Catasrophe Bond Pricing: An Application of Extreme Value Theory

This research aim to (1) apply catastrophe (CAT) bond pricing model by Zimbidis, Frangos and Pantelous onto data of Japan seismic activities (2) explore the option of using generalise Pareto distribution in Zimbidis, Frangos and Pantelous model. Qualitative methods and statistical software R is applied to achieve the objectives of this study. GEV fitted Japan data of earthquake is input to Zimbidis model and found that compared to the original study on Greece, Japan who has higher seismic risk is found to have tail heavier. A 5 year CAT bond is found to be more risky than a one year CAT bond. So the price of a 5 year CAT bond is found to be smaller than that of a one year bond. On the other hand, GPD fitted Japan data of earthquake is input to Zimbidis model and found that the simulated prices are more consistent than those produced by GEV fitted Japan data do. Thus, GPD is said to be more appropriate in distributing Japanese earthquake risk dynamics. Hence, in order to price a Japanese CAT bond, GPD is a better choice given the set of Japan data of earthquake within the same time frame