Abstract

This paper aims to estimate the maximum probable annual losses to the risks of flooding. While examining the frequency of the event, it was observed that, at least one accident occurred periodically, which resulted in maximum losses. This is an empirical study and it is based on maximum losses due to floods per year, as obtained from the International Disaster Database and Munich Re. The analysis is based on 66 extreme flood events in the world between 1906 and 2015. The complementary risk method has been used in deriving a mixed probability distribution, which expresses the number of floods and the maximum losses realized, where zero-truncated Poisson distribution is used for frequency distribution and Last order Weibull distribution for losses. The maximum of flood losses was fitted with compound truncated Poisson-Weibull distribution. Probabilities have been calculated for extreme flood losses, which are less than specific descriptive measures, and expected values have been calculated for extreme flood losses, which exceed the specific descriptive measures. The results from the study indicate that the maximum probable losses are greater than the maximum actual losses. This paper contributes to the risk of extreme floods pricing; and it also helps the governments of the countries in calculating the financial reserves to cover the extreme flood losses.

Key words: Risk management, extreme flood, complementary risk, truncated poisson distribution, last order weibull distribution, expected losses.