報告主題： Earthquake Parametric Insurance with Bayesian Spatial Quantile Regression
During the last two decades, there's a significant increase in economic costs of natural disasters. A variety of parametric insurance has been developed to help the government to withstand the economic losses caused by nature disasters. Among which, earthquake parametric insurance in Yunnan province is a special case. It sets the magnitude of earthquake as the payout trigger. However, as a main limitation of parametric insurance, basis risk is inevitable. To reduce basis risk, a Bayesian spatial quantile regression model is proposed. Earthquake parametric insurance, including its indemnification standards and pricing, is then discussed based on the proposed model and methodologies. The impact of earthquake hazard, risk exposure and vulnerability on economical losses are analyzed and considered in the quantile regression model. Since risk exposure and vulnerability of the epicenter can not be observed and will be treated as latent variables in quantile regression model. Bayesian approaches are applied, and spatial correlation is considered to construct the prior distributions for the latent variables. As a highly seismic region, Yunnan province carried out earthquake parametric insurance since 2014. To obtain a better evaluation of earthquake risk a satisfactory parametric insurance in Yunnan province, the proposed model and methods are applied. With historical earthquake economic losses of Yunnan province from 1992 to 2019, the payments and premiums of 16 different regions in Yunnan Province are calculated. Results show that the basis risk is reduced significantly and the loss ratio is more reasonable compared to the current earthquake insurance.
李云仙，畢業于香港中文大學統計系，現為云南財經大學金融學院保險系教授、碩導。主持國家社會科學基金、國家自然科學基金以及省部級課題多項，參與國家級和省部級課題多項；參與建設省級創新團隊、重點實驗室，以及云南省防災減災智庫；在數理統計與管理，Journal of Multivariate Analysis，Insurance: Mathematics and Economics，Empirical Economics等國內外雜志發表論文多篇。