基于状态空间模型的动态基金业绩评价研究

本文在基金整体业绩评价研究领域对以往经典基金业绩评价指标詹森alpha指数以及基金资产投资的系统风险指标beta的估计方法进行了修正和改良。以往的詹森指数和beta值的估计是将其视为常系数，然而实际中基金的詹森指数和系统风险beta具备时变性。在对常系数下詹森alpha和系统风险beta值的分解式中，本文证明了传统估计值由詹森alpha和系统风险beta的期望值与一系列协方差项组成。随后本文构建了反映动态指标变化的SSM模型，并利用Particle EM算法来估计动态詹森alpha和系统风险beta在各期的估计值，并以此来计算基金在评价时期内平均詹森指数水平和系统风险水平。此外由于获取了各期的系统风险beta，根据择时能力的定义本文构建了反映基金在时期内的择时能力指标。

The Performance Evaluation of Dynamic Mutual Funds Based on SSM

Previous researches always use constant methods to evaluate the mutual fund's performance. In this paper, the constant coefficient estimates are proned to be unable to accurately evaluate the mutual funds' performance by proving that the constant coefficient estimates consist of time-varying coefficient expectation and other component. Then a SSM model is constructed to reflect the time-varying coefficient based on information variables. In addition, the latest Particle EM algorithm is used to estimate the parameters in SSM. Finally, the timing ability of the mutual funds is compared based on the data of six mutual funds between 2005 and 2011.It is found that time-varying coefficient method is more accurate in performance evaluation.