Mainstream and deviating ideologies in Japanese gubernatorial elections

Abstract

The ideological conflicts of Japan's subnational politics have tended to be interpreted as either being largely muted or contained within national dimensions. Following two decades of substantial decentralization and growing local autonomy, however, a diversity of new ideological responses to local issues have appeared. These include neo-liberal parties and executives in wealthier regions such as Tokyo and Osaka or a rising regionalist identity politics such as that found in Okinawa. Nativist right and populist left along with single-issue parties are also now fielding candidates for subnational elections. Despite this increasingly crowded field, there is still no systematic understanding of the divergent ideological worldviews and dimensions of conflict operating at the subnational level. Nor do we know how these worldviews “deviate” from the traditional “norm” of a progressive vs. conservative conflict dimension assumed to characterize Japanese subnational politics. This paper begins to fill this gap by investigating the campaign discourse of gubernatorial candidates both before and after the pandemic outbreak. We find that the language, and underlying ideological orientation, of these candidates can be separated into four clusters: “mainstream”, “old left”, “neo-liberal”, and “fringe”. In addition, “regionalist” and “new left” populism can also be identified in select elections.

Introduction

Traditionally, Japan's subnational politics have largely been understood within the national conservative–progressive (hoshu–kakushin) cleavage. Governors and mayors were, and still often continue to be, classified as either conservative or progressive primarily based on their national party affiliation. Conservatives (those backed by the Liberal Democratic Party [LDP]) are understood to seek “development” in terms of investments in infrastructure and support for business growth, whereas progressives (those backed by socialists and communists as well as various centre–left parties) seek expanded “welfare” along with stronger regulation of business and development (Kabashima and Takenaka 1996, p. 124; Tsuji 2016, pp. 32–34). More recently, since the realignment at the national level and emergence of neo-liberal policies and actors in the 1990s, a new category has been added, the reformist (kaikakuha) governors and mayors (Soga and Machidori 2007, p. 271; Tsuji 2016, p. 35). In contrast to these two traditional groupings which seek to expand, or are at least indifferent to increased, public spending, these reformists prioritize small government and fiscal health.

Looking at recent subnational elections during the Covid-19 pandemic, there are clearly more than just such progressives, conservatives, or neo-liberal positions among candidates. As we demonstrate below, these included numerous candidates calling for expansionary policies supporting businesses and individuals, anti-vax and anti-mask fringe candidates, candidates blaming central government, foreigners, media, and experts, while other candidates demanded more local autonomy.

This paper analyses the language of these diverse candidates and what they reveal about their ideologies and how they “deviate” from the traditional “norm” of progressive vs. conservative vs. neo-liberal ideologies assumed to characterize Japanese subnational politics. Using a unique data set of candidate manifesto data (senkyo kōhō) from gubernatorial candidates, we provide both quantitative and qualitative descriptions of the nature of ideological competition on offer for gubernatorial campaigns. We find that the language, and underlying ideological orientation, of these candidates can be separated into four clusters: “mainstream”, “old left”, “neo-liberal”, and “fringe”. In addition, “regionalist” and “new left” populism can also be identified in select elections. The research is primarily descriptive rather than one seeking to demonstrate any causal argument or answer to a specific research puzzle. The goal is to begin to fill an empirical gap in our understanding of the subnational politics in Japan.

The rest of the paper is organized as follows. We review the literature on ideological competition in Japan's prefectural government. To test this received wisdom, we provide data from gubernatorial campaign discourse to show the kind of language and policy issues characteristic to mainstream candidates (those that win more than 5% of vote share) as well as fringe candidates (those that win less than that threshold). Analysing the text of candidate manifestos and speeches, we try to map the characteristic languages of different partisan candidates. We also determine the frequency by which different types of subnational candidates use words indicative of ideological positions. Finally, we provide qualitative interpretation with excerpts from manifestos (and in some cases campaign speeches) which illustrate diverging ideological responses to the pandemic in Japan. A final discussion will wrap up our findings and link them to broader developments in ideological and partisan conflict nationally and the lack of competition and dynamism in Japanese democracy at both the national and subnational levels of government.

Literature review

Conventional wisdom assumes subnational politics to be largely non-ideological, or at least less ideological than at the national level. It is expected to be de-ideologized since, compared to nation-states, smaller jurisdictions of subnational government tend to be more homogeneous, with residents of similar socio-economic and partisan preferences tending to self-sort into the same communities, especially locally (Tiebout 1956). Moreover, subnational governments in centralized systems tend to lack the authority to raise local taxes or administrative powers to deliver redistributive policies. In most cases, subnational governments do not control borders and are thus unable to prevent the exit of capital and residents dissatisfied with redistributive policies (Peterson 1981), thereby limiting a key distributive dimension of conflict. Since local governments lack authority over substantial matters, redistributive or regulatory, they are seen as primarily responsible for what is assumed to be uncontroversial (and hence non-ideological) public service provision such as garbage collection and infrastructure maintenance. Complementing these structural constraints is a normative view commonly held by politicians, voters, and scholars that subnational politics should be free of the undesirable influence of national-level partisan competition and their accompanying ideological conflicts.

Despite these structural and normative arguments that subnational politics will and should be non-ideological, there is a growing body of work that measures, compares, and identifies the impact of ideologies in subnational elections and governance. For example, we know that the ideological differences of parties which control subnational governments result in diverging local policy outcomes such as the level and content of public expenditure (e.g. Soga and Machidori 2007). Some research has identified uniquely local dimensions of conflict in subnational elections, including divides over regionalism, greater autonomy, and whether a party addresses national issues in subnational elections (e.g. Alonso 2012; Gross and Jankowski 2020). There have also been attempts to measure the degrees of populism, regionalism, and other classes of major ideologies in the discourse used by subnational parties (e.g. Heinisch et al. 2019). Others have explored how parties and partisan candidates alter their ideological content depending on the arena, whether it is in the national or subnational level, and depending on the territory in which the election takes place (e.g. Heinisch et al. 2019; Massetti and Schakel 2016). These studies largely utilize text analyses of various public discourses generated by subnational party branches and politicians – including local manifestos, campaign speeches, party pamphlets and magazines, etc.

Turning to Japan, a few recent studies have sought to describe the nature and estimate the ideological positions of subnational politicians. These include qualitative assessments of the ideologies of individual high-profile subnational politicians, their local political movements, and their voters – e.g. Osaka mayor Toru Hashimoto, Nagoya mayor Takashi Kawamura, and Tokyo governor Koike Yuriko and populism (e.g. Hieda et al. 2021; Yoshida 2020) or Okinawa governor Takeshi Onaga and regionalism. Quantitative studies have estimated the ideological positions of subnational politicians through textual analysis of candidate manifestos and assembly deliberations (e.g. Hakiai 2020) and Twitter account followers (Miwa 2017). These earlier quantitative studies estimate the positions of subnational politicians on an a priori dimension, mainly the traditional left–right (conservative–progressive) dimension. As far as we know, there are few studies which seek to understand the ideological position of subnational Japanese politicians beyond these dimensions or understand them in ideological categories more common in comparative literature such as liberalism, socialism, anarchism, nationalism, fascism, populism, feminism, ecologism, etc.

One feature which has limited understanding of the ideological dimensions of Japanese prefectural and municipal elections is that partisan competition is limited and ambiguous. Non-partisan or pan-partisan (so called mutoha or ainori) candidates are more common than candidates who campaign with a clear partisan affiliation. As ideologies and their programmes are generally tied to parties, the lack of clear partisan conflict has made it difficult to analyse subnational political ideology.

Moreover, aside from the handful of genuinely competitive urban elections, rural and suburban areas have been dominated by conservatives. The competition in these regions is understood not to be primarily between candidates with diverging ideologies, but a “lateral competition” to extract resources from the centre (Muramatsu 1988) among subnational governments of similar conservative leaning. More recently, subnational conflict has been understood as one between the legislative and executive branches of government, rather than between different partisan candidates or groupings within the assembly. Governors and mayors are, regardless of their partisanship, assumed to be generally incentivized to cut spending, thus clashing with legislatures controlled by assembly members who are assumed to be individually incentivized, regardless of their partisanship, to expand particularistic benefits (Sunahara 2011).

How then are we to understand ideologies in a context of limited or ambiguous partisanship?

This paper takes the position that, at its broadest, ideologies are “worldviews” which provide coherent interpretations of the world and guidelines for dealing with collective problems (Knight 2006). Ideologies create political communities, including most importantly political parties (Berman 2006, p. 105). Though parties are key proponents, simplifiers, and repositories of ideologies, they are not the only ones who develop, use, and are shaped by ideology (Freeden 2013). Ideology can be found at all levels of political processes and be meaningfully interpreted and identified through analysis of discourse (Freeden 2013; Van Dijk 2013). Party ideologies are estimated based on their discourse and issue positions, as well as select vocabulary which indicate ideological positions (e.g. Pauwels 2011). Similarly, a politician's ideology can also be estimated from salient themes and assumptions about the political world they exhibit in their discourse or from their issue positions. They are often categorized into major ideological families such as liberalism, conservatism, socialism, anarchism, nationalism, fascism, populism, feminism, and ecologism (Heywood 2021).

In the Japanese literature, political ideology has been synonymous with dimensions of conflict, or in Japanese tairitsujiku. These conflict dimensions can be understood as enduring differences in the ideological orientation, and hence policy positions, between significant parties and thus a characterization of a party system (Lijphart 1981). Such dimensions are estimated from the discourse (manifestos, public statements) as well as policy positions of the parties.¹ The consensus is that the post-war Japanese party system was and in large part still is driven by a conservative–progressive ideological dimension which combine a significant foreign policy dimension of issues related to military security (and reform of Article 9 in the Constitution) as well as economic left–right issues of small/large state, redistribution, trade liberalization, and deregulation (Kabashima and Takenaka 1996; Taniguchi and Winkler 2020). More recent research has identified other dimensions of conflict in recent elections such as differences across parties in attitudes towards traditional LDP 55-system style clientelism, neo-liberal reforms, particularism vs. universalism, or reform vs. anti-reform (e.g. Sakaiya 2011; Shinada 2011). Beyond such national level analysis, there has been no systematic analysis of the kinds of cleavages and conflict dimensions which shape subnational elections, or in terms of mainstream and fringe positions held by subnational actors.

Data and method

To determine the “norm” of conflict and ideological content of subnational elections, we collect and analyse individual candidate manifestos (senkyo kōhō) for gubernatorial elections. This campaign material has been used by scholars to measure the electoral appeals of Japanese politicians to voters systematically and objectively over a series of elections (e.g. Catalinac 2016; Shinada 2001). They may not be reliable indicators of the party's or individual legislator's “true” intentions or actual policy performance once in office, but they are highly effective ways of measuring a candidate's strategy of appealing to voters (Catalinac 2016, pp. 63–75).

The standardization of format and official nature of the campaign material allows for reliable comparisons across elections and regions. Moreover, these kōhō are usually archived by the local electoral commission in booklet form allowing for historical comparison of campaigning strategies. In comparison to other sources of candidate information such as mass media, speeches, and TV, a high proportion of voters is exposed to candidate manifestos, and say they use them to inform themselves of candidates, when casting votes.²

These kōhō contain a wide range of content, typically including: name and picture of the candidate; party affiliation and nomination information; “catch copy” or slogans; personal and/or political philosophy; a list of policies (usually numbered or bullet-pointed) followed by details of the specific policy area; a list of policy accomplishments; a personal profile; and nominations (suisen) by individuals – oftentimes politicians from their own level or other levels of government, including ministers, party leaders, MPs, governors/mayors – or organizations.

We collected 250 candidate manifestos for 78 gubernatorial elections took place between June 2015 and October 2022. The candidate manifestos were downloaded from the Seijiyama.jp website³ and turned into text data through OCR software (ABBYY Finereader). The text data were then read multiple times and corrected by the first author and research assistant for any errors or omissions caused during the digitalization process. The average word count for all manifestos was around 531 Japanese characters, with the range between 40 and 1857.

Candidate data (gender, age, vote share, past career, incumbency, and partisan nominations) and election data (date, electorate, votes cast, win–lose) on the Seijiyama.jp website were also collected as variables. To determine partisanship, we consulted both the Seijiyama.jp endorsement data as well as print media articles announcing the elections to determine if the candidate received official endorsement (kōnin), nomination (suisen), and/or support (shien or shiji) by official announcement of the national or local branch of parties.

All candidates who did not receive any endorsement, nomination, or support from national or local party branches were classified as non-partisan. Those who received both backing from incumbent and opposition parties at the national level (i.e. those receiving backing from both LDP or the Komeito and either the Constitutional Democratic Party of Japan = CDPJ or the Democratic Party for the People = DPP or the Social Democratic Party = SDP or the Japanese Communist Party = JCP) were classified as ainori. Candidates who received backing from either of the parties on the right (LDP and Komeito) were classified as conservative; from either of the parties on the centre–left (CDP or DPP) as centre–left; from either of the parties on the left (JCP or SDP) as JCP/SDP; from Ishin no Kai as Ishin. Candidates backed by other parties (such as Genzei Nihon, N-koku, Reiwa Shinsengumi, Nippon Dai-itto, etc.) were placed in the category of other parties.

The data include 65 non-partisan, 55 JCP/SDP, 39 ainori, 42 conservative, 22 centre–left, 6 Ishin, 2 Japan First, and 17 other parties candidates. Among these candidates, we deemed any candidate with less than 5% vote share as “fringe” or marginal. There were 66 such fringe candidates, of which 44 were non-partisan and 17 were from other parties.

Using this candidate manifesto data we apply the unsupervised learning method of correspondence analysis, using the KH Coder 3.Beta.03i (mac version), to discover how the language of the candidates diverges. A list of the most characteristic terms used by each of the different partisan candidates is also generated through the software.

We also conduct a count of the frequency of select terms in the manifestos which we believe reflect certain ideological positions. This method follows the dictionary-based approach of counting the proportion of certain vocabulary assumed to be indicators of a particular ideology and has been applied to estimate party ideologies as revealed in manifestos and other discourse (e.g. Gründl 2022; Pauwels 2011). Such dictionaries can be constructed deductively through a priori assumptions about what words would be salient and frequently used by a particular ideologue or inductively from existing data (e.g. using a referent text which represents a particular ideology, e.g. using a known populist party's texts and identifying recurrent or key terms from it to create a dictionary) (Pauwels 2011, pp. 103–105).

We take the deductive approach and select some key terms commonly recognized as core themes of specific ideologies. We provide key terms so for the four mainstream ideological groupings of economic left–right and socio-cultural left–right positions. The standard four-quadrant mapping is frequently applied in comparative literature as well as in Japan (Taniguchi 2020; Taniguchi and Winkler 2020). The scheme divides voters and parties into four categories which roughly equate to the so-called macro-ideological families of liberalism/neo-liberalism, socialism, and conservatism. We also test some terms which we argue reflect “thin-centred” ideologies, i.e. nativist, populist, feminist, ecologist, and regionalist positions. These micro-ideologies do not stand alone but complement or overlay the core left–right economic and socio-cultural positions (Freeden 2003, pp. 97–100).

We tested but discarded many terms which did not appear in the manifesto texts or were used in different contexts by all types of candidates, thus not reflective of a particular ideology. The terms which were selected as well as tested, but ultimately discarded for the nine ideological categories are given in Table 1.

Table 1. Dictionary of selected terms indicating ideological positions

Source: Compiled by the author.

Note: Core normative values of ideologies adopted from Heywood (2021); for regionalism, adopted from Heinisch et al. (2019).

Finally, through repeated reading of the manifestos as well as select speeches,⁴ we identified themes and patterns in the campaign texts. We focused on the campaign discourse of candidates during the pandemic crisis to assess what the candidates perceived as problems or challenges; how these issues should be addressed; and use of normative language. We provided excerpts from manifesto texts and some campaign speeches which illustrate these common ways in which different partisan candidates framed the Covid-19 crisis and their varying ideological responses to them.

Correspondence analysis of manifesto texts

The correspondence analysis generates two results (Figs 1 and 2) which illustrate the most characteristic words used by candidates based on partisan affiliation.⁵ Figure 1 excludes fringe candidates with less than 5% vote share, Figure 2 includes them.

Figure 1. Characteristic words for gubernatorial candidates by partisan affiliation excluding “fringe” candidates (2015–2022).

Source: Compiled by the author, using KH Coder correspondence analysis function.

Figure 2. Characteristic words for gubernatorial candidates by partisan affiliation including “fringe” candidates and Reiwa candidate (2015–2022).

Source: Compiled by the author, using KH Coder correspondence analysis function.

Figure 1 shows a three-way clustering of the characteristic language used by different mainstream partisan candidates in gubernatorial election manifestos, when excluding candidates with less than 5% vote share and the single Reiwa candidate.⁶ First, on the positive side of the x-axis is the particular language of JCP/SDP-backed candidates who refer more frequently to welfare matters (which they seek to protect and expand), but equally on national policy issues and politics (which they oppose and seek to stop) as reflected in the characteristic terms: “constitution”, “nuclear”, “consumption tax”, “medical fees”, “livelihood”, “children”, and “welfare” （憲法・消費税・原発・医療費・暮らし・子供・福祉・政治）. On the opposite side of this axis are the ainori, centre–left, and conservative candidates as well as down in the bottom-left the Ishin and non-partisan candidates. The mainstream ainori, centre–left, and conservative candidates share language which primarily reflect emphasis on economic development, regeneration, and demographic issues in terms such as “industry”, “region”, “regeneration”, “roads”, and “town-building”(産業・活性化・地域・道路・まちづくり). The third clustering of Ishin and non-partisan candidates diverges from the other two groups on an emphasis of privatization and administrative reform with characteristic terms such as “private” (sector), “administration” (reform), and “economy” （民間・行政・経済）.

Figure 2, which includes “fringe” candidates of less than 5% vote share and the Reiwa and Japan First candidate, changes this clustering. These non-mainstream candidates use very characteristic language, placing them far out and away from the mainstream candidates crowded around the origin. Other party candidates use terms uniquely such as “cold”, “immunity”, “mask”, “virus”, and “vaccine”（風邪・免疫・マスク・ワクチン） which suggest a preoccupation with Covid-19 in a particular way. They also have characteristic words such as “receiving fee [for NHK]” (受信料). Japan First and Reiwa, two other minor party candidates, are placed away from the mainstream parties towards the fringe group, with characteristic terms such as “Japanese”, “income”, and “individual” (日本人・収入・個人).In this figure, the mainstream parties are separated in terms of language use on the x-axis which appears to capture the traditional conservative–progressive dimension. In Figure 2, JCP/SDP candidates focus on welfare and national policy issues and the ainori, conservative, and centre–left candidates cluster around words of growth, development, and maintaining/governing public services.

A characteristic word analysis,⁷ which lists the words most characteristic to a particular variable, corroborates results from the correspondence analysis, as shown in Table 2.

Table 2. Characteristic word analysis of gubernatorial candidates by partisan affiliation (2015–2022)

Source: Compiled by the author using KH Coder Jaccard coefficient function.

Combining the two types of textual analysis results, we provide the following interpretations.

The language of gubernatorial candidates is largely clustered into four groups: (1) the “mainstream” cluster which includes conservatives, ainori, and centre–left parties which largely emphasize growth, development, and management of existing public services and institutions; (2) the “old left” (communist/socialist) who call for expanded welfare and oppose a range of national policies; (3) the “neo-liberal” Ishin and non-partisan candidates focusing on private sector, deregulation, growth and competition; and a (4) “fringe” group of candidates whose language diverges sharply from the other three groups, including language around the pandemic.

Though it is difficult to extract ideological orientation merely from terms unique to a particular set of candidates, the first mainstream group appears to correspond to traditional conservatives focusing on valence issues of growth and development while avoiding contentious social issues, the second to progressives on the left seeking welfare, the third to neo-liberal reformists, and the fourth to a fringe of anti-establishment, anti-system nativists, populists, and others.

The frequency of key terms in the gubernatorial manifestos indicating specific ideological positions is provided in Table 3. The frequency by which candidates generally use these terms is low (ranging around half a per cent of all words in a manifesto). This level of frequency, however, is comparable to results taken in a similar dictionary-based approach to identify ideological positions which found on average around 2% of dictionary words found in the analysed text (Pauwels 2011, p. 104). We focus instead, as earlier literature has done, on the proportion of words used across the different types of ideological categories. This approach assumes candidates combine various categories of ideological language, but some kinds of language are more pronounced than others.

Table 3. Frequency of key terms in gubernatorial manifestos 2016–2022, by partisanship

Source: Compiled by the author.

Note: The highest frequency of terms per each ideological category is underlined.

We find that the proportion across these ideological categories largely corroborate with our findings in the correspondence analysis and the most characteristic word analysis.

The mainstream candidates from the LDP/Komeito and DPP/CDPJ do not diverge much in terms of the proportion which they use key terms signifying socio-cultural and economic left positions. They do diverge more in terms of terms reflecting socio-cultural and economic right positions, with the conservatives being more prone to conservative and neo-liberal terminology. The mainstream left differentiates itself more in the proportion of terms indicating ecologism and regionalism. The JCP is clearly an outlier in terms of economic left and feminist language, while the Ishin stands out for emphasis on neo-liberal vocabulary and to a lesser extent on populist terms. Nativist language is most prominent among the “other” grouping which includes extreme right fringe candidates.

Representative samples of discourse from the four clusters

A closer analysis of the language – not just a listing of their most unique terms, their spatial estimation vis-à-vis each other, and frequency of key ideological terms – is necessary to interpret the ideological orientations of each of these clusters.

We provide sample excerpts from the campaign discourse which capture broader themes of these four clusters of candidates. This section is interpretive and supplementary to our quantitative findings but should illustrate that the four groupings indeed diverge not just in the language they use, but also in the underlying “worldviews” about collective issues and desired outcomes.

Mainstream cluster

The campaign language from mainstream gubernatorial candidates – those backed by the LDP, Komeito, DPP, and CDPJ – shares a broad array of themes. During the pandemic, these candidates commonly first address countermeasures to protect lives from the pandemic and also an equally strong emphasis to support businesses and local industries affected by the pandemic. These are followed with a range of topics addressing various economic interests and demographic groups: infrastructure/disaster prevention measures; measures to revitalize local economy with new jobs and ways of working, such as telework, support for primary industries and small and medium enterprises, and measures which promise healthy and fulfilling lives for the elderly, young, women students, and children.

For example, an LDP-backed candidate for Yamaguchi prefecture offers the following slogans:

Rebirth and New Challenges. Restoring Yamaguchi's vitality. We will create a new future for Yamaguchi. Protect the lives and health of the people of Yamaguchi. Revitalize the local economy and livelihoods. (Muraoka Tsugumasa/Yamaguchi/manifesto/February 2022)

Similar claims are made by an ainori candidate, backed by both mainstream parties, for Akita prefecture in his manifesto:

Support for prefectural residents. Distribution of gift certificates to support daily life. Support for businesses. Financial support and assistance for business maintenance expenses… We will promote vaccination so that all prefectural residents can live with peace of mind. (Satake Norihisa/Akita/manifesto/April 2021)

The policy content of conservatives, centre–left, and ainori candidates are similar. They talk in equal measure about valence issues of promoting growth, welfare, families, and women in workplace, protecting the environment, and improving education. A mainstream candidate, backed only by the local prefectural CDPJ branch, in Kagoshima echoes these twin priorities of health and wealth:

Yuichiro Itoh will do his utmost to rebuild jobs and livelihoods damaged by the new type of coronavirus with decisive action. Creating a safe, secure, and stable region. A prefecture resilient to disaster, where people can work with peace of mind for the rest of their lives and lead stable lives. (Yuichiro Ito/Kagoshima/manifesto/July 2020)

Short slogans about protecting lives and revitalizing regions, attracting residents, growing the local economy, followed by lists of more specific policy measures and past achievements (if they are incumbents), are characteristic of these mainstream candidates. The discourse of such candidates reflects a shared “big tent” strategy of promising both growth and to as many different groups and interests as possible, while avoiding any controversial issues which may alienate certain groups.

Old left cluster

The communist and socialist candidates in contrast have a more focused, unified message across their manifestos and speeches. The socialist, pacifist, feminist, and ecologist ideological orientation of the party is quite explicit, even in their short candidate manifesto content. Almost all communist-backed candidates state in their manifestos the party goals to oppose constitutional reform (or reinterpretation), war, nuclear power, reduction in welfare, regressive consumption tax, and measures that hurt small businesses. There is emphasis on solidarity and human rights, a society not based on principles of efficiency but solidarity. Although mainstream parties mention promoting women's participation in the work force and providing support to balance career and family, the communist and socialist candidates are more explicit in demanding gender equality itself on principle.

Here is a representative sample by a communist candidate for Chiba prefecture:

Aiming at anti-coronavirus measures and gender equality. Priority on life and livelihoods. No one is left behind. Let's show our anger at the Suga Government [LDP administration at the time] from Chiba. Non-regular workers, students and women are in difficulty due to the Covid-19 disaster… to a city where people can live with peace of mind, against casinos and IRs, and for zero nuclear power plants. Enact gender equality and a prefectural government in which the Constitution can thrive. (Kanemitsu Rie/Chiba/manifesto/March 2021)

Neo-liberal cluster

The language of those in the neo-liberal cluster – such as those backed by Ishin and the Tokyo governor Koike – shares a focus on efficiency, business, innovation, competition, decentralization, and privatization, while criticizing fiscally profligate, ineffective incumbents. They list various pro-growth measures, along with emphasis on investment in education, for the goal of economic growth. They perceive the role of governors as managers of the prefecture, guiding the local economy to leaner, faster growth, by reducing the role of the central state, freeing the dynamic energies of local communities through decentralization and privatization. The candidates, notably from two urban areas, also share a hostility to central government strictures and inefficiencies, and some language which seeks to blame the central government for the spread of Covid.

The following sample of discourse reflects such tendencies. An Ishin candidate for Tokyo claims:

Urban growth strategies to create vitality, including privatization and attracting IRs. Administrative and financial reforms to overcome the financial crisis. Early resumption of economic activities and various schools [restricted by the pandemic] based on scientific evidence by narrowing down the scope of regulation to specific industries. (Ishin/Ono Daisuke/Tokyo/manifesto/July 2020)

Fringe cluster

All candidates in the fringe grouping include those from minor parties such as the Japan First Party or N-Koku, or anti-NHK party, and others who are independents. As such they diverge in their ideological leanings, but their language, as the correspondence analysis reveals, is very different from the mainstream, so they cluster together. Two candidates sampled here reflect a nativist right and an anti-vax, anti-establishment populist position. Though a very short manifesto, the use of the word “Wuhan flu” to refer to Covid, “Japan”, and opposition to “foreigners” receiving livelihood assistance clearly flags nativist xenophobia, in the case of Japan First leader Sakurai Makoto:

Protecting the lives of Tokyo residents from Covid-19 Wuhan flu. Protecting life, livelihood and property. Japan First STRONG JAPAN. Relief for Tokyo residents through significant tax cuts. Zero residence tax. Zero property tax. Zero governor salary. Immediate elimination of livelihood assistance for foreigners to reach those in need. (Japan First Party/Sakurai Makoto/Tokyo/manifesto/July 2020)

With another fringe candidate, the anti-mask, anti-science, anti-expert position is also quite evident in the rambling discussion about immunity.

Covid-19 is just a common cold. It was the media and the government which created the Covid-19 uproar. NO to masks… Self-restraint in going out lowers immunity and makes people sick. (Hiratsuka Masayuki/Tokyo/manifesto/July 2020)

“New left” and “regionalist” candidates

Besides the four clusters of language revealed by Figure 2, we provide examples of discourse from two other ideological orientations – what could be called “new left” and “regionalist” which have not been captured by the correspondence analysis. Yet the two have significant impact on select elections. The first is the language of Yamamoto Taro, leader of Reiwa Shinsengumi and identified by some as a “populist left” party (Eder-Ramsauer 2022). The second group are regionalist, anti-metropole attitudes – which in part overlaps with the decentralist position of the neo-liberals – but are most pronounced in the Okinawan elections.

The main theme of Yamamoto's campaign manifesto (Reiwa/Yamamoto Taro/Tokyo/manifesto/July 2020) for the Tokyo gubernatorial elections is to protect the vulnerable with “love and money”, while criticizing an indifferent central and local government which does not provide adequate support for those falling through the cracks during the pandemic. He criticizes the incumbents at both national and prefectural governments as well as the vested interests which they protect (e.g. companies involved in delivering various government programmes). The combative language is arguably populist in that it pits the many vulnerable groups against the establishment (government and wealthy and vested interest). Yamamoto's discourse shares the socialist concern for redistribution and curbing the baleful influence of neo-liberal capitalism of the “old left” but does not engage with the core traditional “progressive” themes of foreign policy and constitutional reform.

His speech illustrates such ideological assumptions:

Even before Covid-19, it was difficult, there were many people who said they were in an economic emergency. One out of every seven children in this country lives in poverty. One out of three women living alone is poor. Even before the Covid-19 disaster, everyone had been living in poverty for more than 20 years during deflation… Under such circumstances, what are you going to do if the national government or the Tokyo Metropolitan Government does not do something to firmly raise your income level? It is absurd… They are trying to protect only vested interests; their purpose is not to provide benefits to you. (Reiwa/Yamamoto Taro/Tokyo/speech/July 2020)

Finally, we turn to regionalist language and responses to Covid pandemic in Okinawa.

As an incumbent, Okinawa Governor Tamaki in his manifesto and speeches elsewhere during the campaign emphasizes the victimization of the Okinawan people due to failure of the central government, courts, and mainland parties who disregard local autonomy. He has called for the Okinawan voters to come together over “identity and not ideology” to achieve “growth with pride” (Tamaki Denny/Okinawa/manifesto/September 2022).

In his campaign speech, Tamaki, like his predecessor, starts and ends his speech with Ryukyuan to emphasize regional identity, complementing with his appeals to regional pride, identity, and gentle and multicultural Okinawan tolerance (“leaving no one behind”).

Saigusu yo, chuuganabira. Good morning everyone… The people of Okinawa Prefecture will never approve of the way the national government is interfering with local autonomy, which is also the principle of its constitution, and I, for one, will not allow that to happen… I will do everything I can to realize a peaceful, prosperous, and base-free Okinawa with the hearts of the people of the prefecture. Please help us to achieve this goal (Tamaki Denny/Okinawa/speech/September 2022).

Another Okinawa candidate, founder of the regional party Sōzō and former parliamentarian member for Ishin, identified the primary challenge Okinawa faces as internal fighting over the base issue. He calls also for voters to stop being divided over the base issue and focus efforts on development, attracting tourism and investments to the islands, emphasizing both its identity as a victim and also (need for) autonomy and regional pride. (Shimoji Mikiro/Okinwa/manifesto/September 2022)

Both Okinawan candidates emphasize regional identity and strongly frame issues of Covid and economic growth around what is arguably its core dimension of conflict: Okinawa's relationship with the central government, specifically over base issues. The thin ideology of regionalism overlays the economic and cultural left ideology of Tamaki and the more neo-liberal economic ideology of Shimoji.

Discussion and conclusion

Analysis of gubernatorial campaign manifestos and speeches during the pandemic generated three primary findings.

First, correspondence analysis allowed us to systematically demonstrate that the discourse of most candidates diverge into four clusters which reflect emphasis of certain policy issues and underlying worldviews. The language of these groups – “mainstream”, “old left”, “neo-liberal”, and “fringe” – reveals differences in attitudes towards prioritizing public health, economic growth, inequality, foreign policy, markets, deregulation, and political decentralization.

Mainstream candidates focus on valence issues of growth and welfare, while largely sidestepping questions of redistribution, inequality, and controversial cultural and foreign policy questions. This “mainstream” group includes conservative, ainori and centre–left-backed candidates, and are undifferentiated in terms of their language emphasizing both economic development and welfare, making them a different category than the traditional “conservative” camp envisioned in traditional literature. The JCP/SDP-backed “old left” is, as the name suggests, traditional in their pacifist, socialist stance of “progressives”. But their emphasis is not just on expanding local welfare but also on opposing various national policy issues from nuclear power, consumption tax, to constitutional reform, even in the subnational context. The “neo-liberals” also confirm what earlier literature has indicated as a group of reformist governors (and mayors) emphasizing administrative reform, deregulation, and decentralization. Finally, we identified a “fringe” group with very distinct language and policy positions, including nativists and anti-vax populists. These deviants from the mainstream discourse have minimal electoral impact or success (under 5% vote share) and are usually only present in urban area elections. Besides these two groups, we demonstrate the existence of two other ideological orientations – an economic populist left distinct from the “old left” represented by Reiwa Shinsengumi and a regionalism prominent in Okinawan candidates.

Second, the analyses demonstrates that the traditional dimensions of conflict composed of progressive vs. conservative plus reformist/neo-liberal positions are inadequate to understand the dynamics of subnational politics, both on the economic dimension and foreign policy/constitutional reform dimensions. In the face of the pandemic, most mainstream candidates, old and new left, as well as fringe candidates called for expanded support of businesses and individuals, while also strengthening medical care and expanding public services. During the pandemic at least, all types of gubernatorial candidates – aside from the neo-liberals – advocated expanded spending. This finding goes against the claim (Sunahara 2011) that chief executives of subnational governments largely tend to fiscal austerity (while assembly members push for greater spending). The mainstream conservative and progressive candidates also did not diverge on foreign policy positions, with the subnational candidates backed by the LDP, CDPJ, or DPP assiduously avoiding any references to foreign policy or constitutional reform. In contrast, the other non-mainstream candidates diverged and actively engaged on non-economic issues, including foreign policy and constitutional issues as well as those on gender, environment, immigration, and science.

Reading the manifestos and speeches more closely we find the mainstream candidates (LDP, CDPJ or DPP, and ainori-backed candidates) largely share policy positions of pursuing growth, greater welfare, and saving lives without blaming or criticizing any existing policies or groups. Competition appears to centre around valence issues in terms of which candidate can best deliver on these goals. This dynamic mirrors national elections, at least for the 2021 Lower House, which earlier literature has found as one primarily fought over competence in dealing with Covid (Smith et al. 2022, pp. 390–391).

Finally, partisan competition differs largely by urban or rural context, but both tend towards uncompetitive elections with low turnout. In urban areas, there is a melee of candidates besides the mainstream, including neo-liberals and many fringe options. If a challenger outside of the mainstream can capture enough momentum it can generate competitive elections which may lead to an upset, but otherwise the multiplicity of candidates simply generates fragmentation among challengers. In the remaining rural areas, however, there appears to be a lack of candidates challenging the incumbent: most gubernatorial contests end up as one between a mainstream candidate and “old left” candidate (aside from the occasional conservative-split election that pits two LDP-affiliated candidates). Though there may be clear ideological distance between the conservative and JCP/SDP candidates, it is not one which is attractive to voters, as many such contests resulting in uncompetitive and low-turnout elections.

In the seventy-eight gubernatorial elections we sampled, only seventeen elections were competitive with a runner-up ratio (aka sekihairistsu, the ratio which divides the vote share of the runner up with the winning candidate and is a benchmark for the competitiveness) above 0.7. Revealingly, among these competitive elections, six were hoshu bunretsu – a conservative split election between two LDP candidates, eleven were three-way contests involving the LDP, a largest centre–left-backed candidate, and a third-pole party candidate. In contrast, the sixty-one uncompetitive contests were those between incumbents backed by the LDP or by a bandwagon of LDP and centre–left against a lone JCP/SDP candidate. These contests result in the kind of ideologically distinct, but for voters, uncompelling and hence unattractive elections leading to lower turnout. Average turnout for the competitive elections was at 52.5%, while that of the uncompetitive ones was 42.5%.

The uncompetitive dynamic in these prefectures partly mirrors a situation identified by Sakaiya (2023) in his assessment that the national party system is reverting to an uncompetitive “neo-1955” system characterized by a traditional progressive–conservative split over foreign policy/constitutional reform. Unlike at the national level though, however, the mainstream centre–left candidates in gubernatorial elections do not talk about foreign policy or constitutional reform. Instead, the progressive–conservative divide, if there is one, in subnational politics is between the JCP and the mainstream (LDP, ainori, and centre–left parties).

Notably, this convergence of mainstream parties and outlier status of the JCP has been separately identified in other research at the national level. In a correspondence analysis of candidate Tweets for the 2021 Lower House election, Fahey finds that the LDP, DPP, CDPJ, and Ishin largely cluster in the centre, with characteristic terms for each of the parties mostly referring to valence issues while the JCP stands outside of this mainstream discourse, with its characteristic terms reflecting “more divisive issues like nuclear power, US bases in Okinawa, and social justice” (Fahey 2022, pp. 192–193 and Fig. 13.3).

In sum, conservatives and ainori candidates in subnational elections hold an electoral stranglehold, underpinned by ideological convergence, over most gubernatorial elections. The main opposition at the national level – CDPJ and DPP – has difficulty standing candidates sub-nationally (only succeeded in standing candidates in twenty-two out of our seventy-eight sampled gubernatorial elections). They frequently chose to bandwagon onto an ainori candidate or to stay out of the elections on the sidelines. In the few elections they have managed to back a candidate against the conservative incumbent, they have failed to differentiate themselves in language or policy. Voters are not given a clear alternative to conservatives from among the largest opposition parties at the national level, with only the JCP and some fringe candidates providing clear difference in language and policy. Yet this kind of ideological choice does not lead to greater competitive elections.

Beyond the mainstream “norm” of conservatives and progressives along with neo-liberal positions, we find there are other “deviating” ideological orientations in subnational elections, but they are not compelling to most Japanese voters for now. It is unclear if and how such fringe or minor positions can mainstream and make subnational elections more competitive and options on offer more attractive for voters.