Abstract: In a financial market where information is partially observable, a DC pension member can invest in a risk-free asset, a rolling bond, and a stock whose expected return rate is predicted by a predictor that follows a mean-reversion process. The member is ambiguity-averse, and she can only observe the stock price and interest rate, but not the predictor. By virtue of the filtering technique and dynamic programming approach, the analytic expression of the optimal investment strategy for the DC pension with incomplete information and ambiguity aversion is obtained. The sensitivity analysis and comparative static analysis are used to compare the optimal investment strategies in three cases of considering only incomplete information, considering only ambiguity aversion, and considering both incomplete information and ambiguity aversion. The results show that the optimal investment strategy with incomplete information and ambiguity aversion is the most conservative, and the optimal investment strategy with incomplete information is the most sensitive to the change of risk aversion coefficient.

Key words: DC pension, incomplete information, ambiguity aversion, return predictability, filtering technique

摘要： 在信息部分可观测的金融市场中,参与者可投资于一个无风险资产、一个滚动债券和一支股票。其中,股票的预期收益率由一个服从均值-回复过程的预测因子预测。参与者是模糊厌恶的,只能观测到股票价格和利率,却无法观测到预测因子。利用滤波技术和动态规划原理,得到了不完全信息和模糊厌恶下DC型养老金最优投资策略的解析式。进一步,利用敏感性分析和比较静态分析,对比仅考虑不完全信息、仅考虑模糊厌恶以及同时考虑不完全信息和模糊厌恶三种情形下的最优投资策略。结果表明同时考虑不完全信息和模糊厌恶时的最优投资策略最保守,仅考虑不完全信息时的最优投资策略对风险厌恶系数的变化最敏感。

关键词: DC型养老金, 不完全信息, 模糊厌恶, 收益可预测性, 滤波技术