Identification of Coefficients on Ratio Variables in the Cross Section of Stock Returns

This article examines the validity of regressions that use the ratio variables as regressors for analyzing the stock returns. Although price-earnings and price-book value ratios have been often used as regressors, their coefficients rarely find the causal effects due to the omitted variables problem, which is caused by the natural restrictions on the coefficients. Omitted variables also make the prediction less accurate due to the smaller $R^2$. For a correct causal analysis and a better prediction, the variables must be separately included in the model