因子模型与截面股票收益

Factor Models and the Cross-Section of Stock Returns

本文考察了九种因子模型在中国股票市场上的定价表现。基于总体定价能力和大维市场异象投资组合两个维度的研究发现,Liuetal.(2019)三因子模型(LSY3)的表现最优,而Houetal.(2015)q因子模型和Hou etal.(2019,2020)q5模型与Famaand Fenchr(2015)五因子模型的定价表现无显著差异。本文还发现VMG因子定价表现优异的原因是该因子同时包含了价值因子和盈利因子的信息,这说明LSY3模型可统一至股利贴现模型的理论框架下。最后,拓展的股利贴现模型在克服传统股利贴现模型不足的同时也揭示现有因子模型依然存在遗漏重要定价因子问题。

We test the pricing performance of nine popular factor models in the Chinese stock market and find that Liu et al.'s(2019)three-factor model(LSY3)performs best,when comparing the models overall power and their pricing power to span the largest-to-date Chinese data library of market anomalies.In addition,both Hou et al.'s(2015)q-factor model and Hou et al.'s(2019,2020)q'model display similar pricing abilities as Fama and French's(2015)five-factor model.We also find that Value-Minus-Growth(VMG)'s excellent pricing ability stems from the joint information of value and profitability contained in this factor,which in turn implies that LSY3 shares the same theoretical framework as FF5.Finally,we reformulate the traditional dividend discount model and show that there is a new important factor omitted in the existing factor models.