Abstract: Taking into account the current social security fund primarily invested in stocks and bonds, we construct new portfolio on behalf of the portfolio of social security fund by using the CSI 300 Index and Bond Index, in order to measure the dynamic risk of social security fund. Firstly, we model the marginal distribution by FIGARCH model. Secondly, we research the dynamic correlation of the assets in the portfolio based on time-varying Copula, finding that the correlations of CSI 300 Index and Bond Index are heavily persistent. Thirdly, we predict VaR of the portfolio by using the method of Mont Carlo, and carry out the Kupiec test. By comparative study, we find that time-varying model is more effective than constant model on the aspect of VaR prediction; both time-varying FIGARCH-Copula-T model and DCC (Dynamic Conditional Correlation)-MFIGARC-T model can correctly measure the dynamic risk, but the former is better than the latter.