Livelihood Recovery after Natural Disasters and the

11 downloads 0 Views 491KB Size Report
only approximately 20 percent of its total commitment to livelihood recovery ... co-chaired by the Government of Indonesia and the European Commission, with the ... sales of firm i just after the earthquake and Yi,t as the monthly sales of firm i t.
bs_bs_banner

Asian Economic Journal 2012, Vol. 26 No. 3, 233–259

233

Livelihood Recovery after Natural Disasters and the Role of Aid: The Case of the 2006 Yogyakarta Earthquake* Budy P. Resosudarmo, Catur Sugiyanto and Ari Kuncoro Received 14 September 2010; accepted 9 May 2011

The 27 May 2006 Yogyakarta earthquake caused the death of more than 5700 people; more than 60 000 people were injured and hundreds of thousands lost their homes. Bantul District was most severely affected by the earthquake. This paper is an attempt to understand the determinants of livelihood recovery after this natural disaster and, in particular, the role of aid in that recovery process. A panel firm-level survey was conducted by visiting approximately 500 mostly small and micro enterprises in Bantul District on two occasions: 6 and 12 months after the earthquake. This paper argues that: (i) smaller enterprises are more resilient and so are able to recover faster; (ii) an industrial cluster system within a sub-district provides the necessary support for firms to recover; (iii) the quality of village infrastructure could be important; (iv) it is important to distribute aid as early as possible (the faster it is distributed, the better the impact on enterprises affected by the earthquake); and (v) donors should not give too much assurance of financial support to enterprises in cases where the delivery date is uncertain, but rather just provide support when it is actually available. Keywords: micro and small enterprises, industrial organization, development economics, natural disasters, Indonesia. JEL classification codes: L20, O10, R58. doi: 10.1111/j.1467-8381.2012.02084.x

I.

Introduction

The calamity of the December 2004 Indian Ocean tsunami shocked the world into realizing how devastating the impacts of natural disasters can be for people in *Resosudarmo (corresponding author): Indonesia Project, Arndt-Corden Department of Economics, Crawford School of Public Policy, Australian National University, Canberra, ACT 0200, Australia. Email: [email protected]. Sugiyanto: Faculty of Economics and Business, Gadjah Mada University, Yogyakarta, Indonesia, Email: [email protected]. Kuncoro: Faculty of Economics, University of Indonesia, Depok, Jawa Barat, Indonesia, Email: [email protected]. Achmad Maulana was the research assistant for this paper. Sri Giyanti, Wemmy Subiyantini and Retnandari were the field coordinators for the survey. The authors would like to thank the Australian–Indonesian Governance Research Project and the Indonesia Project at the Australian National University for providing support for this research. The authors would also like to thank seminar participants at the Australian National University Arndt-Corden Department of Economics, the 11th East Asia Economic Association International Convention and the 5th Australasian Development Economics Workshop. All mistakes are the authors’ responsibility. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

234

developing countries; in this case, India, Indonesia, Malaysia, Maldives, Myanmar, Somalia, Sri Lanka and Thailand. Since then, the world has started to pay more attention to natural disasters that take place in developing countries and has contributed more generous funding for the recovery process, although not as much as natural disaster funding in developed countries (Athukorala and Resosudarmo, 2005). Of equal importance, further research has been conducted to develop effective policies to mitigate the impacts of natural disasters and to aid post-disaster recovery. Among others are the works by Jayasuriya et al. (2005), Telford et al. (2006) and Nazara and Resosudarmo (2007), as well as those of institutions, such as the World Bank (2008), the ADB (2005) and the UNDP (2005a,b). Various recent publications on post-disaster recovery discuss the need to have proper policies to rebuild the livelihoods of people affected by natural disasters effectively and within a reasonable time period, whereas up to now the major focus of post-disaster recovery has been on building houses and infrastructure. Our understanding of how to develop strategies and how to channel aid to accelerate the recovery of the livelihood of people affected by natural disasters is limited (Christoplos, 2006; Nazara and Resosudarmo, 2007). The present paper is an attempt to understand the determinants of livelihood recovery after a natural disaster and, in particular, the role of aid in that recovery process. It focuses on the recovery of micro, small and medium (both formal and informal) enterprises after the 2006 Yogyakarta earthquake. In most developing countries, such enterprises are the source of livelihood for many poor people, particularly those living in urban and surrounding areas. The 27 May 2006 Yogyakarta earthquake measured 6.3 on the Richter scale and caused the death of more than 5700 people. More than 60 000 people were injured; hundreds of thousands lost their houses and 2 000 000 (half Yogyakarta Province’s population) were affected. Hence, this is one of the more significant natural disasters on a global scale. In particular, the present paper attempts to answer the following questions: 1 What firm characteristics determine the recovery rate of micro, small and medium enterprises? 2 In general, was receiving aid a significant determinant of the recovery rate? 3 Was the expectation of receiving aid a strong motivation to recover faster or did it create a tendency to wait until the receipt of aid, thus maybe slowing recovery? 4 Was an unfulfilled expectation of receiving aid harmful? 5 Was receiving aid in time an important factor?1 A panel firm-level survey was conducted for the present paper by visiting approximately 500 mostly small and micro enterprises in Bantul District on two occasions: 6 and 12 months after the earthquake. Bantul is the district within 1

This paper defines ‘in time’ as within 3 months of the disaster.

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

235

Yogyakarta most affected by the earthquake and home to more than 20 000 small and micro enterprises. The definition of micro, small and medium enterprises is firms with approximately 100 or fewer workers. They can be registered firms (part of the formal sector) or not (part of the informal sector). The definition of aid in this paper is limited to grants (cash or in kind), which can come from the government, local organizations or international donors. The main reasons for this definition are: (i) the amount of grants flowing into a region affected by natural disasters is typically significant and has been the main issue regarding aid for natural disasters; and (ii) information concerning recipients of grants and the definition of grants has been more transparent and better defined than for cheap credit or other subsidies. The paper proceeds as follows. Section II provides a general description of the 2006 Yogyakarta earthquake, its impact and the management of the recovery processes. Section III reviews some of the literature on this subject and outlines the econometric model that will be used. Section IV explains the firm surveys and describes the variables gathered. This paper then uses results from the estimates of the econometric model to answer the five main questions of this paper. Finally, Section VI concludes.

II.

The 2006 Yogyakarta Earthquake

Yogyakarta Province is located in the centre of Java Island. Its population was approximately 3 200 000 in 2005, equating to around 1.5 percent of the Indonesian population. With an average density of approximately 1000 people square kilometer, it is one of the most populated provinces in Indonesia. Yogyakarta’s GDP per capita was approximately US$830 in 2005, around 60 percent of the national average. Yogyakarta Province consists of a municipality and four districts: the city of Yogyakarta and Bantul, Sleman, Kulon Progo and Gunung Kidul Districts.2 Micro, small and medium enterprises dominate business. They are mostly food processing, construction material, textile and garment, furniture and ceramic enterprises. To date, there is no data related to micro enterprises (i.e. how many there are and how many people work for such enterprises). Data is only available for small, medium and large enterprises. In 2005, there were approximately 117 000 enterprises with 650 000 employees. Of these enterprises, 97 percent were small and medium enterprises absorbing 65 percent of these 650 000 workers. Approximately 21 000 of these enterprises were in Bantul District (Bappenas et al., 2006). On Saturday morning, 27 May 2006, an earthquake measuring 6.3 on the Richter scale hit Yogyakarta and some areas of Central Java, causing widespread destruction, loss of life and property (Figure 1). The death toll was estimated at 2 It is important to note that within Yogyakarta Province there is a municipality of Yogyakarta. In this paper, ‘Yogyakarta’ always refers to Yogyakarta Province. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL Figure 1

236

Yogyakarta Province and the 27 May 2006 earthquake epicentre ©Cartography ANU 2008-089/KP

Semarang

Java Sea INDONESIA

7°S

Banda Sea

Java CENTRAL JAVA

Indian Ocean

0

50 kilometres

SLEMAN KULUN PROGO

Cilacap

Yogyakarta

Klaten

BANTUL

Provincial capital District boundary Provincial boundary

GUNUNG KIDUL

8°S

Earthquake epicentre

INDIAN OCEAN 109°E

Surakarta

110°E

over 5700, and up to 37 900 were injured. At least 156 700 buildings were totally destroyed, and over 200 000 suffered varying degrees of damage (Bappenas et al., 2006). Bantul District was the most severely affected by the earthquake; the death toll was approximately 4100 and 12 000 were injured. The Bappenas (2006) joint team estimated that the total physical damage cost in Bantul reached approximately 246 percent of its gross domestic regional product and affected more than 14 000 micro, small and medium enterprises. The response from local and international donors to help people affected by the earthquake was overwhelming. Within a week, Yogyakarta was inundated by many of these organizations. During the second week after the disaster, preparations for recovery activities began. Initially, there was not much coordination among organizations involved. Soon, however, the local government in cooperation with the United Nations Development Programme (UNDP) registered all participating organizations and grouped them into several clusters dealing with similar activities. The recovery activities themselves started 1 month after the disaster. In Yogyakarta Province, commencing in October 2006, approximately 70 percent of households affected were given Rp15m (US$1613) per household to construct or © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

237

renovate their houses. Households were expected to manage the reconstruction themselves using the funds provided. A few were given new houses constructed by donors. Those whose houses were considered to be minimally damaged were not given any housing support. Some school and public reconstruction activities commenced in the second month after the earthquake. Most of these activities were conducted by government and donor agencies. A recovery of livelihood program was also introduced in the second month after the earthquake. In general, the programs aimed to provide financial support, in cash or in kind, linked to technical assistance for micro and small enterprises, to support defaulting lenders to develop effective strategies for viable enterprises, and to establish soft-loan mechanisms to rehabilitate damaged medium-sized business infrastructure and capital equipment (JFR, 2008). The government has not been able to fully implement these objectives because many organizations being involved. For example, not all micro and small enterprises received financial support in cash or in kind; some medium enterprises did, although not many. By June 2008, the government had spent approximately Rp5.4tn (US$570m) on housing, and the Java Reconstruction Fund (JRF)3 as much as US$60m on various activities (mostly housing). No public information is available to trace how much funding has been allocated to livelihood recovery, but, most likely, it has been much less than for housing. For example, the JRF planned to allocate only approximately 20 percent of its total commitment to livelihood recovery programs (JRF, 2008).

III.

Methodology

The literature on the impact of large shocks on firm/enterprise performance is abundant. Recently, many works have been established to analyze the impact of the 1997/1998 Asian crisis on firm performance in countries most affected by the crisis. Examples include the works by Fukuchi (2000), Sato (2000) and Narjoko and Hill (2007) in the case of Indonesia, by Rungsomboon (2005) and Dekle et al. (2005) in the case of Thailand, by Lim and Hahn (2003) and Oh et al. (2008) in the case of Korea, and by Dwor-Frecaut et al. (2000), Mitton (2002) as well as Chen and Hsu (2005) for multi-country comparative analysis. The typical model used for this Asian crisis is one that observes the relationship between firm performance, measured by either output, value added or total factor productivity on the left-hand side, and firm and industrial/market characteristics, on the right-hand side; that is, a model that emerged out of the structure, conduct 3 The Java Reconstruction Fund is a multidonor reconstruction fund pledged by the European Commission, the Netherlands, the UK, Canada, Finland and Denmark. It is governed by a Steering Committee and co-chaired by the Government of Indonesia and the European Commission, with the World Bank as Trustee. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

238

and performance literature (Shepherd, 1972; Scherer and Ross, 1990). Most of this literature aims to explain why the impacts of the crisis vary across firms, even within the same industrial category. It also mostly uses datasets of medium and large firm-level surveys to achieve this goal. In general, it concludes that the variation of impacts across firms can be explained by firm characteristics, such as ownership, size, financial pressure, age, location, economy of scale and export orientation, and industrial/market characteristics, such as industry factor intensity, product market competition and protection. It is important to note that little of this literature actually focuses on the determinants of firm recovery or on small and micro enterprises. Literature on the role of aid (broadly defined to include government and non-governmental organization interventions) in the development of micro, small and medium enterprise (MSME) has been relatively plentiful, although most studies are not in the context of recovery after a natural disaster. The focus is mostly on the role cheap credit provision and input subsidies play in a firm’s performance (World Bank, 1994; Batra and Mahmood, 2001; Levine, 2006). So far, the conclusions are ambiguous. Some literature supports the argument that aid will develop MSME on the basis that they are typically productive but that they face constraints, such as access to credit, some material inputs and proper information. Hence, if aid can be delivered to eliminate these obstacles, MSMEs will grow even faster. However, some argue that aid might not effectively support MSME development, at least in the medium to long term. Aid and intervention could reduce the competitiveness of MSME. Furthermore, when the business environment is bad (e.g. due to too much regulation or the existence of entry barriers), MSME will not develop anyway, with or without aid (Levine, 2006). The firm/enterprise model used in this paper is as follows. Let us define Yi,–1 as the average monthly sales of firm i before the earthquake, Yi,0 as the first month sales of firm i just after the earthquake and Yi,t as the monthly sales of firm i t months after the earthquake. A firm’s initial damage due to the earthquake can be defined as

IDi =

(Yi ,−1 − Yi ,0 ) Yi , −1

(1)

and the damage level remaining at t months after the earthquake as

Di ,t =

(Yi ,−1 − Yi ,t ) Yi ,−1

.

(2)

The recovery rate for firm i t months after the earthquake, hence, can be calculated as: © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

Ri ,t =

( IDi − Di ,t ) IDi

⋅100% for

( IDi − Di ,t ) IDi

239

F = 0.45 for model 1 and Prob > F = 0.10 for model 5). The control group among donors is the foreign NGO, which tends to provide the highest amount of funding to a given type of firm. By February 2007, it is evident that both domestic government organizations and domestic NGOs provided significantly lower amounts of funding. Meanwhile, although foreign governmental organizations tend to give smaller grants than foreign NGOs, the difference is not significant. By August 2007, domestic government organizations were able to deliver more funds to their grantees and so the amount per firm that they gave matches the 6 Appendix 1 shows the results when using monthly sales as a measure of firm size. The results in Appendix 1 are relative similar to those in Table 2, where number of workers is used as a measure for firm size. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS Table 2

245

Estimation results for size of grants given February 2007

Donor characteristics Domestic NGO Foreign government Domestic government Firm level of damage Building damage (%)

Model 1

Model 2

-8.982*** (1.776) -2.382 (3.838) -3.159* (1.724)

-8.942*** (1.772) -2.759 (3.854) -3.172* (1.718)

-2.515* (1.296) -4.493 (4.211) -1.623 (1.303)

-2.486* (1.288) -4.760 (4.229) -1.633 (1.298)

0.042*** (0.011)

0.037*** (0.011) 0.015 (0.009)

0.054*** (0.013)

0.051*** (0.013) 0.010 (0.010)

-0.001 (0.035) 0.029 (0.022) -0.003 (0.090) -0.827 (1.222) 0.000 (0.014) 0.049 (0.043)

-0.000 (0.035) 0.032 (0.022) -0.016 (0.086) -0.615 (1.249) -0.003 (0.014) 0.055 (0.043)

-0.019 (0.030) 0.004 (0.022) 0.067 (0.116) -2.204* (1.241) 0.009 (0.013) 0.017 (0.033)

-0.018 (0.031) 0.006 (0.022) 0.058 (0.111) -2.054* (1.220) 0.006 (0.013) 0.021 (0.034)

-0.389 (0.937) 0.162 (0.114) -0.004** (0.002) 1.021 (0.785)

-0.237 (0.946) 0.151 (0.112) -0.004** (0.002) 0.991 (0.779)

-0.597 (0.872) 0.026 (0.097) -0.001 (0.002) 0.859 (0.888)

-0.489 (0.873) 0.018 (0.098) -0.001 (0.002) 0.837 (0.887)

0.024 (0.048) -0.048 (0.042) -0.254 (0.368) 0.019 (0.027)

0.030 (0.048) -0.042 (0.042) -0.274 (0.368) 0.009 (0.027)

-0.054 (0.057) -0.003 (0.041) -0.504 (0.374) 0.055** (0.025)

-0.050 (0.057) 0.001 (0.042) -0.518 (0.373) 0.049* (0.027)

Included -1.513 (2.254)

Included -1.707 (2.253)

Included 1.277 (2.371)

Included 1.140 (2.388)

5.785 (5.187) 248 0.380

5.201 (5.135) 248 0.385

10.531** (4.596) 248 0.271

10.117** (4.630) 248 0.274

Sale reduction (%) Firm characteristics Number of worker Assets per worker Loan per worker Living area of most worker (Bantul = 1) Portion of product sent outside Yogyakarta Number of years the firm has been established Owner characteristics Owner’s gender (male = 1) Owner’s experience Owner’s experience squared Has other source of income (yes = 1) Village characteristics Distance to the centre of Yogyakarta City Village head’s age Village head’s education Average level of building distroyed in the area Industrial characteristics Dummies of 1-digit ISIC Size of 3-digit ISIC cluster in a sub–district Other Constant N R2

August 2007 Model 3

Model 4

Note: Size of grant (Rp million) is the dependent variable and numbers in parentheses are standard errors. ***, ** and * denote significance at the 1, 5 and 10% levels, respectively. ISIC, International Standard Industrial Classification; NGO, non-government organization.

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

246

amount from foreign NGOs. Hence, to date, the amounts given by domestic government, foreign NGOs and foreign governments are not significantly different. The amounts of grants given by Indonesian NGOs were lower than the amounts given by other institutions. These results suggest that domestic governments and NGOs took relatively longer time to deliver their funding. Furthermore, in the end, the average size of the grant from domestic NGOs was usually smaller than that from other sources. The quality of government bureaucracy in Indonesia is relatively low, and so it is not surprising that they were slow to deliver funds. Domestic NGOs mostly do not have their own funds. They typically work as partners of foreign NGOs and governments who are not able to contribute otherwise. Therefore, it is important to note what an important role they played in making funds available to people affected by the earthquake. However, they do need to wait for funds to be delivered to them and then decide on worthy recipients. This process adds another layer of bureaucracy and creates additional delays. In addition, there is a cost associated with supporting these domestic NGOs, which is deducted from the amount granted. In contrast, people who receive support from domestic NGOs are usually willing to accept that they receive a smaller grant for two reasons. First, they know that domestic NGOs are not as rich as foreign NGOs or any government and, second, they realize that they might not receive support from any other source. Another important determinant of the size of a grant given to a firm is the degree of building damage done to a firm by the earthquake. First, donors tend to give larger grants to firms with a greater building damage. The amount by which sales were reduced due to the earthquake, which in this paper is defined as initial damage (see Equation 1), is not an important determinant. This is understandable. Donors do not have much prior knowledge and information about firms in the area and are unable to acquire this information quickly. Building damage is something that outsiders can easily see. No conclusion can be drawn as to which institutions tend to promise to provide a grant within the first month after the earthquake, particularly because there is no information as to how many firms were approached by a donor but were not promised grants and, in the end, did not receive any. However, the survey reveals that almost all institutions that later provided grants approached firms in the first month after the earthquake and promised these firms that they would provide a grant. Of the foreign NGOs that made such a promise, 17 percent were able to start delivering their support within the first 3 months after the earthquake. The figures are 13 percent for domestic NGOs, 33 percent for foreign government organizations and 9 percent for domestic government organizations. V.2 Firm recovery within 6 months (February 2007) Data from the first firm-level survey is used to facilitate OLS estimates of equation (7). There is no serious multicollinearity issue with these estimates. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

247

Misspecification bias is also not likely to be important, as indicated by the Ramsey RESET test (Prob > F = 0.37). The industrial/market characteristics (size of industrial cluster and dummy for 1-digit ISIC) are expected to work well to control variation across types of industry and so eliminate correlations among errors from firms with the same industrial characteristics. There is concern as to whether there is a simultaneity problem, particularly whether the size of the grant (gi,t in Equation 7) is determined by the expected rate of the firm’s recovery. The variables for being promised a grant and receiving the grant on time most likely do not suffer any simultaneity problems, because within this period it is most likely too early for donors to develop any expectations regarding the rate of a firm’s recovery. Firm and industrial characteristic variables are not expected to suffer any simultaneity problem either as they are measured as averages the 4 months before the earthquake. This paper uses the level of the firm’s building damage and donor characteristics (e.g. foreign NGO, domestic NGO, foreign government or domestic government organization) as instrumental variables (IV), because they do not relate to the firm’s recovery rate as well as the error terms of the OLS estimates, but they are significantly related to grant variables (see Table 2). IV estimation is then conducted; these results can be seen in Table 3 (model 1).7 The p-value of the Wu–Hausman test (0.42) indicates that the OLS estimates are consistent and asymptotically efficient.8 Because firms in the same location might have access to the same information and thus behave similarly, OLS estimates with robust standard errors (model 2 in Table 3) and OLS estimates with error terms are clustered around the kecamatan (sub-district) where the firms are located (model 3 in Table 3), are conducted.9 The present paper focuses its analysis on the result of the OLS cluster estimation. Initial damage, measured by percentage of sales reduction (Equation 1), is a significant determinant of the sample firms’ recovery rate. Those that were affected the most had the strongest interest in recovery. However, this can also mean that, in general, the recovery trajectory is concave rather than linear. Everything else being equal, there is a diminishing marginal rate of recovery. Among the firm characteristic variables, firm size is a significant determinant (at 10-percent significance level) of firm recovery. These results indicate that the smaller the firm, the faster it bounces back to its original level of production. Other firm characteristics do not seem to influence the rate of a firm’s recovery.

7 Appendix 2 shows the results when using monthly sales as a measure of firm size. The results in Appendix 2 are relative similar to those in Table 3, where number of workers is used as a measure for firm size. 8 This is consistent with our interview with several donors. According to these donors, the expectation that a firm will recover faster is not a criteria in providing help, as this is too difficult for them to predict. 9 Sub-district dummies were not included in the model because sample sizes of affected enterprises in some sub-districts are small. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

Has other source of income (yes = 1)

Owner’s experience squared

Owner’s experience

Owner characteristics Owner’s gender (male = 1)

Number of years the firm has been established

Portion of product sent outside Yogyakarta

Living area of most worker (Bantul = 1)

Loan per worker

Assets per worker

Firm characteristics Number of worker

Firm level of damage Initial damage (%)

5.530 (4.210) -0.627 (0.604) 0.009 (0.012) -3.539 (4.159)

-0.393** (0.162) 0.228 (0.140) 0.333 (0.816) 2.114 (6.863) -0.065 (0.060) 0.081 (0.193) 5.362 (3.969) -0.821 (0.624) 0.014 (0.011) -4.184 (6.381)

-0.369* (0.211) 0.209 (0.130) 0.469 (0.356) 2.543 (9.225) -0.053 (0.051) 0.070 (0.122)

0.571*** (0.151)

Model 2

Model 1

0.573*** (0.092)

OLS robust

Feb 2007

5.362 (3.766) -0.821 (0.545) 0.014 (0.010) -4.184 (4.236)

-0.369* (0.206) 0.209 (0.140) 0.469 (0.476) 2.543 (7.630) -0.053 (0.062) 0.070 (0.156)

0.571*** (0.137)

Model 3

OLS cluster

Estimation results of a firm’s recovery

IV

Table 3

2.834 (4.596) 0.160 (0.597) 0.004 (0.011) 0.128 (4.472)

-0.389** (0.171) -0.180 (0.149) -0.053 (0.858) -2.488 (7.386) 0.021 (0.062) -0.068 (0.208)

0.248** (0.099)

Model 4

IV

3.256 (4.610) 0.141 (0.645) 0.004 (0.011) -0.298 (4.653)

-0.383* (0.216) -0.174 (0.185) -0.029 (0.549) -2.176 (7.221) 0.021 (0.064) -0.065 (0.197)

0.241* (0.138)

Model 5

OLS robust

Aug 2007

3.256 (4.415) 0.141 (0.470) 0.004 (0.009) -0.298 (3.636)

-0.383* (0.210) -0.174 (0.121) -0.029 (0.671) -2.176 (8.595) 0.021 (0.070) -0.065 (0.186)

0.241** (0.120)

Model 6

OLS cluster

ASIAN ECONOMIC JOURNAL 248

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

57.757** (22.765) 297 0.250

-0.658 (1.403) -1.250 (13.193) 1.393* (0.833)

Included 21.855* (11.220)

-0.293 (0.250) -0.198 (0.212) 1.723 (1.917) -0.082 (0.114)

57.136** (26.692) 297 0.270

0.457 (0.347) -11.267** (5.202) 0.936** (0.393)

Included 22.251* (11.445)

-0.321 (0.211) -0.150 (0.099) 2.100 (2.830) -0.076 (0.113)

57.136** (23.049) 297 0.270

0.457 (0.364) -11.267** (5.343) 0.936* (0.487)

Included 22.251* (12.350)

-0.321 (0.253) -0.150 (0.210) 2.100 (1.983) -0.076 (0.109)

62.359** (24.819) 295 0.146

1.086 (6.053) 0.312 (0.650) -0.847* (0.513)

Included 35.519*** (12.197)

-0.366 (0.267) -0.128 (0.220) 1.905 (2.007) 0.249* (0.130)

62.479** (24.926) 295 0.149

-1.396 (5.439) 0.281 (0.520) -0.554 (0.360)

Included 34.836*** (13.378)

-0.357 (0.291) -0.114 (0.222) 1.990 (2.245) 0.225* (0.129)

62.479* (37.963) 295 0.149

-1.396 (4.641) 0.281 (0.485) -0.554* (0.290)

Included 34.836*** (12.007)

-0.357 (0.296) -0.114 (0.259) 1.990 (2.117) 0.225 (0.198)

Note: Rate of recovery is the dependent variable and numbers in parentheses are standard errors. ***, ** and * denote significance at the 1, 5 and 10% levels, respectively. ISIC, International Standard Industrial Classification; IV, instrumental variable.

N R2

Other Constant

Grant by August 2007

Received grant in time (yes = 1) ¥ grant

Promise to receive grant (yes = 1)

Aid characteristics Grant by February 2007

Industrial characteristics Dummies of 1-digit ISIC Size of 3-digit ISIC cluster in a sub–district

Average level of building distroyed in the area

Village head’s education

Village head’s age

Village characteristics Distance to the centre of Yogyakarta City

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS 249

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

250

With reference to owner characteristics, none seem to be important determinants of firm recovery. Concerning village characteristics, distance from the centre of Yogyakarta city and the age of the village head are statistically significant at the 15-percent level. The further the village is from the central city of Yogyakarta, the slower the recovery of its firms. This seems to be logical. One argument could be that various infrastructure in a village far from the city of Yogyakarta might have experienced a slower pace of reconstruction, thus affecting the performance of enterprises in that village. Younger village heads seem to be more active in supporting the recovery of firms in their area. With regard to industrial characteristics, although not reported, four of the six parameters of the dummies for the 1-digit ISIC code are significant at the 5-percent level; the other two are significant at the 10-percent level. The cluster size of an industry within a sub-district is a significant determinant of a firm’s recovery. The positive sign of the parameter for the cluster size indicates that the greater the concentration of similar enterprises in a sub-district, the faster these enterprises recover. More information supporting the recovery processes seems to be available when similar enterprises are located in the same area. During the first 6 months, providing a promise to enterprises that they will receive some financial support significantly affects their recovery rate; that is, there is an indication of an ‘announcement effect’ in this case. The expectation of receiving a grant creates either an incentive to wait until the grant is delivered or a disincentive to exert greater effort to recover faster (by finding alternative funds through commercial financial markets). Although only significant at a 20-percent confidence level, receiving grants improves their rate of recovery. The larger the grant, the faster the recovery. Furthermore, those receiving the grants within 3 months after the earthquake show significantly superior performance. The main message to donors is probably as follows. First, it is important to ensure that support arrives as early as possible. Second, given the ‘announcement effect’ could be negative, when the delivery date is uncertain, it is probably better not to make promises to victims of a natural disaster but rather just supply the support unannounced when it is ready to be delivered.10 V.3 Firm recovery within a year (August 2007) Equation (7) then estimated using data from the second firm-level survey, using a strategy that was similar to estimates performed with data from the first survey. This time the Wu–Hausman test (p = 0.48) suggests that OLS estimation produces consistent and asymptotically efficient estimates. There is no need to conduct IV estimation (model 4 in Table 3). Because firms in the same location probably behave similarly, OLS estimates with robust standard errors and OLS estimates with error terms are clustered around the kecamatan (sub-district) 10 Models inserting different types of donor dummy were also conducted and all donor dummies are not significant. During interviews with firms, it was also revealed that recipients don’t care what type of organization provides them with support. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

251

where the firms are located (models 5 and 6, respectively in Table 3) are likely to produce relatively reliable results. The analysis below is based on the results of the OLS cluster estimation. After 1 year, initial damage and firm size are still significant variables in determining a firm’s recovery rate. Assets per worker is now an important determinant of a firm’s recovery at a 17-percent level, supporting the argument that the smaller the firm, the faster it recovers. As was the case with the first survey, none of the owner characteristics are significant. All village characteristics, including its distance from the centre of Yogyakarta, are no longer important. This could be because infrastructure reconstruction progressed well even in areas that were far from the city of Yogyakarta. Industrial characteristics are all still important. With regard to aid related variables, within 1 year, the ‘announcement effect’ diminishes. After a while, those who were late in receiving grants are able to catch up, and those who were promised a grant but did not receive one gave up and figured out how to recover without a grant. Similarly, enterprises that were neither promised nor received a grant found a way to survive without one. They have, overall, been successful, as the sign for the grant has become negative and weakly significant. V.4 Comparison with non-affected enterprises This subsection compares the performance, measured by monthly sales, of firms affected by the earthquake and unaffected firms. A firm’s monthly sales is used as the dependent variable in an equation similar to Equation (7), and the independent variables measure conditions either before the earthquake or before August 2007 (see below). Figure 3 shows the average monthly sales per worker (in million rupiah). This figure indicates that those affected by the earthquake have not been able to match the performance of those not affected. Table 4 shows the determinants of firm performance from the August 2007 sample. Because the August 2007 sample is relatively well distributed across sub-districts, Table 4’s estimates also include dummies for sub-districts in the model. Models 1 and 2 in Table 4 are results from OLS estimations without and with dummies for sub-districts, respectively, using information gathered in the first firm-level survey regarding the state of the firm before the earthquake. Models 3 and 4 make use of the whole dataset from the second firm-level survey. The samples include both firms affected and not affected and both those that received and those that did not receive grants. In models 5 and 6, firms not affected but that received grants were dropped from the samples. This analysis below focuses on OLS estimates with dummies for sub-districts. It is evident from model 2 that the dummy variable representing firms that will be affected by the earthquake is not significantly different from zero. In other word, before the earthquake, the performance of affected and unaffected firms was © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL Figure 3

252

Average firm’s monthly sales per worker (Rp million)

3

2.5

2

1.5

1

0.5

0 Before

Just after Affected

February 2007

August 2007

Not affected

similar. Model 4 shows that, a year after the earthquake, whether a firm was affected by the earthquake or not was a weakly significant and negative determinant of firm performance. In other words, there was a weak tendency for affected firms to have lower sales than affected firms. When those that were not affected but that received a grant are dropped from the dataset, the variable ‘affected by the earthquake’ is no longer significant (model 6). These results indicate that firms affected by the earthquake, whether they received grants or not, were able to match the performance of those not affected and that did not receive any grant within 1 year after the earthquake (i.e. cases where there was no mistargeting). They were not able, however, to match the performances of those not affected but that did receive a grant (the mistargeted cases). VI.

Conclusion

This paper attempts to understand the determinants of micro, small and medium enterprise recovery after the 2006 Yogyakarta earthquake and the role of aid. In particular, we investigate what firm characteristics determine the recovery rate of micro, small and medium enterprises; whether receiving aid is a significant determinant of the recovery rate; whether promising to provide aid creates an ‘announcement affect’ and, if it does, the effect on firms’ recovery rate; and whether receiving aid in time is an important factor. Two firm-level surveys were conducted to gather information related to this issue in Bantul District 6 and 12 months after the earthquake. There are two major © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS Table 4

253

Estimation results of a firm’s output Before Earthquake All Sample

Affected by the earthquake dummy (yes = 1) Firm Characteristics Current number of worker Current assets per worker Current loan per worker Current living area of most workers (Bantul = 1) Current portion of product sent outside Yogyakarta # of years the firm has been established Owner Characteristics Owner’s gender (male = 1) Owner’s experience Owner’s experience squared Has other source of income (yes = 1) Village Characteristics Distance from the centre of Yogyakarta city Village head’s age Village head’s education Dummies of sub-district Industrial Characteristics Dummies of 1 digit ISIC Size of 3 digit ISIC cluster in a sub-district Aid Characteristics Grant by Aug 07

Take Out Mistargeting

Model 2

Model 3

Model 4

Model 5

Model 6

0.219 (3.320)

0.910 (3.853)

-4.618 (3.723)

-6.917* (4.104)

-2.098 (4.516)

-1.605 (4.992)

1.386*** (0.152) 0.051 (0.106) 2.874*** (0.678) 1.832 (5.670) 0.158*** (0.052) -0.185 (0.172)

1.345*** (0.155) 0.035 (0.109) 2.897*** (0.687) 2.972 (5.838) 0.170*** (0.054) -0.104 (0.173)

1.201*** (0.149) -0.026 (0.086) 3.971*** (0.512) 5.513 (3.902) 0.081 (0.064) -0.024 (0.184)

1.151*** (0.153) -0.023 (0.087) 3.928*** (0.518) 5.005 (3.964) 0.106 (0.065) 0.047 (0.186)

1.204*** (0.140) 0.053 (0.081) 0.976* (0.534) 10.123*** (3.896) 0.182*** (0.062) -0.024 (0.174)

1.122*** (0.145) 0.046 (0.081) 1.033* (0.537) 10.136** (3.952) 0.216*** (0.063) 0.047 (0.176)

6.955** (3.409) -0.470 (0.426) 0.007 (0.008) -3.081 (3.487)

7.990** (3.552) -0.503 (0.430) 0.005 (0.008) -2.519 (3.654)

7.226** (3.626) -0.385 (0.451) 0.005 (0.008) -1.364 (3.719)

8.972** (3.766) -0.412 (0.457) 0.004 (0.008) -1.217 (3.899)

2.668 (3.484) -0.271 (0.431) 0.004 (0.008) -2.471 (3.604)

4.576 (3.608) -0.313 (0.436) 0.003 (0.008) -2.637 (3.767)\

-0.339* (0.200) 0.140 (0.176) 1.145 (1.552)

-0.759* (0.400) (0.189) -0.190 -0.190 (1.965) included

-0.363* (0.212) (0.187) 1.886 1.886 (1.666)

-0.388 (0.432) (0.203) 1.119 1.119 (2.133) included

-0.297 (0.207) (0.180) 2.391 2.391 (1.617)

-0.191 (0.431) (0.194) 1.338 1.338 (2.075) included

included 16.132* (9.348)

included 24.847** (10.127)

included 9.185 (9.797)

included 21.804** (10.657)

included 10.883 (9.523)

included 24.404** (10.355)

-0.012 (0.275) -4.276 (4.354) -0.305 (0.648)

-0.096 (0.292) -5.206 (4.522) -0.228 (0.679)

-0.123 (0.282) -4.548 (4.439) -0.369 (0.583)

-0.235 (0.299) -4.450 (4.588) -0.267 (0.610)

-0.918 (17.365) 423 0.470

-3.702 (24.865) 423 0.498

-6.398 (16.161) 372 0.327

-10.257 (22.031) 372 0.365

Received grant in time (yes = 1) x grant

N R-squared

All Sample

Model 1

Promise to receive grant (yes = 1)

Other Constant

August 07

-12.165 (16.430) 423 0.346

26.721 (23.108) 423 0.384

Note: Monthly output (Rp million) is the dependent variable and numbers in parentheses are standard errors. ***, ** and * denote significance at the 1, 5 and 10% levels, respectively. ISIC, International Standard Industrial Classification.

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

254

weaknesses in these surveys. First, because prior information is not available as to how many micro, small and medium enterprises existed in Bantul District and their exact location, it is difficult to generate a high quality sample frame. The community survey conducted for this paper does help, but the quality of information provided by village offices varies according to the quality of administration in those villages. Second, the survey samples are rather small. Thus, dummy variables for village clusters or narrowly defined industries (ISIC 3-digit or more) cannot be used in the model. Taking these weaknesses into account, several conclusions can be drawn from the analysis of this paper. The first group of conclusions is related to donor characteristics. First, the results of this paper indicate that donors probably do not have prior information about affected firms and are not able to acquire this information quickly. Thus, the distribute support based on something that is easily observable. In this case, the degree of building damage was an important factor, even though it may not actually reflect how a firm’s performance (i.e. its level of sales) was affected by the natural disaster. This paper also illustrated that some firms whose level of sales was not affected received aid. Second, in general, foreign NGOs’ governments are faster in delivering their support to firms than other institutions; in the first 6 months, those firms with foreign NGO and government support received a larger grant than those supported by other institutions. The low quality of domestic government bureaucracy and the fact that most domestic NGOs have to wait for funds to come from other sources could explain this slow delivery. A year after the event, domestic government institutions had been able to deliver funds to firms to the same extent as foreign NGOs and governments. The situation was not the same for domestic NGOs. The second group of conclusions is related to recovery, initial damage, and firm, owner and industrial characteristics. First, the initial damage level is a determinant of a firm’s recovery. Not much can be done about this, because the extent of damage is a random process. Second, smaller enterprises turn out to be more resilient to the natural disaster’s impacts than larger firms; that is, smaller enterprises recover faster. This is not surprising because larger firms tend to have more capital than small firms (and, therefore, can suffer heavy damage when there is a disaster) and have more staff so that the organization of large firms is often disrupted. Large firms, therefore, should perhaps be more ready to consider taking out disaster insurance. Third, owner and village characteristics are not always important. The analysis in this paper indicates that the recovery rate of firms depends on the quality of some of the facilities available in the area where the firms are located. Fourth, the idea of clustering a certain type of industry in the same location is not a bad one. Clustering can help provide the support needed by firms to recover because similar firms are often willing to provide mutual support to each other in Indonesia in times of crisis. The third group of conclusions is related to recovery and aid provided to enterprises. In the short term, it is important to distribute aid as early as possible. The faster the aid is distributed, the better the impact on enterprises affected by the © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

255

natural disaster. However, when the timing of aid delivery is uncertain, it is probably better not to make too many promises of financial support to enterprises. The larger the amount of aid, the faster the firm recovers. In the long term, donors should anticipate that the effectiveness of aid will probably diminish. As time goes on, firms have to survive and improve performance by themselves. Firms will eventually have to fund their recovery through commercial financial markets and push themselves to be more competitive. In any case, most aid is only meant to ease the post-disaster transition. The Yogyakarta earthquake experience shows that affected firms are usually able to compete effectively with firms not affected by disaster, except when mistargeting occurs. Unaffected firms that receive mistargeted aid tend both to make use of it and to expand their sales (or market share) to the disadvantage of weakened competitors that have been affected by the disaster. This indicates that mistargeted aid has the effect of disadvantaging firms affected by natural disasters and should be avoided. Although the time factor is critical, donors should develop clear and appropriate criteria to deploy aid.

Appendix Table A1

Estimation results for size of grants given (using monthly sales as a measure of firm size) February 2007

Donor characteristics Domestic NGO Foreign government Domestic government Firm level of damage Building damage (%)

Model 1

Model 2

-8.898*** (1.731) -2.432 (3.812) -3.019* (1.681)

-8.860*** (1.730) -2.794 (3.847) -3.030* (1.680)

-2.475* (1.332) -4.109 (4.089) -1.609 (1.358)

-2.445* (1.325) -4.396 (4.121) -1.617 (1.354)

0.040*** (0.011)

0.036*** (0.011) 0.013 (0.009)

0.054*** (0.013)

0.051*** (0.012) 0.011 (0.010)

-0.019 (0.014) 0.048** (0.022) -0.242*** (0.078) -0.610 (1.223)

-0.018 (0.014) 0.049** (0.022) -0.238*** (0.078) -0.446 (1.245)

0.002 (0.017) 0.011 (0.019) -0.003 (0.065) -2.170* (1.226)

0.003 (0.016) 0.012 (0.019) -0.000 (0.065) -2.040* (1.204)

Sale reduction (%) Firm characteristics Size of sale Assets per worker Loan per size of sale Living area of most worker (Bantul = 1)

August 2007 Model 3

Model 4

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

256 Table A1

(continued) February 2007

Portion of product sent outside Yogyakarta Number of years the firm has been established Owner characteristics Owner’s gender (male = 1) Owner’s experience Owner’s experience squared Has other source of income (yes = 1) Village characteristics Distance to the centre of Yogyakarta city Village head’s age Village head’s education Average level of building distroyed in the area Industrial characteristics Dummies of 1-digit ISIC Size of 3-digit ISIC cluster in a sub–district Other Constant N R2

August 2007

Model 1

Model 2

Model 3

Model 4

0.009 (0.013) 0.048 (0.042)

0.006 (0.013) 0.054 (0.042)

0.005 (0.013) 0.015 (0.033)

0.003 (0.013) 0.020 (0.034)

-0.059 (0.959) 0.146 (0.113) -0.004* (0.002) 1.156 (0.767)

0.066 (0.962) 0.135 (0.111) -0.004* (0.002) 1.129 (0.763)

-0.623 (0.879) 0.034 (0.097) -0.001 (0.002) 0.892 (0.908)

-0.523 (0.880) 0.026 (0.097) -0.001 (0.002) 0.871 (0.906)

0.027 (0.047) -0.047 (0.041) -0.225 (0.369) 0.022 (0.026)

0.032 (0.047) -0.041 (0.041) -0.242 (0.369) 0.012 (0.027)

-0.055 (0.057) -0.004 (0.041) -0.498 (0.376) 0.055** (0.024)

-0.051 (0.056) 0.001 (0.042) -0.512 (0.375) 0.048* (0.026)

Included -1.317 (2.199)

Included -1.483 (2.197)

Included 1.546 (2.255)

Included 1.415 (2.268)

4.938 (5.141) 248 0.392

4.436 (5.103) 248 0.397

10.433** (4.653) 248 0.269

10.035** (4.691) 248 0.273

Note: Size of grant (Rp million) is the dependent variable and numbers in parentheses are standard errors. ***, ** and * denote significance at the 1, 5 and 10% levels, respectively. ISIC, International Standard Industrial Classification; NGO, non-government organization.

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS Table A2

257

Estimation results of a firm’s recovery (using monthly sales as a measure of firm size) February 2007

Firm level of damage Initial damage (%) Firm characteristics Size of sale Assets per worker Loan per size of sale Living area of most worker (Bantul = 1) Portion of product sent outside Yogyakarta Number of years the firm has been established Owner characteristics Owner’s gender (male = 1) Owner’s experience Owner’s experience squared Has other source of income (yes = 1) Village characteristics Distance to the centre of Yogyakarta City Village head’s age Village head’s education

August 2007

IV

OLS robust

OLS cluster

IV

OLS robust

OLS cluster

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

0.593*** (0.092)

0.590*** (0.136)

0.590*** (0.153)

0.268*** (0.099)

0.260* (0.134)

0.260** (0.122)

-0.073 (0.047) 0.316** (0.136) 0.102 (0.785) 1.288 (6.720) -0.094 (0.058) 0.054 (0.195)

-0.067 (0.059) 0.294** (0.139) 0.359 (0.302) 1.940 (7.443) -0.083 (0.062) 0.041 (0.162)

-0.067 (0.057) 0.294** (0.132) 0.359* (0.193) 1.940 (9.457) -0.083 (0.059) 0.041 (0.131)

-0.146*** (0.050) -0.101 (0.142) 0.630 (0.761) -3.457 (7.130) 0.013 (0.060) -0.117 (0.207)

-0.145 (0.105) -0.094 (0.170) 0.644 (0.400) -2.980 (6.784) 0.013 (0.063) -0.114 (0.199)

-0.145 (0.101) -0.094 (0.112) 0.644* (0.337) -2.980 (8.316) 0.013 (0.067) -0.114 (0.175)

5.795 (4.263) -0.591 (0.598) 0.009 (0.011) -3.384 (4.232)

5.523 (3.855) -0.760 (0.556) 0.013 (0.011) -4.177 (4.273)

5.523 (3.947) -0.760 (0.665) 0.013 (0.012) -4.177 (6.433)

3.419 (4.590) 0.182 (0.596) 0.003 (0.011) -0.352 (4.462)

3.948 (4.692) 0.159 (0.644) 0.003 (0.012) -0.909 (4.664)

3.948 (4.643) 0.159 (0.430) 0.003 (0.008) -0.909 (3.692)

-0.350 (0.249) -0.175 (0.212) 2.108 (1.909) -0.084 (0.115)

-0.371 (0.254) -0.128 (0.212) 2.332 (1.995) -0.076 (0.112)

-0.371* (0.218) -0.128 (0.113) 2.332 (2.821) -0.076 (0.111)

-0.478* (0.264) -0.082 (0.219) 2.109 (1.995) 0.236* (0.129)

-0.464 (0.286) -0.064 (0.225) 2.210 (2.209) 0.206 (0.130)

-0.464* (0.282) -0.064 (0.262) 2.210 (2.039) 0.206 (0.202)

Included 26.626** (12.138)

Included 26.626** (11.106)

Included Included Included 40.807*** 39.766*** 39.766*** (11.726) (12.819) (10.171)

0.494 (0.368) -12.258** (5.403)

0.494 (0.347) -12.258** (4.845)

Average level of building distroyed in the area Industrial characteristics Dummies of 1-digit ISIC Included Size of 3-digit ISIC 27.107** cluster in a sub–district (10.892) Aid characteristics Grant by February 2007 -0.603 (1.377) Promise to receive grant -2.375 (yes = 1) (13.013)

0.528 (6.007)

-2.588 (5.324)

-2.588 (4.465)

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

ASIAN ECONOMIC JOURNAL

258 Table A2

(continued)

February 2007

Received grant in time (yes = 1) ¥ grant Grant by August 2007 Other Constant N R2

August 2007

IV

OLS robust

OLS cluster

IV

OLS robust

OLS cluster

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

1.274 (0.818)

0.835* (0.498)

0.835** (0.401)

0.251 (0.644) -0.901* (0.509)

0.214 (0.504) -0.533 (0.353)

0.214 (0.473) -0.533* (0.299)

53.727** (22.772) 297 0.242

53.276** (23.040) 297 0.262

59.924** (24.556) 295 0.158

60.083** (24.668) 295 0.161

60.083 (38.346) 295 0.161

53.276* (27.484) 297 0.262

Note: Rate of recovery is the dependent variable and numbers in parentheses are standard errors. ***, ** and * denote significance at the 1, 5 and 10% levels, respectively. ISIC, International Standard Industrial Classification.

References ADB (Asian Development Bank), 2005, From Disaster to Reconstruction: A Report on ADB’s Responses to the Asian Tsunami. Asian Development Bank, Manila. Athukorala, P. and B. P. Resosudarmo, 2005, The Indian Ocean tsunami: Economic impact, disaster management, and lessons. Asian Economic Papers, 4, pp. 1–39. Bappenas, Provincial and Local Governments of D.I. Yogyakarta, Provincial and Local Governments of Central Java and International Partners, 2006, Preliminary Damage and Loss Assessment: Yogyakarta and Central Java Natural Disaster. Bappenas, Jakarta. Batra, G. and S. Mahmood, 2001, Direct support to private firms: Evidence on effectiveness. World Bank Policy Research Working Paper 3170. World Bank, Washington, DC. Chen, H. J and H. T. Hsu, 2005, The role of firm size in controlling output decline during the Asian Financial Crisis. Journal of Economic Development, 30, pp. 103–29. Christoplos, I., 2006, Links between Relief, Rehabilitation and Development in the Tsunami Response. Tsunami Evaluation Coalition, London. Dekle, R., C. Karnchanasai, P. Hoontrakul, 2005, The Thai currency crisis: Financing constraints, high fixed costs, and corporate governance. Asian Economic Papers, 4, pp. 60–87. Dwor-Frecaut, D., F. Colaco and M. Hallward-Driemeier (eds), 2000, Asian Corporate Recovery: Findings from Firm-Level Surveys in Five Countries. World Bank, Washington, DC. Fukuchi, T., 2000, Econometric analysis of the effect of Krismon shocks on Indonesia’s industrial subsectors. The Developing Economies, 38, pp. 490–517. Jayasuriya, S., P. Steele, D. Weerakoon, 2005, Post-tsunami recovery: Issues and challenges in Sri Lanka. Asian Development Bank Institute Discussion Paper No. 39. Asian Development Bank Institute, Tokyo. JRF (Java Reconstruction Fund), 2008, Two Years after the Java Earthquake and Tsunami: Implementing Community Based Reconstruction, Increasing Transparency. World Bank: Jakarta. Levine, R., 2006, Should government and aid agencies subsidize small firms? In: Transforming the Development Landscape: The Role of the Private Sector (eds Brainard L.), pp. 66–73. Brooking Institution Press, Washington, DC. © 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd

LIVELIHOOD RECOVERY AFTER NATURAL DISASTERS

259

Lim, Y. and C. H. Hahn, 2003, Bankruptcy policy reform and total factor productivity dynamics in Korea: Evidence from micro data. National Bureau of Economic Research Working Paper No. 9810. National Bureau of Economic Research, Cambridge, MA. Mitton, T., 2002, A cross-firm analysis of the impact of corporate governance on East Asian financial crisis. Journal of Financial Economics, 64, pp. 215–41. Narjoko, D. and H. Hill, 2007, Winners and losers during a deep economic crisis: Firm-level evidence from Indonesian manufacturing. Asian Economic Journal, 21, pp. 343–68. Nazara, S. and B. P. Resosudarmo, 2007, Aceh–Nias reconstruction and rehabilitation: Progress and challenges at the end of 2006. ADBI Discussion Paper No. 70. Asian Development Bank Institute, Tokyo. Oh, I., J. D. Lee and A. Heshmati, 2008, Total factor productivity in Korean manufacturing industries. Global Economic Review, 37, pp. 23–50. Rungsomboon, S., 2005, Deterioration of firm balance sheet and investment behavior: Evidence for panel data on Thai firms. Asian Economic Journal, 19, pp. 335–56. Sato, Y, 2000, How did the crisis affect small and medium-sized enterprises? From a field study of the metal-working industry in Java. The Developing Economies, 38, pp. 572–95. Scherer, F. M. and D. Ross, 1990, Industrial Market Structure and Economic Performance, 3rd edn. Houghton Mifflin, Boston. Shepherd, W. G., 1972, The Economics of Industrial Organisation. Prentice-Hall, New Jersey. Telford, J., J. Cosgrave and R. Houghton, 2006, Joint Evaluation of the International Response to the Indian Ocean Tsunami: Synthesis Report. Tsunami Evaluation Coalition, London. UNDP (United Nations Development Programme), 2005a, The Post-Tsunami Recovery in the Indian Ocean: Lessons Learned, Success, Challenges and Future Action. UNDP, New York. UNDP (United Nations Development Programme), 2005b, Post-tsunami Recovery and Reconstruction Strategy. UNDP, New York. World Bank, 1994, Can Intervention Work? The Role of Government in SME Success. World Bank, Washington, DC. World Bank, 2008, An Evaluation of World Bank Assistance for Natural Disasters. World Bank, Washington, DC.

© 2012 The Authors Asian Economic Journal © 2012 East Asian Economic Association and Blackwell Publishing Pty Ltd