Examining the Statistical Properties of Financial Ratios

This paper focuses on two popular predictive variables that are often used to forecast stock market returns: the dividend yield and the price earnings ratio. First, we question the theoretical premise that the dividend yield ought to have predictive power for the aggregate stock market. By referring to the Modigliani and Miller (1961) theorem that dividend policy is irrelevant, but allowing for uncertainty, we show that the dividend yield might not be the appropriate variable to forecast returns on equity. In addition, we show that the dividend yield for stock indexes tends toward the dividend yield of the fastest growing firm in the index, thereby introducing a downward trend in the dividend yield. On the other hand, the earnings yield provides a good proxy for the expected return under mild conditions: if the return on aggregate investment coincides with the expected market return and expected returns form a martingale sequence. Our theoretical findings are confirmed in the S&P 500 data: prices seemingly overreact to dividends, even in the long-run, as suggested by the long-run elasticity of prices with respect to dividends of about 1.5 as first observed by Barsky and DeLong (1993). The long-run elasticity of prices with respect to earnings, however, is insignificantly different from one, suggesting that the focus on dividends alone might be rather misguiding in judging the efficiency of the market. Accounting for these facts, simple tests of market efficiency can be constructed. We find that, in the short-run, the efficient markets hypothesis cannot be rejected.