Stock return predictability before the First World War

This paper studies the predictability of stock returns using monthly data on eight markets over the period 1876-1913. In contrast to much of the existing literature I find broad predictability across stock markets. Market interest rates and seasonal dummies generally have predictive power, and in almost all of series studied there is a statistically significant autoregressive component. These relationships appear to be stable over the sample period. Testing returns from multiple indices for the same market indicates that the compilation of the index does not systematically affect its predictability. Finally, the results are robust to the exclusion of extreme observations.