Prediction Of Chinese Population Mortality Rate Based On Lee-Carter Mode

In recent years,the life expectancy of the Chinese population has significantly increased.The increase in average life expectancy and decrease in mortality rates have led to a rapid growth in the aging population,exacerbating China’s aging population problem,which has brought new challenges to the government,businesses,and other entities.In order to address the risk of longevity arising from the gap between actual and expected lifespan,measures should be taken proactively.Using a stochastic mortality model to improve the accuracy and robustness of long-term forecasts can effectively help economic entities cope with the impact of longevity risk.At present,some research results have been achieved on the prediction models of China’s population mortality,but the discussion on the applicability and prediction effects of the models is not sufficient.In this paper,the Lee-Carter model is selected to forecast China’s population mortality considering the simplicity and applicability of the model.However,the original data of China’s population mortality are short and missing,and the reliability of the prediction results cannot be guaranteed by applying the traditional two-stage estimation method.Therefore,to address the problem of short years and serious missingness of Chinese population mortality data,this paper explores a Bayesian MCMC method for predicting mortality,and applies this method to the Lee-Carter model.Under this framework,the parameter estimates,predicted values and prediction intervals of the model are given at once by considering the effects brought by all different perturbation terms to avoid the error accumulation problem of the two-stage estimation method of the traditional Lee-Carter mortality prediction model.In this paper,using China’s population mortality data for the period 1997-2019 to project 2020-2030,compare the projected values with the actual observed values,and calculate the residuals as well as the evaluation indicators.The results show that by introducing the Bayesian MCMC method into the Lee-Carter model,the method has a more uniform distribution of residuals and better evaluation indicators than the two-stage estimation method,which effectively reduces the adverse effects of data quality problems and improves the robustness of parameter estimation as well as the accuracy of mortality prediction.