Investment Portfolio Analysis Of China's Pension Fund

The state pension fund of China is currently in the transition between the pay-as-you-go and the partial accumulation system. However, its payoff is still too low even with the consideration of its high safety level, which urges the investment portfolio research on the maintenance and increment of its value.If the mortality rate is in the high side of the life period table, the expected remaining lifetime of an individual will be underestimated, which may cause the underfunding of the pension fund. Given the importance of the mortality rate of China's population, my thesis starts with the forecast and parameter estimation according to its age and sex conditions. Followed by the estimation of the initial state space model with Kalman filter technique, forecast is conducted in the generated model. By comparison, the expected remaining lifetime at the youth stage is highly underestimated. Hence, the concept of LR-type fuzzy numbers is adopted to characterize the return level of the fund with its transaction cost in consideration. In the same time, the longevity risk is treated as a control factor as the aging peak of China's population approaches. In accordance with minimum investment requirements of policies, the lower limit is set to increase the practical operability. Therefore, a portfolio optimization model of longevity risk is established based in the fuzzy environment. Furthermore, the returns of indices of bonds, funds and stocks are used to simulate the situations of the corresponding investments from the pension insurance fund.Finally, I can not only obtain an optimal investment portfolio model considering longevity risk in a fuzzy environment, but also analyze the VaR of investment portfolios of China's pension fund. A GARCH(1,1) model is selected to fit a marginal distribution, and parameters of various Copula functions are estimated. After comparative analysis, an optimal fitted tCopula model is found with respect to the available data. The model can be applied to calculate the VaR of investment portfolios of indices of bonds, funds and stocks. The obtained results by regression tests are satisfactory both statically and dynamically.