Research On Asset And Liability Management Of Dividend Life Insurance Products Based On Multi-stage Stochastic Programming

The new regulation puts forward higher requirements for the assetliability management ability of life insurance companies,and asset-liability management by products will be more effective.Under the background of asset-liability management by product,the huge premium scale of dividendpaying life insurance products challenges its asset-liability management.The asset-liability management of dividend-paying life insurance products is dynamic and continuous,and it is necessary to control risks while pursuing profits.Multi-stage stochastic programming model is suitable for solving such problems.This paper constructs a multi-stage stochastic programming model for the asset-liability management of dividend-paying life insurance products,taking the maximization of the expected wealth value at the end of the planning period as the objective function.On the basis of initial investment budget constraint,cash flow account balance constraint,asset portfolio adjustment,investment proportion constraint and dividend distribution constraint,the minimum interest rate guarantee constraint and solvency constraint are innovatively added.The random variables of the model include the return on assets on the asset side and the death payment,surrender expenditure,expense expenditure,premium and reserve evaluation value on the liability side.The initial investment amount of various assets,the holdings and changes in each period are taken as the decision variables.By establishing the vector autoregressive model and Monte Carlo simulation,the scenario tree of asset return rate is generated,which is used as the input value of the random variable of asset return rate in the future.According to the actuarial model of life insurance and historical data,the realized values of these random variables,such as death payment,surrender expenditure,expense expenditure,premium and reserve evaluation value,are generated,that is,the input values of the random variables on the liability side of the model,and the random parameters in the stochastic programming model are transformed into deterministic parameters,thus transforming the stochastic programming problem into a general nonlinear programming problem.The results show that the large-scale asset allocation scheme in a stable state is obtained,and the effectiveness of multi-stage stochastic programming model in asset-liability management of dividend-paying life insurance products is proved,which provides a reference for the large-scale asset allocation decision of asset-liability management of dividend-paying life insurance products.At the same time,it is also found that gradually relaxing the investment restrictions on corporate bonds within a certain range is beneficial to profitability but not conducive to safety.Life insurance companies can build a dynamic asset-liability management program for dividend-paying life insurance products to dynamically adjust their assetliability management objectives and realize dynamic asset-liability management.