Dreaming "Independence" by Riding on Globalization: The Setback of Okinawa's Reformist Politics in the 1990s

Takashi Yamazaki Associate Professor, Department of Geography, Osaka City University
Ph. D. Candidate, Department of Geography, University of Colorado at Boulder
E-mail: yamataka@lit.osaka-cu.ac.jp

Please do not cite or quote without permission

GAD Conference "Responding to Globalization: Societies, Groups, and Individuals" Institute of Behavioral Science, University of Colorado, Boulder. Hotel Boulderado, Boulder, Colorado, April 6, 2002


Introduction

Okinawa (Okinawa Prefecture of Japan) is a group of 108 islands stretching between the Japan Islands and Taiwan (Figure 1). It currently has a population of 1.3 million and an area of 2,267 square kilometers and has been the least developed prefecture in Japan. The per capita income of Okinawa is about 70% of the national average. Currently, 40 military installations are located in Okinawa and occupy approximately 20% of the land of Okinawa Island (Figure 2). Okinawa is traditionally divided into four administrative sub-prefecture regions: the south, central, north, and Sakishima regions. The former three regions are in Okinawa Island. U.S. military bases are most concentrated in the central region of Okinawa Island as a result of the U.S. military occupation and governance from 1945-72 (Table 1).

There are three distinct phases in the modern history of Okinawa. Okinawa's predecessor, the Kingdom of Ryukyu, was annexed to Japan (the Japanese state) in 1879. From 1879-1945, Okinawa was one of the prefectures in Japan and was colonized by and socio-culturally integrated into Japan. From 1945-72, Okinawa was governed by the U.S. military force and transformed into a security "keystone" under the Cold War. Okinawa Island became filled with newly built U.S. military bases to contain the communist bloc in the Asia-Pacific region. During this period, Okinawa was put under a less democratic regime and a series of political mobilizations against the U.S. military rule took place, reflected through mass protest and voting for "reformist" (leftist) candidates and parties. Under the U.S. military rule, Okinawa's society and identity were profoundly transformed. In 1972, Okinawa reverted to Japan and again became one of its prefectures. Even after 1972, however, the concentration of U.S. military bases did not significantly change because the Japan-U.S. security relations established after World War II did not change their basic functions. Currently, 75% of the area of the U.S. military bases located in Japan is concentrated in Okinawa and the military bases occupy 20% of the area of Okinawa Island (Figure 2).

Since the 1972 reversion, the U.S. military presence, economic dependency, and socio-cultural integration into Japan have been the major stimuli in the continuation of political mobilization protesting the U.S. and Japanese governments (Nakano and Arasaki 1976; Arasaki 1996). For Okinawans (natives in Okinawa), U.S. military bases have been not only the sources of problems such as forcible land seizure, obstacles to effective development, physical and environmental damage, and violent crimes, but also the sources of income through job opportunities and rent payment. For the Japanese government, Okinawa is indispensable due to the fact that Japan highly depends for its territorial security on the U.S. military force stationed in Okinawa. Playing the role of the mediator between Okinawa and the U.S., successive Japanese governments have attempted to mitigate the grievances of Okinawans by providing a large amount of developmental subsidies for Okinawa. This has reproduced Okinawa's economic dependency on Japan. The U.S. government, on the other hand, has been able to take advantage of this situation in Okinawa in order to exercise its geopolitical influence over East Asia. Thus, the existence of U.S. military bases, U.S.' and Japan's policies toward Okinawa, and Okinawa's political mobilization have developed a complex political nexus and produced the "political space" of Okinawa as a spatial manifestation of those political processes.

Based on an understanding of these facts, this study explores the development of political mobilization in Okinawa in the 1990s by analyzing the spatio-temporal patterns of collective action and voting. The 1990s is one of the most important politico-historical turning points in Okinawa for the following reasons. First, the end of the Cold War in 1989 raised the expectation that Okinawa's long-term suffering would finally end in the near future. However, the expectation was not met because the Japan-U.S. security relations were redefined and re-secured in 1997. Second, the end of the Cold War also enhanced anticipation that raiding on globalization would bring about a new stage of economic prosperity. For Okinawa, globalization became a key word for overcoming excessive economic dependence on Japan's national economy and public finance. However, the decline of Japan's economy in the 1990s had a significant impact on Okinawa. Third, the rape of an Okinawan girl by U.S. servicemen in 1995 drastically increased Okinawans' grievances and led to a series of mass protests in the second half of the 1990s. While these political uprisings created a new stage of the reformist politics seeking pacifism in Okinawa, Okinawans were forced to make a difficult choice between political idealism and economic realism. Fourth, Masahide Ota was elected governor in 1990. Ota was the third reformist governor in post-reversion Okinawa. In his second term from 1994-98, he actively attempted to promote anti-war, pacifist policies by taking advantage of the political uprisings after 1995 and frame economic policies based on globalization for economic self-supportiveness. Despite his tactical strategy to mobilize Okinawans against the Japan-U.S. security relations, Ota failed to win the election for his third term in 1998, indicating the setback of Okinawa's reformist politics or political idealism.

In the above-mentioned incidents, political mobilization such as collective action and voting played a significant role in changing the course of politics. More generally, the case of Okinawa is a good example in describing how the concept of globalization is incorporated into sub-state political idealism, or regionalism, resisting the oppression of the nation-state.

Conservative-reformist cleavage in Okinawa

Unlike mainland Japan, there has been a very clear cleavage in Okinawa between conservative and reformist parties since the 1950s. This is because Okinawa has directly faced the oppression of a nation-state, be it Japan or U.S., and because party politics have been shaped for or against the state. Elections for a governor and prefecture assemblymen have reflected such political cleavage. While conservative candidates have tended to emphasize developmental policies based on state subsidies and security cooperation with the state, reformist candidates have usually been opposed to the U.S. military presence and the Japan-U.S. security relations. Political parties in Okinawa had become affiliated with those in mainland Japan before the 1972 reversion. Therefore, the controversies between these parties have often represented those between the Liberal Democratic Party (LDP) and reformist parties such as the Social Democratic Party of Japan (SDPJ) and the Japan Communist Party (JCP). Although the LDP has been dominant in the Japanese Diet, the conservative parties in Okinawa have not always been so. For this reason, political struggles over the Okinawa Prefecture Office and Assembly have tended to be intense and voter turnout has been higher than in mainland Japan.1 The tension of the struggles has been well reflected in the results of gubernatorial elections in which the governor has been replaced at similar intervals.2

As previously mentioned, votes for reformist candidates, parties, or policies (reformist votes) have represented Okinawans' discontent with the U.S. military bases and the Japan-U.S. security relations and can be considered an expression of protest identity (Idaka 2001). In Okinawa, reformist-party-affiliated labor unions have constituted the major social movement organizations (SMOs) that organize various kinds of collective actions such as political rallies and public meetings. Such SMOs have played a role in connecting collective actions with reformist votes. Therefore, in order to understand the role of political mobilization in Okinawa in the 1990s, it becomes necessary to examine the interrelationship between collective action and voting by focusing on the activities of SMOs. However, the degree of the interrelationship has changed due to the decrease of labor union membership rate and the increase of floating votes or the voters who do not support any particular party or ideology. Furthermore, such an organizational approach does not necessarily address the function of a "place" as a mediator of political mobilization (Agnew 1987). Okinawa is not a unitary space, nor does it experience a unitary time. The development of political mobilization in any place is spatio-temporarily differentiated. Rather than describing in detail how SMOs connect collective action with voting, this study examines the interrelations between these political behaviors and their spatio-temporal differentiation. By doing so, this study attempts to reveal the spatio-temporal temporal discontents expressed in protest actions and reformist votes. The clear conservative-reformist cleavage in Okinawa will help with this examination.

The terms of Governor Masahide Ota

The first term

In 1990 Masahide Ota was elected governor after the twelve-year term of Junji Nishime. Ota had been Professor in Journalism at the University of the Ryukyus and was one of the well-known leftist opinion leaders in Okinawa. His predecessor Nishime was a powerful conservative politician who served as governor from 1978-90. As mentioned before, successive governors in Okinawa have been replaced at similar intervals according to the political circumstances of the times. Seemingly tired of "the logic of development based on public works" promoted by the conservative governor (the Okinawa Times 11/24/1990: 3), Okinawan voters expected the new reformist governor to change the course of prefecture politics (the Okinawa Times 11/19/1990: 1). The few years around 1990 were the most recent peak of economic prosperity in Okinawa as well as Japan. For the past two decades, the increase rates of GDP and the Gross Prefecture Products were the highest while the unemployment rates of Japan and Okinawa were the lowest (Figures 3 and 4). After the conservative Prefecture Office completed its role of economic development during Nishime's terms, the voters expressed their political preference for more pacifist, anti-war politics which Ota pledged to embody.

In Ota's first term from 1990-94, the national political environment was drastically changed following the end of the Cold War. The LDP dominance in the national Diet came to end in the 1993 election for the House of Representatives. The LDP was forced out of office for the first time since 1955 when it was founded. The SDPJ also faced a significant decline of voter support at the same election. In the early 1990s, the conventional conservative-reformist cleavage in national politics started to disappear. In the campaign for the 1994 gubernatorial election, the pledges between Ota and his conservative opponent were not so different as in the previous election (the Okinawa Times 11/20/1994: 2). Neither was there any significant issue or incident in favor of one candidate over the other. Protest actions during the period were also fewer than in Ota's second term (Figure 5a). After all, the 1994 election resulted in Ota's victory by a huge margin and the lowest voter turnout (62.54%) in the gubernatorial elections (the Okinawa Times 11/21/1994: 1).

The second term

Ota's second term, however, was very different from his first term in many respects. The most important incident in his second term was the rape of an Okinawan girl by U.S. servicemen in September 1995. This crime led to a series of large-scale political rallies in the second half of the 1990s (Figure 5b) and had a significant impact on the Japan-U.S. security relations. The political rallies on October 21, 1995 attracted approximately 85,000 participants in Okinawa Island and 3,000 in Sakishima Islands (according to the host organizations, the Okinawa Times 10/22/1995: 1). These rallies denounced the notorious crime by U.S. servicemen, blamed the Japanese and U.S. governments for their continuing unfair treatment of Okinawa, and required the governments to revise the Japan-U.S. Status of Forces Agreement (SOFA)3 and reduce U.S. military bases (the Okinawa Times 10/22/1995: 1-3). Unable to ignore the explosion of the Okinawans' grievances, the Japanese and U.S. governments started to revise SOFA and make a plan to return the land for U.S. military bases to Okinawa.

Taking advantage of this opportunity, Ota promoted a series of pacifist policies in 1996. First, he refused to sign the land lease contracts for the U.S. military bases instead of the Okinawan landowners refusing to sign the contracts.4 The Japanese government sued Ota for the refusal and this lawsuit was finally brought to the Supreme Court, which decided against Ota in August 1996. Second, according to the petitions collected from prefecture residents, Ota carried out the prefecture-wide referendum in September 1996 asking if SOFA should be revised and if the U.S. military bases should be reduced. Although the result of the referendum had no legal power to change prefecture policies, votes for the revision and reduction were far more than those against them.5 These policies were carried out in parallel with increasing protest actions supporting Ota and protesting the Japanese and U.S. governments. Third, the political environment in favor of Ota allowed him to frame a new grand design of Okinawa in January1996, which was embodied in the Action Program for the Reduction of U.S. Military Bases and the New Industrial Promotion Policy for Cosmopolitan City. One of the reasons Ota was able to take such decisive measures is that the political opportunity structure at the state level was in favor of reformist policies in Okinawa. From 1993-96, the Japanese government was not led by the LDP but by non-LDP parties including the SDPJ or a coalition of the LDP and the SDPJ. The SDPJ was one of the parties supporting Ota in Okinawa. In this political environment, Ota attempted to implement reformist policies to reduce the U.S. military bases and the excessive economic dependency on mainland Japan so that Okinawa could become a group of peaceful islands with no military base and with a self-supportive economy.

A new grand design of Okinawa and globalization

The Okinawa Prefecture government led by Ota published the Action Program for the Reduction of U.S. Military Bases and the New Industrial Promotion Policy for Cosmopolitan City in January 1996. These policies were drawn up after the Japanese and U.S. governments formed the Special Action Committee on Okinawa (SACO) in November 1995. The objective of SACO was to make a plan to consolidate and return the land used for eleven U.S. military installations from the Japanese and U.S. governments' point of view. The prefecture government adopted the Action Program as its own plan for the phased elimination of U.S. military bases by 2015. For Okinawa, the existence of vast U.S. military bases has been an obstacle to effective development and city planning. The implementation of the Action Program, therefore, was a necessary premise for the further development of Okinawa as well as the restoration of a peaceful life for Okinawans. On the other hand, the New Industrial Promotion Policy for Cosmopolitan City was a concrete measure for the development following the reduction and elimination of U.S. military bases by the Action Program. As the title of the policy implies, the Promotion Policy attempted to develop and locate Okinawa in the context of globalization in the Asia-Pacific region. The implementation of the Promotion Policy meant utilization of the vast land that would be returned from U.S. military bases according to the Action Program. Since the elimination of the U.S. military bases meant the loss of job opportunities and income for many Okinawans,6 it became necessary to create new industries to counterbalance the loss. Thus, these two policies complemented each other to constitute "Okinawa's grand design for the 21st century" (Okinawa Prefecture Government 1997: Preface)

Even though the term of globalization appears only twice in the text of the Promotion Policy, the idea of globalization in the Asia-Pacific context is fully expressed. As the text states:

The Cosmopolitan City Formation Concept is Okinawa's grand design for the 21st century and its goal is the promotion of regional characteristics which will contribute to the self-supportive economic development of Okinawa, and continuous development of the Asia-Pacific Region, as well as assisting in maintaining peace.

The concept also aims to transform a military-based island into a peaceful island, and to positively promote various policies based on three basic principles: peaceful exchange, technological cooperation, and economic / cultural exchanges.

Above all, it proposes to implement decisive measures based on a thorough review of the rigid economy of Okinawa today and its progress toward globalization inside and outside the territory to achieve the goal of "creating and promoting new industries suitable for the 21st century." (Ibid, emphasis added)

The Promotion Policy has three general directions: the advancement of the free trade zone (FTZ), the integration and enhancement of information and communication industry, and the formation of a hub for international tourism and destination industry. Among the three policy directions, the advancement of FTZ was the core strategy to promote Okinawa's economic self-supportiveness and competitiveness in the Asia-Pacific context and thus was most closely related to the idea of globalization. As the text puts it:

Okinawa's problems in relation to the promotion of industry include the limited availability of land due to the existence of vast U.S. military bases, delayed improvement of locations for industrial development as a direct result of having been excluded from various post-war industrial promotion policies, and comparatively high transportation costs because of its geographical distance from other parts of Japan and being a prefecture of many relatively small islands.

The industrial promotion of Okinawa has been relatively slow due to the above factors, and relatively weak economical structure which depends largely on mainland Japan still remains. Okinawa is in economically critical condition since Okinawa's per capita income remains the lowest in Japan and the unemployment rate is two times as high as the national average.

Other economic factors include the reduction of public investment and a decrease in base-related income which is a result of the consolidation and reduction of U.S. military bases in Okinawa. Thus, it is feared that further economic sluggishness and decreased job opportunities will worsen.

On the other hand, Japan's present economic condition, including an industrial migration to foreign regions and rapidly growing neighboring countries in Asia, should be taken into consideration for meeting the needs and adapting to changes of the times, such as advanced internationalization and information technology so that Okinawa's self-supportive development will be realized.

Thus, it is necessary to develop new industrial promotion policies which utilize Okinawa's regional characteristics and resources in a positive way. Improvements designated to attract domestic as well as foreign industries need to be expanded by introducing a free trade zone with tax incentives, focusing on deregulation, and constructing an infrastructure which includes an international airport, harbors, and information and communication facilities. (Okinawa Prefecture Government 1997: Chapter I, emphasis added)

From the contents of the Promotion Policy as well as the Action Program, it seems clear that the Okinawa Prefecture government attempted to simultaneously "liberate" Okinawa from U.S. military bases and Japan's national economy and public finance. In other words, the prefecture government sought to break away from the long-time yoke of the nation-state.

However, for the Japanese government as well as the U.S. government, it was not acceptable to set a rigid time limit to the land utilization for the U.S. military bases since they shared the same concern that potential security crises still existed in East Asia. From a geopolitical and geo-strategic point of view, both governments thought that China and North Korea could become threats to the Asia-Pacific region and that U.S. military bases located in Okinawa could not be drastically reduced for that reason. This concern was clearly expressed in the redefinition of the Japan-U.S. security relations, that is, the New Guidelines for Japan-U.S. Defense Cooperation of 1997. Although the Japanese government, which was led by the LDP again in 1996, overtly supported the Promotion Policy, it did not make any clear statement regarding whether of not the Action Program could be implemented on time. Therefore, it was doubtful that Ota's policies could be realized as planned.

In terms of the future prospect of Okinawa's economy, Ota's policies had a weakness. Due to his strong reformist stance against the Japanese and U.S. governments, the LDP-led Japanese government stopped providing subsidies and this antagonistic relationship started affecting the material life of the Okinawans. In addition, even though neither the Action Program nor the Promotion Policy addressed Okinawa's political independence, they sought its "relative" independence as seen in Hong Kong. This prospect, however, was too optimistic. As the text of the Promotion Policy points out, the decline of Japan's economy was becoming clear in the second half of the 1990s. Accordingly, Okinawa's economy was weakened, meaning that it was firmly integrated into Japan's. Approaching was the time Okinawans needed to make a decision for a new direction of Okinawa's politics, which was the next gubernatorial election in 1998.

In the following sections, according to the above-mentioned political environment of Ota's terms, this study examines how Okinawans expressed their grievances by focusing on the spatio-temporal development of protest actions and reformist votes. An emphasis is placed on describing the spatio-temporal differentiation of those political behaviors and analyzing how the rise of collective action affected voting at which geographical scale.

Collective actions in the 1990s

Data and methodology

Since there are no official statistics for collective actions such as rallies in Japan, this study used articles of a local newspaper in Okinawa (the Okinawa Times) to gather information about collective actions (rallies, public meetings, demonstrations, marches, and sit-ins) protesting the U.S. military presence and Japan's related politics. The number of collective actions thus collected in the 1990s is 300. Each collective action was coded with regard to the following six components of protest action: year, object or issue, organizer, place of protest, site of protest, and size (Table 2). "Year" is the year when the protest action took place. "Object or issue" is the object or issue over which the protest action was organized. This component was classified into eleven categories. "Organizer" is the organizer(s) of the protest action consisting of eleven categories. The categories in this component combine the kind of protest organization (six sub-categories) and its scale of activity (five sub-categories). According to the name of the organization and the contents of the newspaper article, the scale of activity was classified into prefecture scale, sub-prefecture scale, the combination of the two, local scale, and others. "Place of protest" is the municipality where the protest action took place while "site of protest" distinguishes on-site protest from an off-site one. Off-site protest was further classified into six sub-categories according to the characteristics of the site of protest such as capital or local city. "size" is the number of participants in the protest action that was reported in the newspaper article. The Okinawa Times often uses the number announced by the organizer of the protest action. The number thus reported tends to be exaggerated compared to the number announced by the local police department. For this reason, this study employs the following categories: 0-99, 100-999, 1000-4999, 5000-9999, 10000-50000, and more than 50000.

If we define a general trend of collective action as the category whose component ratio is more than 10% (30 cases), protest actions have the following characteristics:

As explained in the previous section, political processes over Okinawa became complicated after 1995. In addition to the traditional protest against military maneuvers, new issues such as human rights violations (sexual assaults) and the removal of U.S. military bases induced protest actions in the second half of the 1990s (Figures 5a and b). Since these actions were mostly a reaction against the conducts of the U.S. military force or the decisions of the Japanese and U.S. governments, their rise and fall were not parallel to Okinawa's economy (Figure 3). Rather, political opportunity structure such as the continuation of the reformist prefecture government and the decline of the LDP in the national Diet seems to have contributed to the rise of protest actions.

This type of summary, however, is too general and does not represent the actual pattern of collective action because the cross tabulation breaks up the "meaningful" links among the six inter-linked components. These links are preserved in each case of protest action. It is thus necessary to use a different way of summarizing the data set without breaking up the inter-component links. For this reason, this study used a multivariate statistical analysis called Quantification Theory Type-3 (hereafter referred to as QT3) formulated by Chikio Hayashi (Hayashi 1950; see also Yamazaki 1994). In order to use QT3, the data set of protest actions needed to be reconstructed into a 0/1 binary data matrix in which each of the 300 cases (called items in QT3) had six 1s in the above-mentioned six components. Therefore, the characteristics of each protest action were transformed into (reduced to) the arrangement of six 1s. The similarity of the characteristics among protest actions (items) is interpreted as that of the arrangement of six 1s. The meaningful links among categories can thus be preserved.

QT3 mathematically rearranges the order of the matrix so that similar items and the related categories can be located as closely as possible in the newly arranged matrix (for illustration, see Yamazaki 1994). In this process, QT3 gives a numerical value (called score in QT3) to each category and a correlation coefficient between rearranged items and categories to the new matrix. By reiterating this process, QT3 shows possible rearrangements of the original matrix with their correlation coefficients. According to the correlation coefficients, statistically significant matrixes can be chosen. Each matrix thus chosen has a newly given score for each category. As mentioned above, the meaningful links among categories are converted into the scores, meaning that categories with similar scores can be interpreted as having meaningful links to each other. In QT3, such a rearranged matrix constitutes an axis. Each axis contains all the categories with new scores (see Table 3). Another function of this axis construction is to separate unique categories on an axis since the unique categories are given higher scores (larger absolute values). By constructing a coordinate space with different axes, each category can be located at a particular point in the coordinate space. Therefore, a group of the categories having meaningful links (i.e. sharing similarities) appears as a cluster in the coordinate space. Hierarchical cluster analysis can be applied to the axes with statistically significant correlation coefficients in order to group categories according to the distance among them.

As a result of the analysis by QT3 (Table 2), this study employed Axis 1 to 4 as statistically significant (more than 0.6) and conducted hierarchal cluster analysis by the group average method. The results of the cluster analysis are shown in Figure 6. According to the distance between clustering stages, the optimal step was found when sixteen clusters were created. The second significant step appeared when nine clusters were formed. Thus, this study examines these nine clusters and the sixteen clusters as their sub-clusters. The characteristics of each cluster are explained below.

Cluster A: on-site protest against military presence

This cluster consists of three sub-clusters (A1, A2, and A3). According to the original data set based on newspaper articles, cluster A represents on-site protest action against military maneuvers (A1 and A2) and the arrival of atomic submarines (A3). On-site meeting or demonstration is the most common repertoire of protest in Okinawa (171 items). Cluster A1 shows small-scale on-site protest by PU (labor-union related SMOs active at the prefecture scale). This type of protest action was typically seen at Kin and Onna in the early 1990s. Since PU is the most dominant category as organizer, cluster A1 represents one of the typical collective actions in Okinawa. On the other hand, cluster A2 indicates resident/municipality-led local protest at Yomitan, which was not necessarily common in the 1990s in Okinawa. Although a single type of SMO did not lead protest actions at Katsuren, labor unions were often involved.

Cluster B: major places of protest in the central region

Cluster B includes the largest number of categories (19) but only one category of issue (National events). Three major cities in the central region of Okinawa Island and seven categories of year are grouped in this cluster. Kadena, Ginowan, and Urasoe often became places of protest through the 1990s (27, 24, and 9 items respectively). Otherwise, this cluster does not have clear characteristics. The result of cluster analysis by the Ward method did not form this cluster. It can be said from this that categories in cluster B were not unique enough to be significantly separated on the axes with the exception of the three cities.

Cluster C: the political centrality of the prefecture capital

This cluster is formed around Naha. Many protest actions were organized at Naha (77 items) because it is the capital of Okinawa Prefecture. Thus, CAP (capital) and PH (Prefecture Hall) are included in cluster C. NDFAB (Naha Defense Facility Administration Bureau) is included for the same reason. Legal issues such as land for military use (Land) and the Japan-U.S. security treaty or agreement (Agreement) are specialized in Naha. Women's groups (WG) often organized protest actions in Naha. Naha was also used for political rallies with a large size of participants (5000-9999). In short, cluster C demonstrates that the political centrality of the prefecture capital attracted various kinds of protest actions.

Cluster D: off-site mobilization at local cities

In cluster C, Human rights (human rights violation) and Referendum (prefecture and city referenda) were added to a sub-cluster (C1). C1 represents protest actions at local cities, especially remote cities in the Sakishima region such as Hirara and Miyako. These protest actions were usually organized by labor-union related SMOs active at the sub-prefecture scale (SPU). Since there is no U.S. military base in the Sakishima region, on-site protest was rare. So was it, too, in Okinawa City (coded Okinawa) despite the existence of the Kadena Air Base and its population size (115,336). According to the original data set, the links between C1 and the other two categories cannot be considered significant. The component ratios of Sakishima and Okinawa as the places of protest are 6.7% (20 items) and 2% (6 items) respectively.

Cluster E: new localized protest against the intra-prefecture removal of military bases

Cluster E represents protest actions against the removal of the Futenma Air Station to Nago near the end of the 1990s. There has never been an intra-prefecture removal of an U.S. military base in Okinawa. After Ota was defeated by his conservative opponent in 1998, the new governor decided to accept the removal in 1999. The governor's decision went against the result of the Nago City referendum in 1997. This shift of prefecture politics caused new localized protests at Nago (14 items in 1999). Protest actions at Nago were often led by the new type of SMO combining political parties, labor unions, and citizens' groups active at the sub-prefecture or local scale (SPO). This is a new trend of collective action in Okinawa.

Other clusters

With the exception of cluster F, the rest of the clusters found in this analysis do not represent meaningful links among categories. The reason cluster F was formed is that MCBH (the U.S. Marine Corps Bases Headquarters) is located in Kitanakagusuku.

Protest actions in Okinawa

As the results of cluster analysis, protest actions in Okinawa have the following features. First, on-site protest actions against military presence were fairly common (171 items) and more than half of them were led by labor-union related SMOs (100 items). On-site protest actions led by local residents or municipalities were fewer (43 items). Labor-union related SMOs were more politically reactive and sensitive than local residents or authorities to the problems directly emerging from U.S. military bases. In other words, many of these actions may not have been firmly rooted in the localities. The case of Yomitan, however, is in contrast to that tendency. In both cases, the size of participants was small (mostly fewer than 100).

Second, there were various kinds of protest actions within Okinawa Island through the 1990s. The predominant places of protest were Naha (77 items), Kin (38), Onna (26), Kadena (27), Ginowan (24), and Nago (25). Naha attracted many off-site protest actions due to its political centrality as the prefecture capital while Kadena and Ginowan were the major places of on/off-site protest against the vast U.S. military bases in the central region. The size of participants in these cities was sometimes very large (more than 10,000). As previously mentioned, labor-union related SMOs organized on-site protest against military maneuvers at Kin and Onna in the early 1990s. Nago became one of the major places of protest at the end of the 1990s. According to the varying political environment of the 1990s, protest actions in Okinawa were spatio-temporarily differentiated.

Third, local cities in the Sakishima region were not so politically active as in the other part of Okinawa due to the lack of U.S. military bases. However, labor-union related SMOs organized protest actions in the cities. These SMOs were in most cases the local branches of prefecture-wide SMOs. Nago, the central city in the north region, had also been politically less active until the issue of the removal of the Futenma Air Station came up in the late 1990s. Protest actions organized by new locally based SMOs took place in the previously quiet city. The existence or creation of these local SMOs allowed such remote cities to be politically mobilized although the size of participants tended to be small (mostly fewer than 1000).

In conclusion, the above-mentioned features constituted protest actions in the 1990s in Okinawa. In order to understand the protest actions in the larger political context, it becomes necessary to examine voting as another important political process. Grievances against the U.S. military presence were expressed not only through collective action but also through voting. In the next section, this study analyzes the spatio-temporal pattern of reformist votes as an expression of Okinawans' political preferences.

Reformist votes in the 1990s

General trends

This study defines reformist votes as votes cast for reformist candidates, parties, or policies. Reformist in this study means the candidates, parties, or policies that opposed the Japan-U.S. security relations due to their unconstitutionality and militarism and attempted to take measures to eliminate U.S. military bases from Okinawa. In this sense, reformist votes do not necessarily mean the votes cast by reformist voters but the votes mobilized by reformist politicians or political actors. Before the 1972 reversion to Japan, political parties in Okinawa had been affiliated with those in mainland Japan with the exception of the Okinawa Social Mass Party (SMP). In the 1990s, ideologically reformist (leftist) parties were the JCP, SDPJ, and SMP while ideologically conservative parties were the LDP, the Japan Renewal Party, and the New Frontier Party. Parties such as the Komei Party (Komei), the Japan New Party, and the Democratic Party of Japan were centrist and usually supported reformist candidates in the gubernatorial elections. When these centrist parties supported reformist candidates in the gubernatorial elections, this study regarded the votes for the parties in other elections as part of reformist votes. Therefore, the reformist votes counted in this study includes some centrist votes, most of which was for Komei.

Using newspaper articles in the Okinawa Times, this study collected voting data for the following nine elections and one referendum: the gubernatorial elections in 1990, 1994, and 1998, the national assembly elections for the House of Representatives in 1990, 1993, and 1996, the national assembly elections for the House of Councilors in 1992, 1995, and 1998, and the prefecture referendum in 1996 (Table 4). As mentioned above, since reformist votes were cast for candidates, parties, and policies, the spatial patterns of the votes varied from one election to another. For example, the ratio of reformist votes was high in the 1995 election for the House of Councilors in which two reformist candidates ran for the election. Reformist votes were also predominant in the 1996 referendum which was boycotted by some amount of conservative voters and whose result had no legal authority to change prefecture policies. Thus, reformist votes were not necessarily cast by a particular group of voters called reformists but were mobilized by various factors.

As Table 4 shows, reformist votes seem to have been mobilized in the middle of the 1990s (1994, 1995, and 1996). Voter turnouts during the period, however, were not necessarily high. Although the high ratio of reformist votes in the 1996 referendum was partially the result of a series of protest actions, voters as a whole were not necessarily mobilized by these opportunities to vote. After this apparent increase of reformist votes, Ota was defeated in the 1998 gubernatorial election at the lowest ratio of reformist votes (47.2%) in the 1990s. Why did reformist votes change in such a way?

Assumptions for analysis

In order to analyze the spatio-temporal development of reformist votes, this study used multiple regression analysis for 53 municipalities (observations). Using the Okinawa Prefecture Statistical Yearbook (Okinawa Prefecture various years), this study examined what kind of socio-economic and regional variables could better explain the variance of reformist votes. Among the aggregate data collected from the Yearbook were demographic indexes, per-capita income, gross municipal products, employment rate, employees by industry, subsidies, revenue and expenditure, and the area of U.S. military bases. After the examination of the significance and malticollinearity among those candidate variables, this study finally employed the following indexes as independent variables:

Selected independent variables are as follows (see also Table 4):

The society of Okinawa is often said to be dependent on three Ks: kichi (military bases), kokyo-jigyo (public works), and kanko (tourism). The three Ks illustrate that Okinawa has been dependent on external income sources. Thus, the regression equation should have been devised so as to fully specify this economic dependency. This study, however, was unable to create a variable for tourism due to the lack of the appropriate census data for the tourist industry. Finally, assuming that reformist votes were affected by the degree of economic dependency of the municipality where voters lived, this study chose INCOME, USBASE, CONST, and DEVEX. These variables represent income level, the existence of U.S. military bases, the status of the construction industry, and the size of public works respectively. There was no significant multicollinearity among these variables. USBASE, CONST, and DEVEX partially indicate how each municipality constructed its built environment such as infrastructure using developmental subsidies. DEVEX is highly correlated with the ratio of external subsidies in the revenue of each municipality. Thus, the three variables have geographical implications.

As mentioned above, the existence of U.S. military bases in Okinawa has had ambivalent effects on Okinawans' political behavior. U.S. military bases have been the sources of income as well as problems for Okinawans. External income has been brought to Okinawa not only from U.S. military bases but also from the Japanese government as developmental subsidies. The latter income has been much bigger than the former.10 Due to the historical burden and vast U.S. bases that Okinawa has had to bear, the Japanese government has been obliged to support Okinawa's development but, at the same time, has maintained its dependency on mainland Japan. As a result, public works subsidized by the Japanese government have become one of the key elements in this "dependency trap." Another key element is the construction industry which benefits directly from such public works. Therefore, USBASE, CONST, and DEVEX represent the effects of the dependency trap on reformist votes. In other words, if voters wish to continue to receive state subsidies by maintaining U.S. military bases, they will not support the reformist attempts to reduce or eliminate the bases. If reformist policies do not provide material or non-material substitutes for reduced income sources and if voters employ the "instrumentally rational" calculus, reformist political idealism will lose its basis.

Another important assumption in this analysis is that reformist votes are spatially clustered. U.S. military bases occupy 20-25% of the central and north regions of Okinawa Island and their political economic impacts are not spatially even (Figure 2 and Table 1). There are few or no U.S. military bases in other regions. Voters in such diverse regions are not likely to be mobilized in a uniform way. In order to examine this regional differentiation of reformist votes, four regional dummy variables were included in the regression equation. Excluded were the islands (11 municipalities) geographically isolated from Okinawa Island and Sakishima Islands. The reasons for including the regional dummy variables are threefold.

First, in the regression analysis of geographical data, it is necessary to take into account the existence of spatial autocorrelation. Since it has been commonly known that reformist votes are stronger in the central region, it was anticipated that some variables in this analysis were spatially correlated in that region. Instead of excluding or eliminating the effect of spatial autocorrelation, this study attempted to reveal the effect by including regional dummy variables in the regression equation. If some of the regional dummy variables are significant, somehow "regional" effects on reformist votes may exist in Okinawa.

Second, the regression equation thus formulated represents the effects of the dependency trap and sub-prefecture regions on reformist votes. In other words, the equation assumes that reformist votes are affected by the economic dependency of each municipality and by the geographical "context" in which the municipality is located. The concept of such context has been discussed in political geography (Political Geography 1987, 1996). A shared understanding of the concept is that there are local geographical settings that direct the people living there toward a particular political behavior (e.g. voting for a reformist candidate). This study attempts to show the spatio-temporal development of reformist votes using the regional dummy variables and to indicate the contextual effect on reformist votes. However, the statistical significance of the regional dummy variables does not directly mean that the contextual effect exists but that further examination is necessary to clarify its existence.

Finally, the inclusion of the regional dummy variables significantly improved the adjusted R2s of the regression equations in 1990, 1992, 1993, 1994, 1995, 1996 (the national assembly election), and 1998 (the gubernatorial election) (Table 5). Thus, the regional differentiation of reformist votes needs to be considered and specified.

Results of regression analysis

The results of the multiple regression analysis show that different variables become significant at different times while some variables are consistently significant (Table 5). INCOME is significant only twice in 1990 (the national assembly election) and 1993 and becomes insignificant in later years. Its positive sign indicates that higher income contributed to the increase of reformist votes. On the other hand, DEVEX is significant in 1994, 1996 (the national assembly election), and 1998. DEVEX is high for the municipalities with high expenditure for public works and high dependency on external subsidies, suggesting that reformist votes did not increase in such municipalities in the second half of the 1990s. CONST represents not only one of the most important industries but also the most anti-reformist sub-group in Okinawa. CONST is significant with a minus sign in 1990 (the gubernatorial election), 1992, 1993, 1994, 1996 (the prefecture referendum), and 1998. In most cases, the coefficient of CONST is smaller (i.e. its absolute value is larger) than other negative coefficients. Since the construction industry has been dependent on subsidized public works, reformist policies seeking economic self-supportiveness have been threats to the industry's income source. Another fact to note is that USBASE is significant with a minus sign in 1990 (the gubernatorial election), 1992, 1993, 1994, and 1996 (the prefecture referendum). Unlike conventional wisdom, the existence of U.S. military bases did not necessarily contribute to the increase of reformist votes in the first half of the 1990s. Even in the second half, USBASE continues to have a minus sign although it is statistically insignificant. As previously mentioned, the meaning of U.S. military bases is not straightforward but ambivalent for the voters living close to the bases. Their lives are interwoven with the bases in a complex way. This complexity of the meaning of U.S. military bases will be discussed below.

With regard to the regional dummy variables, CENTRAL is significant in 1990, 1992, 1993, 1994, 1995, 1996 (the national assembly election), and 1998 (the gubernatorial election). The coefficient of CENTRAL is almost always larger than those of the other regional dummy variables, meaning that the central region most contributes to the increase of reformist votes. One exception is the 1996 prefecture referendum in which the coefficient of SOUTH is larger than those of the other regions, but the inclusion of the regional dummy variables in the equation is insignificant. It can be said from this that the central region has been a stable basis for reformist votes. However, unlike the rise of collective action after 1995, the effect of such regional differentiation on reformist votes have decreased after 1996. Furthermore, in the 1998 gubernatorial election in which Ota was defeated, the reformist votes in Okinawa Island were significantly lowered compared to the other islands.11

With regard to the relationship between protest actions and reformist votes, let us take a look at some municipalities. As I have examined, there were concentrations of protest actions on Naha (71 items), Kin (38), Kadena (27), Onna (26), Nago (25), and Ginowan (24). With the exception of Nago which became the major place of protest in 1999, neither Naha, Kin, Kadena, nor Onna was significantly mobilized for reformist votes (Figure 7). Possible reasons for this are that protest actions in those places were mostly organized by the SMOs which were not related to the voters living there and that smaller sizes of on-site protest did not necessarily have an impact on over all the municipality where it took place. Unlike these places, locally based protest actions were seen in Yomitan (11 items) in which the ratio of reformist votes was usually high. Ginowan had been the site of various kinds of protest actions through the 1990s and shows high scores of reformist votes. Furthermore, as far as these cities are concerned, the post-1995 uprisings do not seem to have a significant impact on reformist votes at the sub-prefecture or local level.

To the contrary, the result of the 1996 referendum revealed the prefecture-wide support of Ota's reformist policy. Since the referendum had no legal power to change policies, voters were able to show their political preferences more straightforwardly. The spatial pattern of the referendum votes has a striking contrast with other elections (Figures 8a to c). Unlike the 1995 and 1996 elections, the referendum has a very unique voting pattern (Figure 8b) in which the south, north, and Sakishima regions were more mobilized for the reformist policy than the central region. As mentioned previously, the patterns seen in the two elections (Figures 8a and c) are the "normal" one with higher scores of reformist votes in the central region.

The results of the 1996 referendum are also very paradoxical. The referendum was the very first formal occasion for Okinawans to express their accumulated grievances. It attracted media attention from all over Japan. The question was fairly simple; do you agree or disagree with the Japan-U.S. Status of Force Agreement and the reduction of U.S. military bases? The question indirectly asked if Okinawans would be able to endure the problems caused by the U.S. military presence such as the 1995 rape. Although some amount of conservative voters boycotted the referendum, the referendum was successful in revealing Okinawans' political preferences. However, the reformist votes in the central region were not so mobilized as in the other regions. The normal voting pattern did not appear and usually conservative regions were mobilized for the reformist policies. This is probably because floating votes in the relatively conservative regions were mobilized while the voters in the central region were more careful about the results of Ota's policy. The Okinawa Times explained this paradox by saying, "While voters in the municipalities with military bases expressed their objection to the consolidation or reduction [of the bases], they seem to have wavered due to the economic factors of employment and income form the bases" (the Okinawa Times 9/9/1996: 3).

As mentioned before, right before the referendum, Ota clearly expressed his reformist vision of Okinawa in the Action Program for the Reduction of U.S. Military Bases and the New Industrial Promotion Policy for Cosmopolitan City. However, Ota's vision does not seem to have been so persuasive to the voters in the central region who had to face the ambivalent reality of the U.S. military presence. On the other hand, under the political atmosphere following the 1995 rape, voters in the other regions seem to have been mobilized more easily. In addition, they were not so dependent on the U.S. military bases as to worry about the direct results of Ota's policy. It can be said from this that the "abnormal" results of the 1996 referendum were caused by the temporary political attitudes of the voters in the other regions. Instead, the voters in the central region had to make a very delicate political decision between political idealism (pacifism) and economic realism (dependency). To that extent, reformist votes were "fixed" in the central region; the swing of the votes was smaller there.

Unless this critical position is shared by Okinawans as a whole, the mobilization of reformist votes will be localized and inconsistent due to the spatio-temporarily differentiated impacts of the U.S. military presence. If riding on globalization does not make any realistic alternative to economic dependency and if Okinawa's economy keeps declining, the basis of reformist votes will continue to be undermined. I argue that this is the logic behind the replacement of Ota in 1998.

Conclusion: after the reformist era

According to the results of this study, as protest actions against the U.S. military presence increased after the 1995 rape, Ota framed a new grand design of Okinawa to promote the reduction of U.S. military bases and a new industry based on globalization. In order to achieve these political goals, Ota needed to gain the voters' support. However, reformist votes did not increased so much as protest actions did. At the sub-prefecture or local level, the analysis by QT3 indicates that the geographical concentration of protest actions did not have a direct impact on the increase of reformist votes with the exception of a few municipalities. In voting, Okinawans seem to have made political decisions more carefully. The regression analysis in this study shows that economic realism affected Okinawans' voting behavior. In Okinawa, economic dependency tended to hinder the increase of reformist votes. In other words, while collective action was an important political channel for Okinawans to directly reveal their grievances and political ideals, voting represented a more careful consideration of Okinawa's reality. Ota was successful in mobilizing Okinawans for political idealism through the lawsuit over land for military use, the prefecture referendum, and the new grand design of Okinawa, but unsuccessful in providing a realistic alternative to Okinawa's economic dependency on mainland Japan under the nation-wide economic recession.

After a conservative governor replaced Ota in 1998, the Japanese government resumed provision developmental subsidies for Okinawa, and municipalities such as Urasoe, Kin, and Nago started to offer their land for the new U.S. military bases removed from other part of Okinawa. The rewords for accepting the bases were developmental subsidies from the Japanese government; economic realism started to prevail. Although riding on globalization is still a keyword for economic self-supportiveness, the keyword continues to be idealistic under the serious economic recession of Japan's economy. In addition, Okinawa's geopolitical role as a keystone in the Asia-Pacific region does not seem to change. Under the circumstances of such a dependency trap, Okinawa still shows the weakness of local protest and the strength of the nation-state in the Japanese context. However, from a different point of view, Okinawans seem to have more than one option for their future. The swings in their political behavior between political idealism and economic realism and between reformist and conservative ideologies are the results of their thoughtful choices. If globalization can provide Okinawans with a realistic alternative to economic dependency, reformist votes will be mobilized again.


Notes

1 Okinawa's voter turnouts (%) in the recent national assembly elections are 78.5 (73.3 in Japan) in 1990, 58.5 (50.7) in 1992, 70.8 (67.3) in 1993, 55.3 (44.5) in 1995, 56.8 (59.7) in 1996, and 59.0 (58.8) in 1998.

2 The terms (years) of successive governors are 8 (Yara, reformist), 4 (Taira, reformist), 12 (Nishime, conservative), 8 (Ota, reformist), and 4 (Inamine, conservative incumbent as of 2002).

3 SOFA was concluded in 1960 to determine the status of the U.S. military force stationed in Japan. SOFA provides for Japan's obligation to offer facilities and areas to the U.S. military force, the U.S. authority to control its military installations, and the jurisdiction over U.S. military servicemen. The jurisdiction over the U.S. servicemen who commit a serious crime has been a focal point of protest actions in Okinawa. For Okinawans, SOFA unfairly protects the status of the U.S. military force and servicemen.

4 According to the law, in order to allow the U.S. military force to use the land of Japanese citizens, the Japanese government leases the land and offers it to the U.S. military force. If Japanese landowners refuse to rent their land, the Japanese government has the legal power to force them to do so. In this coercive process, the head of the pertinent local government must sign the land lease contracts instead of the landowners. Since Ota refused to sign as the head of Okinawa Prefecture, the Japanese government sued him for violating the law.

5 The results were as follows: affirmative votes amounted to 88.0%; negative votes 8.24%; and voter turnout 59.5%.

6 Base-related economy in Okinawa consists of base-related employment (8,400 workers in 1999), construction, rent, national subsidies, and other economic impacts. The ratio of base-related consumption, employment, and rent in the Gross Prefecture Expenditure is estimated as 8.1% in 1997 (Okinawa Prefecture 2000a: 45).

7 The municipal developmental expenditures consist of those for projects in civil engineering, agriculture, forestry, fishery, commercial, and manufacture.

8 Since the formerly single electoral district was divided into three in the 1993 election, votes for proportional representation (votes for parties, not candidates) were used.

9 The votes for the candidate affiliated with Jiyuu Rengou (the Liberal League) were counted as reformist votes since the party supported Ota.

10 The average ratio of such external subsidies in the revenue of each municipality is 30.9% in 1999 (Okinawa Prefecture 2000b: 306-307).

11 Ota was born in Kume Island which is one of the isolated islands, the ratio of reformist votes tended to be higher in those islands in the gubernatorial elections. These islands are excluded from the regional dummy variables to avoid perfect collinearity.


References

Agnew, J.A. 1987. Place and Politics: The Geographical Mediation of State and Society. Allen & Urwin, Boston.

Arasaki M. 1996. Okinawa gendaishi (Contemporary History of Okinawa), Iwanami-shoten, Tokyo.

Hayashi C. 1950. On quantification of qualitative data from the mathematico-statistical point of view. Annals of the Institute of Statistical Mathematics 2: 35-47.

Idaka H. 2001. Soto no Okinawa: Aidentiti kuraishisu (Bicephalous OkinawaIdentity Crisis), Gendaikikakushitsu, Tokyo.

Nakano Y. and Arasaki, M. 1976. Okinawa sengoshi (Postwar History of Okinawa), Iwanami-shoten, Tokyo.

Okinawa Prefecture 2000a. Okinawa no beigun oyobi jieitai kichi (The bases of theU.S. military force and the Self Defense Forces in Okinawa), Okinawa Prefecture.

Okinawa Prefecture 2000b. Okinawa-ken tokei nenkan (The Okinawa Prefecture Statistical Yearbook), Okinawa-ken.

Okinawa Prefecture various years. Okinawa-ken tokei nenkan (The Okinawa Prefecture Statistical Yearbook), Okinawa-ken.

Okinawa Prefecture Government 1997. New Industrial Promotion Policy for Cosmopolitan City Formation. Available at www.pref.okinawa.jp/97/FTZ/kokutoshi/index-e.html (accessed March 2, 2002)

Okinawa Times 1990. Soko no zen yato kyoryoku (The successful joint struggle of opposition parties). November 19.

Okinawa Times 1990. Nishime kensei 12 nen no kiseki (The twelve-year tracks of Nishime's prefecture policies). November 24.

Okinawa Times 1994. Wan pointo seisaku (Brief reviews of policies). November 20.

Okinawa Times 1994. Ota-shi taisa de tosen (Mr. Ota won the election by a huge margin). November 21.

Okinawa Times 1995. Kichi shukusyo shi kyotei minaose (Reduce military bases and revise the agreement). October 22.

Okinawa Times 1996. Nijimu datsu-kichi eno negai (Appearing wishes to eliminate military bases). September 9.

Political Geography 1987. Political geography debates No 1: place, context, and voting. Political geography Quarterly 6(1): 7-52.

Political Geography 1996. Various articles. Political Geography 15(2): 129-167.

Yamazaki T. 1994. Political geography in post-war Japan: publication tendencies. Political Geography 16(4): 325-344.