This paper adopts the extreme value and VaR approach to investigate
the amount of rice damaged due to extreme events and
analyzes the collective risk model as a feasible scheme for estimating
annual aggregate losses. The results show that the annual
frequency of rice damage caused by typhoons is shown to fit
well the Poisson distribution with one parameter. The generalized
Pareto distribution (GPD) with two parameters outperforms
the log-normal fit with respect to the tail-related risk measures,
e.g., VaR, ES, and EAS. GPD allows easy estimation of the high
quantiles and the maximum probable loss from the data. The
threshold value can be used as reference in decision making for
setting grant-in-cash relief. We believe that, given different confidence
intervals, these high-quantile measures can provide useful
information in reviewing the applicable loss compensation
regulations and for adjusting natural disaster relief budget plans
or insurance pricing on the non-insurance plan.
