1 Introduction

The tsunami of 2004 devastated huge areas along the coastline of the Andaman Sea. Tourist areas as well as fishing villages along the western coastline of Southern Thailand were hit by the tsunami and destroyed (Siripong 2006). Five years after the disaster, most parts of the area are back to normal and can be considered as having fully recovered. A large segment of the population feels safe again thanks to the installation of the tsunami warning system and the introduction of regular drills, although this disaster still remains very present in the memory of the community (TRAIT Household Survey 2009). Since the area has now fully recovered, the whole process of recovery can be analysed, stretching from first aid provided directly after the event to the rebuilding process and the very latest mitigation measures and strategies. By studying the development of three coastal areas in southern Thailand, it can be seen that this process of recovery varies greatly between different groups of households in terms of time required.

The framework of vulnerability offers different perspectives important in examining the degree of vulnerability of the affected systems (Luers et al. 2003). Based on the idea that vulnerability is determined by two groups of factors, a distinction is made between the external factors of shocks to which a system is exposed and internal factors that represent the ability to cope with and recover from the external shock (Chambers 2006). In later frameworks, such as Turner, Blaikie, and Kelly and Adger (Turner et al. 2003; Blaikie 1994; Kelly and Adger 2000), three critical components were identified, namely the exposure to hazards, the sensitivity to harm caused by the shock, and the capacity to cope. The key elements of vulnerability frameworks and the interrelations between them have been studied intensively (Zou and Wei 2010). The interactions between these factors are often not quantified and are weighted equally in order to derive a vulnerability measurement system (Adger 2006; Briguglio 2003).

Using data from the recovery process after the tsunami of 2004 in Southern Thailand, this study attempts to quantify the interrelations within the vulnerability model in order to assess the contribution of the different perspectives on the recovery process of households in this area. The tsunami of 2004 allows for a study of the effects of rapidly occurring disasters and their covariant impact on a huge number of households living in the affected area. To assess the economic vulnerability of households, questionnaire-based survey data and remote sensing are integrated in a structural equation model (SEM) to reflect the process. This should allow for the derivation of a predictive vulnerability model for non-surveyed households from the ex-post perspective and for the identification of the factors contributing to a fast recovery and therefore a reduction in vulnerability.

2 Theoretical framework

Vulnerability can be seen as the degree to which a system or a system component is influenced by external stress, for example by a natural disaster or any other external shock (Turner et al. 2003). This approach allows a holistic view and encompasses the economic, social and environmental spheres of an area at risk (Birkmann 2006). The theoretical construct of vulnerability consists of three interconnected dimensions. These three dimensions are exposure, sensitivity, and coping capacity and/or resilience capabilities (Turner et al. 2003; Clark et al. 2000). In the following discussion, we will refer to coping capacity/resilience simply as resilience capability.

The BBC Framework of Bogardi, Birkmann and Cardona combines these three elements of vulnerability into one closed system and emphasises that vulnerability assessment does not stop at the assessment of disasters in the past (Birkmann 2006). In fact, it is crucial to assess the process of vulnerability and to include the efforts made to reduce the vulnerability after a disaster by influencing vulnerability, coping capacity or interventional systems such as risk management or early warning systems (cp. Fig. 1). This paper uses the interpretation of vulnerability proposed by Bogardi, Birkmann and Cardona, and Turner et al., who aim to analyse the elements of vulnerability as mentioned earlier (Birkmann 2006; Turner et al. 2003). The framework developed by Turner et al. in particular seeks to quantify and link the elements of vulnerability with each other, as well as with other scales, and offers the opportunity to transfer this framework to a statistical model (Turner et al. 2003; Adger 2006).

Fig. 1
figure 1

Vulnerability framework (Birkmann 2006)

Full size image

2.1 Exposure

Exposure in general can be understood as elements in a certain area that are at risk when a specific stress factor impacts upon this area (UNDP-BCPR 2004). In the case of assessing the economic vulnerability, this refers to the value of economic assets that are at risk in this area or that must endure the certainty of loss (European Spatial Planning Observation Network 2003; MunichRe 2002). For the purposes of assessing the economic vulnerability of households in the Phang-Nga Province, the personal and household belongings located in the area represent the exposed elements.

2.2 Sensitivity

The sensitivity of these exposed elements is represented by the location of the households, namely the height above sea level and the distance to the coastline. The second dimension of vulnerability, sensitivity, defines the degree to which an exposed element is affected by the impact of a natural disaster or external shock (Clark et al. 2000). In the case of tsunami vulnerability assessment, the sensitivity is represented by the location of a household and its housing conditions, which can involve differing degrees of resistance to the impact of the waves. The sensitivity towards indirect effects should also be included in this dimension. Interruption of infrastructure or the loss of income may have a strong impact on the situation of the household in the case of natural disasters. In addition, Turner et al. (2003) refer to the economic structures as an element of vulnerability from which a sensitive status of the household can arise. Limited livelihood options in particular can lead to a susceptible economic system. This can be identified in a monostructured economic system such as Khao Lak, where a strong focus is placed on the tourism sector, leading to a less-diversified income situation for households.

2.3 Resilience

The final dimension of vulnerability is the resilience, i.e. the ability of an exposed element to absorb or resist the impact that could result from a disaster by using available resources and capabilities (UN/ISDR 2004; Adger et al. 2005). Billing and Madengruber (2006) divide resilience into internal and external factors. The internal side refers to a household’s individual strategies and capacities to cope with the effects of the natural disaster, while the external side is determined by capacities provided by the government, non-governmental organisations and other institutions. The concept of sustainable livelihoods refers to the idea that only households that possess the required capabilities and assets can cope with external shocks (Chambers and Conway 1991). These capabilities and assets are divided into five categories of capital upon which the household can rely. Zou and Wei (2010), in their metastudy on vulnerability assessment, identified a vast number on indicators used to describe household vulnerability. According to their study on household level, the most important driving factors of vulnerability result from the lack of financial and human capital. Therefore, these capabilities are operationalised using indicators such as income level and savings to represent the financial capital and level of income. In addition to this, the internal resilience capabilities of a household are also determined by the education level, which may allow the household to react flexibly to external shocks. On the external side of resilience, financial support from either governmental agencies or non-governmental organisations is integrated into the model.

2.4 Outcome of vulnerability

Since all households in the sample had different levels of losses and needed different amounts of time to recover after the impact, the reconstruction time and the time with losses of income can symbolise how the factors of exposure to the tsunami, sensitivity to this impact and resilience interact and allow the households to recover from the shock (cp. Fig. 2).

Fig. 2
figure 2

Effects of covariant rapidly occurring natural disasters on the recovery time of households

Full size image

This can mean reaching either a state similar to the situation before the disaster or an improved condition that the household accepts or states themselves as being recovered (UN/ISDR 2004).

2.5 Combined elements of vulnerability

The discussion above leads to a combined vulnerability model linking the elements exposure, sensitivity and resilience (cp. Fig. 3).

Fig. 3
figure 3

Framework for vulnerability assessment

Full size image

From this discussion, hypotheses have been derived to test the relations between the single elements.

H1 :

A higher level of direct damage caused by a tsunami increases the outcome of vulnerability

H2 :

A higher level of indirect effects caused by a tsunami increases the outcome of vulnerability

H3 :

Greater internal resilience capacities reduce the outcome of vulnerability

H4 :

Greater external resilience capacities reduce the outcome of vulnerability

3 Transferring the theoretical framework to a statistical model

3.1 Statistical method

In the following analysis, the theoretical framework for assessing the vulnerability of households in the region will be transferred to a structural equation model (SEM) in order to investigate the relations between the different aspects of vulnerability. This method is a combination of different statistical methods, which integrates path analysis and factor analysis into a single model. In contrast to exploratory factor analysis, factors and causal relations are specified by the researcher in advance, guided by theoretical assumptions and tested on a specific dataset (Arbuckle 2006). The modelling is designed by the user in the form of a scheme representing the hypothetical relations between latent variables and between the manifest variables and their latent variables (Wold 1982). The latent constructs can thereby be designed as either a reflective or as a formative measurement model. Reflective manifest variables or effect indicators are reflected by the latent variable. The manifest variables influence the particularities of the indicators. For this purpose, they need to be highly correlated corresponding to the quality measures applied in factor analysis. Formative manifest variables, on the other hand, follow the idea of regression analysis and are assumed to determine the manifest variable. The latent variables are set up as a linear combination of the indicators and can therefore be understood as a hypothetical construct determined by the indicators (Diamantopoulos and Winklhofer 2001; Weiber and Mühlhaus 2010; Bagozzi and Yi 1988; Cassel 1999; Hulland 1999).

For this study, the partial least square approach (PLS) was chosen to test the theoretically driven model with the empirical data. This approach tends to be more robust for small sample sizes. The estimation of the final model aims to maximise the explained variance for indicators and latent variables in several iterations. Firstly, the values of the latent variables are estimated, providing the basis for the estimation of the inner and outer weights. In the final step, the relations between the latent variables are estimated using an ordinary least square (OLS) regression method (Chin et al. 1996).

As a quality criterion for this approach, the R² value can be used, which results from the regression models between the latent variables representing the percentage of explained variance of the latent endogenous variable. Chin (1998) suggested labelling an R² value of not less than 0.67 as ‘substantial’, while values not less than 0.33 and 0.19 should be called ‘average’ and ‘weak’, respectively. A second quality criterion for the inner and the outer model is measured using the bootstrapping procedure to calculate t-values for the path coefficients and the loadings between the indicators and their factors. These enable the measuring of the significance of the estimated model. In contrast to other structural equation modelling approaches, such as the covariance analysis for the PLS approach, no goodness-of-fit criteria is available which evaluates the whole model. Since the PLS approach aims to minimise the error terms in all exogenous constructs, the success of the model fitting the data can be observed by examining the R² values of the dependent exogenous variables (Hulland 1999).

For the evaluation of the model, the path coefficients can be used. They are expressed as standardised values as used in OLS regression models. Lohmöller (1989) uses path coefficients with a threshold value of 0.1 in his model. Besides the path coefficients, which represent the direct effect of an independent variable on the dependent one, the indirect effects of variables not directly linked can also be evaluated.

Another helpful criterion is the F² value showing the power of the effect of one exogenous variable on the latent endogenous variable in models using multiple exogenous variables (Weiber and Mühlhaus 2010). The effect of one independent variable on the dependent variable is tested by excluding the variable from the calculation and measuring the difference in the R² value. According to the definition of Cohen (1988) for the use of multiple regression analysis, F² values of 0.02, 0.15 and 0.35 show a minor, average or major effect on the endogenous variable. The PLS method was originally developed in social sciences and is not widely used in vulnerability analysis. Atkins et al. (2000) used components of the structural equation modelling in their study. In the first step, factor analysis was used to estimate factor values that were then used in a regression analysis in a second, separate step. As a dependent variable, income volatility of countries has been used to create the commonwealth vulnerability index.

3.2 Study region

Of all the tsunami-affected countries in Asia, Thailand was chosen because of the vast impact, of the tsunami on the country’s economy. The affected area along the Andaman coast is dominated by the tourism and agriculture sectors, which contribute 4 and 10%, respectively, to Thailand’s GDP (National Economic and Social Development Board (NESDB) 2009). In terms of direct damage, Thailand, with an estimated damage of € 380 million, was less affected than Indonesia, which suffered direct damage to property of nearly € 2.4 billion. On the other hand, Thailand accounted for the highest losses caused by business interruptions and sales declines of all other countries affected (United Nations and World Bank 2005). Of the six tsunami-affected Provinces in Thailand, the Phang-Nga Province was chosen as the study area for this investigation. This province was the worst affected area in Thailand in terms of damage to buildings, infrastructure and property (Vorlaufer 2005). In the next step, Khao Lak and its surrounding villages were chosen as a representative area for the provincial economy. Khao Lak itself has been developing into a tourist “lighthouse” in the Phang-Nga Province since the early 1990s. It is positioned in the global tourism market as a quality tourism area with a strong focus on eco-tourism in contrast to the busy tourism hotspot on Phuket Island (Phang-Nga Tourism Association 2009). Tourism in Thailand was recognised early as a way of diversifying the economy and generating a flow of external income for the country. The business sector has been promoted in several development plans, also being seen, as a tool for overcoming regional disparities in the country (McDowall and Wang 2009; United Nations 1991). The bordering villages to the north and south of Khao Lak are mostly small fishing villages and virtually untouched by tourism. Agricultural areas and aquaculture are dominant along this part of the coastline (TRAIT Company Survey 2009). Despite this, there are strong linkages within the region based on demand from tourism-oriented businesses in the core region and on the commuting of workers to Khao Lak, which ranges from 25% of the total working population in Ban Nam Khem to 33% in Thai Muang. As a consequence, the household income in this tourism-oriented region is much higher than in the rest of the country (cp. Table 1).

Table 1 Key economic facts for the study regions
Full size table

On the basis of a field mapping surveying the building structure of the villages, a quota sample plan was developed. This plan assumes that the greatest influence on the sensitivity comes from the location and the type of building. Within this plan, the households were chosen randomly. If the household had been affected by the tsunami, directly or indirectly, the head of the household was interviewed. The survey was conducted at the beginning of 2009 among households in three villages, namely Khao Lak, Ban Nam Khem and Thai Muang, in the Phang-Nga Province in Southern Thailand (cp. Fig. 4). In total, 225 households in the study area were interviewed by a team from Chulalongkorn University Bangkok. The households were chosen based on a stratified random sample according to their housing situation. The data were gathered using a questionnaire containing questions about demographics, education, employment, income, expenditures as well as the tsunami impact. The categorisation was chosen according to the standards used by the governmental agencies of Thailand.

Fig. 4
figure 4

Study region in the tsunami-affected provinces in Southern Thailand (National Statistical Office 2005)

Full size image

Comparing the housing situation of the study area and the Phang-Nga Province as a whole, especially the type of dwelling and the construction material, Cramers V shows an average relation between these areas (cp. Table 2). While the study area displays some differences in terms of building structure, it can still be considered representative for the situation in the Phang-Nga Province.

Table 2 Housing situation—study area and Phang-Nga Province
Full size table

4 Vulnerability analysis of households in the Phang-Nga Province

For analysing the vulnerability process and attaining the goals mentioned earlier, the factors contributing to the vulnerability and its proxy factor recovery time are ordered as a causal chain to achieve the explanation of the outcome of this process. Using the recovery time as a proxy indicator for the vulnerability, the households were asked in an intentionally loosely worded question to state the time required before they felt they had recovered from the disaster. The model consists of three parts explaining the amount of damage in the household according to the pre-disaster condition and its location. In the second part, the recovery process of the direct damage is evaluated after which the indirect effects are assessed (cp. Fig. 5).

Fig. 5
figure 5

Structural equation model of household vulnerability

Full size image

For the construction of the factors, the approach of formative latent variables was used. The constructs in this model represent explanatory combinations of indicators determined by their combination of variables (Fornell and Bookstein 1982). These factors are considered as index values representing their related indicators in the form of a linear combination (cp. Table 3; Diamantopoulos and Winklhofer 2001). For example, the factors describing the value of household assets are combined with a weighting of three quarters for the property value and one quarter for the long-term consumable goods. For the factor describing the amount of damage, the weighting shifts to around 60% for the property value and 40% for long-term consumable assets (cp. Table 3). This corresponds to the relations shown in descriptive evaluations of the survey data. The factor of recovery time shows a contrary weighting indicating a higher value for the recovery time needed for long-term consumable goods than for property damage. This is explained by the fact that the households started to rebuild the houses before restoring their assets.

Table 3 Outer weightings
Full size table

4.1 Vulnerability model for the Phang-Nga Province

To represent the loss suffered by a household during the tsunami of 2004, household possessions consisting of property value and long-term consumable assets were chosen. The degree of the tsunami impact on the exposed household assets, and thus the sensitivity dimension is represented by the location of the household. This factor is determined by the height of the household above sea level, the distance to the coastline and an indicator representing the degree of damage to the building derived from the statement as to whether a total rebuilding was necessary after the wave impact. Both household damage determining factors show a strong significant influence on this factor (cp. Table 3). While a greater value of assets increases the damage, the location factor reduces the amount of damage. The strongest influence comes from the location factor, since it defines the degree of total household property affected, while the factor representing the assets of exposed households stands for the maximum possible damage. The explanatory power of these factors for the damage of the tsunami can be deemed ‘weak’, since the corrected coefficient of determination shows a value of 0.27 (Chin 2009). If one takes into account the fact that only four items are taken to represent the total value of exposed household assets, this ‘weak’ R² can be accepted (Table 4).

Table 4 Direct damage caused by the tsunami
Full size table

Focusing first on the factor describing the recovery time from the direct impact of the tsunami on the households, it is obvious that there is no significant influence of the amount of damage sustained by the household on the time required to recover. This impact is fully mediated by other influences such as the external or internal resilience capabilities upon which the household can rely. In total, this factor shows an acceptable coefficient of determination with a R² value of 0.49. A reducing effect on the time needed to recover can be found in the capability of households to use their own resources for repairing damage. Using the partial correlation method with the amount of damage as a control variable, both indicators, i.e. a household’s own resources used for rebuilding the property and those used to restore the household interior, show a negative coefficient.

Looking at the external resilience capabilities represented by the amount of help received by governmental or non-governmental agencies, one finds an extending influence on the recovery time, suggesting that the dependence on external help leads to an increased recovery time. Evaluating the relations between each of the indicators setting up the factors reveals differing effects. The partial correlation—controlled by the amount of damage—between the amount of external help provided for the recovery of personal belongings and the restoration of the household interior on the one hand, and the time needed to recover these items on the other hand, is negative. This points to a faster recovery of households that received external help to restore personal belongings. However, looking at the time needed to recover from damage to buildings, the correlation shows a positive influence. Combining these two elements of household property affected by the tsunami to a single factor leads to an overall extending effect of external help on the recovery time according to the greater weighting that the damage to property value is given by the algorithm (cp. Tables 3, 5). A strong F² value indicates a large influence of these two factors. Furthermore, the recovery time of direct effects is influenced by the effects of the damaged infrastructure, indicating that people from highly damaged villages were moved to temporary shelters while the infrastructure was reconstructed, extending the time period before a household could start to rebuild its own property. Personal injuries among the household members are also likely to extend the time period needed to recover from direct effects (cp. Table 5).

Table 5 Recovery time from direct damage
Full size table

According to Smith and Ward (1998), effects of natural disasters on tangible assets can be divided into two groups: direct and indirect effects. Examples of indirect tangible effects of natural disasters include the disruption of supply chains or reduced income for households, even for households that are not directly affected by the natural disaster. One indirect effect that showed a significant influence on the recovery time of households is the loss of income after the tsunami. The duration of the influence of indirect effects on the household is mainly determined by household characteristics concerning its employment situation. Because the tourism sector suffered from a significantly higher loss and needed much more time to reach its original production level than other sectors characteristic for this area, such as agriculture and fishing, a significant impact can be found in the influence from this sector. A reducing influence on the indirect effects results from the diversification of income within a household. To measure this, the Shannon Equitability Index was used, putting the number of income sources and their share of the total household income in relation and combining this in one index value (Schwarze and Zeller 2005). The reducing effect of a diversified household income is higher than the result of a high level of total household income and education. This factor is mostly influenced by the level of education among the working household members, while the total amount of income plays only a minor role. The duration of the recovery period from indirect effects is also determined by the disruption of infrastructure and its negative effect on the household livelihood (cp. Table 6). The length of time needed to rebuild the road network and the water and electricity systems particularly affects the household’s income situation. Since the majority of people possess a mobile phone, the telephone network could be fixed within 2 weeks, and no extending influence could be found resulting from this indicator.

Table 6 Recovery time from indirect effects
Full size table

5 Discussion

The structural equation model developed in this paper shows the relations and interactions between the three different components of vulnerability mentioned in the vulnerability frameworks discussed earlier. Integrating all elements of vulnerability, it allows for a quantification of the relations in the vulnerability framework and for an understanding and reconstruction of the status of vulnerability and the process of recovery that represents it in a most comprehensive way. The first key element of vulnerability, the exposure of household assets, could be strongly linked to the location of the household. The coastal terrain in the study area of Khao Lak and Ban Nam Khem, with its flat areas stretching along the coastline and extending almost 2 km inland, contributed greatly to the total damage sustained by households. On the other hand, the community of Thai Muang, protected from the direct wave impact by a peninsula that reduced the total damage per household, showed a similar recovery time to that of the highly exposed areas of Khao Lak and Ban Nam Khem, indicating a higher vulnerability level of this community. The majority of households are located in areas roughly eight metres above sea level, which still constitutes a tsunami-endangered zone according to the 2004 tsunami wave heights ranging from 5 m in Thai Muang to 12 m in Ban Nam Khem and Khao Lak (Department of Disaster Prevention, Mitigation (DDPM) 2007). Households that had to rebuild their property suffered from a loss of personal belongings of up to € 8,000, while households whose houses were able to withstand the wave impact reported a loss of about half this figure.

The road system and lifelines such as the water system in particular showed an extending influence on the recovery time of households both from direct and indirect effects. The road system in the Phang-Nga Province was severely damaged by the tsunami, requiring an investment of more than € 2.6 million (Nidhiprabha 2007). Due to the enormous efforts undertaken by the government and the Thai military, the road system was restored after 2 months and the telecommunication network after only 2 weeks (TRAIT Household Survey). In some parts of the study region, it took up to 2.5 years before all infrastructure was restored and household members could reach their workplaces and manage their everyday lives without limitations.

The investigation by Calgaro and Lloyd (2008), based on the vulnerability framework introduced by Turner et al. (2003), examines the vulnerability of Khao Lak in a qualitative study. One important result of the study was that the sensitivity of the household livelihoods in this region is notably determined by the limited livelihood options. Particularly, the focus on tourism in the core region of Khao Lak and its surrounding villages and the resulting lack of income diversification leads to a higher sensitivity. Concerning these indicators of vulnerability, it was found that there is no direct linkage to the recovery time of household assets. These indicators are mediated by the indirect results of the tsunami, especially the loss of income, but still show a significant effect on the outcome of vulnerability. Findings from the survey show that the households in the region generate between around 28% (in Thai Muang) and 44% (in Khao Lak) of their income from tourism or related businesses. As a result, households that were not dependent on tourism recovered on average 1 month earlier. Households that generate a higher share of income from tourism-related industries show a longer period of time with losses of income than other households. From the theory of vulnerability, the hypothesis was derived that internal as well as external resilience capacities reduce the vulnerability of households and therefore decrease the time needed to overcome the effects of this natural disaster. The structural equation model indicates that at least the external resilience capabilities represented by the help provided by non-governmental and governmental organisations lead to an increased recovery time. The government paid compensation of nearly € 45 to all individuals affected by the tsunami, regardless of the amount of damage sustained or any other circumstances of these individuals. If the house had been destroyed completely, a compensation of € 430 was paid. NGOs did not follow this restriction and provided an amount of compensation per household which was nearly three times higher. The 2004 tsunami-related funding broke many records and reached nearly € 10 billion from international sources. As a consequence, the recovery process in the tsunami-affected countries could be financed without any budget constraints (Flint and Goyder 2006). According to findings of the Tsunami Evaluation Coalition, the activities that took place during the recovery process showed a lack of assessment, and the international response was ‘insufficiently evidence-based’ (de Ville Goyet and Morrinière 2006). The report focuses on the poor coordination and the lack of exchange of assessment reports between the national coordination efforts and international agencies. This lack of coordination resulted from the large number of participants involved in the recovery process. Furthermore, the national government of Thailand declined international help. As a consequence, only international agencies that had had partnerships with local organisations before the disaster were allowed to act in Southern Thailand. Therefore, a huge number of organisations were involved in the recovery process, which were not specialised in this kind of work, such as religious or missionary organisations (Korff 2009). Governmental coordination for allocating the recovery efforts lacked knowledge, skills and adaptive capacities (Kamolvej 2006). Therefore, the recovery time was extended, while an oversupply of help among the population was also evident due to competition among the donors (Telford and Cosgrave 2007). This reduced self-initiative led to a dependency on external help (Korff 2009). The structural equation model showed that households that were able to finance and organise their own recovery process were able to return to normal significantly faster.

6 Conclusion

The structural equation model developed in this paper shows the relations and interactions between the three different components of vulnerability. Using data from the study area in Khao Lak, it allows for a quantification of these relations in the vulnerability framework. The model showed that the first hypothesis, stating that the amount of direct damage caused by the tsunami has a significant influence on the recovery time, needs to be rejected. Of particular interest is the external help provided by government and NGOs, which showed an extending effect on the recovery time. Therefore, the hypothesis concerning this element of resilience needs to be rejected as well. This may be due to the special situation in Southern Thailand after the tsunami that is less comparable to other natural disasters because of the high amount of donations and the disorganised coordination and allocation of supporting measures. Internal resilience capabilities positively influence the recovery from direct and indirect effects. The indirect effects thereby mediate some of the internal resilience capabilities, which show a significant and strong impact on the households. Further research should focus on the question of whether the lessons learned from this major natural disaster match the ongoing mitigation strategies in the region—especially whether the problem of poorly diversified regional economies, which increases the vulnerability of households, is being dealt with. This would reduce the population’s vulnerability and allow them to recover more quickly.