Advertisement

Economics of Disasters and Climate Change

, Volume 2, Issue 2, pp 203–224 | Cite as

The Impact of the 2006 Yogyakarta Earthquake on Local Economic Growth

  • Aloysius Gunadi Brata
  • Henri L. F. de Groot
  • Wouter Zant
Original Paper
  • 399 Downloads

Abstract

This paper focuses on the local economic growth impacts of a specific natural disaster, viz. the Yogyakarta earthquake in 2006 employing data at the sub-national level in two provinces in Java, Indonesia. Specifically, we are concerned with the heterogeneity in the response of the various economic sectors to the earthquake, the spatial economic spill-overs from the affected regions to the non-affected districts, and the overall implications of the earthquake on the relative position of the local economies of the affected districts. We find that the earthquake did affect the growth of some sectors in the affected districts, but the shock did not change the (industrial) structure of the local economy. All sectors still had positive growth some years after the earthquake, which indicates the existence of a recovery processes following the shock.

Keywords

Economic growth Yogyakarta earthquake 

JEL Classification

Q54 • R11 

Introduction

In recent years, there has been growing interest in the economic impact of natural disasters. However, the available results still show contrasting findings, which suggests that the literature on the impacts of natural disasters on economic growth is still inconclusive. For instance, Skidmore and Toya (2002) and Kim (2010) find that countries that were subjected to natural disasters showed faster economic growth after the disaster, suggesting a Schumpeterian ‘creative destruction’ effect. But Cueresma et al. (2008) conclude that the creative destruction dynamics most likely only occur in developed countries only. Cavallo et al. (2013) also do not find any statistically significant long-run effect of disasters, not even for very large disasters (with the exception of very large disasters that were subsequently followed by radical political upheaval).1

The common view on the contrasting findings of previous studies is that growth theory about the impact of natural disasters does not provide clear predictions, especially regarding the long-term impacts (Chhibber and Laajaj 2008; Loayza et al. 2012).2 It then implies that the impacts need to be identified on empirical grounds. Reviewing the literature on the macroeconomic impacts of natural disasters, Cavallo and Noy (2011) conclude that the emerging consensus is that natural disasters have only a negative impact on short-term economic growth. However, no consensus can be identified with regard to the long-run impacts of natural disasters. Cavallo and Noy (2011) also underline the need of more evidence on the channels of transmission and extension of natural disasters on economic growth. Meanwhile, two recent studies using meta-analysis seem to arrive at different conclusions (Klomp and Valckx 2014; Lazzaroni and Van Bergeijk 2014).3 Klomp and Valckx (2014) find that climatic and geological disasters have a short-run negative impact (which supports the expectation that the long-run output returns to its original growth path), while they find both short- and long-run negative impacts with regard to hydro-meteorological disasters (which suggests that the new long-run growth path is at a lower level than the initial path). In contrast, Lazzaroni and Van Bergeijk (2014) find no evidence that macroeconomic impacts differ owing to geographical factors and the climatic versus geological nature of disasters. They do not find significantly negative impacts of disasters in terms of indirect costs (income or output). One of the important implications of their study is that it is necessary to consider the heterogeneity of the impact of disasters on growth.

In view of this state of affairs, the economic impacts of natural disasters are still a challenging field to be explored. One of the important early contributions is the study by Loayza et al. (2012), who investigate not only the average aggregate impacts of different natural disasters, but also look at the sectoral level of the economy, motivated by the argument that different disasters can affect different economic sectors through different transmission channels. Their findings confirm that disasters do affect economic growth — but not always negatively, and that the effects differ substantially across disaster types and economic sectors. In addition, moderate disasters can have a positive growth effect on a certain sector, while severe disasters do not (see also Fomby et al. 2013; Noy and Vu 2010). However, recent studies argue that it is possible that different disasters, but with a similar level of disruptive impact, can affect growth differently. Using the synthetic control method,4 Barone and Mocetti (2014) find that similarly disruptive earthquakes in two different Italian regions resulted in large and contrasting long-term effects on the Gross Domestic Product (GDP) per capita. They argue that this reflects differences in total factor productivity related to the institutional setting or the role of initial conditions. Their other important conclusion is that the earthquake might intensify regional disparities in both economic and social development.

In this paper we focus on the local economic growth impacts of a specific natural disaster, viz. the Yogyakarta earthquake in 2006. While Loayza et al. (2012) use cross-country data in investigating the sectoral impacts of different natural disasters, we employ data at the sub-national level (district/city level) in two provinces in Java (Indonesia).5 Specifically, we are concerned with the heterogeneity in the response of the various economic sectors to the earthquake (Mehregan et al. 2012), the spatial economic spill-overs from the affected regions to the non-affected districts (Xiao and Nilawar 2013; Dekle et al. 2014), and the overall implications of the earthquake on the relative position of the local economies of the affected districts (see Cavallo and Noy 2011; Lazzaroni and Van Bergeijk 2014; Barone and Mocetti 2014). So far, studies that use sub-national data of developing countries are still scarce (see a study on Mexico [Rodriquez-Oreggia et al. 2013], and a study on Vietnam [Noy and Vu 2010]). Both studies do not study the economic shocks caused by a specific natural disaster. Therefore, we contribute to studies on a specific developing country using a single natural disaster event.

The next section discusses our theoretical framework. It is followed by a brief description of the Yogyakarta earthquake. We then discuss our empirical findings. The conclusions are presented in final section.

Theoretical Framework

Economic growth, an increase of the productive capacity on an economy, is important as this also reflects an increase in the standard of living. Likewise, different patterns of economic growth will affect the standard of living. Volatility and instability of economic growth is substantially larger in developing countries than in developed countries (Pritchett 2000). Based on historical data, Pritchett identified several growth patterns: steady growth (hills and steep hills), rapid growth followed by stagnation (plateaus), rapid growth followed by decline (mountains) or catastrophic fall (cliffs), continuous stagnation (plains), or steady decline (valleys). Shocks may affect these patterns, but the growth path itself, may influence how an economy reacts and recovers from shocks (Martin and Sunley 2015). A particular shock – the impact of a natural disaster – has recently receive increased attention in the literature. Natural disasters may also have major implications for the pattern of economic growth, including such growth patterns in developing countries.

Growth theory does not provide clear predictions with regard to the impacts of natural disasters on long-term growth rates (Chhibber and Laajaj 2008; Loayza et al. 2012). For instance, according to the neoclassical growth model, growth of per capita output is a function of the growth of the per capita capital stock which is a function of the savings rate and the population growth rate. The capital stock, the population growth rate, or both, can be affected by a natural shock. This capital stock can be broadly defined to include human and social capital that are important in explaining variation in economic growth. Natural shocks will then affect the composition of the factors of production which will affect the level of productivity. When natural disasters promote the adoption of new production technologies by local firms, we may expect that natural disasters will positively influence the growth of productivity at the local level. Replacing the damaged machines and other production facilities are examples of the adoption of new production technologies. However, when natural disasters destroy physical capital stock, that reduces the marginal productivity of labour (MPL) but labour demand increases, and then natural disasters will negatively affect economic growth. Most disasters cause many deaths and heavily injured people. This will affect the labour force. Once both the capital stock and the labour force decreases with a disaster, it is not clear what happens to the marginal product of labour. Using a simple Cobb-Douglas production function, this depends on the relative size of decrease of the capital stock and the labour force. For example, if the labour force decreases more than the capital stock, then the MPL will increase. These two possible effects mean that the total impact of natural disasters on the long-run economic growth can be positive or negative.

Chhibber and Laajaj (2008) outline four possible scenarios for the long-run impact of natural disasters (see also Klomp and Valckx 2014).6 As our focus is on economic growth, we use the annual growth of Gross Domestic Regional Product (GRDP) rather than the level of GRDP (see Fig. 1). This figure shows that all scenarios have the same rate of pre-disaster growth reflecting a normal growth path. It also shows the same short-term negative impact of the natural disaster in terms of a significant drop of economic growth. But some years after the disaster, the four scenarios show different growth patterns.
Fig. 1

Four scenarios of the long-run growth impact of natural disasters. Source: Modified from Chhibber and Laajaj (2008)

The Scenarios A and B indicate that disasters only have a temporary impact on economic growth. In other words, the natural disasters only affect the short-run growth negatively. To be specific, in these scenarios the destruction of the capital stock per worker will accelerate growth temporarily in the short run by increasing its marginal return because capital becomes relatively scarce, or because of a drop in capital-labour ratios. It then stimulates savings and investment inflows. Reconstruction programmes can increase the capital-labour ratio again. It should be noted that reconstruction investment in developing countries often takes a long time to implement due to the problem of capacity constraints and also of excess demand for certain skills that are in limited supply. Massive temporary flows of external aid assistance can also accelerate economic growth, but it then stops and moves back to its normal growth path.

Scenario C clearly shows that a natural disaster has a permanent impact on economic growth, whereby the new growth path is established at a lower level than the pre-disaster growth path. The reason is that the disaster has permanently reduced the stock of capital, as households and the private sector face difficulties in reinvesting. In Scenario D the replacement of capital with a better technology enhances the long-run growth rate of the economy. This scenario is known as the Schumpeterian ‘creative destruction’, in which disasters provide an opportunity to update the stock of capital and adopt new technologies that can improve total factor productivity.

As already mentioned, different types of disasters can create different scenarios. For instance, Chhibber and Laajaj (2008) note that an earthquake is more likely to be associated with scenarios B or D, as this disaster is usually followed by considerable reconstruction which may trigger expansion, and eventually technological change. Meanwhile, drought corresponds to scenarios A and C because the loss is generally restricted to the annual and seasonal production. Drought can be related to D, for instance, only if this disaster leads to a major investment in irrigation or other drought-reducing technologies. But this scenario is less applicable for deadly droughts that reduce land fertility. In short, different natural disasters can influence the pattern of economic growth differently.

Studies show, however, that it is possible that different economic sectors are affected differently by the earthquake. Mehregan et al. (2012) find that the 2003 Bam earthquake in Iran had no negative impact on the total employment of Bam County, but the growth varied among the major economic sectors. Loayza et al. (2012) ind that earthquakes lead to higher industrial growth not only because a reduction of capital increases the marginal return on investment, but because there is a growth boost from the demand for capital reconstruction. However, they note that severe earthquakes can result in opposite economic impacts. In addition, Von Peter et al. (2012) note that, although there were physical similarities between the Haiti and the New Zealand earthquakes in 2010, their economic impacts on the real economy were different due to the difference in their financial preparedness for dealing with natural disasters.

The differences between economic sectors in responding to the shock can be influenced by the sectoral composition of the economy (Groot et al. 2011; Ridhwan et al. 2014) and the linkages between these sectors in an economic system or by its supply and demand (see Loayza et al. 2012; Dekle et al. 2014). As the affected regions are no closed economies, the shock may then have positive or negative spill-over impacts on other areas, depending on the spatial linkages between the regions. For instance, Dekle et al. (2014) find that the decline in intermediate inputs produced in Tohoku largely explained the impact of the Tohuku earthquake on other regions. Therefore, the shocks to Tohuku caused by the earthquake are propagated spatially to other regions. Using data of income and employment in counties/parishes in the states of Alabama, Lousiana, and Missisippi, Xiao and Nilawar (2013) find a spatial demand shift from the core disaster area to the edge area, within three years of Hurricane Katrina.

Overall, the disaster may change the relative position of the local economy of the affected regions. Okuyama (2014) indicates that the Kobe earthquake in 1995 accelerated the hollowing-out process in the Kobe economy. Meanwhile, based on their study on the impact of two earthquakes in two Italian regions, Barone and Mocetti (2014) conclude that in the long term, the earthquake might exacerbate regional disparities in economic and social development. This reflects differences in total factor productivity mainly due to the differences in the institutional setting between the two regions. Lazzaroni and Van Bergeijk (2014) note that this institutional setting is one of important resilience factors that may influence the indirect costs of natural disasters.

An Overview of the 2006 Yogyakarta Earthquake

The Yogyakarta earthquake occurred on May 27, 2006, at 05:52 local time (or May 26, 2006, at 22:54 GMT) with its epicentre in the Indian Ocean at about 33 km south of the Bantul district in the Yogyakarta province (Bappenas 2006; see Fig. 2). According to the Bappenas report, this earthquake, which had a strength of 5.9 on the Richter scale (or 6.3 according to the USGS), was one of the most costly natural disasters in the developing world. It affected five districts in Yogyakarta province and six districts in Central Java province (see Table 1). With regard to the size of the impact, Bantul, Klaten, Yogyakarta, Sleman, Kulon Progo, and Gunung Kidul were heavily affected by the earthquake.
Table 1

Human and housing impacts of the Yogyakarta Earthquake

District

Province

Human impact

Housing impact

Death toll

Injured

Totally destroyed

Damaged

Bantul

Yogyakarta

4121

12,026

46,753

33,137

Klaten

Central Java

1041

18,127

65,849

100,817

Sleman

Yogyakarta

240

3792

14,801

34,231

Yogyakarta (city)

Yogyakarta

195

318

4831

3591

Gunung Kidul

Yogyakarta

81

1086

15,071

17,967

Kulon Progo

Yogyakarta

22

2179

6793

9417

Magelang

Central Java

10

24

499

729

Boyolali

Central Java

4

300

715

825

Sukoharjo

Central Java

1

67

1185

488

Purworejo

Central Java

1

4

144

760

Wonogiri

Central Java

0

4

23

70

Bappenas (2006: Table 2, Table A.1)

Bappenas (2006) notes that these six districts are very densely populated with 4.5 million inhabitants. Most of the people in the affected areas are relatively poor especially in Bantul, Klaten, Kulon Progo and Gunung Kidul. In addition, the report informs that most of the people in the affected areas, especially in Bantul and Klaten, share very similar characteristics and living conditions. Per capita income of the six affected districs is reported to be about 60% of the national average. Agriculture is not an important activity in the city of Yogyakarta as it is a high-density urban area. However, this sector still contributes more than 20% of regional GDP in other affected districts. Other important sectors in these areas are (1) manufacturing, (2) trade, hotels, and restaurants, (3) financial services, and (4) other services. It is important to note that the Yogyakarta province is a major Indonesian centre of learning as it has a high concentration of schools and universities.

According to Bappenas (2006), the scale of damage caused by this earthquake was larger than the damage caused by the Aceh tsunami in 2004, especially because the population densities in Yogyakarta and Central Java were higher than in Aceh and even in Java as a whole. This earthquake caused damage and losses of Rp 29.1 trillion (US$ 3.1 billion) or 41% of regional GDP (see World Bank and GFDRR 2012). The most affected sector was the private sector (households and private companies). Housing was the sector most impacted by the earthquake. The Bappenas report notes that most of the affected houses were between 15 and 25 years old, and less than 3% were houses of traditional design. The earthquake has had limited impact on public and private infrastructure. Based on the preliminary estimates, the reduction in economic activity was expected to cause around 130,000 job losses, equal to 4% of the total pre-earthquake employment in the affected areas (Bappenas 2006: pp. 74–75). Home-based industries (small-medium enterprises) were concentrated in the areas affected by the earthquake, especially in Bantul, which had the largest decline in manufacturing sector activity (see World Bank 2012; Resosudarmo et al. 2012). Several areas in the Central Java and Yogyakarta provinces are known as the centres of production for furniture, ceramics, handicrafts and batik. In short, considering its impacts, this earthquake was a large exogenous shock for Central Java and Yogyakarta province.

The rehabilitation and reconstruction fund from government sources was estimated at Rp 11.7 trillion or 40% of the estimated total damage and losses (Bappeda Provinsi DI Yogyakarta et al. 2008). The rehabilitation and reconstruction fund (including contributions from donors, NGOs, and other communities) has been allocated to four activities: housing and settlements (84.4%), public facilities (10.6%), economic recovery (4.9%), and disaster risk mitigation (0.1%). Until 2007, Rp 7.9 trillion has been delivered for post-earthquake recovery programs in which the largest contributor was the Government of Indonesia (85%). About two-thirds (63%) of the fund from the central and local government for the first two years (2006–2007) has been allocated to Yogyakarta province and 37% to the Central Java province. The two most affected districs, Bantul and Klaten, received the largest portion of funds in the Yogyakarta province the and Central Java province, respectively.

It is interesting that the reconstruction in Java is noted to have procedeed faster than in Aceh and Nias after the Indian Ocean tsunami (World Bank 2007; see also Leitmann 2007). Two years after the disasters in Aceh and Nias, 45% of the required houses had been rebuilt. Meanwhile, the percentage of houses completed in Central Java and Yogyakarta up to March 2007 was 52% (in Yogyakarta 62.5% and in Central Java 33%). This achievement can be attributed to several factors. First, the Yogyakarta earthquake did not sweep away functional materials, so they could be reused. Second, transitional housing provided material inputs for the reconstruction of permanent housing. Third, the capacity of the government and the condition of infrastructure were less severely affected in Java than in Aceh and Nias. Fourth, prices of goods and services in Yogyakarta in the months after the earthquake were relatively stable. The mobility of goods and services from outside the region was facilitated by the relatively undamaged infrastructure and proximity to sources of labour and material inputs. These four factors are important explanations for the success of physical or infrastructure reconstruction. However, the Java Reconstruction Fund (JRF) reports that one year after the earthquake the economy of the region had not yet fully recovered as business activity was characterised by significantly lower production capacity, with sales falling far below the pre-earthquake levels (World Bank 2007).

Investigating the Economic Impact of the Earthquake

The 2006 Yogyakarta earthquake is an exogenous shock that we consider to be a natural experiment with a number of affected districts and a number of unaffected districts. The different economic performance of these two groups measures the impact of the earthquake (see Xiao and Nilawar 2013, Xiao and Feser 2014). We have 40 district in our data set: 35 districts in Yogyakarta province, and 5 districts Central Java province. We simply classified six districts (Klaten, Bantul, Sleman, Kulon Progo, Gunung Kidul, and Yogyakarta city) as the affected group. Details on the effect of the earthquake by district are already discussed in the previous section. Other districts in the Central Java province are classified as the non-affected group.

In order to investigate the economic impact of the Yogyakarta earthquake in 2006, we focus on economic performance measured by real Gross Domestic Regional Product (GRDP at constant price 2000) compiled by the World Bank (The World Bank Indonesia Database for Policy and Economic Research).7 We will focus on the period of 2000–2009 in order to isolate the impact of the earthquake from another shock in the following years. This other shock is the eruption of the Merapi volcano that is located close to Yogyakarta, in October–November 2010 that affected Klaten, Magelang, Boyolali and Sleman), killed 386 people, destroyed more than 3000 houses, and caused losses and damage amounting to Rp 29.1 trillion or about US$ 403 million (Bappenas and BNPB 2011).

For our empirical purpose, local economic performance is represented by the annual GRDP growth, instead of the growth of per capita GRDP or the level of GRDP, for several reasons. First, natural disasters rarely cause a significant decrease in the population (see Von Peter et al. 2012). The BPS data confirm that there was no substantial change in the migration rate in the most affected province Yogyakarta. Recent migration in this province was 11% in 2000 and 8% in 2010. Second, we need employment data if we want to measure the sectoral productivity. Unfortunately, data on the number of people employed at the district level based on the Labour Survey (SAKERNAS) are available only for 2007–2012. In addition, as shown in Fig. 3, the economic impact of the earthquake is more evident when we use annual growth of GRDP (Panel B) instead of the GRDP level (Panel A). We would probably see its impact on the GRDP level if we used quarterly data, but we only have annual data.
Fig. 3

Growth of gross regional domestic product. Notes: AD: Affected districts (6 districts); NAD: Non-affected districts (34 districts); All: All districts (40 districts)

Both GRDP level and growth in Fig. 3 have been normalised using their 2005 figure. We present the economic performance of three groups: all districts (40); the affected districts (6); and the non-affected districts (34). Note, as shown in Table 1, the earthquake largely affected six districts (Bantul, Klaten, Kulon Progo, Gunung Kidul, Sleman, and the city of Yogyakarta), so we treat them as the most affected districts. Bappenas (2006) identified these districts as the heavily affected districts. Panel B suggests that the earthquake negatively affected GRDP growth, but there was a different response between the six affected districts and the rest of the districts. This indicates that the economic growth of Bantul, Klaten, Kulon Progo, Gunung Kidul, Sleman and Yogyakarta were more negatively influenced by the earthquake compared with the other 34 districts. Compared with the 2005 growth, the decrease in the affected districts was quite large, almost 70% lower than the growth rate just before the earthquake. The pre-earthquake growth rate was almost restored in 2008, but it then dropped again in 2010. One possible reason is that this was caused by the Merapi eruption in 2010 that also had a negative impact on the local economy. It implies that to isolate the impact of the earthquake from the possible impact of the eruption we need to focus on the period before the eruption. Another possible explanation is the global financial and economic crisis in 2008/2009. However, it is probably not the case for Indonesia as there is no difference in industry growth between the pre-crisis period and the crisis period (see Monroe and Mirzaei 2016).

Furthermore, the fact that the growth rate of the non-affected districts was lower than their growth rate in 2005 indicates the possibility that there may have been a spatial spill-over effect of the earthquake impact from the six affected districts to the other districts. Theoretically, non-affected districts could have a positive spill-over effect to the affected districts. It is then important to investigate the spatial influence of the earthquake. We note that the most-affected province, Yogyakarta, contributes only about 10% of the total GRDP of Central Java and Yogyakarta provinces, but some districts in these provinces have some specific economic roles. The city of Yogyakarta is well known as a tourist destination (the second most popular in Indonesia after Bali), and as the home of many universities. Other districts, especially Bantul, have many handicraft centres. Therefore, it is possible that the earthquake shock to districts in Yogyakarta province (and the Klaten district in Central Java) negatively (or positively) affected the growth of other districts, especially their neighbours. Since we focus on the sectoral growth, sectoral spill-overs within a district also have to be investigated.

The first aspect we aim to investigate is the inter-temporal impact of the earthquake. In order to measure the long-term economic impact of the 2006 Yogyakarta earthquake, we employ a dynamic panel model, adopted from Von Peter et al. (2012), as follows:

$$ {Y}_{si,t}=\kern0.5em {\propto}_i+\sum {\beta}_j{Y}_{si,t-j}+\sum {\gamma}_j{EQ}_{i,t-j}+{\varepsilon}_{si,t} $$
(1)
where Y si,t is the annual GRDP growth of sector s, in district i and in year t; EQ i,t is an earthquake dummy (1 for the six districts that were affected by the earthquake in 2006 for the year 2006, and 0 otherwise). The districts affected by the earthquake are Bantul, Klaten, Yogyakarta, Sleman, Gunung Kidul and Kulon Progo. The cumulative (or long term) effect of the earthquake is (% GRDP or GRDP sectoral) = (γ0 + ... + γj) /(1–β1 –...–β k ). This model will be estimated using panel district fixed effects8 with robust standard errors clustered by district not only for the growth of aggregate GRDP but also for the GRDP growth at the sectoral level for the period 2000–2009.

Since growth behaviour is not the same in every sector, there are possibly heterogeneous responses to the earthquake. In order to investigate this heterogeneity, we employ a general framework, emphasising that natural disasters may affect the demand for or supply of the output of a specific sector. For instance, reconstruction programmes certainly may have a positive impact on the growth of the construction sector, but this probably also applies to the mining and quarrying sector, as the latter provides housing materials. Another example is that the damage to production facilities in the manufacturing sector will reduce the demand for energy or other utilities. Inter-sectoral linkages are likely to affect responses to the earthquake, and are a possible source of the sectoral heterogeneity. The negative relationship between sectors may, however, also reflect substitution effects, especially between traditional and non-traditional sectors. Generally, traditional sectors, especially agriculture tend to lose their role as modern sectors become more prominent; but these traditional sectors can play a role as a buffer against shocks that hit the local economies.

Ideally we need detailed data like regional Input-Output matrices, in order to investigate these inter-sectoral relationships. As we do not have this type of data, we decided to use a simple model. The model is based on the hypothesis that the growth of a certain sector is influenced by the growth of other sectors measured by the total output of other sectors. For instance, the annual growth of the service sector can be affected by the fluctuation in manufacturing output. This type of approach has been used in Groot et al. (2011) on the crisis sensitivity of local economies in Europe. In addition, focussing on the regional heterogeneity of the impact of monetary policy in Indonesia, Ridhwan et al. (2014) determine the role of cross-regional industrial composition (as represented by the share of manufacturing output). The panel model is formulated as follows:

$$ {Y}_{si,t}=\kern0.5em {\propto}_i+{\beta Y}_{nsi,t}+\gamma {EQ}_{i,t}+{\varepsilon}_{i,t} $$
(2)
where Y si,t is annual GRDP growth of sector s of district i in year t; Y nsi,t is annual GRDP growth, excluding sector s, of district i, in year t; EQ i,t is a time dummy for the earthquake (1 for the six affected districts from 2006 until 2009, 0 otherwise). Note that the model is applied for GRDP at the sectoral level only for the period of 2000–2009.

The above models do not capture the spatial spill-over from the affected districts to the non-affected districts. As already noted, the Yogyakarta earthquake was a large natural disaster, so it could indirectly affect other districts, especially those which share a border with the most affected districts. It is then interesting to find out whether there was a spill-over impact from the affected districts, especially to their neighbouring districts. We will investigate this by estimating the following panel model:

$$ {Y}_{si,t}=\kern0.5em {\propto}_i+{\beta TY}_{ad,t}+\gamma {Border}_{i,t}+{\varepsilon}_{i,t} $$
(3)
where Y si,t is the annual GRDP growth of sector s of non-affected district i in year t; TY ad,i,t is the annual growth of total GRDP of the six affected districts in year t; Border is a dummy variable for non-affected districts that share administrative borders with the most affected districts. This dummy variable takes the value 1 for these districts from 2006 until 2009, and 0 otherwise. Coefficient β captures the impact of the spill-over effect from the affected districts to their neighbours. A positive sign of this coefficient indicates that there was a complementary relationship between the affected and non-affected districts, while a negative sign can be interpreted as evidence of a substitution relationship between the two groups. This type of dummy variable has also been used in Xiao and Nilawar (2013) to investigate the spatial demand shift from the core areas affected by Katrina to the edge areas. Coefficient γ reflects the indirect impact of the earthquake on the non-affected districts that share borders with the most affected districts. Looking at the administrative map of Java, we can identify five districts that share their border with the affected districts. These are Boyolali, Magelang, Purworejo, Sukoharjo, and Wonogiri. The expectation is that these five districts had different growth behaviour, which reflects the possibility of a spatial spill-over from their neighbours that were largely induced by the earthquake.

We have discussed our approaches to find out the local economic impacts of the earthquake. The findings may raise a further question regarding whether the earthquake changed the relative position of the affected districts in the spatial system of economic activity. The answer to this question will basically reflect the overall impact of the earthquake on the affected districts that probably also influenced the non-affected districts.

To describe this overall impact, we will compare the pre- and post-earthquake economic growth of the two groups of districts at the sectoral level. The pre-earthquake growth is the average annual growth between 2001 and 2005, while the post-earthquake growth is the average annual growth between 2006 and 2009. We then relate these growth rates to their initial comparative advantage level measured by the location quotient (LQ)9 for the respective sector. LQ is an analytical statistic that measures a region’s industrial specialisation relative to a larger geographic unit. It is used in spatial economics to measure the ‘revealed’ locational advantages of certain regions to attract and develop certain industries (Hoen and Oosterhaven 2006). On the one hand, a sector with a high LQ has a chance to grow faster than other sectors. However, if the sector makes an important contribution to the local economy and this sector is affected by natural disasters, then the local economy will also be affected. In case of a natural disaster, vulnerability of a leading sector to natural disasters will affect the whole local economy. Therefore it is important to have a focus on sectors that not only have a high LQ, but that also make a large contribution to the local economy. This contribution can be reflected by the size of the sector. When considering this issue we weight the growth variable by the economic size (GRDP) in 2005 and 2009 for, respectively, the pre-and post-earthquake period.

Empirical Results

The Inter-Temporal Impacts

First, we investigate the lag structure of the annual GRDP growth of the affected districts. Using the standard procedure, we find that almost all of the annual growth at sectoral level of the six affected districts have four years lag, except Trade, Hotels, and Restaurants in Klaten (three years lag) and total GRDP in Kulon Progo (two years lag). Therefore the optimal lag of the annual GRDP growth that we used is four years. Applying the same procedure to the earthquake variable, we find that we do not need a lag for this variable since its optimal lag is 0. This is in line with Fig. 3 which shows that in 2007 the economic growth of the affected districts was already in the recovery process. The results are presented in Table 2.
Table 2

The Inter-temporal impacts of the earthquake

Sector

Constant

Lag 1

Lag 2

Lag 3

Lag 4

Earthquake 2006

Cum. effect

R 2

Adj. R2

N

Agr

3.85***

−0.17**

−0.17

−0.10

−0.07

0.21

0.14

0.10

0.08

200

 

(0.30)

(0.06)

(0.10)

(0.06)

(0.06)

(0.87)

    

Min

10.96***

−0.30*

−0.35

−0.05

−0.43**

3.65

1.71

0.20

0.18

185

 

(2.43)

(0.15)

(0.19)

(0.11)

(0.13)

(8.77)

    

Man

5.14***

−0.02

−0.04

−0.07

−0.11

−3.49

−2.81

0.12

0.10

200

 

(0.70)

(0.06)

(0.07)

(0.07)

(0.07)

(2.41)

    

Util

9.34***

−0.12

−0.20*

−0.13

−0.03

−4.20*

−2.84

0.13

0.11

200

 

(1.48)

(0.08)

(0.09)

(0.09)

(0.07)

(1.78)

    

Con

8.86***

−0.01

−0.08

−0.31***

−0.06

9.33*

6.39

0.30

0.28

200

 

(1.64)

(0.10)

(0.10)

(0.06)

(0.09)

(4.29)

    

Trade

5.12***

−0.07

−0.03

−0.01

0.07

−1.73**

−1.66

0.10

0.08

200

 

(0.55)

(0.09)

(0.07)

(0.04)

(0.04)

(0.60)

    

Tran

7.33***

−0.02

−0.03

−0.18*

−0.19**

−1.14

−0.80

0.11

0.08

200

 

(0.78)

(0.11)

(0.07)

(0.07)

(0.06)

(1.59)

    

Fin

6.29***

−0.07

−0.07

0.01

−0.04

−7.60***

−6.50

0.34

0.33

200

 

(0.62)

(0.06)

(0.06)

(0.06)

(0.09)

(1.88)

    

Other

5.32***

0.18**

−0.10

0.01

−0.05

0.18

−0.19

0.07

0.04

200

 

(0.55)

(0.06)

(0.07)

(0.04)

(0.04)

(0.33)

    

GRDP

3.27***

0.18**

0.07

0.09

−0.03

−1.18**

−1.71

0.14

0.12

200

 

(0.36)

(0.06)

(0.08)

(0.09)

(0.05)

(0.42)

    

The panel data covering 40 districts between 2000 and 2009. Agr (Agriculture), Min (Mining and Quarrying), Man (Manufacturing), Util (Utilities), Con (Construction), Trade (Trade, Hotels and Restaurants), Trans (Transportation and Telecommunication), Fin (Financial Services), Other (Other Services), GRDP (Total GRDP including oil and gas). The dependent variable is the annual growth of GRDP of the respective sector. Two districts (Magelang city and Pekalongan city) have no output from the Mining and Quarrying sector. Robust standard errors are in parentheses (clustered by district)

* significant at the 5% level; ** significant at the 1% level; *** significant at the 0.1% level. The cumulative effect of the earthquake is (% GRDP) = (λ0 + ... + λj) /(1–β1 –...–βk). The same formula applies for GRDP at sectoral level

The results show that the earthquake had a significantly negative impact on the aggregate output (see row GRDP). At the time of the earthquake, this natural disaster made growth in the six affected districts 1.18% points lower than in non-affected districts (Lag 0 of the earthquake variables). The cumulative growth effect of the earthquake on the aggregate output is −1.71%, which suggests that, overall, the earthquake had a negative impact on the growth of the aggregate output of the affected districts.

Looking at the sectoral level, the results show that there was heterogeneity in the response to the earthquake across different sectors. At the time of the event, the earthquake negatively and significantly influenced the growth of financial, utilities, and trade, hotels and restaurants sector in the six affected districts. The construction sector received a positive influence from the earthquake in the year of the event, and it was statistically significant, indicating that there was an economic benefit from the rehabilitalion and reconstruction. According to Bappenas (2006), housing and public utilities are priorities for the first two years after the shock, while the revitalisation of the local economies is expected to be finished at the end of 2008. The earthquake shows a negative effect on the manufacturing growth, but its coefficient is statistically insignificant. Based on the cumulative effects, the three most negatively affected sectors were financial services, utilities, and manufacturing. On the other hand, the construction sector received the largest positive cumulative effect. The cumulative effect on the mining and quarrying sector is probably related to the large positive effect of the earthquake on the construction sector. Meanwhile, the cumulative effects on agriculture and other services are close to zero, indicating that these sectors were relatively unaffected by the earthquake.

To sum up, the findings so far confirm previous studies on the impact on growth of natural disasters (see, e.g., Loayza et al. 2012). Specifically, first, the results show that a natural disaster negatively affects economic growth specifically at the time of the event, especially for districts that were directly affected by the earthquake. Second, the results confirm the negative consequences of a severe natural disaster. It should be recalled that the 2006 earthquake has been categorised as a large natural disaster in Indonesia, and we estimate the earthquake impacts on the economy at the sub-national level that make more nuanced impacts. Third, different sectors have responded differently. As indicated by the cumulative impacts, the earthquake had no negative impacts on the primary sectors (the impact was even positive). But this earthquake had strong negative effects on the secondary and services activities. In other words, the non-primary sectors are more vulnerable to earthquakes than the primary sectors. Fourth, the results generally show that scenario A and scenario B of the impacts of the earthquake are the most likely for the affected districts (see Fig. 1). The results indicate the temporary impact of the earthquake suggesting that there was a chance for the economies of the affected districts to return to their pre-earthquake growth path.

The Sectoral Spill-Over Impacts

The results using panel district fixed effects with robust standard errors clustered by district are presented in Table 3. The table shows that the two sectors whose annual growth was significantly influenced by the growth of other sectors were agriculture and trade, hotels and restaurants. In other words, these sectors were highly sensitive to the growth of other sectors. The negative impact of the growth of other sectors on agricultural growth reflects a substitution effect between the agricultural and non-agricultural sector indicating that agriculture still can play a role as a buffer sector. This is not surprising, as it indicates the common response in developing countries. Therefore, when other sectors experienced negative or slow growth, agriculture received a positive spill-over. At the opposite side, growth of other sectors gave negative impact on agriculture. This impact especially came from construction sector, the only sector in the affected districts that had large growth after the earthquake as the benefits of post-earthquake reconstruction activities. Reconstruction activities needed a large number of workers and they were mainly agricultural workers.
Table 3

The sectoral spill-over impacts of the earthquake

Sector

Constant

Growth of other sectors

Earthquake

Interaction (Growth of other sectors X Earthquake)

R 2

Adj. R2

N

Agr

5.42***

−0.66*

1.16

−0.03

0.05

0.05

360

(1.27)

(0.26)

(3.03)

(0.68)

   

Min

2.76*

0.35

22.87

−3.63

0.07

0.06

333

(1.35)

(0.31)

(14.94)

(3.01)

   

Man

5.23***

−0.22

6.72

−1.95

0.10

0.09

360

(0.49)

(0.11)

(4.40)

(0.98)

   

Util

7.81***

−0.19

−15.31

2.79

0.03

0.02

360

(1.42)

(0.33)

(8.09)

(1.82)

   

Con

5.66***

0.11

21.61***

−5.43***

0.23

0.23

360

(1.50)

(0.36)

(3.46)

(0.60)

   

Trade

2.72***

0.42***

1.41

−0.26

0.10

0.09

360

(0.43)

(0.10)

(2.52)

(0.61)

   

Trans

5.77***

−0.10

−7.09

1.55

0.02

0.01

360

(0.66)

(0.16)

(3.69)

(0.91)

   

Fin

5.70***

−0.09

−13.11

2.51

0.08

0.07

360

(0.55)

(0.13)

(6.87)

(1.60)

   

Other

5.36***

−0.05

−3.22

0.59

0.00

<0

360

(1.23)

(0.29)

(1.88)

(0.40)

   

The panel data cover 40 districts between 2000 and 2009. Agr (Agriculture), Min (Mining and Quarrying), Man (Manufacturing), Util (Utilities), Con (Construction), Trade (Trade, Hotels and Restaurants), Trans (Transportation and Telecommunication), Fin (Financial Services), Other (Other Services), GRDP (Total GRDP including oil and gas). The dependent variable is the annual growth of GRDP of the respective sector. Two districts (Magelang city and Pekalongan city) have no output from the Mining and Quarrying sector. Robust standard errors are in parentheses (clustered by district)

* significant at the 5% level; ** significant at the 1% level; *** significant at the 0.1% level

It should be noted that looking at the interaction variable, we can say that the recovery of other sectors tended to decrease the post-earthquake growth of the construction sector in the affected districts. The positive coefficient of the earthquake on the growth of trade, hotels and restaurants indicates the role of this sector in supporting other sectors.

The Spatial Spill-Over Impact

The results using panel district fixed effects with robust standard errors clustered by district for the period of 2000–2009 are presented in Table 4. The results indicate that there was no statistically significant relationship between the growth rate of the aggregate GRDP of the non-affected districts and the growth rate of the affected districts (see, row GRDP). But the Border dummy variable shows a statistically significant positive coefficient, which indicates the positive spatial impact from the affected districts on their neighbours. At the sectoral level, we see that there were substitution effects between the affected districts and the non-affected districts. This substitution effect (indicated by a negative coefficient) is statistically significant for the utilities and construction sectors. Meanwhile, there was no difference in sectoral economic growth between districts that are neighbours of the affected districts and other districts.
Table 4

The spatial spill-over impacts of the earthquake (A)

Sector

Constant

Growth of total GRDP of the affected districts

Border

R 2

Adj. R2

N

Agr

1.12

0.30

1.08

0.00

<0

306

 

(2.00)

(0.46)

(0.84)

   

Min

6.09*

−0.21

1.26

0.00

<0

279

 

(2.68)

(0.61)

(1.41)

   

Man

5.13***

−0.18

0.68

0.01

0.00

306

 

(1.18)

(0.27)

(0.78)

   

Util

13.54***

−1.55*

−3.05

0.02

0.02

306

 

(2.87)

(0.64)

(1.58)

   

Con

11.16***

−1.19*

−0.39

0.02

0.01

306

 

(2.47)

(0.56)

(0.51)

   

Trade

3.71***

0.15

1.05

0.02

0.01

306

 

(0.85)

(0.19)

(0.58)

   

Trans

6.51***

−0.31

−0.55

0.00

<0

306

 

(1.30)

(0.30)

(0.52)

   

Fin

4.33***

0.17

−0.07

0.00

<0

306

 

(0.97)

(0.22)

(0.58)

   

Other

7.61***

−0.56

1.00

0.01

<0

306

 

(1.66)

(0.38)

(0.82)

   

GRDP

3.68***

0.12

1.00***

0.03

0.03

306

 

(0.58)

(0.13)

(0.21)

   

The panel data cover 40 districts between 2000 and 2009. Agr (Agriculture), Min (Mining and Quarrying), Man (Manufacturing), Util (Utilities), Con (Construction), Trade (Trade, Hotels and Restaurants), Trans (Transportation and Telecommunication), Fin (Financial Services), Other (Other Services), GRDP (Total GRDP including oil and gas). The dependent variable is the annual growth of GRDP of the respective sector. Two districts (Magelang city and Pekalongan city) have no output from the Mining and Quarrying sector. Robust standard errors are in parentheses (clustered by district)

* significant at the 5% level; ** significant at the 1% level; *** significant at the 0.1% level

Replacing the growth of the total GRDP of the affected districts with their sectoral growth gives different results (Table 5). As we can see, there were statistically significant relationships between the growth of agriculture, mining and quarrying, utilities, trade, hotels and restaurants, and transportation and telecommunication of the non-affected districts, on the one hand, and the growth of the same sector of the affected districts, on the other. These relationships were positive, except for transportation and telecommunication, which suggests the complementarity relationships between the two groups of districts. However, as in Table 4, there was no evidence of a difference in post-earthquake sectoral growth between the neighbours of the affected districts and those of other districts. A possible reason is that the local economies both in Yogyakarta province and Central Java province are already interconnected as a result of transportation development across the Java island, especially road, while the distance between main cities is relatively short (see Dick 2000). In addition, although the earthquake in 2006 destroyed some roads and bridges, but transportation was not the most affected sector. This sector needed rehabilitation and reconstruction, but the main transportation modes still existed. Since the two factors, good transportation and short distance between districts, affect the degree of economic mobility across districts, it is also difficult to fully figure out the influence of border dummy variable that is only measured by administative boundary.
Table 5

The spatial spill-over impacts of the earthquake (B)

Sector

Constant

Growth of GRDP of the respective sector of the affected districts

Border

R 2

Adj. R2

N

Agr

1.31**

0.41*

0.40

0.03

0.02

306

(0.43)

(0.15)

(0.86)

   

Min

4.71***

0.13*

0.95

0.01

0.01

279

(0.27)

(0.06)

(1.37)

   

Man

4.33***

0.00

0.75

0.01

<0

306

(0.20)

(0.06)

(0.82)

   

Util

4.86***

0.25**

−1.49

0.04

0.03

306

(0.63)

(0.08)

(1.61)

   

Con

5.81***

0.02

−0.01

0.00

<0

306

(0.32)

(0.04)

(0.51)

   

Trade

2.86***

0.31**

1.00

0.03

0.03

306

(0.54)

(0.11)

(0.55)

   

Trans

6.37***

−0.18*

−0.63

0.01

0.01

306

(0.54)

(0.08)

(0.53)

   

Fin

4.88***

0.04

−0.06

0.00

<0

306

(0.17)

(0.03)

(0.58)

   

Other

3.66***

0.37

1.22

0.02

0.01

306

(0.98)

(0.24)

(0.85)

   

GRDP

3.68***

0.12

1.00***

0.03

0.03

306

(0.58)

(0.13)

(0.21)

   

The panel data covering 40 districts between 2000 and 2009. Agr (Agriculture), Min (Mining and Quarrying), Man (Manufacturing), Util (Utilities), Con (Construction), Trade (Trade, Hotels and Restaurants), Trans (Transportation and Telecommunication), Fin (Financial Services), Other (Other Services), GRDP (Total GRDP including oil and gas). The dependent variable is the annual growth of GRDP of the respective sector. Two districts (Magelang city and Pekalongan city) have no output from the Mining and Quarrying sector. Robust standard errors are in parentheses (clustered by district)

* significant at the 5% level; ** significant at the 1% level; *** significant at the 0.1% level

The Overall Impact of the Earthquake on Economic Growth

Figure 4 presents the change of economic growth at sectoral level for the two groups of districts. Panel A (the affected districts) shows that the sectors that had a high initial comparative advantage generally tended to have a high annual growth during the pre-and post-earthquake period. The most important sectors in the affected districts based on their contribution to the local economy, as indicated by their economic size were: agriculture, manufacturing, trade, hotels and restaurants, and other services, although their comparative advantage was relatively low, except for other services. The annual growth of the manufacturing sector dropped from 4% to about 1%, but its economic size was relatively stable. This figure confirms previous analysis on the temporal impacts: that is the earthquake had a significant impact on the growth of the manufacturing industries, especially the home-based industries that were concentrated in this affected area.
Fig. 4

The Overall Impact on Local Economic Growth. Notes: Panel A: 6 affected districts. Panel B: 34 non-affected districts. Sectoral codes: 1. Agriculture, 2. Mining and Quarrying, 3. Manufacturing, 4. Utilities, 5. Construction, 6. Trade, Hotels and Restaurants, 7. Transportation and Telecommunication, 8. Financial Services, 9. Other Services. Periods: before the earthquake (b); and after the earthquake (a). Average annual growth of GRDP: before the earthquake (2001–2005), after the earthquake (2006–2009). Location quotient: before the earthquake (2000), after the earthquake (2005). Weight variable: before the earthquake (GRDP in 2005), after the earthquake (GRDP in 2009). Arrow (→): from before to after

We also see that the growth rate of the financial sector was negatively affected by the earthquake. In contrast, the figure shows that the post-earthquake growth rate of the agricultural sector was slightly larger than its pre-earthquake growth. This may indicate that agricultural activities had a positive contribution in absorbing the negative impacts of the earthquake on the other sectors. Meanwhile, there were no significant changes in the annual growth of trade, hotels and restaurants, and other services, as well as in their economic size. In other words, the important sector that was negatively affected by the earthquake was manufacturing, while trade, hotels and restaurants, and other services survived at their pre-earthquake growth level.

Looking at Panel B (the non-affected districts) in Fig. 4, we do not find evidence that the sectors with a strong revealed comparative advantage tended to grow faster than other sectors. For instance, the pre-earthquake annual growth of manufacturing and financial services was relatively stable at 5.5%. Three important sectors in the non-affected districts were agriculture, manufacturing, and trade, hotels and restaurants. The post-earthquake growth of agriculture was slightly larger than its pre-earthquake growth, while the opposite trend occurred in the manufacturing sector. Another important sector, trade, hotels and restaurants, did not show a significant change in its annual growth. All sectors with low comparative advantages showed an increase in their annual growth.

In order to check whether the growth changes were related to the earthquake, we need to compare the two groups of districts. We have identified that the most important sectors for the affected districts were agriculture, manufacturing, trade, hotels and restaurants, and other services. Three of them were also important for the non-affected districts (agriculture, manufacturing, and trade, hotels and restaurants). The growth trend of these important sectors was relatively stable. The difference was only in terms of the rate of change in growth between the two periods. For instance, the annual growth of manufacturing in the affected districts dropped by 3 percentage points (from 4% to 1%), while the decrease in the annual growth of the same sector in the non-affected districts was 1.7 percentage points (from 5.5% to 3.8%). Meanwhile, the annual growth rate of agriculture in the non-affected districts and the affected districts changed from 2.5% to 3.5% and from 2.3% to 3.7%, respectively. These figures suggest that the increase of agricultural growth and the decrease of manufacturing growth were caused by the earthquake and other factors. To be precise, the earthquake accelerated the decrease of manufacturing growth and the increase of agricultural growth in the affected districts. At the same time, the earthquake had no impact on the growth of trade, hotels and restaurants in both the affected districts and in the non-affected districts. One of the possible reasons for the slow growth of manufacturing was the global financial and economic crisis in 2008/2009. The most-affected industries were textiles, clothing and footwear, for instance, those in Pekalongan city, one of the important producers of textiles/batik in Central Java. The growth of these industries slowed down as a result of their weakened export demand and the increasing price of imported cotton (see Yusrina and Akhmadi 2013).10

Based on our analysis, the general conclusion is that the earthquake did affect the growth of some sectors in the affected districts, but this shock did not change the (industrial) structure of the local economy. The post-earthquake growth of some important sectors tended to be smaller than the pre-earthquake growth, which partially reflects the impact of the earthquake. However, it is important to note that after the earthquake all sectors still had positive growth, which indicates the existence of a recovery processes. As a result, the earthquake did not change the spatial distribution of economic activities between the affected districts and the non-affected districts.

Conclusion

The focus of this paper has been on the local growth impacts of the Yogyakarta earthquake in 2006. We identified some important findings based on our empirical analysis using output data at the sectoral level for 40 districts in the provinces of Yogyakarta and Central Java.

Investigating the temporal impact of the earthquake, we find that different sectors have responded differently to the natural disaster. In particular, the non-primary sectors were more vulnerable to the earthquake than the primary sectors. This finding indicates that there was a chance for the economy of the affected districts to return to their pre-earthquake growth path.

With regard to the spill-over impact of the earthquake across the sectors, our estimates show that different sectors had different sensitivity to the fluctuation of other sectors. We find that the most sensitive sectors were agriculture and trade, hotels and restaurants. However, focussing on the affected districts we find that the most sensitive sector was construction, as the recovery of other sectors reduced the growth of the construction sector, which suggests the temporary benefit of the reconstruction and rehabilitation programmes.

We have also investigated the spatial impact of the earthquake from the affected districts to the non-affected districts. There were spatial spill-over effects mostly in the form of complementary relationships between the affected districts and the non-affected districts. The post-earthquake aggregate growth of the neighbours of the affected districts tended to higher than that of other districts, but there was no difference at the sectoral level.

Overall, the earthquake did affect the growth of some sectors in the affected districts, but this shock did not change the (industrial) structure of the local economy. All sectors still had positive growth some years after the earthquake, which indicates the existence of the recovery processes. In other words, the earthquake did not change the spatial distribution of economic activities between the affected districts and the non-affected districts.

Footnotes

  1. 1.

    These radical political revolutions are the Islamic Iranian revolution after the 1978 earthquake, and the Sandinista revolution a few years after the 1972 earthquake in Managua, Nicaragua (see Cavallo et al. 2013).

  2. 2.

    Different possible scenarios are discussed in next section. See also Martin (2012) on regional economic resilience.

  3. 3.

    Klomp and Valckx (2014) use 25 studies (until 1 April 2013), while Lazzaroni and Van Bergeijk (2014) also study the direct effects of disasters (such as damage and fatalities), and build on 34 studies for the indirect effects (income or output) of disasters.

  4. 4.

    The synthetic control method is a counterfactual approach for comparative analysis. It is proposed by Abadie and Gardeazabal (2003) and Abadie et al. (2010).

  5. 5.

    For simplification, the term ‘district’ in this paper also includes cities.

  6. 6.

    Martin (2012) also provides similar scenarios, although he uses the framework of regional resilience.

  7. 7.
  8. 8.

    We use fixed effects to control for the spatial heterogeneity among the districts. This spatial heterogeneity is related to, for instance, the difference in the geographical position (North Coast vs. South Coast, coastal districts vs. inland districts) or in the industrial stage of development (the three biggest economic centres based on GRDP per capita are Kudus, Semarang city, and Surakarta city, see Brata 2009).

  9. 9.

    We used this formula: LQ s,i,t  = (GRDP s,i,t / GRDP i,t ) / (GRDP r,i,t / GRDP rt ), where s is the sector; i is the district (or local economy); t is the year; and r is the reference area or total Yogyakarta province and Central Java province (40 districts). The result of this formula is also referred to as a coefficient of specialisation. An LQ of 1.0 in manufacturing, for example, means that the district and the province have an equal sector share in mining. An LQ of 1.8 in agriculture means that the district has an 80% higher share in agriculture than the province.

  10. 10.

    This issue is beyond the scope of this paper.

Notes

Acknowledgements

The authors would like to thank two anonymous reviewers and Editor-in-Chief Prof. Ilan Noy for valuable suggestions and comments. Brata acknowledges a scholarship from the Indonesian Directorate General of Higher Education (DIKTI). The usual disclaimer applies.

References

  1. Abadie A, Gardeazabal J (2003) The economic costs of conflict: a case study of the Basque Country. Am Econ Rev 93(1):113–132CrossRefGoogle Scholar
  2. Abadie A, Diamond A, Hainmueller J (2010) Synthetic control methods for comparative case studies: estimating the effect of California tobacco control program. J Am Stat Assoc 105(490):493–505CrossRefGoogle Scholar
  3. Bappeda Provinsi DI Yogyakarta, Bappeda Provinsi Jawa Tengah, Bappenas (2008) Laporan Pemantauan dan Evaluasi Dua Tahun Pelaksanaan Rehabilitasi dan Rekonstruksi Pascabencana Gempa Bumi 27 Mei di Wilayah Provinsi DI Yogyakarta dan Provinsi Jawa Tengah (Monitoring and Evaluation of Two Years Implementation of Rehabilitation and Reconstruction in Post-disaster Earthquake May 27 in the Province of DI Yogyakarta and Central Java). Jakarta: BappenasGoogle Scholar
  4. Bappenas (2006) Preliminary damage and loss assessment Yogyakarta and Central Java natural disaster. Bappenas, JakartaGoogle Scholar
  5. Bappenas, BNPB (2011) Rencana Aksi Rehabilitasidan Rekonstruksi Pascabencana Erupsi Gunung Merapi Provinsi D.I. Yogyakarta dan Provinsi Jawa Tengah Tahun 2011–2013 (the action plan for rehabilitation and reconstruction in post-eruption of the Merapi volcano in the province of DI Yogyakarta and Central Java, 2011–2013). Jakarta: Bappenas & BNPBGoogle Scholar
  6. Barone G, Mocetti S (2014) Natural disasters, growth and institutions: a tale of two earthquakes. J Urban Econ 84(C:52–66CrossRefGoogle Scholar
  7. Brata AG (2009) Do geographic factors determine local economic development. Econ Manag Finan Markets 4(3):170–189Google Scholar
  8. Cavallo E, Noy I (2011) Natural disasters and the economy — a survey. Int Rev Environ Res Econ 5(1):63–102CrossRefGoogle Scholar
  9. Cavallo EA, Galiani S, Noy I, Pantano J (2013) Catastrophic natural disasters and economic growth. Rev Econ Stat 95(5):1549–1561CrossRefGoogle Scholar
  10. Chhibber A, Laajaj R (2008) Disasters, climate change and economic development in Sub-Saharan Africa: Lessons and directions. J Afr Econ 17(Supplement 2): 7–49Google Scholar
  11. Cueresma JC, Hlouskova J, Oberstreiner M (2008) Natural disasters as creative destruction? Evidence from developing countries. Econ Inq 46(2):214–226CrossRefGoogle Scholar
  12. Dekle R, Hong E, Xie W (2014) The regional spill-over effects of the Tohoku earthquake. University of Southern California, Department of EconomicsGoogle Scholar
  13. Dick H (2000) Representation of development in 19th and 20th century Indonesia: a transport history perspective. Bull Indones Econ Stud 36(1):185–207CrossRefGoogle Scholar
  14. Fomby T, Ikdea Y, Loayza NV (2013) The growth aftermath of natural disasters. J Appl Econ 28(3):412–434CrossRefGoogle Scholar
  15. Groot SPT, Möhlmann JL, Garretsen JH, De Groot HLF (2011) The crisis sensitivity of European countries and regions: stylized facts and spatial heterogeneity. Cambridge J Reg Econ Soc 4(3):437–456CrossRefGoogle Scholar
  16. Hoen AR, Oosterhaven J (2006) On the measurement of comparative advantage. Ann Reg Sci 40(3):677–691CrossRefGoogle Scholar
  17. Kim C (2010) The effects of natural disasters on long-run economic growth. The Michigan. J Bus 41:15–49Google Scholar
  18. Klomp J, Valckx K (2014) Natural disasters and economic growth: a meta-analysis. Glob Environ Chang 26(1):183–195CrossRefGoogle Scholar
  19. Lazzaroni S, Van Bergeijk PAG (2014) Natural disasters’ impact, factors of resilience and development: a meta-analysis of the macroeconomic literature. Ecol Econ 107(C:333–346CrossRefGoogle Scholar
  20. Leitmann J (2007) Cities and calamities: learning from post-disaster response in Indonesia. J Urban Health 84(3 Suppl):144–153CrossRefGoogle Scholar
  21. Loayza NV, Olaberŕia E, Rigolini J, Christiansen L (2012) Natural disasters and growth: going beyond the averages. World Dev 40(7):1317–1336CrossRefGoogle Scholar
  22. Martin R (2012) Regional economic resilience, hysteresis and recessionary shocks. J Econ Geogr 12(1):1–32CrossRefGoogle Scholar
  23. Martin R, Sunley P (2015) On the regional economic resilience: conceptualization and explanation. J Econ Geogr 1(1):1–42CrossRefGoogle Scholar
  24. Mehregan N, Asgary A, Rezaei R (2012) Effects of the bam earthquake on employment: a shift-share analysis. Disasters 36(3):420–438CrossRefGoogle Scholar
  25. Monroe T, Mirzaei A (2016) The impact of the global financial crisis on industry growth. Manch Sch 84(2):159–180CrossRefGoogle Scholar
  26. Noy I, Vu TB (2010) The economics of natural disasters in a developing country: the case of Vietnam. J Asian Econ 21(4):345–354CrossRefGoogle Scholar
  27. Okuyama Y (2014) Disaster and economic structural change: case study on the 1995 Kobe earthquake. Econ Syst Res 26(1):98–117CrossRefGoogle Scholar
  28. Pritchett L (2000) Understanding patterns of economic growth: searching for hills among plateaus, mountains, and plains. World Bank Econ Rev 14(2):221–250CrossRefGoogle Scholar
  29. Resosudarmo B, Sugiyanto C, Kuncoro A (2012) Livelihood recovery after natural disasters and the role of aid: the case of the 2006 Yogyakarta earthquake. Asian Econ J 26(3):233–259CrossRefGoogle Scholar
  30. Ridhwan MM, De Groot HLF, Rietveld P, Nijkamp P (2014) The regional impact of monetary policy in Indonesia. Growth Chang 45(2):240–262CrossRefGoogle Scholar
  31. Rodriquez-Oreggia E, De La Fuente A, De La Torre R, Moreno HA (2013) Natural disasters, human development and poverty at the municipal level in Mexico. J Dev Stud 49(3):442–455CrossRefGoogle Scholar
  32. Skidmore M, Toya H (2002) Do natural disasters promote long-run growth? Econ Inq 40(4):664–687CrossRefGoogle Scholar
  33. Von Peter G, Von Dahlen S, Saxena SC (2012) Unmitigated disasters? New evidence on the macroeconomic cost of natural catastrophes. Working Paper No. 394. Bank for International SettlementsGoogle Scholar
  34. World Bank (2007) One year after the java earthquake and tsunami: reconstruction achievements and the results of the java reconstruction fund. Java reconstruction fund progress report. World Bank, Washington, DC, p 2007Google Scholar
  35. World Bank (2012) Rekompak: rebuilding Indonesia's communities after disasters. Financial Literacy and Education Russia Trust Fund. Washington DC, World Bank GroupGoogle Scholar
  36. World Bank, GFDRR (2012) Advancing disaster risk financing and insurance in ASEAN member states: framework and option for implementation. IBRD/World Bank, Washington, DCGoogle Scholar
  37. Xiao Y, Feser E (2014) The unemployment impact of the 1993 US midwest flood: a quasi-experimental structural break point analysis. Environ Hazards 13(2):93–113CrossRefGoogle Scholar
  38. Xiao Y, Nilawar U (2013) Winners and losers: analysing post-disaster spatial economic demand shift. Disasters 37(4):646–668CrossRefGoogle Scholar
  39. Yusrina A, Akhmadi (2013) Impact of the global financial crisis on households in Kota Pekalongan. SMERU Working Paper November 2013Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of EconomicsAtma Jaya Yogyakarta UniversityYogyakartaIndonesia
  2. 2.Department of Spatial EconomicsVrije Universiteit AmsterdamAmsterdamNetherlands
  3. 3.Tinbergen InstituteAmsterdamNetherlands
  4. 4.Department of EconomicsVrije Universiteit AmsterdamHVNetherlands

Personalised recommendations