.
Further complicating this equation, presidential campaigns tend to be supra-partisan. As Linz (1994) has pointed out, presidential candidates often tend to be ‘outsiders’ with few links and little experience in party organization. They often appeal to the electorate on personal qualities, even campaigning against parties (Linz, 1994: 26–29). Assuming that chief executive candidates – unless they are strongly ideological partisans – are motivated primarily by re/election, they will seek to appeal not only to their partisan supporters, but to a broader swath of the public.
The attractiveness of such ‘outsider’ candidates in presidential contests makes it difficult for parties, both locally and nationally, to conduct partisan gubernatorial campaigns. If the national party leadership insists on pursuing partisan contests locally, the local branch may be faced with a difficult choice of backing a partisan gubernatorial candidate against a more attractive non- or supra-partisan candidate. Moreover, such ‘outsider governors’ are potentially less faithful to the party and could turn against their partisan supporters post-election, as they are elected separately from the party in legislature and are not dependent on its confidence for survival. Legislative parties constantly face the risks that their presidents (governors) act beyond their control (Samuels and Shugart, 2010: 47).
Structures surrounding ‘local presidents’
Parties, at national or at local level, might not always perceive gubernatorial/mayoral elections as being critical contests in which they must engage. Depending on the particular configuration of institutional factors in a political system, the contest for governors and mayors may attract either little or considerable attention and involvement from different branches of the party.
The significance of such gubernatorial/mayoral elections is shaped primarily by how much power such figures have in local and national decision-making. This is a function of how centralized/decentralized the state is and how much power the governor or mayor has relative to the local legislature. The more decentralized a state and more powerful a governor or mayor in terms of its executive power, the higher the stakes of such contests for the party at all levels.
Secondly, public perception of such gubernatorial contests will shape how parties value these elections. If local contests are viewed as referendums of the performance of national parties (as so-called ‘second-order’ elections), national party leaderships will likely be concerned with its outcomes. This is assumed to be particularly so if such local elections occur close in time to a national-level election, as local election results could affect public perception of party strengths and impact their future voting behaviour. Whether the elections are concurrent with national ones may also affect how much effort national parties would commit to gubernatorial campaigns.
In sum, contests for the post of relatively weak local chief executive in highly centralized states may be considered indifferently by national party leaderships. On the other hand, high-profile governors and mayors with wide discretion are likely to attract considerable party interest leading to potentially serious conflicts of interest between the national and local party organization.
Empirical exploration
Japan’s sub-national politics
Japan in the post-war period has generally been characterized as fiscally and politically centralized, yet administratively decentralized (Scheiner, 2006). Despite such centralization, chief executives have exercised considerable discretion over the implementation of national projects within their locality. These include the construction of major public works projects such as nationally funded dams or roads and various sensitive issues such as the stationing of US military bases or nuclear power plants. In these instances, national parties in government become engaged in campaigns to ensure a gubernatorial candidate sympathetic to its project wins in these particular regions.
Furthermore, gubernatorial elections are often spun by the media and perceived by party members, including the national leadership, as ‘second-order’ elections with the potential to impact the ‘momentum’ of its national contests. 2 The 2009 gubernatorial election for Shizuoka, two months prior to a Lower House election, was described by LDP leadership as a crucial ‘preliminary skirmish’ that was to form a ‘breakwater’ to block mounting popular opposition against the party nationally. 3 Kataoka (1994: 174) has observed party leaderships engage closely in contests seen as vital for national party fortunes or image, such as high-profile gubernatorial elections in major cities like Tokyo, Kyoto and Osaka.
For the local party branch and its legislators, local chief executive contests are arguably even more vital for their political fortunes. First, Japan’s local chief executive has dominant control over budget and policy formation as well as the ability to veto and dissolve the assembly. This gives them a central role in agenda-setting and local decision-making (Soga and Machidori, 2007: 47–48). Secondly, local assembly members are elected in multiple-member or at-large districts under a single non-transferrable vote system (MMD SNTV). This means that assembly members can often be elected representing only a specific area or support group within their multi-member district. This generates an incentive for local assembly members to deliver private goods to their specific supporters in their constituencies (Soga and Machidori, 2007: 46–47). Local assembly members thus seek to be in the legislative majority and maintain supportive ties with the governor who controls the flow of such resources within the locality.
These structural considerations in Japan illustrate how gubernatorial campaigns are important for the party organization’s national and local branches. The article now turns to examine how Japan’s two largest parties have dealt with these contests.
Liberal Democratic Party
The LDP’s candidate selection (both legislative and chief executive) for local elections has been described as being conducted ‘bottom-up’ from lower to higher tiers of the party organization (Kataoka, 1994; Kawamura, 2008: 10). Local party executives have been responsible for selecting and nominating candidates and forwarding their decision to the national party leadership, which formally ratifies the choice (Kataoka, 1994: 197–216).
Such ratification was in most cases a formality, with few instances of party headquarters rejecting local nominations. According to Tsuji (2010), a pattern of agreement between party centre and local branches resulted in relatively conflict-free selection of gubernatorial candidates during the 1960s to 1990s for the LDP. In the whole post-war period, Tsuji has observed only four elections (out of 626 elections) in which the local LDP branch endorsed a gubernatorial candidate separate from the national leadership.
In only a handful of cases has the LDP leadership rejected a candidate put forward by the local branch. Such instances have occurred for the most high-profile gubernatorial elections – such as those for governor of Tokyo or Osaka. In three cases (Nara 1963, Tokyo 1971 and 1991) observed in the post-war period, the ‘parachuted’ candidate lost as a result of local branches refusing to cooperate. In the case of Tokyo 1991, the secretary general of the national LDP resigned to take responsibility for the loss of the candidate he had backed against the wishes of the Tokyo metropolitan party branch (Kataoka, 1994: 214–215).
In the gubernatorial campaign in Osaka for 2000, The LDP leadership ‘parachuted’ an official candidate – a former career bureaucrat named Ota Fusae – into the district. 4 The prefectural branch of the LDP refused to accept the centrally selected candidate and recruited their own, backing him throughout the election. The prefectural branch further demanded that local LDP MPs support their alternative candidate or else lose electoral support in upcoming national level elections. 5 Though such a threat to withhold campaign support did not materialize, tensions between the local LDP legislators and Ota persisted even after her successful election. The chairman of the Osaka prefectural branch – a local MP – resigned for being unable to maintain unity in the ranks of the local branch. 6
The LDP leadership, however, has occasionally intervened when internal rifts within the local prefectural branch required headquarters to act as arbitrator in the gubernatorial candidate selection (Kataoka, 1994: 211–213). Such internal rifts are problematic for the national party leadership on two accounts: first, multiple candidates from the same party may result in all of them losing because the party’s vote-share has been divided. Secondly, such splits over gubernatorial campaigns could reduce the unity of the local branch in mobilizing for national elections in the future.
In some instances, the national leadership has opted not to nominate any official candidate when faced with differing groups in a local branch fighting over the choice of a gubernatorial candidate. In these cases, the party leadership permitted multiple candidates to compete as independents, presumably in order not to favour any one faction of the branch over another. Tsuji (2010) notes that the LDP central leadership appears increasingly beset by such split elections, permitting (or being unable to control) multiple candidates to run for governorship since the 1990s. This observation parallels Reed’s (2009) work, which demonstrates the LDP leadership being incapable of controlling the number of parliamentary candidates running in multi-member districts prior to the introduction of single-member districts in 1995.
Democratic Party of Japan
Like the LDP, the DPJ leadership in its early years adopted a similar permissive approach to local chief executive elections. From its inception in 1998, the DPJ leadership did not provide clear instructions to local party organizations in how or whether to conduct gubernatorial campaigns.
Through this period, DPJ local branches often endorsed incumbents or newcomers who were also backed by the LDP, 7 their main opposition at national level or simply sat out elections altogether. 8 As it sought to expand its electoral strength across Japan, the DPJ leadership likely felt that allowing local branches to bandwagon with the opposition or simply sit out gubernatorial contests was problematic.
In May 2006, the central party leadership announced that local branches had to endorse a party candidate not backed by the opposition party (LDP) for local chief executive elections of all prefectures and the largest cities. Ozawa Ichiro, the DPJ leader at the time and architect of the policy, explained why the party wished to contest local elections: ‘The DPJ will lose its reason for existing if it supports the same candidate as the LDP [in local elections]. The process of finding and endorsing a candidate serves as a large force in strengthening the party’s foundation.’ 9 Moreover, supporters and media had become critical of occasions when the DPJ failed to contest local gubernatorial elections. An editorial – ‘Why does the DPJ avoid a fight?’ – criticized the party for not putting up gubernatorial candidates and argued that it was failing to serve as an opposition party. 10
Despite these pressures, the DPJ’s record in putting forward their own party candidates to compete against a candidate backed by the main opposition has been mixed. 11 The DPJ has managed to stand their own candidates against a LDP candidate in roughly a third of all major local chief executive elections, four years since it announced its policy to do so. Though an improvement from the prior period, these results illustrate the difficulty party leadership faces in steering local branches to act as instructed.
What could explain this failure of the DPJ to stand candidates in so many gubernatorial elections? Put bluntly, the party’s long-standing weakness in the regions has made local branches hesitant to contest partisan elections. Though the DPJ has made gains at national level, ultimately taking government in 2009 for the first time since its formation, their presence in local legislatures has been thin. During the period 2006 to 2009, the DPJ controlled only 15 percent of prefectural assembly seats compared to LDP’s 48 percent.
With such a shortage of local assembly members, the DPJ does not have a network of local politicians to mobilize votes for their own gubernatorial candidate. Nor can the DPJ provide support post-election in the legislative arena even if their candidate were to win. Hence in regions where the DPJ has been weak, the local party organization has generally been reluctant to back their own partisan candidate. Attractive candidates, in turn, have also been unwilling to stand as a candidate on the DPJ label in such regions, knowing they may receive limited support in electoral campaigns as well as within the legislature.
The Wakayama gubernatorial election of 2006 is a case in point. The DPJ could expect little local traction and support. The party had no MPs representing the prefecture, only two of 46 prefectural assembly seats, and very few party members in the region. The national party leadership had been pushing for a local DPJ municipal politician to stand for the election. This met with resistance from local assembly members as well as unions in the prefecture that announced their support of the LDP-backed incumbent. 12 In such an environment, the Wakayama local organization abandoned plans to establish its own candidate and instructed local party members to vote freely. The chairman of the DPJ prefectural branch – a local MP – resigned because of having failed to stand a partisan candidate.
Similarly in Kagawa, a prefecture with equally weak DPJ presence, the local party organization abandoned plans for endorsing its own candidate in the 2006 gubernatorial elections. At first, the prefectural DPJ branch had endorsed the incumbent (also endorsed by the LDP), but the national party leadership rejected this decision. Ultimately, the local branch failed to endorse any candidate. As the next gubernatorial elections approached, a DPJ prefectural assembly member explained that there was little point in pushing for a DPJ candidate in this prefecture as they were a minority in the assembly and would not be able to provide enough support to get their preferred candidate’s policies passed even if he/she were to win the election. 13 In neither election were local branches reportedly punished for their failure to stand a partisan candidate.
The DPJ, however, has succeeded in endorsing its own gubernatorial candidates against opposition rivals in about a quarter of all gubernatorial elections between 2006 and 2010. Revealingly, the local branches in these cases were well established and had greater reason to believe that their candidate would win. In 11 out of 12 gubernatorial contests between 2006 and 2010 in which the DPJ local branch stood their own partisan candidate, the party had won a larger share of proportional representation votes than the opposition LDP for the most recent upper house elections. 14
However, there have been prefectures (for example, Mie or Nara in 2007) in which the DPJ have gained a considerably larger share of PR votes than the LDP, but still failed to back a partisan candidate in the subsequent gubernatorial contests. Strong party support in national elections is therefore a necessary, but not sufficient, condition for the DPJ prefectural branch to contest gubernatorial elections. Furthermore, despite sitting out in the campaign nominally (in Nara, Mie, Wakayama, Kagawa and elsewhere), local DPJ legislators have supported the successful governors post-election, effectively joining the legislative majority and securing access to policy influence.
Supra-partisan candidates
Japanese parties face another challenge in engaging in gubernatorial campaigns: attractive candidates tend to avoid partisan labels and could potentially betray the party post-election. The dilemma and selection risks facing Japanese parties engaging in local presidential campaigns is not unique. As Samuels and Shugart argue: ‘… the skills most useful for winning a presidential election include proven vote-drawing ability and an appealing, supra-partisan public image. These skills may only weakly correlate with the skills that make one a faithful executor of the party’s will’ (2010: 63).
As this presidential theory predicts, candidates have shown a tendency to avoid strong affiliation with one party, preferring to run as ‘independents’ or supra-partisan representatives of the whole prefecture. These candidates often call themselves members of a ‘prefectural party’ (kenminto), representing themselves as supra-partisan (chotoha) candidates. They either seek endorsements from a pan-party coalition (usually excepting the Japan Communist Party) or refuse official endorsements from parties.
This supra-partisan predilection of candidates generates difficulties for party organizations. To illustrate: in Shiga prefecture, Kada Yukiko – a citizen activist and professor with no affiliations to the national parties – sought the endorsement of all major parties in her bid for governor in 2006. Both DPJ and LDP refused endorsement and jointly backed the incumbent. Kada defeated the incumbent in a landslide after which the DPJ supported and LDP blocked her in the legislature. In her re-election in 2010, she sought to maintain ‘equidistance’ to all the parties and spurned the DPJ offer to become their official candidate. Kada asked only for ‘support’ (shiji) from the prefectural DPJ, rather than ‘endorsement’ (suisen), a strategy aimed at avoiding being too coloured by partisan affiliation. 15
The main opposition in the Shiga prefectural legislature, the LDP, were split internally about whether to stand a candidate against the highly popular Kada. Ultimately, the LDP local prefectural branch chairman – a party insider and former Lower House Diet member – announced candidacy. He at first resigned from the party to enter into the campaign officially as an ‘independent’. Though fully backed by local LDP politicians and party organization, this candidate claimed to be non-partisan. 16 The LDP ‘insider’ lost against the popular ‘outsider’ DPJ candidate.
Such attractive ‘outsider’ candidates also pose a post-election risk to the party as they cannot be de-selected if they act against its interests. To illustrate, Osaka’s governor Hashimoto Toru – a celebrity lawyer and party outsider – received LDP support in his bid for governor in 2008. One year later he formed his own local party in direct competition with the LDP, halving it through actively encouraging defections from its local branch to his party. Neither local nor national leaders of the party were punished for having selected and nurtured a governor who turned against the LDP.
For the party leadership, these local candidates’ unwillingness to fully accept partisan affiliation is problematic. The national leadership wishes to increase its exposure and profile in regions where it is weak and the gubernatorial election is an opportunity to do so, but in these areas candidates are particularly inclined to refuse to be strongly affiliated with the party. For the local party organization, a candidate unwilling to fully commit to the party may diverge from their policy preferences once in office. Yet insisting on outright party affiliation may drive attractive candidates away. When national party leadership insists on local partisan contests, as with the DPJ, local branches face a hard choice: to back an unpopular partisan ‘insider’ against a potentially unfaithful but attractive supra-partisan ‘outsider’ candidate. As the DPJ record shows, local units have generally preferred to sit out elections rather than contest such tough contests.
Failed delegation and loose franchise
Why have the DPJ and the LDP leadership apparently been so ineffective in controlling their local branches, as evidenced in the above examples? Both the LDP and DPJ internal statutes give the party leadership at the centre formal powers to overrule local branch decisions on gubernatorial campaigns. From a delegation model perspective, local units are nominally ‘delegated’ the discretion to act in local campaigns. Though the party leadership does not spell out the schedule of sanctions it would apply if local units chose to act against its perceived interests, theoretically, the party as principal could take punitive actions against its local agents.
Delegation theory posits various mechanisms to keep agents in line: ex ante constraints and incentive structures as well as post ante sanctions to induce or threaten agents to act in the principal’s interest. In terms of intra-party relations, these could include the expulsion of local branch leaders and members, withdrawing financial support to local units, and furthering or hindering careers of local party branch members (Van Houten, 2009: 147). Of these, hardly any have been observed in our cases. None of the local politicians were reported as having been refused party endorsement or evicted from the party for supporting their own preferred gubernatorial candidate.
Local politicians who had refused to support nationally ‘parachuted’ candidates, as in the cases of Tokyo and Osaka above, were not punished for backing their own preferred candidate. Instead, sanctions by the party leadership have usually been taken against the MP in the local district who had not managed to keep local branches and politicians in line or failed to back a successful candidate. 17 Such sanctions usually involve the local MP nominally in charge of the prefectural branch resigning from his/her branch chairmanship. These parliamentarians have continued their careers unabated and been given party nominations in subsequent national elections.
Such reluctance to coerce local branches reflects the costliness and ineffectiveness of such sanctions, at least in the case of gubernatorial elections. The greatest cost of such top-down coercion is the loss of electoral support from local branches in national-level elections. It is well documented that Japanese parliamentarians, particularly of the LDP, have depended heavily on a network of prefectural and municipal politicians for their electoral campaigns (Inoue, 1992; Park, 2000). Fiscally centralized clientelist relations between the centre and regions have encouraged the formation of a network of local politicians campaigning on behalf of parliamentarians in their respective districts (Scheiner, 2006). The national party is loathe to punish or alienate these ‘arms and legs’ of their own members’ parliamentary campaigns.
The stringency of Japan’s electoral campaigning rules has also meant that national-level candidates have to depend heavily on local party organizations for their campaigns. The short campaigning period and various limitations on, for example, the use of broadcast or Internet media has meant that candidates have had to rely on ground-level campaigns heavily dependent on local party organization. Local party organizations hence possess a powerful resource on which national party candidates (and leadership) depend considerably. Considering the rarity of top-down candidate selection, threats (overt or otherwise) to cut off this support may have been effective deterrents to unilateral action by party leadership.
Researchers have elsewhere found evidence that the LDP have punished districts (and thereby indirectly punished local politicians) by reducing national subsidies to localities that did not sufficiently support the party in national elections as well as de-selecting local politicians who failed to mobilize sufficient voters in their district during national elections (Scheiner, 2006: 72–74). There seems little merit, however, for the LDP (or DPJ) leadership to reduce subsidies to localities in which prefectural branches failed to unite under a gubernatorial candidate or refused to back a ‘parachuted’ candidate. Such measures would backfire as they would punish local voters not responsible for actions taken by local branches and potentially reduce party support in future national elections.
This is not to say that the national leadership of Japanese parties is unwilling to act to control local units in other areas and situations, or that both LDP and DPJ organizations are decentralized and bottom-up in all of the parties’ key activities. In recent years, the LDP leadership has taken a more top-down approach to selecting national-level candidates. In the 2005 lower house election, the party executive rejected rebellious veteran candidates backed by local branches and parachuted in hand-picked, more loyal, candidates (Asano, 2006). A number of prefectural legislators (e.g. those in Gifu and Hiroshima prefecture) who backed parliamentary candidates against official LDP candidates in this election have been punished by eviction from the LDP.
Yet, in the case of gubernatorial campaigns, national party leaderships appear generally to accept the primacy of local branches’ decisions. The dynamics, however, between the two Japanese parties differ. In the case of the LDP, the discretion provided to local branches appears to be one of looser ‘franchise’, whereas for the DPJ it appears to be one of failed ‘delegation’. In attempting to force local units to contest partisan campaigns, the DPJ headquarters was neither able nor willing to control its local branches which had diverging preferences from the party at centre. The DPJ leadership in this particular area could not act as an effective principal, though its instructions to local branches suggest it perceived itself to be so.
In contrast, the preferences of LDP headquarters and local branches generally converged over gubernatorial campaigns, and headquarters provided full discretion to local branches in this activity. This dynamic could be interpreted as successful delegation: preferences for gubernatorial contests within the LDP were aligned, so headquarters could delegate completely without needing to steer its local branches with various control mechanisms.
Yet certain features of LDP’s dynamics over gubernatorial campaigns suggest that the LDP organization may better be described as a franchise model. Unlike the DPJ, LDP headquarters never instructed its branches to contest gubernatorial contests in a particular manner. This reluctance to pursue top-down control is revealing, especially since the 1990s when many LDP local branches were beset by factional splits over gubernatorial candidates (Tsuji, 2010), which could potentially be damaging to the party at the centre. In rare cases where LDP headquarters interfered in gubernatorial campaigns in specific prefectures, local branches have openly and vigorously resisted such imposition. In these cases, the branches appear to behave less as subordinate agents and more as equal ‘franchise’ outlets resisting attempts by the head office to overstep and interfere in local operations.
What, then, accounts for why divergent party models seem to fit the LDP and DPJ organizations? 18 For most of its years as the ruling party in a one-party dominant system, the LDP has been a decentralized clientelist organization: local politicians affiliated with local LDP Diet members to gain access to pork from the centre in exchange for mobilizing votes at national elections. So long as this bargain was kept, the party at the centre and local MPs did not interfere with local decisions, including the choice of gubernatorial candidates. A top-down approach would have been counter-productive, as both arms of the party mutually benefited from the common LDP brand as provider of pork to certain clients.
The DPJ, however, which spent most of its years in opposition and unable to access levers to pork, positioned itself as a programmatic party. Unable to gain support in regions through clientelist promises like the ruling party, the DPJ leadership needed to push local units to contest local elections, including gubernatorial campaigns. As a programmatic party, the DPJ also had greater reason to seek partisan contests and ensure unity of its partisan image in local contests. The party centre, seeing itself as a principal in the relationship, instructed its local ‘agents’ on how to contest gubernatorial elections. Yet lacking effective sanction mechanisms – or rather unwilling to use them for fear of driving members to exit the party – the DPJ has been unable to steer local units. Hence the DPJ headquarters’ attempt to ‘delegate’ instructions to local agents in how to contest gubernatorial campaigns failed.
Intra-party conflicts over the Mayor of London
How do party organizations in other countries with differing institutional environments, political cultures and party systems engage in elections of local chief executives? As a counterpoint to the Japanese cases, the article looks briefly at the high-profile Greater London Authority elections of 2000 and 2004. In these elections, the selection of mayoral candidates generated intra-party tensions and emergence of a popular ‘outsider’ candidate similar to those discussed in Japan.
For the first mayoral election of the GLA in 2000, the Labour Party leadership expressed antipathy to the candidacy of Ken Livingstone, and pushed for their preferred candidate, Frank Dobson. A controversial electoral college which favoured the votes of the party at the centre (MPs and MEPs and union leaders) over rank-and-file members was designed to select the candidate, with the expectation that it would favour Dobson. Dobson won this primary with a majority of votes, while Livingstone lost despite securing the majority of votes from party members in London. 19 Livingstone refused to accept the results and ran as an independent, which led the Labour leadership to oust him from the party. Livingstone went on to win the election by a considerable landslide, while Dobson finished a weak third after the Conservative candidate. As Van Biezen and Hopkin argue, this illustrates ‘party dissidents’ success in appealing to the grassroots supporters and non-party members in order to overturn the decisions of the formal party organization’ (2006: 28).
For the London mayoral elections of 2004, Labour’s original candidate – Nicky Gavron – was trailing in the polls and the party risked losing again to Livingstone, who intended to run as an independent. As elections approached, Gavron stepped down to act as deputy for Livingstone in a united campaign for the mayoralty. London Parliamentary Labour members, unions and MPs representing London pushed for Livingstone’s re-admission to the party against the opposition of party leaders such as the then Chancellor Gordon Brown and Deputy Prime Minister John Prescott. 20 Labour Party leadership eventually did re-admit Livingstone to the party and nominated him as their official candidate.
Despite attempts at top-down control, the Labour leadership failed to successfully back its candidate against local preferences. This is surprising considering Britain has a highly centralized party system, disciplined and entrenched local party politics, and the selection of national and local candidates tends to be closely controlled by the party at the centre (Webb, 1994). Yet devolutionary trends and expectations that candidates for local positions should be selected bottom-up prevailed. As one study describes it, ‘Ken Livingstone’s victory as an Independent candidate was one of the most substantial blows struck against the conventional party system in post-war British politics’ (Curtice et al., 2001: 33).
Conclusion
The conflicts over London mayoralty elections echo the dilemmas faced by Japanese parties in contesting gubernatorial campaigns. Top-down selection led to local resistance, splits in the local party and negative press for the party. At the same time, the party leadership felt it could not fully delegate this selection process to local organizations as such elections were critical for the image and fortunes of the party as a whole.
In Japan and Britain, as in other systems with high-profile mayors and governors, elections to these local posts are significant events for national and local party organizations. Yet both central party leadership and local party branches have structurally diverging priorities and preferences in terms of whether and what sort of candidate to nominate for these chief executive elections. Adding to the difficulties, as observed in London and some of the Japanese cases, candidates for high-profile ‘local presidencies’ have attracted ‘outsiders’ to the party who are often not equally attractive to local and national arms of a party, and hence become another source of intra-party conflict.
The party’s national leadership could seek to avoid these situations by asserting control over local branch decisions. Yet such top-down actions, when attempted on rare occasions, have tended to backfire: local branches withhold electoral support for the parachuted candidate or threaten to boycott support in national level elections. Both Japanese parties appeared to have few sanctions at its disposal to bring recalcitrant local branches to heel. In the cases observed, national party headquarters did not punish local branches or politicians directly for not heeding its demands – instead MPs from the localities were nominally punished through resignation from prefectural party branch executive positions.
Instead of top-down control, national party leadership has granted wide discretion in decisions relating to the local chief executive to local party branches. Such an approach has not been without risks and costs for the national party leadership. Local branches may be hesitant to contest local elections, preferring to bandwagon around attractive ‘outsider’ candidates not fully committed to any party. This means a party which hopes to expand its foothold in regions through gubernatorial elections will not be able to do so easily in districts where it is weak.
The key insight of the article – that local presidentialism disrupts intra-party relations due to differing preferences of national headquarters and local branches – should be applicable to all states with powerful, high-profile governors and mayors. London’s mayoral elections illustrate this generalizability. The UK party system is highly centralized, party labels are very strong, and party competition is programmatic – all in contrast to Japan – and yet local presidentialism threw a surprising spanner in the works of regular top-down party control. The growing prominence of mayors in Europe (Magre and Bertana, 2007) – especially in prominent high-profile cities – is likely to have similar impact on intra-party relations in these regions.
Further research in relations between party organization and local chief executives remains. Comparative questions to explore may include: when does party headquarters effectively control local branches in terms of gubernatorial campaigns? Does the existence of local presidents contribute to intra-party conflicts over policy preferences? Does it lead to the rise of local parties, eventually generating incongruence between local and national level party systems?
The article, however, has demonstrated that the process of selecting candidates for the posts of governor and mayor – even the decision to contest them – generates numerous intra-party conflicts. Researchers of states with directly elected local chief executives would do well to consider this aspect as a major factor shaping the dynamics of party organization.
Footnotes
Funding
Research for this article was conducted during a Postdoctoral Fellowship for Foreign Researchers provided by the Japan Society for the Promotion of Sciences (JSPS) at Osaka City University between 2010 and 2011.