태풍피해위험 분산을 위한 초과손해액 추정

Estimated Excess Losses to Disperse the Risk of Typhoon Damage

This study estimates the magnitude of excess loss based on the parametric model using South Korea’s typhoon damage. The data used for the analysis consists of annual typhoon damage (the number of observations are 47) in South Korea from 1958 to 2015. First of all, we conduct goodness-of-fit test and use graphical method to find a suitable model for typhoon damage. Extreme Value Theory (in short, EVT) is used to consider the probability of observing rare and extreme events corresponding to the tail of the distribution, such as a typhoon. Visual inspections and statistical tests show that the Generalized Pareto distribution is more appropriate than the other models (i.e., log-normal, burr, weibull, and normal distribution). Then we select appropriate thresholds under the selected model and calculate the excess losses that can occur based on the criteria. Furthermore, we have proposed a reinsurance pricing method to determine not only the insurer’s retention level but also to transfer the risk to potential losses exceeding the retention level. The results of this analysis will be helpful for the decision of the typhoon damage assessment, the determination of the appropriate reinsurance price for the risk transfer, the fiscal support decision that the government takes.