多维地震风险模型与保险基金测算

Multivariate Modeling for Earthquake Risks and Earthquake Insurance Fund Projection

本文选取我国大陆地区1990~2019年的地震灾害数据为研究样本,以地震造成的直接经济损失、死亡人数和受伤人数三个维度为研究对象,建立了复杂数据结构下的多维地震风险模型。在不同期限结构和不同偿付水平下,运用蒙特卡洛模拟算法测算了我国地震保险基金的风险规模和保费水平。与传统的巨灾精算模型相比,多维地震风险模型不仅解决了连续型、离散型和半连续型数据的联合建模问题,还能够更好地刻画多维地震风险之间非对称的相依结构。实际分析表明,在巨灾风险度量中考虑复杂的相依结构能够更好地分散地震的尾部风险,有效地减少政府的财政负担,提高地震保险的风险覆盖水平,为完善我国地震巨灾保险制度提供理论支撑。

Using the earthquake damages during 1990-2019 in China as the sample, the paper established a multivariate earthquake risk model with a complex statistic structure, aiming to study direct losses, casualties, and number of injured people caused by earthquakes.The Monte Carlo simulation is used to calculate the Value-at-Risk(VaR) of the earthquake insurance fund, along with the premiums based on different duration structures and solvency levels.In comparison with the traditional actuarial models for catastrophes, the multivariate earthquake risk model not only solves the problems of correlated modelling for continuous, discrete and semi-continuous data, but also better characterizes the complex interdependence structure among multi-dimensional catastrophe risks.The empirical result shows that the tail risk can be dispersed by introducing the dependence structure in the earthquake risk model, thus effectively reducing government’s fiscal burdens and increasing insurance coverage.It also provides theoretical support for improving the earthquake insurance system.