Multiple Population Mortality Jointly Forecasting in China Using PCF Model

Abstract

Based on the data of population mortality in China since 1994, this paper studies the changing trend of sex differences of mortality by proposing and implementing some novel approaches. First of all, this paper proposes a new method using the Poisson common factor (PCF) model to forecast the mortality jointly by both sexes. The study finds that the PCF model effectively captures the common trend of mortality between two sexes and uses additional factors to reflect differences of between two sexes, which can reduce the errors caused by low quality or large fluctuation of the mortality in China. Meanwhile, the forecasting values of mortality based on the PCF model can abide by the human biological law well, and the sex ratio of mortality converges to a fixed constant in the long run without increasing too much statistical error. Second, this paper improves the parameter estimation method of PCF model, and the innovative two-step method is used to estimate the model, which can make the maximum likelihood estimator converge more easily. Finally, as an application of the novel methods which are proposing in this study, we measure the longevity risk of pension by using the PCF model and find that the PCF model can make up the underestimation of longevity risk from traditional models and provide more scientific information to the sponsor of pension plan.