上市公司综合盈利水平与股票收益

本文基于2000-2017年上市公司的财务及股票交易数据,研究了上市公司综合盈利水平与股票收益之间的关系。我们使用目前资产定价文献中较新的偏最小二乘法和组合预测法,从12个衡量公司盈利能力的指标中提取了一个测度上市公司综合盈利水平的指标。研究结果显示,上市公司综合盈利水平能够显著预测未来股票收益。使用单因子偏最小二乘法、取12个月斜率的平均值构造的综合盈利水平最有效,以其构建的多空对冲投资组合能产生15%的年平均收益,夏普比率达到0.75。与此对应,组合预测法提取的上市公司综合盈利水平的预测能力稍低,但依然显著。在控制了其他公司特征变量后,综合盈利水平对于股票收益的解释能力依然稳健。本文还从经济机制的角度出发,探讨了综合盈利水平对收益的预测来源。我们发现,上市公司综合盈利水平与股票预期回报的正向关系在投资摩擦更低的组中更高,而在错误定价程度更高的组通常更低。这些结果支持了基于投资摩擦的Q理论,而与行为金融的错误定价理论相悖。

Composite Profitability of Chinese Firms and Stock Returns

The question of why different assets deliver different returns is a fundamental problem in finance. In this regard, the literature has mainly focused on the relationship between the profitability and subsequent stock returns of firms. Profitability is also an important factor in the newly proposed asset pricing models. Furthermore, the empirical research on asset pricing has shown that a large number of firm characteristics can be used to forecast a cross-section of stock returns. However, because some of these factors have lost their predictability after being identified in academic papers or learned by the market, the question of how to extract the commonalty of the predictors and aggregate the effective information has become a key issue in empirical finance. Different from the literature, which explores the predictability of individual profitability-related proxies, in this paper, we aggregate a composite profitability measure of Chinese firms from a set of individual profitability-related indicators. We then investigate the relation between a firm's composite profitability and stock returns in the Chinese stock market.
Specifically, we use the partial least squares (PLS) and forecast combination (FC) methods to aggregate a composite profitability measure from 12 individual profitability related proxies. Composite profitability provides a comprehensive measure of a firm's profitability, and may provide the basis for a new asset pricing model. We obtain data from the China Stock Market and Accounting Research (CSMAR) database from 2000 to December 2017, including accounting data, monthly stock returns, Fama-French (1993, 2015) common factors, and the Chinese risk-free rate. We find that firms with high composite profitability always have high future stock returns. Using the single factor PLS method and taking the 12-month average slopes is the most efficient way to aggregate the composite profitability. The long-short portfolio generates 15% average annualized returns, with a Sharpe ratio of 0.75. In comparison, using the FC method to calculate the composite profitability generates lower subsequent stock returns. The main objective of PLS is to extract a common factor from a set of predictors that has the highest covariance with the predicted variable, which is a “disciplined” dimension reduction technique. The FC approach averages the univariate predictive regression values of firms' profitability equally, but it ignores the multivariate information structure and interaction between firms' profitability. Hence, the PLS approach is more effective in aggregating information for cross-sectional analyses, and makes more accurate future return predictions.
We use different asset pricing models to calculate the abnormal returns generated by the composite profitability, including the Fama-French five-factor model. The results show that when using the PLS single factor model, the abnormal returns of the monthly long-short portfolios are 1.27% (t=3.07), 1.50% (t=3.16), and 1.22% (t=2.94) based on the capital asset pricing model, and the Fama-French three-factor and five-factor models. After controlling for other firm characteristics and risks, such as firm size, book-to-market ratio, and reversal, the positive relation between composite profitability and stock returns is still significant and robust. We then investigate why firms with a high composite profitability have higher future stock returns. The results indicate that the composite profitability premium is stronger among firms with low investment friction, which is consistent with the implications of the investment-based q-theory asset pricing models. However, the premium is not stronger among firms with high mispricing, which contradicts the behavioral mispricing explanations.
Our results differ from the findings on the U.S. market, which suggests that investors in the Chinese market also have to focus on the rational expectation-based model. Our findings also indicate that the research on the international markets cannot adequately explain what happens in the Chinese market. Furthermore, reducing the investment friction helps the market to value its composite profitability more precisely. Future studies should focus on other aggregated information on firm characteristics, such as the investment and trading frictions. Moreover, the economic links and information structure of these factors should also be explored to understand the uniqueness of the Chinese stock market.