Pricing Of Catastrophe Bonds Based On Mixed Weight LogGED-GPD Model

The recent occurrence of major natural disasters in China is very serious.The loss of insurance claims caused by catastrophe risks has remained high for six consecutive years,and the solvency of insurance and reinsurance companies has become increasingly prominent.The market capitalization of China's capital market is estimated to be as high as 80 trillion yuan.The risk of catastrophe risk will be dispersed into the capital market.Financial products such as catastrophe bonds will be used to disperse the insurance risks caused by major natural disasters and reduce the catastrophe insurance industry.The compensation pressure has effectively compensated for the shortcomings of the reinsurance company's ability to compensate.The catastrophe bonds themselves have a series of characteristics such as strong liquidity and low credit risk,which is an ideal way to transfer catastrophe risks to the capital market.This paper takes China's flood catastrophe bond as the main research object and has completed the following work:Firstly,the distribution model of catastrophe loss amount is studied.On the basis of summarizing the distribution model of catastrophe loss amount proposed by the predecessors,the logarithm-based generalized error distribution is a natural extension of the lognormal distribution and is compared with the lognormal distribution.On the basis of a better fitting effect,a logarithmic generalized error distribution is used instead of the commonly used lognormal distribution.Since the catastrophe loss data generally has thick tail characteristics,the generalized Pareto distribution is introduced to fit the thick tail part of the catastrophe loss data,and the mixed weight method is added to construct the combined distribution model,so that the weight of the model is two.The parameters of the distribution model are determined together,and the logarithmic generalized error distribution-generalized Pareto distribution(LogGED-GPD)combined distribution model of mixed weights is obtained.Secondly,the National Flood Control Report issued by the Office of the National Flood Control and Drought Relief Headquarters was reviewed.The data on the amount of losses and the number of annual floods in China from 1997 to 2017 were extracted,and EM was used in the data preprocessing stage.The algorithm performs missing values on the sample data.The descriptive statistics analysis and thick-tailed test of the data after missing missing values prove that the sample data has thick tail characteristics,and the distribution fitting,parameter estimation and distribution are based on the mixed weight LogGED-GPD combined distribution model.Hypothesis testing,the conclusion that the fitting effect is better than the other nine commonly used distribution fitting models is obtained.The negative binomial distribution is used again to replace the Poisson distribution commonly used by the predecessors,and the distribution,parameter estimation and hypothesis testing of the annual occurrence of flood catastrophe losses in China are obtained,and the distribution of the annual occurrences of flood catastrophe losses in China is obtained.The model provides a theoretical basis for the calculation of the price of China's flood catastrophe bonds.Finally,based on the CAPM capital asset pricing model,the expected yield,yield to maturity and pricing formula of the catastrophe bond are given.According to the distribution model of the loss amount and the number of losses established above,the Monte Carlo simulation method is used to simulate the flood in China.The trigger point probability of catastrophe bonds has been empirically priced for China's flood catastrophe bonds.