Predictability of stock returns: is it rational?

Abstract

This study identifies three anomalies in the British capital markets. It is statistically proven that the logs of six stock prices, in the British stock market, are cointegrated with the logs of a market index, a bond price, and an exchange rate. This means that the lagged residual of each cointegration regression appears in the regression of the first-differences of the above variables in what has been called the error-correction regression. This anomaly means that past information is helpful in predicting current stock returns and this may be due either to the presence of a fad, or to forward-looking rationality. Two other anomalies are 1) the correlation of cointegration residuals across stocks, which may be explained by a common factor absent from the regressions and 2) the fact that it is the first-difference of the interest rate, instead of the level, that explains stock returns, which is consistent with the evidence that the level of the interest rate is non-stationary.