特征因子、资产定价模型与股票回报的可预测性

Characteristic Factors,Asset Pricing Models and Predictability of Stock Returns

在风险厌恶的假设下，股票回报具有可预测性。问题是资产定价模型可以解释可预测性吗？本文通过对我国A股市场的研究发现，具有小规模，低流动性，低账面市值比，盈利能力差与投资保守特征的股票组合回报存在更强的可预测性。同时，包含六个因子的定价模型解释了长记忆以及多数组合的非线性依赖。虽然经调整后的股票回报仍具有显著的序列相关，但序列相关大幅度减小。因此，可预测性很大程度上来源于定价模型，风险与可预测性是相关的，可预测性不是“免费的午餐”。

Stock returns have predictability under the assumption of risk aversion. The question is that can asset pricing model explain the predictability？. In this paper, through analyzing Chinese A-share market we find that stock portfolio returns having characteristics including small size, low liquidity, low B/M ratio, weak profitability and conservative investment exist higher predictability. Besides, six-factor asset pricing model explains long-term memory and nonlinear dependency of most portfolios. Although adjusted stock returns still have significant serial correlation, the correlation greatly decreases. Therefore, predictability is largely derived from the pricing model, risk and predictability are correlated, and predictability is not a free lunch.