存在离群值情形下行业股票的风险特征分析

Risk Characteristic Analysis of Stocks of Different Industries in Cases with Outlier Values

资产定价是金融研究的核心问题。著名的资本资产定价模型(CAPM)给出了使用β系数来衡量资产风险的方法。实践中常使用拟合线性模型的方法来估计β系数。当样本数据中有离群值时,经典的最小二乘法可能会有较大的估计误差。这种情况下,采用稳健估计方法是一个较好的选择。我们通过模拟的数据,发现当数据存在离群值时,稳健估计方法的偏差比最小二乘法小得多。使用稳健估计方法来分析A股市场不同行业股票的风险大小,发现银行、钢铁、交通运输业的β系数相对较小,而电子和房地产行业有较大的β系数。

Asset pricing is a central issue in financial research.The famous Capital Asset Pricing Model(CAPM)usesβcoefficient to measure asset risk.In practice,methods of fitting linear models are often carried out to estimate theβcoefficient.The classical least squares method may have large estimation errors when the sample data contain outliers.In this case,adopting a robust estimation method is a better option.We find that the robust estimation methods have much smaller bias than the least squares method when there exist outliers based on simulated samples.Using the robust estimation methods to analyze the risk of stocks in different industries in the A-share market,we find that theβcoefficients of the banking,steel and transportation industries are relatively small,while the electronics and real estate industries have largeβcoefficients.