This master thesis deals with pricing models for catastrophe bonds. We present the probabilistic setup based on the work of Cox/Pedersen 2000 in which financial and catastrophic risks are treated independently. The change of numéraire technique to a zero-coupon bond price process is introduced and short rate models are derived. Catastrophic risks are modeled with the hazard function approach or a non-homogeneous compound Poisson process. Different cat bond pricing methodologies are presented and partially extended. The simple payoff profile of Ma/Ma 2013 is extended to the more general stepwise or piecewise linear function in Nowak/Romaniuk 2013. A simple closed form solution with LIBOR coupon payments is developed by Jarrow 2010. In the second part of this thesis an alternative probabilistic model is developed resting upon the theory of enlargement of filtrations from default risk. The argumentation of Cox/Pedersen 2000 is examined. The important assumption that the use of the real-world measure for catastrophic risks leads to arbitrage-free prices is justified. By the use of forward measures the closed form pricing approach of Jarrow is extended to a solution avoiding two approximations Jarrow had to make. In the computational part – for which we use the software R and MATLAB – a left-truncated NatCat linked loss index from 1985 to 2011 is inspected. Intensity functions are expanded by a linear trend and a random periodicity. Further, the peaks over threshold claim size distribution and parametric survival functions of NatCat linked cat bonds are estimated. The extended pricing approach of Jarrow is implemented in order to show sensitivities of cat bond prices.
«