natural disasters and participation in volunteer activities: a case study of the great hanshin-awaji...

Annals of Public and Cooperative Economics 84:1 2013 pp. 103–117

NATURAL DISASTERS AND PARTICIPATION INVOLUNTEER ACTIVITIES: A CASE STUDY OF THE

GREAT HANSHIN-AWAJI EARTHQUAKE

byEiji YAMAMURA*

Seinan Gakuin University, Japan

ABSTRACT: The Great Hanshin-Awaji (Kobe) earthquake in 1995 has had a signif-icant detrimental effect on the economic conditions of southern-central Japan. However,the earthquake also led people to acknowledge the importance of the many volunteeractivities in Japan at that time. Using a large sample of individual-level data from1991 and 1996, this study investigates how and the extent to which the earthquakeincreased the participation of students and house-workers in volunteer activities. Aftercontrolling for various individual characteristics, a Heckman-Tobit model was usedand the following key findings were obtained: (1) the probability of students’ partici-pating in volunteer activities was 2% higher after the earthquake than before, and (2)the number of days that students spent participating in volunteer activities was 4.38days longer after the earthquake than before. However, the same did not hold true forhouse-workers.

Keywords: natural disasters, social capital, volunteer activities

JEL classification: N35, Q54, Z13

Naturkatastophen und Beteiligung an ehrenamtlichen Aktivitaten:Eine Fallstudie des großen Hanshin-Awaji-Erdbebens

Das große Erdbeben von Hanshin-Awaji (Kobe) im Jahr 1995 hat erhebliche negative Auswirkun-gen auf die wirtschaftlichen Rahmenbedingungen im sudlichen Teil von Zentraljapan gehabt.Jedoch hat das Erdbeben die Menschen auch dazu gebracht, die Bedeutung der vielen ehre-namtlichen Aktivitaten in Japan zu dieser Zeit anzuerkennen. Unter Verwendung einer großenStichprobe mit auf Individualebene bezogenen Daten aus den Jahren 1991 und 1996 wird in dieserStudie untersucht, wie und in welchem Ausmaß das Erdbeben die Beteiligung von Studenten undHausfrauen/manner an ehrenamtlichen Aktivitaten gesteigert hat. Nachdem man verschiedene

∗ Through the Research Centre for Information and Statistics of Social Science Institute ofEconomic Research, Hitotsubashi University, the National Statistics Center provided me with theanonymized data of the ‘Survey on Time use and Leisure Activities’. Based on this data, I processedand constructed the statistics used in this paper. In addition, I gratefully acknowledge financialsupport from the Japanese Society for the Promotion of Science (Foundation (C) 22530294 (PI:Eiji Yamamura)). E-mail: [email protected]

© 2013 The AuthorAnnals of Public and Cooperative Economics © 2013 CIRIEC. Published by Blackwell Publishing Ltd, 9600 Garsington Road,Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

104 EIJI YAMAMURA

individuelle Merkmale gepruft hatte, wurde ein Heckman-Tobit-Modell angewandt, und es wur-den die folgenden wesentlichen Erkenntnisse erzielt: (1) Die Wahrscheinlichkeit der Beteiligungvon Studenten an ehrenamtlichen Aktivitaten war nach dem Erdbeben um 2% hoher als davor. (2)Nach dem Erdbeben verbrachten Studenten 4,38 Tage mehr mit ehrenamtlichen Aktivitaten alsbisher. Allerdings galt das nicht fur Hausfrauen/manner.

Catastrophes naturelles et participation aux activites de volontariat :une etude de cas sur le tremblement de terre de la region du Great

Hanshin-Awaji

Le tremblement de terre de Great Hanshin-Awaji en 1995 a eu des consequences nefastes signi-ficatives sur les conditions economiques du Japon du Centre-Sud. Cependant, ce tremblement deterre favorisa aussi la prise de conscience de l’importance des nombreuses activites benevoles auJapon a cette epoque. A partir d’un grand echantillon de donnees individuelles de 1991 a 1996,cette etude examine comment et dans quelle mesure ce tremblement de terre a augmente la partici-pation d’etudiants et d’hommes/femmes au foyer a des activites volontaires. Apres un controle deplusieurs caracteristiques individuelles, les auteurs utilisent un modele Heckman-Tobit et obtien-nent les resultats essentiels suivants : 1) la probabilite de volontariat chez les etudiants a augmentede 2% apres le tremblement de terre et 2) le nombre de jours que les etudiants consacrent a desactivites benevoles a augmente de 4.38 jours suite au tremblement de terre. En revanche, cela n’apas ete verifie pour les hommes/femmes au foyer.

Catastrofes naturales y participacion en las actividades de voluntariado:un estudio del caso en el terremoto de la region del Great Hanshin-Awaji

El terremoto de Great Hanshin-Awaji en 1995 tuvo consecuencias nefastas significativas en lascondiciones economicas del centro-sur de Japon. Sin embargo, este terremoto favorecio tambien latoma de conciencia de la importancia de las numerosas actividades benevolas en Japon duranteesta epoca. A partir de una gran muestra de datos individuales de 1991 a 1996, este estudio examinacomo y en que medida el terremoto aumento la participacion de los estudiantes y de las amas de casaen las actividades de voluntariado. Despues de un control de varias caracterısticas individuales,los autores utilizan un modelo Heckman–Tobit y obtienen los relevantes resultados siguientes: 1)la probabilidad de voluntariado entre los estudiantes aumento un 2% tras el terremoto y 2) elnumero de dıas que los estudiantes consagran a actividades benevolas aumento en 4,38 dıas tras elterremoto con relacion a la situacion anterior. En contraposicion, esto no ha podido ser verificadopara las amas de casa.

1 Introduction

A growing number of works have focused on natural disasters, and a popularissue now in social science is how to cope with unforeseen events (e.g., Albala-Bertrand1993, Tol and Leek 1999, Skidmore and Toya 2002, Toya and Skidmore 2007, 2010,Cavallo et al. 2010). For instance, the Great Hanshin-Awaji (Kobe) earthquake occurredin 1995, and in response to the earthquake, economic researchers analyzed the out-comes of the earthquake and suggested various policy implications (e.g., Horwich 2000,Sawada 2007, Sawada and Shimizutani 2007, 2008). The function of the market andthe role of government are considered important in disaster prevention and in coping

© 2013 The AuthorAnnals of Public and Cooperative Economics © 2013 CIRIEC

NATURAL DISASTERS AND PARTICIPATION IN VOLUNTEER ACTIVITIES 105

with disaster. In addition, the level of damage caused by disasters obviously dependson institutional conditions (Kahn 2005). Hence, social capital such as social networksand community participation appear to contribute to the prevention of and resilience tonatural disasters (Chamlee-Wright 2010, Yamamura 2010). Participation in volunteerwork can be considered to contribute to the formation of social capital in modern soci-ety. Researchers have paid much attention to the role of informal cooperative activities,such as voluntary disaster control organizations, in reducing the damage arising fromnatural disasters (Tierney and Goltz 1997).

In the wake of the Great Hanshin-Awaji earthquake, many people, especiallyyoung people (generally students), arrived in Kobe to participate in volunteer activities.This was the first time in Japan that such a large number of people had come forwardas volunteer workers. Therefore, 1995 is regarded as ‘the first year of volunteer activ-ity’ in Japan. The earthquake has resulted in Japanese residents acknowledging theimportance of the role played by volunteer workers (Waseda University Social ScienceInstitute 1996). This may be a reason why so many people without question joined thevolunteer activities in response to the eastern Japan earthquake in 2011. Thus, un-foreseen disastrous events have triggered cooperation and collective action for disaster-prevention and resilience.1 After the Hanshin-Awaji earthquake, students and house-workers were especially encouraged to take action. From an economic viewpoint, thismay have been because their opportunity cost was relatively smaller than workers inpaid employment. However, the argument that the Hanshin-Awaji earthquake becamethe catalyst for volunteer participation has not been assessed based using close statis-tical analysis.2 Hence, this paper used a ‘Survey of Time Use and Leisure Activities’(STULA) to explore how and the extent to which the Hanshin-Awaji earthquake im-pacted on individuals’ volunteer activities. This survey provided individual-level dataand consisted of 375,676 observations. This paper’s analysis has the advantage of usinga rich data set.

The remainder of this paper is organized as follows. Section 2 provides an overviewof the Great Hanshin-Awaji earthquake. Section 3 explains the data set and the empiri-cal method used. Section 3 provides the estimation results. The final section offers someconclusions.

2 Overview of the Great Hanshin-Awaji earthquake

According to the Cabinet Office, Government Of Japan (2007), 21% of earthquakesof magnitude 6 and over occurred in Japan, although Japan landmass is only 0.25% of

1 In Japan, homeownership and neighbours influence the degree of participation in commu-nity activities (Yamamura 2011 a, b).2 Fukushige (2010) used individual-level data to ascertain the determinants of participationin volunteer activities for middle-aged and advanced-age respondents. However, that data waslimited to people over 40 years old and in selected areas. Furthermore, the sample size wassmaller than 1,000 in each area. In addition, Fukushige (2010) used aggregated-level data forexamination purposes, and was constructed using whole generations in Japan. However, becauseof data limitations, individual characteristics were not controlled.


106 EIJI YAMAMURA

Figure 1 – Map of Japan showing Kobe’s location (the area suffering the most damage inthe 1995 earthquake).

World’s.3 On 17 January 1995 an unprecedented devastating earthquake occurred insouthern-central Japan (the Hanshin-Awaji area). Kobe, a densely populated city andan important hub port in western Japan, suffered the greatest damage above all otherareas. Figure 1 shows Kobe’s location in the south-eastern area of the Hyogo Prefec-ture,4 Hanshin-Awaji. Japan’s earthquake scale ranges from level 1 (weak) to level 7(devastation) – most of Kobe was categorized as level 7 in the Great Hanshin-Awajiearthquake (Ministry of Land, Infrastructure, Transport and Tourism 1996:3). As aconsequence of the earthquake, the death toll reached 6,308. Further, approximately100,000 homes were destroyed and 33,000 suffered partial destruction. The loss of hous-ing was valued at more than US$60 billion, and that of capital stock exceeded US$100billion (Horwich 2000, Sawada and Shimizutani 2007, 2008). Turning our attention toother disasters, Hurricane Katrina resulted in approximately 5,336 deaths and US$26.5billion in damage (Sawada 2011: 46). The East Japan Earthquake, a more recent dev-astating disaster in Japan caused approximately 15,200 deaths. In this earthquake,thssXe calculated total damage was between US$20,000–30,000 billion (Sawada 2011:46). This damage is comparable to the damage caused by the Hanshin-Awaji earthquake.In contrast, the death toll of the Sichuan earthquake in China in 2008 was approxi-mately 69,000 and that of the 2004 tsunami in Indonesia was approximately 165,700.This suggests that the damage caused by natural disasters in less developed countrieswas far greater than that of the Hanshin-Awaji earthquake. Considering this togetherimplies that the level of damage caused by natural disasters depends in part on thedegree of economic development because antidisaster measures can be put in place once

3 Japan incurred 13% of the total amount of damage resulting from natural disasters world-wide during the past 30 years (Cabinet Office, Government of Japan, 2007).4 A Japanese prefecture is the equivalent to a state in the USA or a province in Canada.There are 47 prefectures in Japan.



a country achieves a certain level of economic development. Apart from an increase inincome level, quality of government is also thought to improve in the process of economicdevelopment.

People anticipate that government plays a leading role in dealing with unforeseenevents such as natural disasters, because government should secure the safety of citizensagainst such events. However, the government’s initial response was slow, and this laterdrew criticism.5 In contrast, young people came to Kobe from various parts of Japan toserve as volunteer workers. It is widely acknowledged that informal volunteer activitiesare more flexible than formal institutions and can make considerable contributions inthe event of natural disasters.

3 Theoretical background

Researchers have shown that natural disasters influence economic developmentvia various channels (Skidmore and Toya 2002, Crespo-Cuaresma et al. 2008). As a directoutcome of natural disasters, capital stock is destroyed and in turn output is reduced.This is a negative effect of natural disasters on economic development. In contrast, asargued by Skidmore and Toya (2002), ‘disasters also provide an opportunity to updatethe capital stock, thus encouraging the adoption of new technologies’ (Skidmore and Toya2002: 665). Through the adoption of new technology, natural disasters are consideredto accelerate economic growth. This is an indirect positive effect of natural disasters oneconomic development. Hence, natural disasters are thought to be a catalyst of creativedestruction (Skidmore and Toya 2002).

Social capital (i.e., cooperation and collective action through social network) playsan important role in economic development, but not in economic growth (Putnam 2000).Social capital is accumulated through people’s active participation in community eventsand volunteer work. In other words, community participation is regarded as an individ-ual’s investment in social capital (Glaeser et al. 2000). Natural disasters seem to destroythe social capital formed through long-term personal interactions because natural dis-asters can cause the death of community members. Hence, social capital is reduced asa consequence of natural disasters. However, the scarcity of social capital increases thevalue of social capital in economic terms. This in turn causes people to realize the impor-tance of social capital in daily life. Furthermore, ‘informal’ social capital is regarded as asubstitute for formal government and the market (Yamamura 2009). Accordingly, ‘infor-mal’ social capital became more important directly after the Hanshin-Awaji earthquakewhen formal government was not effective and the market did not function properly.

5 Even though Japan had previously experienced the Hanshin-Awaji earthquake, the gov-ernment’s response to the East Japan Earthquake was slow. ‘For example, after 11 March it tookthe government more than three months to enact a basic law for rebuilding Tohoku’s coastalcommunities, whereas a similar law came into force only a month after the massive 1995 Kobeearthquake . . . the most dismaying difference between the two catastrophes is the time it took topass a supplementary budget to fund full-scale reconstruction work. After the Hanshin quake,a budget to rebuild Kobe was enacted in around four months. After last March’s disasters, theruling Democratic Party of Japan took twice as long – more than eight months – to enact a 12trillion yen reconstruction budget for Tohoku’ (Hongo 2012: 9).


108 EIJI YAMAMURA

This provides young people with the motivation to participate in community activities orvolunteer work. Hence, I hypothesize that natural disasters are thought to be a catalystfor the accumulation of social capital, leading people to participate in volunteer work.

4 Data and methods

4.1 Data

In this paper, I used individual-level data sourced from STULA. Based on thedata provided, I constructed various variables to be used for statistical estimations. TheJapanese Government (Ministry of Internal Affairs and Communications, StatisticalBureau in Japan) began conducting STULA in 1976. These surveys are held every5 years. The survey includes observations randomly chosen from almost all regionsthroughout Japan. Surveys have been conducted in 1991, 1996, 2001, 2006, and 2011,6

when I look only at STULA in the period after 1990s. The survey is conducted in Octoberof the survey year, and in 1996 STULA was conducted approximately 18 months afterthe earthquake. With the aim of assessing the impact of the earthquake on individuals’behaviour, this paper compares participation rates in volunteer activities before andafter the Great Hanshin-Awaji earthquake. Hence, STULA data from 1991 and 1996 areused in this paper. In addition to issues regarding social activities, STULA asks standardquestions concerning individuals’ characteristics. The data includes information relatingto marital status, age, gender, annual household income, and education level. Combineddata from 1991 and 1996 were gathered from approximately 396,332 respondents agedover 15 years old. However, not all respondents answered all of the survey questions.Inevitably, data regarding some variables used in the estimations were not available. Asa consequence, the number of samples used in the regression estimations was reducedto 375,676.

The key variable, the proxy for the degree of participation in volunteer activities,is defined as follows: in the STULA questionnaire respondents were asked ‘Did youparticipate in any volunteer work within a year?’ The possible responses to this questionwere ‘Yes’ or ‘No’. If respondents choose ‘Yes’, they were asked ‘How many days did youparticipate in volunteer activities within a year?’ There were 7 possible responses tothis question: ‘1–4 days’, ‘5–9 days’, ‘10–19 days’, ‘20–39 days’, ‘40–99 days’, ‘100–199days’ and ‘more than 200 days’. The interval of the responses differed according tothe various categories. The distribution of the participation in volunteer activities ispresented in Table 1. Table 1 shows that 77.8% of respondents did not take part involunteer activities. Further, 14.6% of respondents participated in activities from 1 to4 days. Hence, the share of those who participated in activities for more than 5 dayswas less than 10%. Based on the interval data of the volunteer activities, I constructeda proxy for the degree of volunteer activities as follows: I assumed that the days inwhich everyone in each category participated in the activities as the midpoint value. Forinstance, a value of 2.5 days was assigned for those in the ‘1–4 days’ category. In thecase that respondents did not participate at all, a value of 0 was assigned. In contrast,the ‘more than 200 days’ category results in the problem of top-coding. I solved that

6 Individual-level data could only be accessed for 1991, 1996, and 2001.



Table 1 – Distribution of days spent participating in volunteer activities

Days %

0 77.81–4 14.65–9 3.310–19 2.720–39 0.940–99 0.4100–199 0.2200– 0.2

Figure 2 – Distribution of ages.

problem as follows: one can participate in the activities for 365 days at best within ayear. Hence, for the ‘more than 200 days’ category, I assigned a value of 282.5, which isthe mid-value of 200 and 365.

In the original data set, as illustrated in Figure 2, the various ages were groupedinto 15 categories. The groups representing respondents in their 40s are the largest.In addition to those aged 40–44, 45–49, and 35–39 (the three largest age groups), the15–19 year age group is also large, with a distribution of more than 0.08. The 15–19age group basically represents students. Hence, the 15–19 age group is sufficiently largeto justify analysis. Further, as illustrated in Figure 3, annual earnings were groupedinto 12 categories. In the estimations reported in Tables 4(1) and (2), dummies for agegroup and household income groups were included.7 The other variables used in theregression estimations are shown in Table 2. With the exception of Volunteer, which is

7 Education level was grouped into 9 groups, including current students at junior high school,high school, junior college, university, and graduates from junior high school, high school, juniorcollege, university, and others. Their dummies are also included as independent variables in theestimations.


110 EIJI YAMAMURA

Figure 3 – Distribution of household income.

Table 2 – Definition of variables used for estimation and basic statistics

Definitions

Volunteer Number of days spent participating in volunteer activities 2.2Student A value of 1 is given if respondent is a student, otherwise 0 (%) 8.8House A value of 1 is given if respondent is a house-worker, otherwise 0 (%) 26.9Marry A value of 1 is given if respondent is married, otherwise 0 (%) 63.7Male A value of 1 is given if respondent is male, otherwise 0 (%) 47.4Urban A value of 1 is given if respondent resides in an urban area, otherwise 0 (%) 28.7Owner A value of 1 is given if respondent resides in own home, otherwise 0 (%) 76.7Observations 375,676

Note: Numbers are mean values for Volunteer. The percentage of respondents taking 1 is also reported.

the number of days spent participating in volunteer activities, the variables in Table 2are dummy variables. Definitions and the share of each dummy variable are presented inTable 2. With regard to social position, such as Student and House (house-workers), theoriginal data set showed 6 categories: workers (those who have full-time jobs), student(without job), student (with part-time job), house-workers (without jobs), house-workers(with part-time jobs), others (without jobs). Student consists of both student (withoutjob) and student (with part-time job). House consists of house-workers (without jobs)and house-workers (with part-time jobs). Workers are workers (those who have full-timejobs).

Table 3 compares Volunteer between 1991 and 1996. With respect to the resultsbased on the whole sample, Table 3(1) shows 2.23 days for 1991 and 2.25 days for1996. The difference is not statistically significant. As for the sub-sample of students,Table 3(2) shows 0.43 and 0.59 for 1991 and 1996, respectively. This suggests thatthe number of days that students spent doing volunteer activities is distinctly lessthan those for the whole sample. However, Volunteer for student is 0.16 days morein 1996 than in 1991 and statistically significant at the 5% level. It follows then thatstudents are less likely to participate in volunteer activities than other groups but the



Table 3 – Comparison of mean values of days spent participating in volunteer activities

1991 1996 t-statistics(1) Whole sample

2.23 2.25 0.41

(2) Sample restricted to students0.43 0.59 2.16∗∗

(3) Sample restricted to house-workers2.76 2.95 1.89∗

Note: ∗ and ∗∗ indicate statistical significance at the 10% and 5% levels, respectively.

Hanshin-Awaji earthquake increased students’ volunteer participation rate. Further,concerning house-workers, Table 3(3) shows 2.76 and 2.95 for 1991 and 1996, respec-tively. This suggests that the number of days that house-workers spent doing volunteeractivities is slightly more than those for the whole sample. In addition, Volunteer forhouse-workers is 0.19 more in 1996 than in 1991 and statistically significant at the10% level. The combined results of Tables 3(1), (2) and (3) suggest that overall the par-ticipation rates in volunteer activities did not differ before and after the earthquake.However, when focusing only on students and house-workers, whose opportunity costwas relatively smaller than workers, those two groups were more likely to participatein volunteer activities after the earthquake than before.

4.2 Econometric framework and estimation strategy

The estimated function of the baseline model takes the following form:

Volunteeri = α0 + α1 Studenti∗1996 dummyi + α2 Housei

∗1996 dummyi + α3 Studenti

+ α4 Housei + α51996 dummyi + α6 Marryi + α7 Malei + α8 Urbani + α9 Owneri + ui,

where Volunteeri represents the dependent variable in individual i. Regression pa-rameters are represented by α. The error term is represented by ui. As shown inTable 1, Volunteeri takes 0 for approximately 80% of observations. An individual’sdecision-making is considered in two-steps. First, an individual will participate in vol-unteer activities or not. Second, an individual will consider the degree to which he/shewill participate. In such a situation, selection bias will occur if a simple OLS model isused. Hence, the Heckman Tobit model is more appropriate to control for selection bias.Using the Heckman Tobit model, in the first step, a Probit estimation is used to estimatethe decision to participate or not. In the second step, the OLS model is used to examinethe number of days spent doing volunteer activities for those who decided to participate.

A cross term including the key variables of Studenti∗1996 dummy and Housei

∗

1996 dummyi was incorporated to capture the effect of the Hanshin-Awaji earthquake onparticipation rates in volunteer activities for students and house-workers, respectively.If their coefficient signs are positive, then they are more likely to participate in volunteeractivities after the earthquake than before. In addition, there are a number of factorsthat may affect an individual’s propensity to engage in volunteer activities. Marriedpeople seem to be more integrated into the community and social networks, and hence,Marry is expected to take the positive sign. In urban areas, the opportunity cost ofresidents engaging in volunteer activities is higher because there are opportunities to


112 EIJI YAMAMURA

Table 4(1) – Determinants of Volunteer

(1) (2)OLS Heckman

Student∗1996 dummy 0.22∗∗ 4.38∗∗∗(2.32) (3.21)

House∗1996 dummy 0.09 –0.11(0.80) (–0.21)

Student –0.69∗∗ –10.1∗∗∗(–2.07) (–3.24)

House 0.65∗∗∗ 5.86∗∗∗(7.75) (7.75)

1996 dummy -–.06 –1.34∗∗∗(–1.07) (–3.78)

Marry 0.66∗∗∗ 9.40∗∗∗(9.72) (9.41)

Male 0.84∗∗∗ 5.25∗∗∗(15.8) (12.3)

Urban –0.08 –5.35∗∗∗(–1.51) (–5.89)

Owner 0.03 3.53∗∗∗(0.66) (5.03)

Constant 0.64∗ –59.9∗∗∗(1.70) (–8.19)

R-square 0.01Wald chi-square 1,319Censored observations 293,045Uncensored observations 83,631Observations 375,676 375,676

Note: Numbers in parentheses are t-statistics calculated using robust standard errors. ∗∗ and ∗∗∗ indicate signifi-cance at the 5% and 1% levels, respectively. In all estimations, age group dummies, household income dummies,and education level dummies are included as independent variables but are not reported because of spacelimitations.

obtain higher earnings. Hence, Urban is predicted to take the negative sign. Ownerrepresents house ownership, and is included because people who own their homes aremore inclined to be ‘good citizens’ and participate in community activities (DiPasqualeand Glaeser 1999, Yamamura, 2011a). The expected sign of Owner is positive.

The opportunity cost for full-time workers participating in volunteer activities isvery high. Full-time workers work weekdays, which impose severe time constraints onvolunteer activities. Hence, full-time workers are less likely to participate in volunteeractivities. In addition to Student and House, in the first stage of the Heckman Tobitmodel, Workers is included as an independent variable whereas Workers was excludedfrom the second-stage estimation. Hence, in the first stage, ‘Others (without job)’ is usedas the reference group.

5 Estimation results

In Table 4(1), the results of the simple OLS model are shown in column (1), andthe results of the second-stage estimation using the Heckman Tobit model are reported



in column (2). Table 4(2) shows the results of the first-stage estimation of the HeckmanTobit model.

Column (1) of Table 4(1) shows that the cross term of Student∗1996 dummy yieldsthe positive sign and is statistically significant at the 5% level. In addition, the absolutevalue is 0.22, which implies that the number of days spent participating in volunteeractivities in 1996 is 0.22 days longer than in 1991. However, this result is likely to beinfluenced by selection bias and therefore care should be called for when interpretingthese results. In contrast, House∗1996 dummy is not statistically significant although ityields the predicted positive sign. This implies then that there was no difference in thenumber of days spent by house-workers participating in volunteer activities between1991 and 1996.

As discussed earlier, the OLS regression results appear to suffer from selectionbias. Hence, the bias was controlled for and the second-stage results of the HeckmanTobit estimation are shown in column (2) of Table 4(1). The cross term of Student∗1996dummy yields the positive sign and is statistically significant at the 1% level. What ismore, the absolute value is 4.38, which implies that the number of days spent partici-pating in volunteer activities in 1996 is 4.38 days longer than in 1991. This means thatresult for Student∗1996 dummy in the simple OLS estimation suffered a downwardsbias. House∗1996 dummy is not statistically significant and takes an unpredicted neg-ative positive sign. The other control variables exhibited statistical significance in allcolumns. In line with the prediction, Marry and Owner yield the positive sign. The abso-lute value of Marry is 9.40, and can be interpreted as meaning that married people aremore inclined to participate in volunteer activities for a further 9.40 days than others.The absolute value for Owner, 3.53, means that home owners will participate in volun-teer activities for an extra 3.53 days. In contrast, Urban yields the predicted negativesign. The absolute value for Urban is 5.35, and suggests that residents of urban areasare less likely to participate in volunteer activities, volunteering 5.35 days less thanothers. Turning to results of the first-stage estimation using the Heckman Tobit model,in Table 4(2), the cross term for Student∗1996 dummy yields the positive sign and is sta-tistically significant at the 1% level. Further, its absolute value shows a marginal effectof 0.02, thus, the probability of students’ participating in volunteer activities was 2%higher in 1996 than in 1991. Contrary to the estimation, House∗1996 dummy yields thenegative sign and is statistically significant at the 1% level. Considering Tables 4(1) and4(2) jointly shows that the earthquake promoted the participation of students in volun-teer activities. However, the earthquake did not promote house-workers’ participationin such activities.

After the Kobe earthquake, the creation of informal support systems to deal withdisasters was considered important (Shaw and Goda 2004). As an antidisaster preven-tion policy, the Japanese government has been testing social network services for thepurpose of improving community building and disaster management (Schellong 2007).Furthermore, communities were found to have an advantage over government after theearthquake in the task of gathering detailed information about community requirementsin disaster-stricken areas (Koonno 2007). In other words, local communities can makemore appropriate use of local knowledge for economic resilience than government. How-ever, with respect to the revival of community in the case of the Chuetsu earthquake in arural mountainous region of Japan, there appeared to be a psychological gap between thevictims who continued to reside in the stricken-area and those who evacuated the area


114 EIJI YAMAMURA

Table 4(2) – First-stage analysis of column (2) of Table 4(1).

Heckman(first-stage Probit)

Student∗1996 dummy 0.02∗∗∗(3.09)

House∗1996 dummy –0.01∗∗∗(–2.89)

Student –0.003(–0.28)

House 0.09∗∗∗(25.1)

1996 dummy –0.01∗∗∗(–7.58)

Marry 0.07∗∗∗(41.9)

Male 0.02∗∗∗(12.9)

Urban –0.07∗∗∗(–48.2)

Workers 0.05∗∗∗(17.9)

Owner 0.05∗∗∗(30.3)

Constant –1.22∗∗∗(–20.6)

Observations 375,676

Note: Values are marginal effects. Numbers in parentheses are z-statistics calculated using robust standard errors.∗∗ and ∗∗∗ indicate significance at the 5% and 1% levels, respectively. In all estimations, age group dummies,household income dummies, and education level dummies are included as independent variables but are notreported because of space limitations.

(Fukutome 2007).8 This phenomenon is related to the problem of collective action: if res-idents expect each other to return, they will return. If they do not have this expectation,they will not return. In contrast, the experience of Hurricane Katrina tells us that ‘thedirect material support offered through patterns of mutual assistance helped to over-come the collective action problem by reducing the first mover cost of returning residents’(Chamlee-Wright 2010: 46). This shows that the effectiveness of community depends oncertain conditions. Because of the limitation of the data used in this study, this papercannot examine how the effect of natural disasters on participation in volunteer workand in the formation of social capital differs according to socio-economic conditions.

6 Conclusions

Unforeseen and uncontrolled events such as natural disasters have a significantimpact on socio-economic conditions. This inevitably results in economic loss, which isconsidered a negative outcome. However, in the case of the Great Hanshin-Awaji (Kobe)

8 The remaining closed-community in the Chuetsu region sometimes refused the assistanceof volunteer workers from other regions (Fukutome 2007).



earthquake in 1995, many young people arrived in Kobe to participate in volunteeractivities. The earthquake led people to acknowledge the importance of volunteer activ-ities in Japan. However, the issues surrounding the volunteer response have not beenexamined using detailed statistical analysis based on individual-level data.

This paper used a large sample of individual-level data to explore how and theextent to which the Great Hanshin-Awaji earthquake increased the participation ofstudents and house-workers in volunteer activities. As suggested in the descriptiveanalysis, students are characterized as being less active in volunteer activities whereashouse-workers are considered active. I found, however, via the Heckman-Tobit modelthat students were more likely to participate in volunteer activities immediately after anearthquake than before. Furthermore, the number of days that students spent partici-pating in volunteer activities was greater immediately after the earthquake than before.However, this does not hold true for house-workers. This suggests that the earthquakepromoted the less active students to participate in volunteer activities but did not in-fluence the active house-workers’ participation level although their opportunity cost forthe volunteer activities is lower than workers.

The results of the estimations in this paper can be interpreted differently: themacro-economic shock or institutional change between 1991 and 1996 influenced the be-havior of people causing an increase in participation in volunteer activities. For instance,in Japan, Japanese people enjoyed a business boom, ‘the bubble economy’ from the mid-1980s to the early 1990s. ‘The prolonged period asset inflation showed initial signs ofcollapse in 1991. By October, stock prices had fallen 50 percent from their 1989 peak’(Moriguchi and Ono 2006: 165). After the boom period, Japan entered a long-term eco-nomic recession, considered to have begun in 1991. Such a change of economic conditionpossibly influenced individuals’ behaviour such as participation in volunteer activities.If this holds true, then the effect of the Hanshin-Awaji earthquake on participation involunteer activities is spurious. It is necessary to engage in a closer examination of theeffect of the earthquake. For instance, the Gini coefficient and unemployment rate ineach prefecture could be used to capture the influence of changes to the macro-economicconditions. However, due to a limitation of data, information regarding the prefecturewhere respondents reside was not available. Furthermore, it seems plausible that theimpact of the earthquake on individuals’ behaviour differs between victims of the earth-quake and non-victims. Information regarding residence is necessary to identify victimsof the earthquake. Hence, the data limitation did not allow me to investigate how andthe extent to which the behaviour of the earthquake victims differs from non-victims.This information is necessary to identify the respondents’ residence, and then theseissues can be addressed in future studies.

REFERENCES

ALBALA-BERTRAND J., 1993, Political Economy of Large Natural Disasters, NewYork: Oxford University Press.

CABINET OFFICE, GOVERNMENT OF JAPAN, 2007, ‘A Disaster Preven-tion White Paper’ (in Japanese) Cabinet Office, Government of Japanhttp://www.bousai.go.jp/hakusho/h19/index.htm


116 EIJI YAMAMURA

CAVALLO E., POWELL A. and BECERRA O., 2010, ‘Estimating the direct economicdamages of the earthquake in Haiti’, Economic Journal, 120(546), F298–F312.

CHAMLEE-WRIGHT E., 2010, The Cultural and Political Economy of Recovery: SocialLearning in a Post-Disaster Environment, New York: Routledge.

CRESPO-CUARESMA J., HLOUSKOVA J. and OBERSTEINER M., 2008, ‘Natural dis-asters as creative distruction? Evidence from developing countries’, Economic Inquiry.,46(2), 214–226.

DIPASQUALE D. and GLAESER E.L., 1999, ‘Incentives and social capital: are home-owners better citizens?’ Journal of Urban Economics, 45(2), 354–384.

FUKUSHIGE M., 2010, Kakei ni Yoru Syakai-teki Katsudo no Keiryo Bunseki. (inJapanese), [Econometric Analysis of Social Activities], Kyoto: Mineruva Shobo.

FUKUTOME M., 2007, ‘Chusankan chiiki no Seikatsu Saiken to Komyniti zukuri’(in Japanese), [Mountainous region’s recovery of life and reforming of commu-nity], [Econometric Analysis of Social Activities], in M. Urano , J. Oyane and T.Yoshikawa, eds, Fukko Komyuniti-ron Nyumon [Analysis of Reviving Community],Tokyo: Kobundo.

GLAESER E.L., LAIBSON D. and SACERDOTE B., 2002, ‘An economic approach tosocial capital’, Economic Journal, 112, 437–458.

HONGO J., 2012, ‘One year on, Tohoku stuck in limbo: Anxious locals dismayed, frus-trated that the recovery is still only taking baby steps’, The Japan Times SpecialReport: 3.11 one year on, a Chronicle of Japan’s Road to Recovery, Tokyo: The JapanTimes.

HORWICH G., 2000, ‘Economic lessons from Kobe Earthquake’, Economic Developmentand Cultural Change, 48, 521–542.

KAHN M., 2005, ‘The death toll from natural disasters: the role of income, geographyand institutions’, Review of Economics and Statistics, 87(2), 271–284.

KONNO H., 2007, ‘Toshi inna-eria no shinsai fukko’ (in Japanese), [Recovery fromdisaster in inner areas of an urban city], [Econometric Analysis of Social Activities], inM. Urano, J. Oyane and T. Yoshikawa, eds, Fukko Komyuniti-ron Nyumon [Analysisof Reviving Community], Tokyo: Kobundo.

MINISTRY OF LAND, INFRASTRUCTURE, TRANSPORT AND TOURISM., 1996,‘White Paper on Disaster Management 1996’, Tokyo: Ministry of Land, Infrastruc-ture, Transport and Tourism.

MORIGUCHI C. and ONO H., 2006, ‘Japanese lifetime employment: a century’s per-spective’, M. Blomstrom and S. La Croix, eds, Institutional Change in Japan, NewYork: Routledge.

PUTNAM R. 2000, Bowling Alone: The Collapse and Revival of American Community.A Touchstone Book.

SAWADA Y., 2007, ‘The impact of natural and manmade disasters on household welfare’,Agricultural Economics, 37, 59–73.

SAWADA Y., 2011, ‘Saigai to Keizai: Shizen saigai, gijutsu teki saigai, jinteki saigaino kurosu kantori bunseki kara’ (in Japanese) [Disaster and economics: evidence



from cross-countries analysis regarding natural disaster, technological disaster, andmanmade disaster], Sekai Keizai Hyoron [World Economic Review], 55(4), 45–49.

SAWADA Y. and SHIMIZUTANI S., 2007, ‘Consumption insurance against naturaldisasters: evidence from the Great Hanshin-Awaji (Kobe) earthquake’, Applied Eco-nomics Letters, 14(4–6), 303–306.

SAWADA Y. and SHIMIZUTANI S., 2008, ‘How do people cope with natural disasters?Evidence from the great Hanshin-Awaji (Kobe) earthquake in 1995’, Journal of Money,Credit and Banking, 40(2–3), 463–488.

SCHELLONG A., 2007, ‘Increasing social capital for disaster response through socialnetworking service (SNS) in Japanese local governments’, NCDG Working Paper No.07–005.

SHAW R and GODA K., 2004, ‘From disaster to sustainable civic society: the Kobeexperience’, Disasters, 28, 16–40.

SKIDMORE M. and TOYA H., 2002, ‘Do natural disasters promote long-run growth?’Economic Inquiry, 40 (4), 664–687.

TIERNEY K and GOLTZ J. D., 1997, ‘Emergency response: lessons learned from theKobe earthquake’, University of Delaware, Disaster Research Center, PreliminaryPaper #260.

TOL R. and LEEK F., 1999, ‘Economic analysis of natural disasters’, in T. Downing, A.Olsthoorn and R. Tol, eds, Climate Change and Risk, London: Routledge.

TOYA H. and SKIDMORE M., 2007, ‘Economic development and the impacts of naturaldisasters’, Economics Letters, 94(1), 20–25.

TOYA H. and SKIDMORE M., 2010, ‘Natural disaster impacts and fiscal decentraliza-tion’, CESifo Forum, 11(2), 43–55.

WASEDA UNIVERSITY SOCIAL SCIENCE INSTITUTE, 1996, Volunteer Activities inthe Great Hanshin-Awasji Tarthquake. (in Japanese), (Hanshin-Awaji Daishinsai niOkeru Saigai Volunteer Katsudo). Tokyo: Waseda University Social Science Institute.

YAMAMURA E., 2009, ‘Dynamics of social trust and human capital in the learningprocess: the case of the Japan garment cluster in the period 1968–2005’, Journal ofEconomic Behavior and Organization, 72 (1), 377–389.

YAMAMURA E., 2010, ‘Effects of interactions among social capital, income and learningfrom experiences of natural disasters: a case study from Japan’, Regional Studies,44 (8), 1019–1032.

YAMAMURA E., 2011a, ‘Comparison of the effects of homeownership by individuals andtheir neighbors on social capital formation: evidence from Japanese General SocialSurveys’, Journal of Socio-Economics, 40(5), 637–644.

YAMAMURA E., 2011b, ‘How do neighbors influence investment in social capi-tal?: Homeownership and length of residence’, International Advances in EconomicResearch, 17(4), 451–464.


natural disasters and participation in volunteer activities: a case study of the great hanshin-awaji...

Documents