Prediction Methods Of Chinese Population Mortality

The trend of extending the life of the population has become recognized and accepted, on the one hand this trend offers significant quality of life of people, but it also brought some social problems. In the insurance system, referred to the negative impact of this called longevity risk, with surging of the population aging, longevity risk has become one of increasingly serious social problems which community, government, businesses and households are facing. Longevity risk is the survival of the general population, the average expected life in excess of the survival of life, and such risk is not based on law of large numbers to control systemic risk. This led to the life insurance company life insurance product, company pension plans and Social Security Fund pension of government are all subject to the threat of longevity risk. Therefore,both effectively identify and measure the impact of longevity risk are the important issue of life insurance companies and the social security sector now. And only when longevity risk identification and measurement of time has been made, it is a way to take effective control measures to reduce the risk of negative impact of longevity risk.The one of the main reason that life expectancy is mortality rate declining. There is no doubt that accurate, comprehensive forecast of mortality data has played a very important and basic role in identifying and measuring longevity risk. This article is mainly dedicated to find the way that suit to characteristics of mortality in our country, and predict a comprehensive and more accurate mortality data for the future. In addition, because mortality in the elderly population has a great effect in identifying and measuring the longevity risk, therefore this is still the basis of mortality prediction model used to extrapolate the mortality rate out of the elderly population.In the choice of choosing mortality prediction model, this article prefer to the Lee-Carter model which has advantagement in the prediction accuracy and application of broad aspects, and it is no doubt that it is a key in predicting for the model parameters in the model. Model parameter estimation methods including Ordinary Least Squares (OLS), Weighted Least Squares (WLS), Sngular Value Decomposition (SVD) and Maximum Likelihood Estimation (MLE) parameter estimation method. There are four methods predict mortality in 2009 compared with the actual value, I believe that the maximum likelihood parameter estimation method is more suited to our deaths data characteristics.Model for the extrapolation of mortality of the elderly population mainly in the Gompertz model and the Coale-Kisker model selection, the 2005 mortality data relatively more, so the above two models were used to this article on the 2005 mortality of the elderly population extrapolation. Extrapolation by comparing results, I believe that the Coale-Kisker model is more appropriate for extrapolation of our elderly population. Through the application of Lee-Carter model and the Coale-Kisker model, to get the country in 2010-2020 by sex, by age (form 0 years old to 102 years old) mortality prediction.This is the greatest innovation of combing the elderly population mortality Coale-Kisker extrapolation model and the Lee-Carter forecasting models, resulting elderly population mortality data in the future. Its significance lies in:First, as our country steps into the aging society gradually, the proportion of elderly population is increasing, the use of mortality can not predict the future population of the elderly population of the lack of data; secondly, whether the government's social pension insurance scheme commercial insurance companies,insurance products and pension annuity business,in considering the payment standard and the reserve set aside the issue, the elderly population mortality data is undoubtedly essential. The second innovation point is that in the use of Lee-Carter model, regardless of adopting the original data or predicting the mortality data are age-level, which is different from previous studies that is based on a 5-year-old age group to predict. Its significance lies in the individual age, for some relatively large changes in general instead of using one age mortality of all ages mortality is unreasonable. The third innovation is when estimating Lee-Carter model parameters,investigated four model parameter estimation, and selected a method that is more suitable to China's method of mortality data characteristics.