贝叶斯方法在养老基金资产负债管理中的应用

把一个静态资产负债管理模型———均值方差模型应用到定额给付养老金计划的资产负债管理中,在允许无风险借贷的条件下研究养老金在无风险资产和风险资产间的分配问题,用定量分析的方法求出了最优投资组合的一般形式;又针对投资收益率特征参数未知的情况,提出了矩估计和贝叶斯估计两种方法求解最优资本配置比例,将两种方法的结果与一般形式对比,分析了影响最优投资组合的因素,得知养老基金在风险资产中的投资比例与基金经理对风险的厌恶程度、风险资产的风险益酬、风险资产收益率的波动性成负相关关系;并且随决策者掌握的历史信息增加,在风险资产上的投资比例也随之增加,投资行为逐渐趋于理性化;对上述结果进行仿真,验证了结论的有效性。

The Application of Bayesian Method in ALM for Pension Fund

A static ALM model (mean-variance model) is utilized for a defined benefit pension scheme,following the assumption that the risk-free borrowing and lending are allowed and that the assets are allocated between riskless asset and risky asset and risky asset,the generic form of the optimal investment strategy is derived.For the situation where the characteristic parameters of the investment return are unknown,moment method and Bayesian method are developed to calculate the optimal investment percentage and the results derived by the two methods are compared.The factors that affect the optimal investment portfolio are analyzed and the results lead to the conclusion:the degrees of investor’s risk-aversion,the risk premium of the risky asset and the resky asset and the volatility of investment return all have a negative impact on the percentage invested in the risky asset;and that with the amount of information obtained increasing,the investor behaves more rationally. The simulation is performed to confirm the efficiency of our conclusion.