Decision-making Model of Bank's Assets Portfolio based on Multi-period Dynamic Optimization

Based on the dynamic programming method, by the use of the constraints on VaR, laws, regulations, and operation, a multi-period dynamic portfolio optimal model for banks is successfully developed with the objective of maximizing the portfolio's yield. The characteristics and innovations of this paper are as follows. First, using the Backward Induction Method, the optimal portfolio in the current period is obtained on the basis of the optimal portfolio in the next period. It solves the problem of obtaining the portfolio's yield only in a single period and ignoring the interactions of various periods in the existing studies. Second, this paper considers that the credit rating migration in the previous year has a significant impact on the expected value of the loan's yield in the current year. Then both the loan's yields under different ratings and the one-year credit rating migration matrix are employed to calculate the corresponding annual expected values and standard deviations of the loan' s yields of various corporations. It can objectively reflect the real yield and risk. As a result the problem of simply seeking the expected value of each loan's yield or only considering it as a constant in the recent studies can be solved. Third, from the analysis result, it shows that the portfolio's yield in the near period is larger than that in the long period. And it also shows that the model gets a progressive optimal result and can well reflect the time value of assets. The paper solves the problem of neglect of the time value of assets in the recent studies. Finally, this paper uses VaR to control the multi-period portfolio risk and solves the problem of lack of consideration of the bank's risk tolerance ability and the requirement of capital supervision in recent multi-period studies.