Financial Imbalances in Regional Disaster Recovery Following Earthquakes—Case Study Concerning Housing-Cost Expenditures in Japan

Abstract

In the aftermath of the 1995 Kobe Earthquake, a large-scale effort towards reconstruction of houses damaged by the quake was required. This led to increased mortgage, thereby financially plaguing a number of earthquake victims and inhibiting their long-term sustainability and self-supported recovery. The current framework of housing reconstruction assistance provided by the Japanese government does not account for regional disparities in cost and other socioeconomic factors. This study proposes a technique for estimating the cost of reconstructing household units damaged in an earthquake by considering the effects of construction methods influenced by regional climatic zones. The financial constraints on rebuilding resources have been estimated by considering the annual regional income and household savings, as determined by social factors and employment opportunities. The susceptibility of regions to the occurrence of earthquakes has also been factored in the calculation of recovery costs. Together, these factors are used to provide a more complete picture of economic costs associated with earthquake recovery in different regions of Japan, thereby revealing large disparities in the difficulty and financial burden involved in the reconstruction of household units. Results of this study could be used to develop a robust system for earthquake-recovery assistance that accounts for differences in recovery costs between different regions, thereby improving the speed and quality of post-earthquake recovery.

1. Introduction

Japan is an earthquake-prone nation. As such, even if disaster victims and their families are fortunate enough to survive an earthquake unscathed, the loss of their households effectively marks the beginning of a difficult process of reviving their everyday lives. Many earthquake victims, incapable of bearing the cost of rebuilding their homes in their hometown, are forced into moving to other places, consequently causing sustainable communities to collapse. As a result, the depopulation of local communities may advance to such an extent that the very existence of the village itself may get threatened. A significant portion of costs associated with earthquake recovery gets utilized in the reconstruction of damaged houses. This cost is often borne by the damaged household itself, frequently in the form of additional debt. The prominence of the so-called “double loans” in the aftermath of the 1995 Kobe Earthquake added to existing mortgages and financially plagued many earthquake victims, thereby inhibiting their sustainable, long-term self-supported recovery. For the purpose of restoring communities within their native land for a long time in the aftermath of a natural disaster, the sustainability of disaster recovery can be defined as a means of maintaining peoples’ comfort and economic well-being in their original place of residence. In order to broad the research area of “sustainability,” this paper is positioned as one example to deal with in a broad sense.

Although Japanese homes are typically constructed of wood, the country stretches out geographically to cover a large landmass and thus, includes a variety of climatic zones that must be accounted for when undertaking household reconstruction [1]. In other words, the geographical expanse of Japan from north to south results in differences in climate, thereby calling for regionally varying methods of household construction. Consequently, there exists large regional disparities in construction costs, despite most houses being consistently wood-framed. These disparities, in turn, result in regional imbalances in recovery costs, which have previously not been evaluated or discussed. The cost of household reconstruction, therefore, forms one of the most crucial factors for local communities to lead a sustainable life.

The proposed study was undertaken to reveal regional imbalances in recovery costs and focuses on Japan as an earthquake-prone nation. The study makes use of the method of risk finance [2] to call attention to new problems that result in regional imbalances in the recovery brought about by the intersection of building standards and differences in climate and other natural features.

Support Systems for Rebuilding Lives in Japan

Minami et al. [3], by using results of a survey performed concerning the sense of recovery among residents of the Okushiri Island 20 years after the occurrence of an earthquake and considering the completion of building reconstruction as a rough cutoff point, reported that 42.3% of the survey participants confirmed that they had fully recovered from the earthquake and that household rebuilding and financing thereof were the most important factors of concern during post-disaster life planning. Against this background, the Act Concerning Support for Reconstructing Livelihood of Disaster Victims was enacted in Japan in 1998 and the collected tax money was directed towards rebuilding individual damaged homes. The basic form of support amounts to provision of ¥3 million for purchase of a new home in the event of very heavy irreparable damage caused or total collapse of a victim’s household. The said amount includes ¥1 million towards current livelihood support. In the event of substantial to heavily damage caused to a household, the financial support provided amounts to ¥1.5 million (including ¥0.5 million towards current livelihood support). However, according to a survey performed by the Cabinet Office [4], the amount of support provided by this initial system for distributing public funds for individual assistance is completely inadequate for completely damaged household units. This seems obvious given that the average cost of constructing a detached house in Japan is approximately ¥18 million [5].

Figure 1 shows results of a survey performed by Nakashima et al. [6] to determine household reconstruction costs incurred by individuals residing in households on the Okushiri island in the aftermath of the southwest-off Hokkaido earthquake. The figure shows the distribution of rebuilding and repair cost of houses, the median value of which amounts to ¥10 million or more. This results in a major portion of recovery expenditures is to be borne by households themselves. This problem concerning household reconstruction support has subsequently reappeared after the 2004 Chuetsu [7] and 2007 Noto Peninsula [8] earthquakes, among others. In addition to the previously mentioned tax-based funding, the main resources available to assist disaster victims comprise goodwill donations made on an individual basis. However, not much can be expected from this source as contributions depend on social circumstances of individual donors, thereby adding an element of uncertainty as to how much will be donated. Furthermore, large-scale disasters imply that aggregate donations must be allocated to more households, thereby resulting in a small assistance available to each individual household. Additionally, individual disaster victims may possess security in the form of savings or earthquake insurance but the household enrollment rate for earthquake insurance nationwide is only about 30.5%, which is significantly smaller compared to the 82% enrollment for fire insurance. Additionally, the said earthquake insurance does not cover the full cost of damage [9]. In the face of such circumstances, disaster victims, in the past, have borrowed an average of ¥14.615 million [3] after recent seismic disasters, mainly to rebuild their homes. This approximately equals 3.5 times the average annual income of a salaried worker in Japan in 2017. Despite the support provided in Japan by the Act Concerning Support for Reconstructing Livelihood of Disaster Victims, the actual burden of household reconstruction continues to be quite severe on disaster victims. Further, as described in this study, there exist regional imbalances in the size of this burden. This imbalance is caused not only by different earthquake environments (regional disparities in the probability of an earthquake) but also by differences in construction costs arising from regional disparities in household construction methods due to climate and other natural features as well as social phenomena, such as disparities between assets stemming from different employment conditions.

Extant studies concerning regional disparities have largely accounted for differences in seismic hazards (e.g., Shimizu et al. [10]) without assessing differences in the reconstruction cost of wood-framed housing or economic conditions of regional residents. This study applies regionally dependent factors, such as seismic hazards, construction costs, vulnerability of wood-framed houses and savings with respect to annual income to seismic risk assessment techniques. Ultimately, this study proposes a new method for risk finance that accounts for debt at an individual level and also newly defines debt weight with respect to disaster victim assets, rather than the amount of debt exclusively. This exposes new issues, as discussed in Section 3, concerning regional imbalances in the recovery burden of households when accounting for different economic circumstances of disaster victims. These differences in recovery burden can have a significant impact on the rate and sustainability of disaster recovery.

2. Methods of Calculating Household Reconstruction Cost and Data Sources

2.1. Climate, Natural Features and Housing Construction Costs

In this section, housing construction costs are described in detail. Figure 2 depicts the climatic divisions of Japan in accordance with Sekiguchi [10]. Japan consists of a bow-shaped chain of islands stretching from north to south over a distance of approximately 3000 km. In terms of climate and natural features, Hokkaido in the north (Figure 2, Climatic Divisions Ia and Ib) is classified as a subarctic zone that is cold and experiences heavy snowfall in winter, whereas the Nansei Islands in the south can be classified as experiencing a subtropical climate with a small annual temperature range and heavy annual precipitation. Additionally, the Pacific side of the main island of Honshu (Division III) is humid during summer and dry during winter, whereas the Sea of Japan divisions (Divisions Ic, Id and Ie) experience long, warm daytimes during summer, along with Japan’s heaviest snowfalls during winter.

Figure 3 depicts next-generation energy-efficiency standards for buildings established by the Ministry of Land, Infrastructure and Transport [11]. The northern part of the Japanese archipelago (high-altitude region) is a cold region, whereas a heavy wintertime-snowfall region is unevenly distributed on the Sea-of-Japan side by the backbone of mountains running through the middle of Honshu. These climate differences in Japan demonstrate a significant effect on methods of household construction. Figure 4 depicts the regional distribution of home-foundation types and foundation heights [12]. To accommodate vertical loads, such as those imposed by snow and the advent of soil freezing, continuous reinforced concrete footings are adopted on a large scale in the Hokkaido and Tohoku regions. Correspondingly, foundations in cold-snow regions are typically constructed high above the ground level. Additionally, the use of lightweight sheet metal as a roofing material is extensive in the Hokkaido, Tohoku and Hokuriku regions while in other regions, protection from summertime solar radiation and frequent typhoons is essential, so the use of heavy tiles is more predominant (Figure 5 [12]).

These regional differences in house-construction methods are directly reflected in housing costs. Figure 6 depicts the distribution of the average costs for constructing a detached wood-framed house by prefecture [4]. The highest and lowest costs of ¥22.26 million in the Toyama prefecture and ¥15.27 million in the Miyazaki prefecture, respectively, correspond to a cost difference of roughly ¥7 million, thereby representing a rather large regional disparity of approximately 50%. Overall, the cost of house construction is higher in prefectures lying on the Sea of Japan side compared to those on the Pacific side. During damage recovery, these cost differences directly impact the recovery of disaster victims.

2.2. Computation of Individual Burden of Recovery Costs

In general, the individual amount of debt incurred during household reconstruction in the aftermath of an earthquake may be formulated as described in Equation (1) when modeled with generally high points of commonality. Owing to variability of individual circumstances, donations and other contributions (strongly dependent upon many external factors) were excluded from the formulation and so was earthquake insurance (owing to its relatively low enrollment rate of 30.5%, as of 2016 [8]).

D = L – R – M [unit: yen]

In Equation (1), D denotes the amount of debt incurred during reconstruction of individual households; L denotes the cost of household damage (house-reconstruction cost); R denotes the amount of household reconstruction support provided pursuant to the Act Concerning Support for Reconstructing Livelihood of Disaster Victims; and M denotes the amount of household savings available.

To be noted about this equation is the fact that the amount of house-reconstruction debt D—which has a significant impact on post-earthquake quality of life—is greatly affected by the residential area and there exists a rather high probability that the amount of reconstruction debt would be influenced by regional differences in household reconstruction methods. For example, if a wood-framed house dweller rebuilds a house in the same location as original damaged house, the cost of house damage L in Equation (1) could be expressed as follows.

L = Hazard × Vulnerability × Cost

In the above equation, Hazard represents the direct magnitude of disaster, that is, the size of regional incoming seismic motion (depicted in Figure 8). Hazard represents Peak Ground Velocity in cm/sec or Seismic Intensity defined integrally from 0 to 7 in Japan Meteorological Agency Scale (abbreviated to JMA Intensity in this paper), which in turn, depends on the seismic magnitude and the ground-motion-attenuation relation. Vulnerability represents the influencing factor on the receiving side, that is, the non-earthquake resistance of regional homes and household economic situation. Lastly, Cost represents the regional wood-framed house construction cost, all of which is regionally dependent. The value of R in Equation (1) remains largely constant nationwide for all residents and all regions and structures, since it is determined by the damage level of the house and implies a subsidy of ¥3 million in the event of total collapse of a household. However, M is expressed in terms of regional minimum-wage disparities. Hence, there exist large differences in its value depending on assets held by residents in different regions (as described in Section 3).

2.3. Seismic Risk-Assessment Techniques

The proposed study focuses on the damage caused to wood-framed houses—the standard type of household in Japan—and calculates regional disparities in the expense and amount of debt incurred by individual households during reconstruction per residential area using nationwide municipalities as units. Results obtained were subsequently compared. In this study, the term “municipality” refers to the smallest basic administrative unit in Japan. The boundary lines depicted in the maps of Japan used in this paper represent municipal boundaries. As of April 2018, there exist 1724 municipalities in Japan, including the special wards of Tokyo.

2.3.1. Method for Predicting Damage Cost per Individual Household

Individual households, which form the primary subject of this study, were divided into annual income brackets within each municipality. The cost of damage and debt were calculated using Equations (1) and (2).

Regular seismic damage risk assessments were performed using Equation (2) when estimating the cost of damage. It should be noted that construction costs are often uniformly addressed using nationwide average values. In this study, however, major focus was on regional disparities; consequently, regional disparities in housing prices (Figure 6) were actively considered in the estimation of total cost. First, when using Equation (2), seismic-motion prediction maps provided by the Japan Seismic Hazard Information Station (J-SHIS) [13] were used to calculate the total number of wood-framed houses damaged along with the cost of damage (reconstruction cost) in each municipality. Subsequently, by proportionally distributing the cost of damage among the percentage of damaged households predicted within each municipality, the cost of damage (L) per household was determined. Next, using Equation (1), the average income and average savings per municipality was used to compute the average amount of debt per household.

This section introduces the regular risk assessment (cost of damage) corresponding to circumstances in Japan whilst also describing the calculation method involved. Assuming the size of tremors within a given municipality to be uniform, the incoming seismic motion (Hazard) was represented by the size of tremors experienced at the location of the municipality town hall. The earthquake resistance (or non-vulnerability) of a house was determined based on its seismic score in the given municipality, which in turn, was determined by means of a probability distribution. To determine household-reconstruction expenses (Cost), the average construction cost per region, depicted in Figure 6, was used. In Equation (1), the financial support (R) towards household reconstruction, provided by the government, is the same nationwide and depends exclusively on the degree of damage caused to a house. The value of household savings (M), which can be individually dispersed to cover reconstruction expenses, was determined using average savings per annual income bracket in the subject municipality. Figure 7 depicts the general workflow of this calculation.

2.3.2. Obtaining Seismic Motion

Nationwide seismic hazard information (Hazard in Equation (2)) is provided via internet by J-SHIS [13], which is operated by the National Research Institute for Earth Science and Disaster (National Research and Development Agency). The stochastic seismic-prediction map (2012 edition) provided by J-SHIS [13] was used in this study. Considering the location of the municipality town hall as the reference point, the peak ground velocity PGV_i [cm/s] was computed based on the seismic bedrock velocity and site amplification and subsequently used as the representative value of seismic motion for the municipality. Figure 8a–d depict 1%, 5%, 10% and 50% 50-year occurrence probabilities, respectively, of peak ground velocities (PGV) indicated in the legend. Given a constant assumed term of 50 years, the predicted seismic motion was observed to be large, since the probability of occurrence is small. Throughout Figure 8a–d, the seismic hazard on the Pacific side was observed to be larger compared to that on the Sea-of-Japan side.

2.4. Computation of Number of Wood-Framed Houses

The proposed study considers wood-framed houses as primary subjects susceptible to seismic risk. It is, therefore, necessary to determine the number of households living in wood-framed houses as well as the number of such houses within each municipality. In Japan, however, this statistic is not consistently maintained. The basic statistical information maintained within each municipality includes the population, number of households and number of households living in wood-framed houses. House-building techniques and percentage of houses of each configuration are also publicly available [14,15]. These databases were used to estimate the number of wood-framed houses within each municipality. For statistical processing, houses in Japan were categorized into three types of configurations—detached, row and apartment houses—and two construction types—wood-framed and non-wood-framed. Figure 9 depicts exterior images of the three configurations of representative wood-framed structures investigated in this study. A detached house is a configuration wherein a single household resides independently inside a single building. A row house is a configuration in which multiple households reside whilst sharing the same external building walls but no common spaces. An apartment house is a configuration containing households that share spaces, such as stairs and entrances. Because the number of households residing in a structure differ depending on its configuration, the percentages of houses in each configuration were determined. Subsequently, using a conversion factor M_i,k for each municipality i, the number of households per building method k was converted into the number of wooden buildings H_i,k as follows.

$$H_{i,k}=M_{i,k}\times G_{i,k}$$

(3)

$$G_{i,k}={\overline{w}}_{i,k}\times Z_{i,k}$$

(4)

Here, $G_{i,k}$ denotes the number of households per wood-framed house configuration k; ${\overline{w}}_{i,k}$ denotes the average percentage of households per wood-framed house configuration k; and $Z_{i,k}$ denotes the number of households in wood-framed structures.

2.5. Computation of Wood-Framed House Reconstruction Support and Cost of Damage

The total financial compensation ${{\displaystyle \sum }}_l{{\displaystyle \sum }}_k{R}_{i,k,l}$ citing complete or partial destruction of a wood-framed house within a given municipality i corresponds to a maximum of ¥3 million or ¥1.5 million, respectively. This support amount, Y_l, was used to compute the total support as follows:

$${\displaystyle \sum }_l{{\displaystyle \sum }}_k{R}_{i,k,l}=Y_l\times E\left(I_i,l\right)\times G_{i,k}$$

(5)

Here, $E\left(I_i,l\right)$ denotes the probability of the degree of damage caused to a wood-framed building (l) at seismic intensity $I_i$, which is generally considered a vulnerability function concerning major structural types in each country. The degree of building damage caused (Damage Index) estimated based on the input seismic motion (seismic intensity) for wood-framed houses can be converted into a function using the building seismic score as a parameter (Figure 10 [16,17]).

The cost of damage suffered by a household was assessed in terms of replacement cost using Equation (2). The cost of building damage was determined by the extent of damage caused to a given house. Thus, the total amount of wood-framed house damage ${{\displaystyle \sum }}_l{{\displaystyle \sum }}_k{L}_{i,k,l}$ for building damage (l) in a municipality (i) was computed using Equation (6) derived using the relations Vulnerability × Hazard = $E\left(I_i,l\right)\times H_{i,k}$ and Cost = ${\alpha }_l\times C_i$.

$${{\displaystyle \sum }}_l{\displaystyle \sum }_k{L}_{i,k,l}={\alpha }_l\times C_i\times E\left(I_i,l\right)\times H_{i,k}$$

(6)

In the above equation, $C_i$ denotes the cost of housing construction within municipality i (Figure 6) and ${\alpha }_l$ denotes the percentage of houses damaged by earthquake. Here, scales of very heavy damage to total collapse were defined as 100% while those corresponding to substantial to heavy damage were defined as 50%, in accordance with Takahashi et al. [18].

The cost of damage per household within each residential area was then determined using Equation (6). As an example, the risk curve (damage exceedance probability) within ordinance-designated cities (cities with populations of 500,000 or more, as depicted in Figure 11) in all regions were computed and are compared in Figure 12.

The risk curves for all 17,427 municipalities were computed in a manner identical to that depicted in Figure 12 and the number of damaged houses and corresponding cost of damage per seismic motion exceedance probability are displayed on maps of Japan in Figure 13 and Figure 14, respectively. Figure 15 depicts the cost of damage per household for the 10% 50-year earthquake exceedance probability.

This procedure constitutes normal seismic-risk assessment. In the proposed study, regional disparities in construction costs (Figure 6) were also considered but when damage risk determined in Figure 15 was compared against the hazard distribution in Japan depicted in Figure 8, distributions in regions with large seismic hazard and those with large costs of home damage were observed to be similar. Regional disparities in seismic hazard are quite large when compared against regional disparities in construction costs and thus, the cost of damage is, at present, largely governed by the seismic hazard. If risk curves, depicted in Figure 11, are examined on an individual basis, the expected cost of damage to household dwellings in Sapporo City equals approximately ¥500,000 for a 10% exceedance probability, which equals the maximum probable earthquake loss (PML). At the same time, in the Yokohama and Shizuoka cities (on the Pacific coast), expected costs of damage are close to ¥10 million, thereby indicating that regional disparities in the assumed probable cost of damage can be quite large. Summarizing this trend for the entire country, there exists a distribution of municipalities—mostly concentrated on the Pacific side of Japan—with a cost of damage per household exceeding ¥10 million.

2.6. Computation of Indebtedness of Individual Households

The reality of risk assessment is that risk is largely governed by the probability of earthquake occurrence. In cases of actual damage, the probability no longer carries any real meaning. For disaster victims, the occurrence of actual damage results in ever-increasing debt. Further, in addition to regional disparities in terms of housing costs, hurdles in the way of household reconstruction and overall recovery when facing complete destruction (Figure 6) vary owing to differences between individual economic means, such as household savings. There, therefore, exists a real possibility that the risk in terms of cost of damage, which is typically determined in general seismic risk assessments (using estimation method in Equation (2)) and difficulties associated with house reconstruction (estimated using Equation (1)) may not match owing to differences between assessment criteria. To mitigate the effects of these differences, the proposed study computes the amount of individual debt per region caused by the necessity of household reconstruction after an earthquake in accordance with Equation (1). To facilitate inter-regional comparison with respect to the environment within calculations as well as account for differences in economic circumstances, the proposed study was performed on the premise that the standard Japanese wood-framed houses subjected to earthquake damage are rebuilt at the same location.

In addition to estimating the cost of damage per household within each municipality, the probability of earthquake occurrence was ignored in this study to better clarify regional disparities in the debt burden caused by differences in housing costs and amount of savings. Further, the debt incurred owing to reconstruction of a heavily damaged or totally collapsed house for each annual-income bracket was computed using Equation (1). In this case, the equation for calculating the total cost of damage to wood-framed houses in Equation (6) was modified, as described in Equation (7), by assuming ${\alpha }_l$ = 1, thereby indicating complete destruction and ignoring the vulnerability function E(I_i, l).

$${\displaystyle \sum }_k{L}_{i,k,l=complete destruction}=C_i\times H_{i,k}$$

(7)

To determine the amount of household savings (M) in Equation (1), the relationship between the number of households and amount of savings was deduced considering total savings per annual-income bracket for each prefecture based on data published in the National Survey of Family Income and Expenditure [19,20] as well as the number of households within each annual-income bracket in similar homes and land statistics surveys [14,15].

3. Results and Discussion

The previous section described seismic risk assessment and determined the cost of damage in different regions of Japan based on available data to provide a background. In this section, results of the said risk assessment and cost of damage calculation are evaluated to determine the burden of household recovery depending on municipality, thereby making recommendations for eliminating the inherent inequality of this burden.

3.1. Comparison of Indebtedness of Individual Households

Figure 16, Figure 17, Figure 18, Figure 19, Figure 20, Figure 21, Figure 22, Figure 23 and Figure 24 depict average savings per annual-income bracket (M in Equation (1)) within each prefecture and the amount of debt incurred (D in Equation (1)) during reconstruction of a home after its complete destruction. As previously demonstrated by Nakashima et al. [6], the mode value of the annual income of people residing on the Okushiri island is ¥3 million, whereas the corresponding cost of household reconstruction lies in the range of ¥10–15 million (Figure 1), which is same as the amount estimated through use of the authors’ prescribed method and depicted in Figure 17. This demonstrates the accuracy of simplified assumptions represented by Equation (1), despite its deterministic approach. The amount of savings is largely governed by socioeconomic factors, such as annual income, employment opportunities and family structure and thus, significant regional disparities were observed. For example, average savings were observed to be higher in prefectures in and around the Chubu region, which is centered around Nagoya and was observed to be ¥11 million higher compared to other prefectures. The regional difference in savings when compared to the annual income bracket was observed to be ¥8 million for families with annual incomes of up to ¥2 million and approximately ¥40 million for those with annual incomes of ¥15 million. Generally, average savings tends to decrease and regional imbalance tends to increase the farther the region is from the capital region (around Tokyo), Tokai region (around Nagoya City) and towards the Hanshin region (around Osaka). Thus, a trend can be observed wherein regional disparities in the amount of savings become wider with increase in annual income.

However, given the same degree of damage, public funding of house-reconstruction expenses is uniform nationwide. The amount of debt (D) in each prefecture, therefore, depends exclusively on the housing cost and savings per annual-income bracket within that prefecture. Consequently, prefectures close to the Chubu region, which exhibits high levels of savings, were observed to be under small amounts of reconstruction debt. The amount of debt increases with distance away from this region, thereby illustrating significant effects of regional disparities in the amount of savings. Although savings and house-reconstruction expenses slightly differ depending on the prefecture, they remain largely identical for households with an annual income of up to ¥6–7 million and there exist regions wherein rebuilding may be accomplished solely with savings of households with annual incomes exceeding ¥8 million. Excluding prefectures of Aomori, Kochi and Kagoshima, household reconstruction is possible nationwide solely with individual savings of households with annual incomes exceeding ¥15 million.

Comparing Figure 16, Figure 17, Figure 18, Figure 19, Figure 20, Figure 21, Figure 22, Figure 23 and Figure 24, the regional disparities in the amount of debt incurred owing to housing reconstruction can be observed to be clearly influenced by differences in savings, which in turn, are influenced by differences in annual income. Simultaneously, there exist even greater regional imbalances in debt incurred owing to housing reconstruction because of significant regional disparities in housing costs resulting from climate and other environmental factors, although annual savings remain unchanged. If examined at the individual household level, the amount of debt is greater for households with lower annual incomes, which has a significant impact on their ability to cover subsistence expenses. Further, it is obvious that the lower the annual income, the more generous is the assistance required and the higher is the average household-reconstruction cost for a given region. The degree of potential debt burdens a household in Japan faces after an earthquake is not random. Though the risk of debt burden may significantly differ from previously determined seismic risks, it must not be discarded, since it effectively represents regional socioeconomic disparities. The degree of seismic risk is not the only factor influencing recovery cost and thus, further study of recovery subsidization schemes may be required based on new, broader perspectives. Although such studies have been previously performed on a global level [21,22,23], in the case of Japan, consideration of prefectural differences is also necessary.

3.2. Difficulties Associated with Housing Reconstruction

The previous subsection discussed the amount of debt incurred during reconstruction of a damaged house and it was determined that there exist significant regional disparities in the amount of assets that could be set aside for rebuilding, albeit the cost of damage remained unchanged. Consequently, there exist significant regional imbalances in the amount of debt incurred during housing reconstruction. Presently, the Act Concerning Support for Reconstructing Livelihood of Disaster Victims of Japan provides uniform assistance depending on the cost of damage caused to a household but the economic weight of the damage experienced by disaster victims cannot be neatly measured in terms of cost of damage and thus, determination of appropriate assistance may require an accounting of all assets of disaster victims, defined in this study as the weight of debt per damaged household, as described in Equation (8).

weight of debt = (amount of household debt (D))/(disaster victim economic circumstances (assets) (M))

(8)

Figure 25a depicts the weight of debt calculated using Equation (8). As depicted in this figure, the damage cost was computed using Equation (6) and accounted for the 10% 50-year earthquake exceedance probability. As can be observed, at this probability, there exists a significant distribution of debt, wherein the weight of debt is larger on the Pacific side. It must, however, be noted that excluding certain regions on the Pacific side, many municipalities demonstrate weights of debt of 100% or less, thereby indicating complete coverage of housing-reconstruction costs by household assets. Examining this figure, the sense of economic anxiety felt by residents when facing the possibility of earthquakes is understandable, although care must still be exercised with regards to earthquake occurrence probability. Because the probable degree of damage may be assessed as small in regions with low probability, a lack of preparation, financial or otherwise, may be encouraged by the underestimated degree of risk.

It is important here to emphasize that recent earthquakes in Japan have occurred with hardly any regard to the estimated earthquake probability. In this case, the ratio of debt to actual damage (residence rebuilding expenses and assets) emerges as a real problem as depicted in Figure 25b, which depicts the distribution of weight of debt when reconstructing a very heavily damaged or totally collapsed house. A harsh reality is evident—regions with heavy debt burden are consistent with dual regional imbalance, wherein the greater the cost of rebuilding a vulnerable household within a region (Figure 6), the fewer are the assets available to a household to finance the reconstruction (Figure 16a).

The weight of debt of ordinance-designated cities with 10% 50-year earthquake exceedance probability, as calculated using Equation (8), was subsequently compared nationwide and is represented by blue bars in Figure 26. Notwithstanding larger regional disparities, the conclusion that the cost of damage recovery can be borne within the typical amount of household savings is consistent with that drawn from Figure 25a. However, if the earthquake probability is ignored and weight of debt incurred by a household when their dwelling is very heavily damage or totally collapsed is determined, as represented by red bars in Figure 26, it is clear that the actual cost of recovery significantly exceeds the savings of any household. Of particular importance in this comparison is the fact that the smaller the burden of debt when accounting for earthquake probability within a region, the larger is the burden of actual damage when a dwelling completely collapses.

Till date, in Japan, regional imbalances in seismic-risk assessment have reflected only differences in earthquake probability (Figure 8). However, Figure 27 depicts differences in the weight of debt per municipality, calculated using Equation (8), in cases of complete house destruction with the 50-year earthquake exceedance probability of 10% being taken into account. The figure clearly illustrates that, unlike Figure 8, the lower the predicted earthquake probability within a region, the greater is the burden of recovery debt when actual damage occurs. This supports conclusions previously drawn from Figure 26. From a standpoint of providing relief to those vulnerable to disaster, it is essential that the provided support accounts for these regional imbalances to address the said differences in cost.

4. Conclusions

This study was performed to reveal regional imbalances in earthquake-recovery costs in Japan. Consequently, regional differences in the difficulty of housing reconstruction were demonstrated via regional differences in earthquake hazard probability, annual incomes, savings and housing prices. At present, regional imbalances considered by the existing recovery system are based on the probability of occurrence of an earthquake, which is a natural phenomenon beyond policy control (Figure 8). The Japanese government provides rebuilding assistance to homeowners in the event of an earthquake. However, this assistance is relatively small and is uniform across the country. A Japanese earthquake insurance system exists to protect residents against financial debt but the enrollment rate in the said insurance is rather low. It is, therefore, impossible to rectify regional disparities by depending on the current support policy. However, the proposed study reveals that regional imbalances, including differences in construction costs due to varying climate zones as well as annual incomes linked to social factors and employment opportunities, directly influence the burden of recovery debt incurred by disaster victims, thereby governing the speed and difficulties involved in rebuilding lives after earthquakes. Although this problem is complex because factors underlying the issue are social in nature, the solution to more effectively rebuilding lives after an earthquake is within the control of national policy. Presently, subsidies are determined based nationwide-uniform standards. However, conclusions drawn from this study indicate that a system for economic assistance that accounts for regional imbalances in costs and socioeconomic factors would likely be more successful in alleviating post-earthquake debt and improving the speed and quality of earthquake recovery.

The proposed research is based and largely dependent on several deterministic equations with a high level of uncertainty. It is, therefore, not realistically possible to validate modeling outcomes with actual data except in several cases. The model proposed in this study depends on circumstances wherein it is necessary to construct a substantive model based on simple assumptions owing to insufficiency of on-field and related statistical data even in an earthquake-prone nation, such as Japan. Adding to limitations of this research, there also exist aspects affected by the customs of Japan. For example, many of the inhabitants live in wooden houses in Japan and Japanese reinforced concrete structures and heavy-gauge steel structures are resistant to seismic motions. Consequently, non-wooden structures were excluded from consideration in this study. A further study concerning reconstruction of non-wooden houses must, therefore, be conducted and it is necessary to pay attention to discussions at a global level along with the many inherent uncertainties.