中文摘要：

本文提出了一种因子面板数据随机波动模型（FPSVM）,以便研究可观测因素和不可观测因素对金融资产配置的影响,合理制定投资策略.其中,可观测因素用模型中的解释变量来反映,不可观测因素通过对随机误差项进行因子分解予以体现.由于面板数据随机波动模型包含均值方程和波动方程,而且潜在因子和解释变量之间存在一定的相关关系.本文采用基于贝叶斯估计的MCMC算法对因子面板数据随机波动模型的高维参数进行估计,讨论了其中的先验和后验分布的设定.最后运用中国股票市场数据对资产价格的影响因素进行了分析,其结果可以应用于金融资产配置中.

英文摘要：

In this paper, we propose factor panel data stochastic volatility models to research observable and unobservable factors in asset allocation. These results contribute to develop a reasonable investment strategy. The observable factors and unobservable variables in FPSVM are reflected by explanatory variables and error terms respectively. The FPSVM consist of mean equation and the volatility equation. The weak correlation exists between latent factors and explanatory variables. We estimate high-dimensional feature parameters based on Bayesian MCMC algorithm. The priori and posteriori distribution are studied carefully. Finally, we use Chinese stock market data to analyze the influence factors in asset allocation.