关键词: 人口死亡率, 死亡异质性, 混合泊松模型, 长寿风险

Abstract: Multi-population stochastic mortality model is a frontier issue in the fields of population statistics and actuarial science. It is meaningful to establish a mortality forecasting model in line with the characteristics of Chinese population for population forecast, longevity risk measurement and active management. However, the current research pays less attention to the heterogeneity of population death risk, which can not provide scientific research assumptions for multi-population stochastic mortality modeling. Based on the test of the heterogeneity of death risk in China, this paper constructs a Mixed Poisson Common Factor Model for the joint forecasting of male and female mortality. Furthermore, the iterative algorithm of maximum likelihood parameter estimation is given to jointly forecast the mortality of men and women in China, so as to provide a scientific basis for insurance companies to deal with the risk of longevity. The research shows that the Mixed Poisson Common Factor Model can effectively describe the heterogeneity of population death risk, improve the fitting of the model, effectively avoid the disadvantages of traditional models underestimating the improvement of population mortality, and the change trend of mortality sex ratio is in line with human biological laws. Under the second generation of China’s risk oriented solvency system, this study can provide a more robust additional capital reserve for the longevity risk evaluation of insurance companies.

Key words: Population Mortality, Death Heterogeneity, Mixed Poisson Model, Longevity Risk