A draft of this paper was presented during the session “Natural resource management and environmental governance” at the 2019 annual meeting of The Japanese Society for Public Administration. I would like to thank Yoshiaki Kubo for organizing the session and Michiya Mori, Haruka Kubo, and Toshinori Tanaka for their insightful comments and presentation. Most of the interviews in this study were conducted together with Gaku Mitsumata, Daisaku Shimada, and Kazuki Kagohashi. I would also like to thank Kohei Yokomachi for his research assistance. This study was supported by JSPS KAKENHI Grant Number JP16K16236 and JP19K12458.
1Environmental governance has been studied for several decades across a wide range of disciplines, including political science, economics, sociology, and geography (e.g., Bulkeley, 2005; Mol, 2006; Ueta, 2007; Young, 2013). The term “environmental governance” basically refers to a social function that attempts to steer societies in such a way as to promote sustainability and avoid the adverse effects of resource use (Young, 2013, p.10-11). Since environmental governance is a comprehensive concept, it has often been loosely used in many contexts without explicit definition. Recent environmental governance studies argue that the absence of a prevailing analytical framework in this field hinders the systematic accumulation of knowledge about it (Young, 2009, p.37; Tacconi, 2011; Visseren-Hamakers, 2015; Bennett & Satterfield, 2018; Ohno, 2019b; Hegger et al., 2020; Newig & Rose, 2020).
- 1 This keyword search was conducted on January 10, 2021.
2To facilitate the systematic accumulation of knowledge on environmental governance, studies of policy processes—a central theme in environmental governance studies (Ruseva et al., 2019)—offer several beneficial frameworks or theories (e.g., Weible & Sabatier, 2018) based on research into environmental issues. The Advocacy Coalition Framework (ACF) is a typical example of such a framework (Sabatier & Weible, 2007). Although many empirical ACF studies have targeted natural resources and environmental issues (Weible et al., 2009; Pierce, Peterson, Jones, et al., 2017), this has not been explicitly applied in environmental governance literature. According to the SCOPUS database1, only four papers included both “advocacy coalition framework” and “environmental governance” in their title, abstract, and keywords (e.g., Koebele (2019)).
3In addition to their application in environmental governance, applying the ACF outside of Western Europe and North America is a challenge in ACF studies (Henry et al., 2014; Jenkins-Smith et al., 2017). Although the applicability of the ACF—originally developed under a pluralistic political regime in the United States—in other political regimes was initially questioned (e.g., Parsons, 1995), recent studies have identified that the ACF is basically effective in explaining policy processes under a corporatist regime (Nohrstedt & Olofsson, 2016), a relatively recent democratized regime (Jang et al., 2016; Park & Weible, 2018), and an authoritarian regime (Li & Weible, 2019).
4As the ACF’s application to European countries significantly developed the framework (Sabatier, 1998; Sabatier & Weible, 2007), testing the ACF in nontraditional settings will contribute to the development of the framework. As Weible et al. (2016) used the ACF in cross-country comparative studies, testing the applicability in various social and political settings will contribute to developing the ACF as a framework for comparative studies. Although Japan has a relatively long experience with democracy among non-Western countries and pluralistic policy processes to some extent (e.g., Muramatsu & Krauss, 1987), there exist considerably fewer ACF applications in this country. A recent review of Japan’s ACF studies (Ohno et al., 2021) pointed out the need for further hypothesis testing and methodological development in Japan to develop the ACF as a framework for comparative policy process studies. Thus, this study applies the ACF’s policy change hypotheses and research methods to the policy process that leds to the Arase Dam removal, a critical case of major policy change in Japan’s environmental governance.
5Based on the above background, this paper is structured as follows. First, it contains a review of the ACF literature and puts forth a hypothesis on policy change. Second, after reviewing the outline of the Arase Dam removal case, it explains the methods and data used to test the hypotheses. It then presents the results of the analyses of the advocacy coalition, policy change pathways, and coalition resources accordingly. Based on these results, the paper discusses how unconventional policy changes are undertaken under the complex watershed environmental governance and the applicability of the ACF in environmental governance studies and in Japan’s subnational policy process context.
6As with many disciplines in the social sciences, governance has been a major topic in environment and sustainability studies since the 1990s (e.g., Delmas & Young, 2009; Young, 2013; Margerum & Robinson, 2016). Given the wide-ranging topics from market-based governance to global governance (Lemos & Agrawal, 2006; Armitage et al., 2012; Ohno, 2019b) and the interdisciplinary nature of environmental governance studies (Tacconi, 2011), it is difficult to organize the results of these diverse studies under a unified framework. Newig and Rose (2020) characterized environmental governance studies as what Whitley (2000) termed “fragmented adhocracy.” Although many case studies offer useful insights to uncover the dynamics of environmental governance (e.g., Murota & Takeshita, 2013), it is difficult to find a unified framework that organizes individual case knowledge in a systematic manner within current environmental governance scholarship. Some scholars have proposed general frameworks to understand environmental governance (e.g., Vatn, 2015; Bennett & Satterfield, 2018). Nevertheless, these frameworks have not been frequently used in other studies. The fragmented streams of the environmental governance literature stem from its interdisciplinary nature (Tacconi, 2011). In addressing such fragmented literature, “framework thinking” (Ostrom, 2005) is beneficial. Ostrom (2005) distinguishes a framework, a metatheoretical overall picture of a certain theme across various disciplines, from a model or theory, with a specific hypothetical relationship between variables. A metatheoretical framework can organize various findings about environmental governance across various disciplines. Thus, it will be a vital academic contribution to identify the usefulness and challenges of the framework in environmental governance studies.
7The ACF has developed several distinctive concepts to elucidate the mechanisms of major policy changes. It targets the policy subsystem rather than specific government agencies or actors (Sabatier, 1988; Sabatier & Weible, 2007; Jenkins-Smith et al., 2017). The policy subsystem concept has an advantage for researchers in environmental governance studies. For instance, similar to most environmental issues, the watershed does not fit existing government boundaries and involves many government agencies and stakeholders at multiple levels. A policy subsystem includes diverse agencies and actors involved in environmental governance.
8The ACF assumes that policy actors have belief systems consisting of three tiers: deep core beliefs, policy core beliefs, and secondary aspects (Sabatier, 1988; Jenkins-Smith et al., 2017). The ACF hypothesizes that human beings simplify the complexity of reality by filtering through these belief systems and act accordingly to realize their beliefs (Sabatier & Weible, 2007, p. 194; Jenkins-Smith et al., 2017, p. 140). Given that environmental issues are highly complex and uncertain policy issues, focusing on actors’ belief systems is important in environmental governance studies.
9Advocacy coalitions—another major concept in the ACF—are aggregates of actors who share similar beliefs and engage in nontrivial degrees of coordination in a policy subsystem (Sabatier & Weible, 2007, p. 196). Aggregating many actors into a few advocacy coalitions is useful for understanding complex dynamics in the policy subsystem (Sabatier & Weible, 2007, p. 196; Jenkins-Smith et al., 2017, p. 141).
10The ACF discusses two policy change hypotheses (Table 1) that assume five pathways to policy change.
Table 1. Two Policy Change Hypotheses in the ACF
Policy Change Hypothesis 1 (Bottom-up Policy Change)
Significant perturbations external to the subsystem, a significant perturbation internal to the subsystem, policy-oriented learning, negotiated agreement, or some combination thereof are necessary, but not sufficient, sources of change in the policy core attributes of a governmental program.
|
Policy Change Hypothesis 2 (Top-down Policy Change)
The policy core attributes of a government program in a specific jurisdiction will not be significantly revised as long as the subsystem advocacy coalition that instated the program remains in power within that jurisdiction—except when the change is imposed by a hierarchically superior jurisdiction.
|
Source: Jenkins-Smith et al. (2017); Pierce et al. (2017)
11The first hypothesis, the so-called “bottom-up policy change,” postulates that one of four pathways—external perturbation, internal perturbation, policy-oriented learning, and negotiated agreement—or some combination thereof is a necessary source of policy change (Jenkins-Smith et al., 2017; Pierce, Peterson, Jones, et al., 2017). External perturbation refers to changes in socioeconomic conditions, changes in a governing coalition, and disasters on which actors in a policy subsystem cannot exert their influences (Jenkins-Smith et al., 2017). Conversely, internal perturbation refers to policy failures or scandals within a policy subsystem (Jenkins-Smith et al., 2017). Policy-oriented learning refers to the alteration of actors’ policy beliefs (Sabatier, 1988, p.133). Negotiated agreement refers to reaching agreements that involve policy core changes among previously warring coalitions (Sabatier & Weible, 2007).
12The second hypothesis, the so-called top-down policy change, postulates that impositions by a hierarchically superior jurisdiction lead to major policy changes (Jenkins-Smith et al., 2017; Pierce, Peterson, Jones, et al., 2017). This hypothesis evokes the role of veto players (Tsebelis, 2002), whereby the institutional arrangement of veto points in a policy subsystem or actors’ policy preferences therein can influence major policy change. In contrast to policy change hypothesis 1, only a few studies have been conducted on policy change hypothesis 2; for instance, Ellison (1998) argued that support by the government influences power balances among coalitions, Elliott and Schlaepfer (2001) and Kübler (2001) observed that policy did not change until governing coalitions had changed, and Li and Wong (2019) observed that policies did not change until the leader in a one-party system had changed.
13While the above five pathways are independent variables that lead to policy change, they do not directly lead to policy change on their own. Instead, there are intervening variables such as coalition resources (Jenkins-Smith et al., 2017). Coalition resources include formal legal authority, public opinion, information, mobilizable supporters, financial resources, and leadership (Sabatier & Weible, 2007). Several past ACF studies have studied coalition resources (Weible, 2007; Ingold, 2011; Nohrstedt, 2011; Elgin & Weible, 2013); for instance, Nohrstedt (2011) argued for the relative importance of formal legal authority among coalition resources.
14While the Arase Dam was the first large dam removed in Japan, many dams, including smaller ones, have been removed around the world (Bellmore et al., 2017; Sneddon et al., 2017). The ACF has been applied in the policy process studies of these dam removals. Lowry (2003) discussed that dam removal as a major policy change occurs when external factors support policy change, political receptivity is high, and the physical complexity is low. The importance of external factors corresponds to that of external perturbation in the ACF. Although Lowry’s framework offers a useful lens for understanding the position of individual cases, it offers fewer variables and hypotheses than the original ACF. Thus, difficulties can arise when trying to understand the dynamic interactions among actors, events, and policy changes within a case. Brewitt (2019) also examined the policy process leading to dam removal by referring to the ACF; however, he only referred to some components of the ACF to guide case studies. Concerning the ACF, Brewitt formed the hypothesis that “stakeholders must form mega-coalitions in order to win political backing, raise money, and stave off opposition” (Brewitt, 2019, p.186); however, he did not test it.
15Japan’s policy process has often been characterized by relatively strong bureaucratic power (Iio, 2007), long-term single-party dominance in the Diet (Muramatsu & Krauss, 1984), and somewhat closed political opportunity structures (Schreurs, 2003). Its river policy process has long been characterized as a strong bureaucracy with technical expertise (Takahasi & Uitto, 2004; Kada, 2006; Aldrich, 2010; Chakraborty, 2013). In particular, Japan’s dam policy is the least likely case of major policy change because it has often been regarded as unchangeable due to the closed policymaking created by the iron triangle of politics, bureaucracy, and the construction business (Igarashi, 1999; Ueta, 2007). Japan has almost 2,700 large dams and several new large dams are in the planning stage of construction (Nihon Dam Kyokai, 2021). Although some dam construction projects have been halted (Mita, 2010; Sunahara, 2011), dam removal was not regarded as a realistic policy option several decades ago, just as was the case in the United States (Brewitt, 2019).
16Although several small dams and one large dam have been removed in Japan (Mizoguchi & Kuraji, 2006; Noda et al., 2018), the Arase Dam removal is the first case in terms of its large size, environmental purpose, and location in the mainstream of a major river. This dam, which is 25 m high and 210.8 m wide, was a single-purpose dam built for hydropower generation by the Kumamoto Prefectural Government (KPG) in 1954. It was built in the former Sakamoto Village along the Kuma River in southern Kumamoto Prefecture. The Kuma River, a major river in Kumamoto Prefecture, is 115 km long, with nearly the entire watershed located in the Kumamoto Prefecture jurisdiction. The livelihoods of the people residing in the watershed were closely associated with the river; for instance, some locals made their living via Ayu (sweetfish; Plecoglossus altivelis) fisheries.
17The policy subsystem, a fundamental scope of this study, is characterized by the Kuma River watershed, a geographical dimension, and the policies concerning the Arase Dam removal, a policy dimension. Such an operationalization of the policy subsystem can be regarded as too narrow. However, because dam removal is an emerging policy issue (Sneddon et al., 2017), it can be assumed that its policy subsystem is in a nascent stage and has a limited number of actors (Beverwijk et al., 2008). Brewitt (2019, p.189) noted that the nature of dam removal policy and politics is fluid and that its policy subsystem remains under construction.
- 2 See Sambongi (1988) and Noda (2018) for more details on water rights institutions in Japan.
18Since policy decisions regarding the dam’s removal correspond with ecological boundaries rather than existing political boundaries, this case study enables us to observe the cross-scale or cross-level interactions among various actors across those boundaries, which is an important aspect of environmental governance (Cash et al., 2006). Concerning the institutional landscape within the policy subsystem, both prefectural and national government agencies had legal authority over policies concerning the Arase Dam removal. Since KPG owned and ran the Arase Dam, they could decide whether to remove it by changing the budget and management plan of the Public Enterprise Bureau (PEB), which was in charge of the Arase Dam in the KPG. Since the Kuma River is designated as a Class A river under the River Act, the Ministry of Land, Infrastructure, Transport, and Tourism (MLIT), a national government agency, manages it as the River Administrator. It is the entity in charge of granting permission to build dams or take water from rivers.2
- 3 Sabatier and Jenkins-Smith (1999) defined major policy change as changes in the core aspects of spe (...)
19The policy process leading to the removal of the Arase Dam was not straightforward and experienced three major policy changes3 within almost 10-year period. After the construction of the Arase Dam in 1954, residents started recognizing the adverse effects of the dam on their livelihoods. In response to their requests, in December 2002, Governor Shiotani expressed her decision to remove the dam at the time of the water right renewal (Period one). Although the prefectural government prepared to start the removal work, Governor Kabashima, who was newly elected in 2008, decided to maintain the Arase Dam due to the high cost of removing it in November 2008 (Period two). However, Governor Kabashima was finally forced to express his intention to remove the Arase Dam in February 2010 because of the growing dissent and the national government’s judgment regarding the further renewal of water right. Although I divided the entire policy process into three periods based on the timing of when the governors expressed their policy decisions, such expressions themselves can be conceptualized as policy (Cairney, 2020, p.18), and those three decisions were clearly reflected in the subsequent budget, programs, and organizational reforms of the PEB and KPG. These three major policy changes within a single case study enable us to observe multiple policy processes that led to different policy decisions within similar fundamental social and ecological settings, despite this research only studying a single case.
20Although the Arase Dam removal can be regarded as an individual infrastructure project rather than a national-level policy, which is the type most frequently studied in ACF literature (Pierce, Peterson, Jones, et al., 2017), a policy can broadly be defined as the “total sum of the government action” (Cairney, 2020, p.2). Since the removal of the Arase Dam was decided on and implemented by the local government, this study regards the removal as a policy change. Other similar dam removal studies have conceptualized dam removal as policy change (e.g., Lowry, 2003; Clark, 2009). ACF literature such as Kim (2003) and Yordy et al. (2019) focused on specific infrastructure projects and successfully uncovered the dynamics behind those processes.
21To understand why the Arase Dam policies changed several times, I conducted (1) a content analysis to identify actors and their policy beliefs, (2) an analysis of the events corresponding to the pathways to policy change, and (3) an analysis of the changes in coalition resources. These analyses were conducted for all three periods. However, as described above, since the situation changed considerably after Governor Kabashima’s inauguration during period two, I divided this period into before and after his inauguration (periods 2.1 and 2.2, respectively) to better identify policy beliefs and advocacy coalitions. This study did not analyze the coalition resources in period 2.1 because, as described below, I identified only a single coalition and no major events that might have affected the coalition resources during this period.
22For the analysis of the actor’s policy beliefs and changes in coalition resources, I utilized discourse network analysis, which offers a unified protocol by combining category-based content analysis and network analysis (Leifeld, 2017, p. 304) with open-source software to implement these methods (Leifeld, 2016; Leifeld & Gruber, 2018). A network analysis is crucial not only for understanding the complex dynamics of governance arrangements (e.g., Carlsson & Sandstrom, 2008; Pittman & Armitage, 2019) but also for studying certain key concepts in the ACF such as advocacy coalitions, policy subsystems, and coalition resources (e.g., Weible & Sabatier, 2005; Ingold, 2011; Leifeld, 2013). Hence, discourse network analysis can help operationalize key concepts in the ACF and visualize the interactions among policy actors.
23I identified the actors in the policy subsystem and inferred their policy beliefs using content analysis of newspaper articles. This approach has been used in several ACF studies (Leifeld, 2013; Tosun & Lang, 2016; Olofsson et al., 2018; Heikkila et al., 2019). Using newspaper articles enables us to trace actors’ voices and behaviors across a relatively long period based on certain standards. This enables inferring the policy beliefs and behaviors of actors who might have otherwise been difficult to interview or survey. However, caution should be taken regarding bias because this method may underrepresent actors who are not frequently reported in the media. In addition, it is important to remain cautiously aware that actors’ stances as expressed in articles might depend on their narrative strategies (Heikkila et al., 2014, p. 81). Hence, I confirmed the validity of the content analysis results by cross-checking them with the interview results (as described below) and other publications.
24The newspaper articles targeted for the analysis were collected from the Kumanichi Database Service, which is offered by the Kumamoto Nichinichi Shimbun. The Kumamoto Nichinichi Shimbun, a major local newspaper in the study area, boasts a 69% share of newspaper readership in Kumamoto Prefecture4. I used this database to search for articles that included “Arase Dam” for the period from May 1988 (the beginning of the period for which full-text searches are available) to March 31, 2010 (when the KPG stopped its hydropower generation at the Arase Dam). In total, 703 articles met the search conditions and were collected.5
- 6 Only the articles that reported events were coded. Special feature articles, editorials, and reader (...)
- 7 Although the PEB is a department of the KPG, I regarded it as an independent actor in this analysis (...)
25I read all 703 articles and coded them when they included policy statements, that is, “a text portion where an actor utters his or her policy preferences in a positive or negative way” (Leifeld, 2013, p.173), regarding the removal of the Arase Dam.6 For each statement, I coded (1) the name of the organization making the statement, (2) the policy concept claimed in the statement, and (3) the binary information regarding support of or opposition to the policy concept. Since codes for each variable were not predetermined and had to be fixed inductively, similar to past discourse network analyses (Metz et al., 2018; Buckton et al., 2019), I employed an intercoder agreement approach rather than reporting intercoder reliability scores (Campbell et al., 2013). Another coder, an undergraduate student analyzing the removal of the Arase Dam, independently verified my coding. In the event of any disagreement, we discussed and judged items together to improve our coding reliability. I coded a total of 436 statements by 53 organizations relating to 22 policy concepts. Since the ACF assumes that a policy subsystem is composed of actors who are regularly involved in an issue (Sabatier & Weible, 2007, p.192), I excluded 24 organizations that provided only one statement from the analyses in this study (see Appendix 1 for the list of excluded organizations). I confirmed that this exclusion did not significantly introduce bias into the study’s results by cross-checking other newspaper articles and interview records. The other 29 organizations were included in the analysis and their abbreviations are listed in Appendix 2.7
26In terms of actors’ beliefs, seven policy concepts that directly indicate positions for or against the Arase Dam removal were analyzed: “remove the Arase Dam,” “remove the Arase Dam for flood prevention,” “remove the Arase Dam for the fishery,” “remove the Arase Dam for environmental improvement,” “maintain the Arase Dam for renewable energy,” “maintain the Arase Dam for financial issues,” and “maintain the Arase Dam for water use.” Although the first concept is relatively broad compared to the others, all refer to policy core beliefs, which is a fundamental goal in the policy subsystem. The other fifteen policy concepts pertain to secondary aspects; however, they were rarely stated in the newspaper articles (see Appendix 3). Thus, as in other ACF studies (Weible, 2007; Nohrstedt, 2011), I focused on the policy core beliefs in the subsequent analysis.
- 8 As described above, an advocacy coalition is defined by similar policy beliefs and engagements in n (...)
27Similar to other ACF studies (Weible & Sabatier, 2005; Weible, 2007; Nohrstedt, 2011), I then performed Tabu search clustering using UCINET (Borgatti et al., 2002) to identify advocacy coalitions by the patterns of organizations’ reference to policy concepts.8 I performed clustering several times with different cluster numbers and aggregated organizations into advocacy coalitions according to the number of clusters with the highest goodness of fit index.
28I also described the policy process in each period to identify whether the five pathways to policy change occurred in this process. I read newspaper articles, interview records, and literature sources to identify events corresponding to these five pathways. In addition to Kumamoto Nichinichi Shinbun, I collected articles from other major newspapers: Asahi Shimbun, Mainichi Shimbun, and Yomiuri Shimbun to comprehend related events as comprehensively as possible and checked their validity by comparing each newspaper article’s content.
29Local events within the village were unlikely to be covered in newspaper articles. Thus, I conducted interviews with major actors in Sakamoto Village. Since those interviews aim to complement newspaper articles in terms of local events, I employed purposive sampling to choose interviewees (Kapiszewski et al., 2015, p.212). Based on the literature and the newspaper articles, I selected interviewees who seemed to play important roles in the Arase Dam removal and in issues related to the Kuma River. I attempted to include interviewees from different organizations as much as possible to comprehensively cover past events in the village (see Appendix 4 for the list of interviewees). The interviews were conducted for about an hour and mainly covered interviewees’ relationships with the Kuma River and Arase Dam. Depending on the circumstances of the interviews and the interviewee’s preferences, I chose to either record the interviews or take notes (Kapiszewski et al., 2015, pp.227-229). Recorded interviews were all transcribed. The written notes were typed out and cleaned to maintain the quality of the data. Those interview records, as well as additional newspaper articles other than Kumamoto Nichinichi Shimbun, were not used for discourse network analysis but for identifying pathways to policy change and for confirming the results of discourse network analysis.
30Regarding policy-oriented learning, I relied on the results of the aforementioned clustering to identify the changes in each actor’s policy beliefs. If an actor in one advocacy coalition moved to another coalition, I assumed that there had been time-series changes in their policy beliefs (Elliott & Schlaepfer, 2001, p. 653). However, since I was unable to collect pre-1988 newspaper articles, it was difficult to identify policy belief changes before 1988 using this method. Therefore, I relied on interviews to identify any pre-1988 changes in actors’ policy beliefs.
31Coalition resources were qualitatively described for each period. Based on Nohrstedt (2011), I operationalized five components of coalition resources—formal legal authority, mobilizable supporters, public opinion, information, and leadership—as follows: for formal legal authority, the criterion was whether a coalition had actors with legal authority regarding the issue of concern. In the case of the Arase Dam removal, KPG was the principal actor who had legal authority. MLIT also had legal authority over the Arase Dam in terms of water rights permissions. For mobilizable supporters, the criterion was whether a coalition incorporated new members. Public opinion, that is, public support for a coalition’s policies (Sabatier & Weible, 2007), was clarified through the existing opinion polls on the removal of the Arase Dam. Information, which a coalition can utilize to pursue their policy goal, refers to “information regarding the problem severity and causes and the costs and benefits of policy alternatives” (Sabatier & Weible, 2007, p.203). For leadership, the criterion was whether a coalition had entrepreneurial leaders (Mintrom & Norman, 2009) with attractive visions and who were capable of skillfully exploiting political resources to achieve their policy beliefs (Nohrstedt, 2011).
- 9 I used a Fruchterman Reingold layout with a force-based algorithm for the network graph layouts to (...)
32Since formal legal authority and mobilizable supporters are aspects of the coalitional membership structure, I generated congruence networks (Leifeld, 2013) that enable a visual interpretation of complex coalition structures (e.g., Yordy et al., 2019). I created a network graph for both the first and second halves of each period to identify changes in the network structure.9
33The result of Tabu search clustering indicated that dividing actors into two clusters was optimal for each period. The two clusters in each period were identified as the Removal Coalition (which supported the removal of the Arase Dam) and the Maintenance Coalition (which insisted on maintaining the Arase Dam). As far as the coded policy statements indicated, no actors opposed removing the Arase Dam in period 2.1. Although the Yatsushiro City government (YCG) formed an independent cluster in period 2.1, I regarded YCG as a member of the Removal Coalition because it supported the dam’s removal. Table 2 shows the composition of the advocacy coalitions in each period. Two government agencies (PEB and KPG) joined the other coalition. However, this change does not correspond to policy-oriented learning because it merely reflects the governor change in the KPG rather than the alteration of policy beliefs for the same actor.
Table 2. Advocacy coalitions in each period
|
Removal coalition
|
Maintenance coalition
|
Period 1
|
Residents’ organizations
AAD
NGO
CAKRC
Parties and politicians
KLDP, SVMA
Government
KPG
Fisheries
SVFA, YFA, 37FCAs, KRFCA
|
Government
PEB
|
Period 2.1
|
Residents’ organizations
AAD
Parties and politicians
KLDP, DPJ, SDP
Government
KPG, PEB, YCG
|
―
|
Period 2.2
|
Residents’ organizations
AAD, AADR, VAF
NGO
CAKRC, CNADR, CSRK
Parties and politicians KASDP, KCJCP, KDPJ, SVMA
Fisheries
KRFCA, PFFCAs, PAAFCAs, YFCA
|
Government
PEB, KPG, KUWSA
Farmers
YLID
|
Period3
|
Residents’ organizations
AADR, VAF
NGO
CNADR, CSRK, ECCM
Parties and politicians
AMFADR, KASDP, KCJCP, KDPJ
Government
YCG
Fisheries
KRFCA, PAAFCAs, PFFCAs, YFCA
|
Government
PEB, KPG
Farmers
YLID
|
Note: Each cell entry includes the abbreviated organization names and their attributes, all of which are explained in Appendix 1.
Source: The Author
34Table 3 shows the policy beliefs of advocacy coalitions for each period, indicating that both coalitions’ policy beliefs basically remained stable throughout all three periods. The Removal Coalition continually demonstrated their policy belief that the Arase Dam should be removed to improve both the environment and fisheries and to clear flood damages. The Maintenance Coalition continually demonstrated their policy beliefs that the Arase Dam should be maintained because of financial issues, the importance of renewable energy, and water-use issues. The Maintenance Coalition began stating new policy concept supporting the “maintain the Arase Dam for water use” position after period 2.2. This concept was defended in particular by the downstream local government (YCG) and water users (YLIDs), who were newly involved in the Maintenance Coalition after period 2.2. However, other actors in the Maintenance Coalition did not make statements about this policy concept, indicating that this new policy concept for maintaining the Arase Dam was not a widely held concept within the coalition. These results support the view that policy core beliefs do not easily change (Sabatier & Weible, 2007).
35Table 3 also indicates that two coalitions did not refer to the same policy concepts. While the Removal Coalition advocated for dam removal due to flood damage and environmental/fishery improvements, the Maintenance Coalition did not make statements concerning those issues but rather referred to other issues. On the other hand, the Maintenance Coalition advocated for dam maintenance due to financial, renewable energy, and water-use issues, which the Removal Coalition opposed. This makes it difficult to assume that a negotiated agreement occurred between the coalitions.
Table 3. Changes in policy beliefs in each coalition
|
Coalition
|
Remove the Arase Dam
|
Remove the Arase Dam for environmental improvements
|
Remove the Arase Dam for fishery improvements
|
Remove the Arase Dam for flood prevention
|
Maintain the Arase Dam for financial issues
|
Maintain the Arase Dam for renewable energy
|
Maintain the Arase Dam for water use
|
Period 1
|
Removal
|
12 / 0
|
3 / 0
|
4 / 0
|
2 / 0
|
0 / 2
|
-
|
-
|
|
Maintenance
|
0 / 6
|
-
|
-
|
-
|
-
|
1 / 0
|
-
|
|
|
|
|
|
|
|
|
|
Period 2.1
|
Removal
|
14 / 0
|
1 / 0
|
-
|
-
|
0 / 1
|
-
|
-
|
|
|
|
|
|
|
|
|
|
Period 2.2
|
Removal
|
43 / 0
|
6 / 0
|
7 / 0
|
1 / 0
|
0 / 5
|
0 / 1
|
0 / 3
|
|
Maintenance
|
0 / 19
|
0 / 1
|
-
|
-
|
13 / 0
|
5 / 0
|
3 / 0
|
|
|
|
|
|
|
|
|
|
Period 3
|
Removal
|
50 / 0
|
5 / 0
|
2 / 0
|
2 / 0
|
0 / 4
|
0 / 1
|
-
|
|
Maintenance
|
0 / 12
|
-
|
-
|
-
|
4 / 0
|
-
|
1 / 0
|
Note: The number in each cell indicates the total number of positive statements given for each concept by actors in each coalition (left side of “/”) and the total number of negative statements given for each concept (right side of “/”).
Source: The Author
36In period one, I observed events that correspond to external perturbation, internal perturbation, and policy-oriented learning. Regarding external perturbation, there was a change in the macro socioeconomic situation concerning the position of hydropower generation within the prefecture. Although the Arase Dam initially generated about 16% of the electricity used in the prefecture, it only generated 0.7% of the prefecture’s electricity in 2002 (Kumamoto Nichinichi Shimbun, October 30, 2002 [morning edition]).
37In the first half of period one, the various harmful effects of the dam emerged (internal perturbation), and the residents’ perceptions of the dam changed (policy-oriented learning). After the Arase Dam was constructed, residents started recognizing its adverse effects such as qualitative changes in floods caused by sludge accumulating in the dam reservoir, flood damage due to rising water levels in the dam reservoir, offensive odors due to the eutrophication of the reservoir, and vibrations arising from the dam’s discharge (Interviews 3, 5, 6, and 7). For instance, Interviewee 3 said: “The dam changed everything. It changed the shape, width, depth, and color of the river.” He also said: “We used to catch fish during floods. Even when flooded, we washed a little and cleaned up the house. But after the dam was constructed, there was the muddy sediment that human power couldn’t move.” While the Arase Dam was initially recognized as a symbol of development (Ohno, 2019a), residents living along the dam learned of such negative impacts of the dam from their experiences and altered their beliefs about it, which corresponds to policy-oriented learning. Several residents’ groups were formed and they requested compensation for their damages from the KPG (Interviews 3 and 5), but this had a minor impact on the the removal of the dam movement because of “the political pressure on the campaign’s leader” (Interview 3).
38Another external perturbation—the influence from another dam policy subsystem, namely the influence from the Kawabegawa Dam construction policy—substantially changed the situation. MLIT planned to construct the Kawabegawa Dam in a tributary of the Kuma River upstream of the Arase Dam. The controversies over the Kawabegawa Dam attracted national attention from the late 1990s to the early 2000s. In 2001, some residents of Sakamoto Village who opposed the Kawabegawa Dam construction circulated a petition calling for the establishment of a referendum ordinance in Sakamoto Village. Although the village assembly ultimately rejected it, this movement stimulated the villagers’ awareness of dam issues. At the same time, several fishermen established the Sakamoto Village Fisherman Association (SVFA) as a voluntary association opposed to the construction of the Kawabegawa Dam, while also tackling the Arase Dam problem (Kimoto, 2011).
- 10 The AAD changed its name to the Association for Arase Dam Removal (AADR) in 2008.
39Furthermore, the timing for the water right renewal was a significant internal perturbation. The water right, which is necessary for hydropower generation at the Arase Dam and had been granted to KPG by MLIT at the time of the dam’s construction, was going to expire at the end of March 2003. Although KPG tried to renew this right, a number of Sakamoto Village residents opposed this renewal and requested the removal of the dam. In June 2002, the Association for the Arase Dam (AAD) was established by merging SVFA and other existing resident groups dealing with the Arase Dam issues, which accelerated the dam removal movement (Interviews three and six).10 Interviewee six said: “fishermen alone could not influence the village mayor’s decision. So, we gathered the residents who were suffering from the dam and established AAD.” Four resident organizations including the AAD petitioned the village assembly to publish its opinion statement on the dam’s removal. The Sakamoto Village assembly passed this petition and issued an opinion statement calling for the dam’s removal in September 2002.
- 11 Several newspaper articles reported that the KLDP proposal was related to their intention to build (...)
40Governor Shiotani thought that the prefectural government should take the Sakamoto Village assembly’s request to remove the dam seriously. The Kumamoto prefectural chapter of the Liberal Democratic Party (KLDP), a ruling party in the prefectural assembly, also advocated the dam’s removal.11 In response, Governor Shiotani stated in December 2002 that the prefectural government would start the removal project after allowing the hydropower generation business to continue for seven more years.
41In period one, there were favorable changes to the Removal Coalition’s formal legal authority, mobilizable supporters, leadership, and information. The Removal Coalition, which was only comprised of resident groups and fishermen in the first half of period one (Figure 1a), evolved into a coalition that included important actors such as the Sakamoto Village mayor and assembly members, KLDP, and KPG—all potential veto players—in the latter half of the period (Figure 1b). These actors’ formal legal authority served as a beneficial resource for the Removal Coalition to realize their policy core beliefs. Another dam controversy in the same watershed helped the Removal Coalition significantly expand its network. The collaboration with anti-Kawabegawa Dam movements (i.e., The Citizen’s Association for Kuma River Conservation) served as an advantageous resource for the Removal Coalition in terms of mobilizable supporters. Such network expansion largely resulted from the AAD leaders’ efforts in Sakamoto Village. Several association members and village assembly members personally mobilized supporters within the village (Interviews three and six). Interviewee six said: “I went throughout Sakamoto Village to gather signatures for the petition and do various activities” and “three village assembly members in the AAD were eagerly lobbying for other assembly members.”
Figure 1. Changes in actors’ congruence networks (Period 1)
Note: The color of the node indicates each organizational attribute. Blue indicates fishermen, green indicates farmers, pink indicates NGOs, orange indicates residents' organizations, purple indicates parties and politicians, and red indicates government agencies.
Source: The Author
42In contrast to the coalition resource changes advantageous to the Removal Coalition, no such changes took place in the Maintenance Coalition in period one; specifically, the Maintenance Coalition received no support from either the governor or the assembly, it did not utilize information resources to realize its policy beliefs, and it saw no increase in its mobilizable supporters.
43Two external perturbations occurred during period two. First, a new Kumamoto prefectural governor was elected, leading to a change in the governing coalition, which may be perceived as an external perturbation. After Governor Shiotani’s declaration on the Arase Dam removal, KPG was about to begin the dam removal with careful preparations. Following the expiration of Governor Shiotani’s term, Ikuo Kabashima, supported by the Liberal Democratic Party (LDP), was elected in March 2008. In June 2008, shortly after his inauguration, Governor Kabashima announced that the Arase Dam removal would be halted. This decision was mainly financial—he estimated that the removal would cost ¥5.4 billion JPY, which exceeded the initial estimate by ¥700 million JPY. He believed that the only way to prevent a financial crisis in the prefecture was to maintain the dam, take measures to prevent the environmental deterioration of the dam reservoir, and return the income generated by the dam to the area’s residents (Kabashima, 2014, p. 141).
44Another external perturbation during this period was the municipal merger. Sakamoto Village, where the Arase Dam was located, was an individual municipal government. However, in August 2005, it was merged into Yatsushiro City. After the municipalities were amalgamated, Sakamoto Village became a mere district under the jurisdiction of the YCG. This amalgamation changed the number and area of municipalities in the policy subsystem. While the area affected by the Arase Dam nearly corresponded to the original municipality area, the newly expanded post-amalgamation municipality area increased the gap between the local government’s jurisdiction and the area damaged by the dam.
45In response to strong opposition to his plan by residents near the dam, Governor Kabashima put together a team to examine whether the dam should be removed. The team suggested that the removal would cost over ¥9 billion JPY—much more than the cost of maintaining it (Kumamoto Nichinichi Shimbun, November 12, 2008 [morning edition]). Thus, Governor Kabashima eventually decided to maintain the Arase Dam (Kumamoto Nichinichi Shimbun, November 27, 2008 [evening edition]).
46In period 2.2, the coalition resources changed favorably for the Maintenance Coalition and unfavorably for the Removal Coalition. In terms of formal legal authority, the Removal Coalition lost a number of important actors who were potential veto players such as KPG, KLDP, and YCG (Figure 2). The election of a new governor made the KPG’s position favorable toward maintaining the Arase Dam. KLDP, which had clearly stated its support for the removal in period one, refrained from expressing a clear stance on the issue, thereby revealing its tacit “understanding” of the governor’s policy decision to suspend the removal (Kumamoto Nichinichi Shimbun, June 18, 2008 [morning edition]). Additionally, the Yatsushiro City mayor did not clarify his position regarding the Arase Dam. In contrast to period one, no municipal government explicitly supported removal in period 2.2 (Figure 2). Conversely, the Maintenance Coalition included a new government organization, the Kamiamausa Uki Water Supply Authority, which advocated the dam’s maintenance to ensure their drinking water supply sourced from the Kuma River. Furthermore, farmers in Yatsushiro City began to support the maintenance of the dam to ensure there would be sufficient water for irrigation. Although the Arase Dam did not store irrigation water or drinking water in its reservoir, water users were afraid of losing their nearby water reservoir if it was removed. The Yatsushiro Land Improvement District, a farmer’s irrigation association, submitted a petition advocating the Arase Dam’s maintenance to the Yatsushiro City assembly; this petition was ultimately successful. A similar opinion statement was also passed by the Yatsushiro City assembly’s majority (Kumamoto Nichinichi Shimbun, September 20, 2008 [morning edition]). These changes in formal legal authority benefited the Maintenance Coalition but were detrimental to the Removal Coalition.
Figure 2. Changes in actors’ congruence networks (Period 2.2)
Note: The color of the node indicate each organizational attribute as explained in Figure 1.
Source: The Author
47Unlike in period one, public opinion and information did not serve as advantageous resources for the Removal Coalition. Opinion polling administered by Kumamoto Nichinichi Shimbun in September 2008 showed that 33.9% of the residents in Kumamoto Prefecture supported the removal of the Arase Dam, whereas 51.5% supported its maintenance. The Maintenance Coalition was also superior in terms of information resource utilization, such as basing its decision to maintain the dam on revised cost estimations for its removal.
48In terms of mobilizable supporters, although the Removal Coalition had more supporters in the first half of period 2.2 than in period 1, there was no network expansion in period 2.2. As argued above, although downstream water users supported the maintenance of the dam, the Maintenance Coalition did not successfully mobilize these water users. Regarding leadership, the former Sakamoto Village mayor and various village assembly members vigorously conducted pro-dam-removal activities and such leadership served as a resource for the Removal Coalition.
49External perturbations, internal perturbation, and imposition by a superior jurisdiction were salient features in period three. Substantial changes in the governing coalition, namely external perturbations, occurred in both YCG and the national government in 2009. In Yatsushiro’s mayoral election in August 2009, the incumbent, who had no clear stance on the dam removal issue, lost to a challenger who clearly supported its removal. After the election, YCG started lobbying for the dam’s removal. At the national level, the Democratic Party of Japan won the House of Representatives election in 2009 and subsequently formed a new coalition government with the Social Democratic Party instead of the former long-lasting LDP government. Since both parties suggested that the national government should subsidize the removal cost before the election, KPG had higher expectations of financial support from the national government.
50As an internal perturbation, the deadline for renewing the dam’s water right became a critical issue again. The water right that had been renewed in 2003 was going to expire in March 2010. However, there were no clear rules for further renewal terms because renewal in 2003 had been permitted based on the presumption that the dam would be removed. While KPG tried to renew the water right and maintain the dam, MLIT judged that renewing the water right would be impossible, which corresponds to an imposition by a superior jurisdiction in the second policy change hypothesis. In January 2010, the minister of MLIT announced that the existing water right would be revoked by the end of March 2010. Since KPG was thrust into a difficult situation, it announced a new policy on January 29, 2010 and stated that it would begin the dam removal project after continuing the hydropower generation business for two more years to offset the removal costs (Kumamoto Nichinichi Shimbun, January 30, 2010 [morning edition]). However, neither the Fishery Cooperative Association nor the prefectural assembly supported the continuation of hydropower generation. On March 24, 2010, Governor Kabashima stated that KPG would withdraw its application for a new water right relating to the continuation of the hydropower generation business.
51During period three, coalition resource changes were relatively disadvantageous to the Maintenance Coalition and advantageous to the Removal Coalition. The first half of period three had a similar network structure to period 2.2 (Figure 3). However, several important changes occurred in the latter half of period three (Figure 3). First, YCG joined the Removal Coalition. The inclusion of the municipal government of the region in which the dam was located was advantageous to the Removal Coalition in terms of formal legal authority. The electoral victory of a pro-removal candidate in the Yatsushiro mayoral election suggests that public opinion was a favorable coalition resource for the Removal Coalition.
Figure 3. Changes in actors’ congruence networks (Period 3)
Note: The color of the node indicate each organizational attribute as explained in Figure 1.
Source: The Author
52The Removal Coalition’s network expanded during this period. Voluntary assembly members organized the Yatsushiro City Assembly Members Federation for Arase Dam Removal in November 2009, which developed into a federation of approximately 60 assembly members (including parliament and prefectural assembly members) in December 2009 (Kumamoto Nichinichi Shimbun, December 15, 2009 [morning edition]). The residents’ organizations also expanded their network. In November 2009, approximately 900 people participated in the “Citizen’s Meeting for Kawabegawa Dam Suspension and Arase Dam Removal,” which was led by an executive committee comprising members from 37 organizations. These events suggest that the Removal Coalition increased its mobilizable supporters.
53Although not reflected on the network graph, in period three, MLIT’s stance negatively impacted the Maintenance Coalition’s resources, particularly its formal legal authority. As the water right renewal deadline approached again, MLIT’s authority over water rights became a critical factor in deciding the dam’s fate. Its declaration that KPG’s water right would not be extended placed the latter into a difficult situation. Failure to obtain support from MLIT adversely affected the Maintenance Coalition during this period.
54The results of this study show two facets of the utility of the ACF in environmental governance studies, namely, the utility of policy change hypotheses on the one hand and on the other, research methods within the ACF scholarship. This study confirmed two policy change hypotheses: as suggested in the first policy change hypothesis , several pathways to policy change such as external perturbations, internal perturbations, and policy-oriented learning acted as necessary sources of policy change. As suggested in the second policy change hypothesis, an imposition by a superior jurisdiction, namely MLIT in this case study, ultimately led to major policy change. Given that superior jurisdiction often has formal and legal authority, the second policy change hypothesis may be better understood by associating it with a coalition resources perspective. The pathways to policy change in each period confirmed the causal relationship between external perturbations, changes in coalition resources, and policy change as presented by Nohrstedt (2011). The importance of external perturbations and coalition resource changes as triggers of policy change corresponds to the impact of the external conditions and the significance of political receptivity as suggested by a previous dam removal study (Lowry, 2003). The importance of coalition resources also supports the mega-coalition hypothesis (Brewitt, 2019). Regarding the application of the ACF to nontraditional settings, similar to previous studies (Sato, 1999; Takao, 2016), this study found no critical obstacles to applying the policy change hypotheses to Japan.
55Although the study confirmed the policy change pathways observed in past ACF scholarship (e.g., Nohrstedt, 2011; Jenkins-Smith et al., 2017), it also revealed the need for the further refinement of policy change hypotheses in the ACF. First, the interrelationship among the five pathways to policy change should be examined further. Although the ACF literature tends to observe multiple pathways (Pierce, Peterson, & Hicks, 2017), this study indicates the need for a closer look at, and theorization of, the interrelationships between pathways. For instance, in period one, influence from another policy subsystem acting as an external perturbation altered the coalition resources and led to policy changes. However, such a causal mechanism would not have occurred without prior internal perturbation and subsequent policy-oriented learning. Similarly, although the imposition by a superior jurisdiction finally led to major policy change in period three, it would not have occurred without the prior external and internal perturbations. Further examination of the cumulative effects of the pathways to policy change is required.
56Second, from a similar perspective, the importance of timing should also be examined further. This study revealed that two perturbations (the Kawabegawa Dam policy subsystem and the deadline for renewing the water right)—both of which coincidently occurred in period one—accelerated the dam’s removal. Similarly, two perturbations (the change in the governing coalition and the water right renewal deadline), both of which coincidently occurred in period three, also triggered the change from maintenance policy to removal policy. The importance of the influence of multiple independent events at the same time on the policy change has been discussed in the Multiple Stream Framework (Herweg et al., 2017). In the ACF literature, Heikkila et al. (2014) confirmed the importance of timing in the policy process. Taking into account the timing and sequence of events in the analysis is a challenge that future research should address in both ACF studies and environmental governance studies.
57Third, the results of this study indicate that the influence of policy venue shifts when policy changes. In the series of Arase Dam removal policy processes, the principal policy venue shifted from the Sakamoto Village government (in the first half of period one) to KPG (the latter half of period one and period two) and finally to the MLIT (period three), making it difficult for the Maintenance Coalition to realize its policy beliefs. The effects of venue shifts on policy change are discussed in the Punctuated Equilibrium Theory (Baumgartner et al., 2017). Several ACF studies have also focused on policy venues as intervening variables that associate external and internal perturbation with policy change (Nohrstedt, 2011). However, changes of policy venue in this study occurred due to the timing of water right renewals determined by the institutions concerned rather than because of venue shopping (Baumgartner & Jones, 1991) in which actors strategically search for a supportive venue for their policy proposal. Thus, future research should examine how institutional factors affect policy change.
58Fourth, in the context of policy change in the cross-scale environmental governance, we should not overlook the boundaries of local governments. The fact that the Sakamoto Village assembly passed an opinion statement calling for the Arase Dam’s removal affected KPG’s decision in period one. However, the Removal Coalition lost the support of the municipal government in period two because Sakamoto Village was merged into the larger Yatsushiro City. This suggests that the congruence between the jurisdiction of the local government and the damaged area, which pertains to a problem of the fit in environmental governance (e.g., Young, 2002), is crucial for understanding the policy change within the nested structure of environmental governance (Ostrom, 1990), such as watershed governance in this study.
59In terms of research methods, this study confirmed the usefulness of the ACF’s research methods to systematically collect and analyze data in environmental governance studies. I identified advocacy coalitions, policy beliefs, and coalition resources following the methods developed by previous ACF studies (Weible & Sabatier, 2005; Nohrstedt, 2011), and indicated their utility in different study contexts. When applying these methods, it is difficult to use the same data source (e.g., the testimony of stakeholders in Congress) due to the characteristics of Japan’s policymaking process, namely the lack of transparency in the policy process (Ito, 2005). However, this study could systematically collect and analyze actors’ beliefs and coalition structure by utilizing the discourse network analysis of the newspaper articles. Given the lack of a unified framework in environmental governance studies, employing the ACF is a prominent research strategy for systematically accumulating and utilizing knowledge in the field.
60In this study, I used the ACF to analyze the policy process of the Arase Dam removal, an example of major policy change in Japan’s environmental governance, and examined the two policy change hypotheses and the roles of the coalition resources therein. Consequently, I demonstrated the validity of two policy change hypotheses in Japan. However, this study suggests that the ACF needs further refinement in terms of the interrelationship between pathways to policy change, the effect of timing and sequence on policy changes, the effect of the institutional settings of policy venues on policy change, and the effect of the boundaries of local government.
61The use of the ACF enables us to unpack the complex policy process in the environmental governance systematically rather than explaining it in an ad hoc manner. Thus, even a single case study can contribute to cumulate the evidence and to develop theories in environmental governance studies.
62Future comparative case studies must be conducted to confirm the validity of this study’s findings. In particular, it would be useful to explore a comparison with a case that has a different dependent variable (e.g., a case in which dam removal was not ultimately realized despite the presence of a removal movement) to advance our knowledge of the policy process of major policy changes. Aside from Japan, there are many cases where dam removal has been carried out and conducting quantitative analyses of such cases may be a promising research strategy.