论文部分内容阅读
从大量的金融资产中提取出的系统风险比基于β系数的单变量方法更为有效,但资产规模的增加会导致“纬数灾难”等问题,难以获得准确估计。本文在将金融资产收益分为公共系统因素和个体特质因素基础上,提出用具有条件异方差形式的动态潜在因子模型(CHDL)估计和预测动态系统因素,用非参数核密度估计系统下跌时的边际期望损失(MES)。本文利用上海证券市场180只样本股进行实证分析,通过IC和Onat检验发现个股和各板块存在显著的系统因子;利用CHDL模型对个股和各板块的系统因子和资产未来收益进行估计和预测,在此基础上计算边际期望损失。Mincer-Zarnowitz回归最优检验法表明,CHDL模型计算的系统风险比常用的市场指数模型具有更高的准确性。
The system risk extracted from a large number of financial assets is more effective than the one-variable method based on beta coefficient. However, the increase of asset scale will lead to such problems as “latitude calamity ” and it is difficult to obtain accurate estimates. Based on the public system factors and individual trait factors, this paper proposes a dynamic latent factor model (CHDL) with conditional heteroskedasticity to estimate and predict the dynamic system factors. The non-parametric kernel density is used to estimate the decline of the system Marginal Expected Loss (MES). In this paper, an empirical analysis is made on 180 sample stocks in the Shanghai stock market. ICs and Onat tests show that there are significant system factors in each stock and each sector. The CHDL model is used to estimate and forecast the system factors and asset future returns of individual stocks and each sector. Based on this calculation, the marginal expectation loss. The Mincer-Zarnowitz regression to the optimal test shows that the systematic risk of the CHDL model calculation is more accurate than the commonly used market index model.