The Local Government Network in Japan: An Analysis of the Cooperation among Local Governments at the Time of Disasters Masahiro ZENKYO Assistant Professor School of Law and Politics Kwansei Gakuin University Paper prepared for presentation at the 23rd International Political Science Association Conference, Montreal, Canada, 22 July, 2014. *This is a draft. Please do not refer to it without the author s permission.
1. Introduction The purpose of this article is to investigate the state and functions of the Agreement for Mutual Support at the Time of Disasters in areas affected by the Great East Japan Earthquake (hereinafter referred to as the 3.11 disaster ). Through empirical analysis, this article answers questions about the types and numbers of agreements that have been concluded by governments for the provision of mutual support, and whether these agreements proved effective in the 3.11 disaster. One of the main questions raised around disaster support is whether intergovernmental networks are effective. Many scholars have argued about the importance of networks between local governments since the occurrence of 3.11 disaster; they have pointed out that the number of local governments that finalized Agreements for Mutual Support at the Time of Disaster (Saigai-ji Sogo Ouen Kyotei) has increased. However, the lack of empirical studies regarding the state and function of those agreements is a serious problem, as there has only been anecdotal evidence as to the success in some cases. To solve this problem, a sample survey was conducted to study the Agreements of Mutual Cooperation at the Time of Disasters in Japan. This data is useful for an empirical and statistical analysis of the state and function of these agreements. Further, these data are valuable because at present, there are little data currently available to allow for an in-depth analyses of the effectiveness of these agreements. As far as could be determined, besides the survey data presented in this report, no other studies exist that have examined the functions of the formal intergovernmental networks in Japan. The findings in this paper provide perspectives for future disaster prevention research. 1
The structure of this article is as follows. In Section 2, the background to the study and the research subjects are explained. Section 3 introduces the survey data and tests the validity of the survey, and Section 4 gives the results of the empirical analysis. Finally, Section 5 presents a summary of the findings and gives guidance as to future research possibilities. 2. Background and Research Question The Great East Japan Earthquake and the Consequences of this disaster On March 11, 2011, the 3.11 disaster and tsunami caused by the earthquake struck the Tohoku region in Japan. This 9.0 magnitude earthquake was the strongest ever recorded in Japan. In addition, this earthquake was also the fourth-strongest earthquake experienced since the beginning of the 20 th century. Over 18,000 people were killed 1, which is greater than in either the Great Han-Shin Awaji Earthquake or the Ise-wan Typhoon 2 (in 1959). Although the 3.11 disaster had a lot of features compared other disasters 3, the most distinguishing feature was the collapse of the municipal governments in Iwate, Miyagi, and Fukushima prefectures. In the coastal areas, not only citizens but also officers and politicians employed in the local governments were killed. For example, Koki Kato, who was mayor of Otsuchi town at the time of the 3.11 disaster, was killed by the tsunami along with many other officers. Further, because the Town hall was also destroyed, the 1 As White Paper on Disaster Management (2013), the total number of deaths and missing is over 18,500. 2 6,437 lives were lost in the Great Han-Shin Awaji Earthquake, and 5,098 lives were lost in the Ise-wan Typhoon. 3 Nishikawa (2011) shows detailed features of the 3.11 disaster. 2
Otsuchi administration was unable to operate for some time after the disaster. In other words, the 3.11 disaster exposed the lack of disaster prevention readiness at the municipal government level in Japan, which is a serious problem as the role and actions of the local government are particularly critical at these times (Col, 2007). Since the Great Han-Shin Awaji Earthquake, the national government has promoted various prevention and mitigation scenarios for disaster readiness. However, during the 3.11 disaster, municipal governments were unable to sufficiently exercise control over the situation 4. The measures for a dysfunctional government have never been discussed because no one believed that local governments could die. Therefore, it appears that at a time when disasters strike, the affected local governments are unable to work independently. Importance of the System of Mutual Support between local governments Although the municipal governments were never deemed to be unable to function, the initial response to the 3.11 disaster was actually carried out because of the support organized by many local governments in the non-affected areas. The necessary measures for the initial response, such as guiding evacuees to shelter, gathering information on the victims or the deceased, and offering treatment, were conducted by officers belonging to local governments in both the affected and non-affected areas. The preliminary feature of the support given by local governments in the nonaffected areas was that it was voluntary and not out of compulsion. In general, after the local government in the affected areas requested support, the municipal governments 4 Kawata (2013) criticizes that the great merger of municipalities in the Heisei era decreased the resources and empowerment of municipal governments for disaster readiness. 3
carried it out. However, clearly, this system does not work effectively because it does not make sense that affected municipal governments need to request support in massive disasters such as the 3.11 disaster. Voluntary and rapid support from non-affected municipal governments without the need for a request is essential in the initial response to disasters. Because of the positive and voluntary support by the municipal governments, the necessary minimum initial response to the 3.11 disaster was carried out. Of course, at the same time, the national government and six local government organizations (Chiho Roku Dantai) also attempted to support the initial response. However, these attempts were too slow and too late. In contrast, the support of local governments was rapid and flexible. After the 3.11 disaster, many scholars highlighted the importance of establishing and expanding the system of mutual support between local governments (Inatsugu ed., 2012; Kamiya and Sakurai, 2013). As Kawamura (2014: 204-205) said, while towns or villages are perhaps appropriate units for their own community planning, they find it difficult to independently control the environment for crisis management. The National government also recognizes these issues, and has therefore specified the importance of systems of mutual support at the time of disasters in the Basic Law for Reconstructing Areas Hit Hard by the Great East Japan Earthquake (Higashinihon Daishinsai Hukkou Kihon Hou). Kamiya and Sakurai (2013) mention that to prepare for massive earthquakes in the near future, local governments need to build and reinforce formal intergovernmental networks based on the Agreement for Mutual Cooperation at the Time of Disasters 5. 5 The Agreement for Mutual Cooperation at the Time of Disasters are generally explained as guidelines which are concluded with governments and corporations or inter-governments for physical or material support related to rescue activities at a disaster site. 4
Lack of Empirical Analysis The systems for voluntary mutual support between the local governments are more important than ever to ensure the provision of a rapid response to disasters and to plan for disaster prevention. The building of a formal intergovernmental network is especially important for the future of Japan. However, little empirical research exists about the state and function of the intergovernmental network for disaster prevention. Most of the evidence is based on impressions only or on anecdotal evidence of successful cases. One reason for the lack of empirical analysis is that many scholars were not interested in these intergovernmental networks, as compared to correlations between public and private sectors, until the 3.11 disaster. There are many benefits in collaborations for disaster prevention, mitigation, and restoration and these facts are also well known from previous studies (Mitchell, 2006; National Infrastructure Advisory Council, 2009; Kapucu, Augustin and Garayev, 2009). In contrast, the importance of the intergovernmental network in Japan has rarely been focused. In addition to this reason, grasping the state and function of intergovernmental networks in Japan is difficult for the following two reasons. First, very few materials have recorded the mutual support between local governments. Second, many local governments, especially those at the town or village levels, do not have information about their collaboration networks online. Hence, even if the objects were limited to a formal network based on the Agreement for Mutual Cooperation at the Time of Disasters, we would not be able to entirely understand them. I suggest that the lack of the empirical analysis is a research subject in itself, but this may be impossible to achieve. The requirement of the hour is the information about the number of agreements between municipal governments and the objectives of these 5
agreements. Furthermore, information about the requirement of these agreements by the local governments and the support provided by the contents to the affected local governments in times of disasters is necessary. Finally, the requirement is to know the basis of difference between these agreements between various municipalities and recognizing the amount of agreements with all the municipalities. For the future, knowing the response to these requirements is necessary. An empirical analysis of the formal networks based on the Agreement for Mutual Cooperation at the Time of Disasters could be extremely useful. Obviously, knowing about all the networks and agreements could be challenging and difficult; however, if the research is limited to formal networks and the relations with affected areas, useful guidance for the future could be extrapolated. The research subject of this article is to examine the state and function of intergovernmental local networks in times of disasters in the affected areas in Japan. Specifically, I seek to answer the two following questions. First, what kind of local government networks are built in Japan to respond to disasters? For this question, I will focus on the state s formal local government network based on the Agreement for Mutual Cooperation at the Time of Disaster. Second, what types of formal local government networks are effective? Not every network was working in the 3.11 disaster. I will answer these questions through an empirical analysis. 3. Data and Survey Validity Survey Data From December 13 th, 2013 to January 31 st, 2014, a research survey was conducted 6
through the mail to study the effectiveness of the Agreements of Mutual Cooperation at the Time of Disasters in Japan. The fieldwork was entrusted to the Survey Research Center Inc. The self-administered questionnaire was sent to a total of 127 sections in charge of concluding the Agreement of Mutual Cooperation at the Time of Disasters 6. Therefore, the respondents to this survey were either chiefs or management grade officers involved in detailed risk management and disaster prevention policies. A total of 86 (67.71%) surveys were returned. The questionnaire in this survey has two main parts. The first part seeks information about the state of the Agreements of Mutual Cooperation at the Time of Disasters. In this part, the respondents were asked to answer the following free description type questions: What kinds of agreements does your government make? How many agreements has your government concluded? And When and why were they concluded? In addition, I also asked the respondents to enclose copies of the concluded agreements from the respondents respective municipal governments. The second part of the questionnaire sought to understand the agreements from the respondents subjective evaluations. Respondents were asked to answer questions about the number of officers dispatched and provide a subjective effectiveness review of the agreements. These questions were selective answer types in most cases. The characteristics of the questions in this part allowed for an exploration of the effectiveness of the agreements in each phase of disaster recovery; from response (7 days after the disaster), to initial restoration (up to 6 months later), to restoration and reconstruction stage (up to 1 year later), and finally to the final reconstruction (> 1 year). 6 There are 127 municipal governments in Iwate, Miyagi, and Fukushima prefecture. 7
Check of Validity Before discussing the empirical analysis based on the survey data, I need to confirm the validity of the survey as over 30% of municipal governments did not respond. If the non-responsive governments are not randomly distributed, the results of this survey would be biased. The validity was confirmed using a cross table analysis (chi-squared test) and an analysis of mean differences (t-test). Table 1 shows the results for the chi-square test to check geographical bias. As shown in this table, there are no great differences. However, apparently, the response rate of the municipal governments in Miyagi prefecture is high, but the differences are statistically insignificant. Additionally, the result of chi-squared test to check the differences between the municipalities in coastal and non-coastal areas is also not significant. Table 1: The results of the chi-squared test: response*prefecture or location n (responded) % prefectural different Iwate prefecture 21 63.6 (χ 2 = 1.013) Miyagi prefecture 26 74.3 Fukushima prefecture 39 66.1 place difference in coastal areas 27 71.1 (χ 2 = 0.276) in non-coastal areas 59 66.3 Note: significant level at *: p < 0.05, **: p < 0.01 Table 2 shows the results for the test of mean difference. The independent variable is a dummy response (response or no-response) and the dependent variables are population, population density, fatalities, and a financial capability index. The differences for population and financial capability index are significant. The values for the population and financial capability index from the responding municipalities are higher than the values for the non-responding municipalities. In other words, this result implies that there is a difference 8
between the urban and rural areas. Table 2: t-test results mean population (2010) responded 55822.59 non-responded 22113.00 population density (2010) responded 520.44 non-responded 381.96 fatalities responded 174.24 (2013.4) non-responded 158.78 financial capability index responded 0.46 (2010) non-responded 0.37 sig. Note: significant level at *: p < 0.05, **: p < 0.01 Sources: National Census (2010) and White Paper on Disaster Prevention (2014) * * To summarize these results, there is a statistically significant difference in the response rate between the city level and the town or village level municipalities. However, this is not a serious problem because it is rare for towns or villages to have concluded many mutual support agreements. If I conducted a survey on the non-responding municipalities again, the results would be unchanged. Therefore, the survey data in this article has sufficient validity for an empirical analysis. 4. Empirical Analysis The State of Formal Networks at the Time of Disasters The Agreements of Mutual Cooperation at the Time of Disasters can be categorized into two types. The first is a pair type agreement, and the other is a multiple parties agreement. Further, the latter type is divided into small area and wide area cooperative agreements. As an operational definition, the agreements that include only two municipalities (one-to-one) are defined as pair type agreements, and the agreements including over three parties are defined as multiple type agreements. Further, agreements 9
that have a cooperation limited to only adjacent prefectures are defined as small area cooperative agreements, while the agreements that build cooperative relations beyond adjacent prefectures are defined as wide area cooperative agreements. Table 3 shows the number and ratio for each type of agreement, before and after the 3.11 disaster 7. Over 50% are pair type agreements, the number of which more than doubled after the 3.11 disaster (from 48 to 132 agreements). However, the multiple type agreements increased moderately. In addition, the ratio of small and wide area cooperative agreements was almost the same before and after the 3.11 disaster. The percentage of small area cooperative agreements was about 65.6% before the 3.11 disaster, and after the 3.11 disaster the percentage fell slightly to 62.5%. This result implies that multiple type cooperative agreements had little change compared to pair type cooperative agreements. Table 3: The number and ratio for each type of agreement The number of municipalities Before the 3.11 disaster After the 3.11 disaster N % N % pair type 2 48 59.3 132 73.3 multiple type 3 7 8.6 12 6.7 4 5 6.2 7 3.9 5 9 11 13.6 14 7.8 10 29 4 4.9 9 5.0 over 30 6 7.4 7 3.9 The mutual support agreements are not equally distributed across all municipal governments, and there is a gap between urban and rural governments. The larger the population size of the governments, the more likely they are to have concluded agreements for mutual support. Table 4 shows a correlation of population size and the number of agreements concluded in each municipality. The parameters are divided on the basis of areas: Iwate, Miyagi, and Fukushima. In Iwate and Fukushima prefectures, 7 Pay attention to the estimated value, in actuality the true value is higher than this. 10
these two variables are related to a positive correlation except for small area cooperative agreements (p < 0.01). In Miyagi prefecture, the relationship between population size and the number of pair or wide area agreements is positive statistically, but the correlation of population size and the number of pair type agreements is not significant. Although there are differences in the relation between population size and the number of agreements by regions or types, city level municipalities tend to conclude agreements, especially pair and wide area agreements, at a higher level than governments at the town or village level. Table 4: Pearson s correlation for the number of agreements and population size for each municipal government in Iwate prefecture in Miyagi prefecture in Fukushima prefecture The number of pair type (in the 3.11 occurrence) 0.505 ** 0.414 * 0.382 ** The number of wide area type (in the 3.11 occurrence) 0.559 ** 0.529 ** 0.509 ** The number of small area type (in the 3.11 occurrence) 0.216 0.117 0.439 ** The number of pair type (at the present) 0.428 ** 0.195 0.423 ** The number of wide area type (at the present) 0.578 ** 0.593 ** 0.508 ** The number of small area type (at the present) 0.270 0.322 0.237 Note: significant level at *: p < 0.05, **: p < 0.01 To summarize the above analysis, the findings are as follows. First, the pair type agreement proportion is large. After the 3.11 disaster, many local governments in the affected areas concluded The Agreements of Mutual Cooperation at the Time of Disasters. Some scholars call this phenomenon the first year of cooperation between local governments in Japan (Kamiya and Sakurai, 2013). However, most of these agreements are pair type cooperative agreements, and the number of multiple type agreements increased only marginally after the 3.11 disaster. Second, there is a gap in the number of agreements between city and town or village level municipalities. Municipal governments at the city level have many agreements, but municipalities at the town or village level 11
have few. Moreover, this gap remains the same after the 3.11 disaster. The Effectiveness of the Formal Networks in Times of Disaster What kinds of agreements for mutual support were effective in the 3.11 disaster? I answer this question through the survey data analysis. There are two useful questions in the survey. The first is a question about the number of dispatched officers, I ask about the support offered based on the Agreement for Mutual Support at the Time of Disaster: How many officers were dispatched to your government after the 3.11 disaster? The answer to this question is a 5-point ordinary scale: from less than 10 people (1) to more than 300 people (5). The second is a question about the subjective effectiveness of the agreements, I ask about the effectiveness of the Agreement of Mutual Support at the Time of Disasters: How effective were these agreements in each phase from recovery to reconstruction? The answer is a 5-point ordinary scale: from very effective (1) to not effective at all (5). Table 5 shows the frequency distributions for these two variables separated on the basis of coastal and non-coastal areas. This table shows that the number of dispatched officers was obviously different in different regions. Almost all respondents in the noncoastal areas answered less than 10 people because the municipal governments in these areas probably did not suffer any critical damage in the 3.11 disaster. On the contrary, many officers were dispatched through all phases from recovery to reconstruction in the coastal areas. The frequency distribution is also different by region. The respondents in coastal areas tend to recognize the agreements as more effective than the respondents in the noncoastal areas. While the number of officers sent based on the Agreement for Mutual 12
Support at the Time of Disasters has gradually decreased over time, the total rate of very effective and effective in coastal areas does not decrease. Table 5: Frequency distributions for the two questions in coastal area in non-coastal area less than 10 people ~7 days the number of officers who are dispatched ~6 ~1 >1 months year year the subjective effectiveness ~7 days ~6 months ~1 year >1 year 68.4 60.0 57.1 61.9 very effective 72.7 63.6 69.6 65.2 10 50 people 21.1 20.0 19.0 19.0 effective 9.1 22.7 21.7 17.4 50 100 people 5.3 10.0 4.8 9.5 neither or not 18.2 13.6 8.7 17.4 100 300 people 5.3 5.0 9.5 0.0 not very effective 0.0 0.0 0.0 0.0 more than 300 not effective 0.0 5.0 9.5 9.5 people at all 0.0 0.0 0.0 0.0 N 19 20 21 21 N 22 22 23 23 less than 10 people 92.1 92.1 89.2 97.1 very effective 67.4 64.4 53.5 42.5 10 50 people 5.3 5.3 2.7 2.9 effective 20.9 26.7 30.2 32.5 50 100 people 2.6 2.6 7.1 0.0 neither or not 7.0 6.7 11.6 20.0 100 300 people 0.0 0.0 0.0 0.0 not very effective 4.7 2.2 4.7 2.5 more than 300 not effective 0.0 0.0 0.0 0.0 people at all 0.0 0.0 0.0 2.5 N 38 38 37 34 N 43 45 43 40 Note: DK/NA is missing. These results imply that the officers in the non-coastal areas did not recognize the function of the mutual support agreements. In other words, this also suggests that the officers in the coastal areas did recognize the agreement s function. From this finding, examining the regional differences in the function of the agreements appears necessary. Tables 6 and 7 show the results of the correlation analysis regarding the effectiveness of the agreements. Table 6 shows the correlations for the number of agreements and dispatched officers, and Table 7 shows the correlations for the number of agreements and the subjective effectiveness. The correlation for the number of the agreements and the number of officers is only statistically significant in the coastal areas. In the non-coastal areas, all correlation 13
coefficients are very low and insignificant. However, the fact that not all agreement types are important is also necessary to consider. The correlation for the number of wide area type agreements and the number of officers is only significant and limited until 6 months after the disaster. The number of officers based on pair or small area cooperative agreements is perhaps smaller than that based on wide area cooperative agreements. Table 6: Correlation of the number of agreements and dispatched officers (τ_c) in coastal areas in non-coastal areas ~7 days ~6 months ~1 year 1 year~ the number of all types 0.539 ** 0.406 * 0.402 * 0.357 * the number of pair 0.091 0.038 0.184 0.177 the number of wide area 0.399 * 0.353 * 0.308 0.212 the number of small area 0.229 0.175 0.048 0.073 the number of all types 0.076 0.091 0.026 0.112 the number of pair type 0.015 0.004 0.044 0.021 the number of wide area 0.039 0.039 0.101 0.100 the number of small area 0.110 0.154 0.113 0.104 Note: significant level at *: p < 0.05, **: p < 0.01 Table 7 also shows some important findings as follows. First, the correlation for these variables is only statistically significant in coastal areas (except in one rare case), the same as the result in Table 6. At the time of initial response or recovery, for respondents in the coastal areas, the correlation for effectiveness and the number of wide area agreements is significant. However, this correlation is not significant 6 months after the disaster, but the correlation with the number of pair type cooperative agreements is significant and seems to take the place of the wide area type cooperative agreements 8. To summarize these analyses, the agreements for mutual support appeared to be working, at least at the time of the initial response. Though limited to coastal areas, the numbers of officers dispatched were larger in municipalities having many mutual support 8 The reason why the correlation with pair type is significant at the recovery and reconstructing phase is not clear. 14
agreements. Furthermore, the respondents in such governments tended to evaluate the function of the agreements favorably. However, not all agreements for mutual support were effective. In many cases, only the wide area cooperative agreements were seen as effective at the time of disaster. There were only a few that were effective in the small area cooperative agreements. Table 7: Correlation of the number of agreements and subjective effectiveness (τ_c) in coastal areas in non-coastal areas ~7 days ~6 months ~1 year 1 year~ the number of all types 0.353 * 0.273 * 0.261 * 0.261 the number of pair 0.192 0.176 0.244 * 0.306 * the number of wide area 0.260 ** 0.248 * 0.176 0.181 the number of small area 0.205 0.217 0.166 0.057 the number of all types 0.032 0.058 0.027 0.139 the number of pair type 0.023 0.068 0.094 0.103 the number of wide area 0.012 0.076 0.173 * 0.132 the number of small area 0.166 0.186 0.006 0.138 Note: significant level at *: p < 0.05, **: p < 0.01 5. Conclusion In this article, I examined the state and function of the intergovernmental network in times of disaster in Japan at the local level. The following are the main findings of this article: 1) although many local governments in the affected area had concluded The Agreements of Mutual Cooperation at the Time of Disasters, almost all of these are pair type cooperative agreements, whereas the number of small and wide area cooperative agreements increased marginally or not at all after the 3.11 disaster. 2) Wide area cooperative agreements were seen to be effective during the 3.11 disaster; however, pair and small area cooperative agreements were not seen as effective as the wide area cooperative agreements in the initial response. On the basis of these results, the wide area cooperative agreement is possibly an 15
effective agreement. Increasing the number of wide area cooperative agreements would be beneficial for an initial response to disasters; this could be an area for future studies. In addition, eliminating the gap between the city, town, and village level municipalities is also a matter of concern. Exploring and finding resources for the concluding of wide area cooperative agreements in small municipal governments is required. These implications from the empirical analysis will contribute to studies about local governance for disaster prevention. However, there are several problems in this article. The results and findings are limited and can really only provide a glimpse of the collaborative agreements in Japan. An in depth understanding of the state and functions of the intergovernmental network is necessary; hence, a further study on the mutual support agreements is necessary. It is hoped that this article will be of some use to disaster prevention practitioners in the future. Acknowledgement: This work has been supported by the Grant-in-Aid for Scientific Research (B) from 2013 to 2015 (Grant Number: 25285047). References Col, Jeanne-Marie (2007) Managing Disasters: The Role of Local Government Public Administration Review, Vol. 67, issue 1, pp.114-124. Inatsugu, Hiroaki (2012) Daikibo Saigai ni Tsuyoi Jichitai-kan Renkei: Genba Karano Houkoku to Teigen. Japan: Waseda Daigaku Syuppanbu. Kamiya, Hideyuki and Seichi Sakurai (2013) Jichitai Renkei to Juen Ryoku: Mou Kuniniha Izon Dekinai. Japan: Kojin no Tomo Sya. 16
Kapucu, Naim, Maria-Elena Augustin and Vener Garayev (2009) Interstate Partnerships in Emergency Management: Emergency Management Assistance Compact in Response to Catastrophic Disasters Public Administration Review, Vol. 69, Issue 2, pp. 297 313. Kawamura, Kazunori (2014) Higashinihon Daishinsai to Chiho Jichi: Hukkyu Hukko ni okeru Hitobito no Ishiki to Gosei no Kadai. Japan: Gyosei. Kawata, Yoshiaki (2013) Higashinihon Daishinsai ni okeru Shichoson Saihen Saigai, Yoshiteru Murosaki and Masaharu Koda eds. Sichoson Gappei ni yoru Bosairyoku Kudoka: Higashinihon Daishinsai de Roteishita Heigai. Japan: Minerva Shobo, pp. 3-22. Mitchell, James K. (2006) The Primacy of Partnership: Scoping a New National Disaster Recovery Policy. Annals of the American Academy of Political and Social Science 604, pp.228-255. National Infrastructure Advisory Council (2009) Framework for Dealing with Disasters and Related Interdependencies. Nishikawa, Satoru (2011) Japan s Preparedness and the Great Earthquake and Tsunami Funabashi Yoichi and Heizo Tkenaka eds. Lessons from the Disaster: Risk management and the compound crisis presented by the Great East Japan Earthquake. Japan: The Japan Times, Ltd., pp.19-47. 17