THE EFFECT OF NATURAL DISASTER ON
and Kronenberg (2016) stated that housing transaction has a long-lasting effect;
consequently, people usually don’t want to make a lot of compromises. Furthermore, Jim and Chen (2007) stated that the fundamental assumption is that homebuyers purchase not only the dwelling unit but also the neighborhood and environmental attributes.
The natural disaster is a part of environmental attributes that should be influencing the residential property values. Houses arguably bear a high proportion of damage due to natural disasters since it has the characteristic of locational fixity. Therefore, these disasters should be one of the considerations when people decide to own a property, especially in the disaster-prone countries like Indonesia. However, this is not always the case. People’s awareness of disaster risk is not always in consideration when housing transaction happens, which is perhaps due to their lack of knowledge. Most people in Indonesia are neglecting the hazardous potency of natural disasters perhaps due to the relatively quiet tectonic movement for more than 200 years (Reid, 2015). Recently, there is a growing awareness in Indonesia related to the effect of natural disaster on people’s lives due to the improvement of information technology, which contributed to the increase of the frequency of disaster occurrence being reported. However, the extent to which this situation has impacted the housing market is quite difficult to measure. Therefore, the research question for this paper is: Does the occurrence of a natural disaster affect household decision to own a house in Indonesia? To answer this question, this paper will adopt the hedonic pricing method coupled with disaster occurrence characteristics.
B. Research Problem and Methodology
Data for this study is obtained from the Indonesia Family Life Survey (IFLS). IFLS is a longitudinal socioeconomic and health survey with the purpose of providing data for behavior and outcome studies (Strauss, Witoelar, and Sikoki, 2016). The IFLS has been conducted five times nation-wide. It has individual, household, and community-level data, with the first wave of IFLS held in 1993. This wave contains data that represents 83% of the population in 13 provinces at that time. The latest wave is IFLS wave five conducted in 2014 for the household level and in 2015 for the community level. This data contains additional data from split-off houses and households that migrated to other provinces. This study will utilize data from the fifth wave of IFLS and combine data from the household level with the community level to see the recent impact of the disaster on the housing market.
It will use housing imputed rent price as the dependent variable as a proxy for willingness to pay for home ownership. This price refers to the value stated by the owner or occupier about how much they are willing to pay for rent if they were renting the house.
Some respondents give monthly rent while others submit a yearly price. This study will combine monthly and yearly data and convert the yearly value into monthly by dividing it with twelve. The explanatory variables can be grouped into three characteristics that are housing physical/structural characteristics, locational characteristics, and disaster characteristics.
C. Data Analysis and Results
The result of the estimation of the R-squared values is 0.20, which is relatively low probably because of the nature of IFLS data. IFLS is a nation-wide questionnaire-type survey with the purpose of capturing people’s behavior. Moreover, its respondents have a diverse socio-economic background; therefore, this survey has a high probability to have volatility in the data. From the estimation, the majority of the explanatory variables are statistically significant at 1% or 5% level. In the housing structural characteristics, all variables show a positive sign, which is as expected. However, the wall material does not significantly affect the imputed house rent price although the sign is positive. In the locational characteristics, all the coefficient of variables are statistically significant. Urban variable shows that if the house is located in the urban area, it is estimated to receive an increase in imputed rent. The distance to market variable shows a negative sign and is significant at 1% level. This indicates that the further the house is from the market, the less amount of rent the respondents are willing to pay.
In the disaster characteristic, which is the main focus of this study, the selected hazard variables have the negative sign, and the coefficient of volcanic eruption is statistically significant. This implies that, in general, the volcanic eruption is perceived as the main issue affecting the decision to own a house in Indonesia compared to other types of disaster. This situation happens because of several factors. First of all, there are many kinds of dangers associated with the volcanic eruption such as lateral blast, debris avalanche, pyroclastic flows, and lahar (Siebert, Glicken, and Ui, 1987). Most of the eruptions expel large amount of lahar, which consisted of hot and liquid substances that destroys everything on its pathway down. The tsunami could also occur if the volcano is located underwater (Siebert et. al., 1987). Volcanic ash is also one of the products of the volcano eruption. This ash could cause a serious health problem, especially the ones related to the respiratory system; furthermore, it can travel a long distance, endangering people who are not in close proximity to the volcano. The eruption also has the ability to alter the temperature of the surrounding areas such as the eruption of Mount Pinatubo Philippines in 1991 (Parker, Wilson, Jones, Christy, and Folland, 1996). Second, the incidents of volcano eruption could last for a longer period compared to other types of disaster.
Some volcanoes erupt violently once and then stop; some erupt continuously for a period that could last for days or weeks, while others erupt irregularly for several years. Third, the events of a volcanic disaster usually get intense media attention and broadcasted nationally, while other hazardous events may get the attention but not as intense as volcanic incidents. However, most of the people don’t know the disaster management process for this kind of disaster.
The volcanic hazard, as shown in Table 3, mostly happens in the Java and Sumatra Island since the majority of volcanoes in Indonesia are located in these islands. However, it is important to note that this doesn’t mean that the other islands are safe from the direct threat of volcanic eruption, except, maybe, Kalimantan and Papua Island, which, to the best of knowledge, doesn’t have any volcano. The other island may have dormant volcanoes that could be active in the future.
The other types of disasters except volcanic eruption do not significantly affect the price due to the low number of households in IFLS wave five data that got hit by them.
From Table 4, we can see that only 131 households that got hit by a landslide, which is less than 1% of the total, and only 255 households were affected by drought. The number of households who got hit by the tsunami is the smallest, which is only seven household.
D. Conclusion
This study provides an overall insight related to the effect of natural hazards on the housing market in Indonesia. The result of the regression shows that decision to own a house is indeed affected by the occurrence of natural hazards, especially volcanic eruptions because it is arguably the disaster with the greatest threat in Indonesia.
To tackle the threat of natural disasters, especially volcanic eruption, this paper proposes two main policy implications that the government of Indonesia could adopt. The first policy implication is to mitigate the decline of housing value related to the impact of a volcanic eruption, and the second policy is to reduce uncertainty in housing market associated with volcanic hazard and promote market transaction. In order to mitigate the declines in housing value, it is important to always remind the households about the threat of natural hazards, like a volcanic eruption, whether they have experienced it or not. A routine and nation-wide educational campaign about the risk of a volcano eruption and how to respond to it is needed. Japan is a good role model since this country has routine drill regarding earthquake response and mitigation. In the case of Indonesia, it’s the response drill of volcanic eruption that is needed the most, especially in Java and Sumatra Island. Moreover, it is highly suggested that the government explicitly provides a simple volcanic eruption hazardous map with the risk zones as well as the routes for evacuation. For the second policy implication, the government could implement a law that requires the appropriate stakeholder to disclosure the risk of various natural hazards to the property or houses. There are two examples of law with a different approach that government could consider. The first is California Natural Hazard Disclosure Law (AB 1195) that requires the developers to disclose the risk of various natural hazards that may hit the property that they have built. Another approach is by employing real-estate transaction specialist, which is inspired from the Japan’s system based on The Building Lots and Building Transaction Business Law. This transaction specialist is designated with the main task to reduce the information asymmetric in the housing market as well as has the authorization to spread information related to the property, including the potential hazard at the location to the buyers.
This paper only utilizes data from IFLS wave five in 2014; therefore, it is highly affected by the recent occurrence of a disaster, which is prominently volcanic eruption such as the eruption of Mount Merapi, Mount Kelud, and Mount Sinabung that happened in the same 5-year period. The other types of disasters are not statistically significant simply because the number of households that got affected by them in the data is very small. The result of the regression shows that volcanic eruption is significantly affecting the house price in recent years, but to claim whether the other types of disaster are truly
not significant for the housing price is beyond this result. Therefore, further improvement could utilize data with a longer period to assess the impact of disaster on the housing market more accurately. Another improvement is to use several variables for the disaster proxy such as the probabilistic value of hazardous event from each type of disaster and the individual household perceived risk of disaster. This approach might capture the individual household behavior related to the disaster a little bit more precise than the result of this paper.
11
► Nama : Muhammad Ridwan
► Unit Organisasi : Kementerian Pekerjaan Umum dan Perumahan Rakyat, Biro Perencanaan Anggaran dan KLN
► Program Studi : Magister Perencanaan Ekonomi dan Kebijakan Pembangunan
► Negara Studi : Indonesia-Jepang
► Universitas : Universitas Indonesia