This paper uses financial data from 1900 to 1925 to run out of sample tests of different asset pricing models. We find that we cannot reject the strong predictions of Sharpe’s (1964) CAPM, but that there are portfolios with significant alphas that violate Sharpe’s CAPM weak predictions. The Black (1972) version of the CAPM performs worse than Sharpe’s counterpart. We also test the Fama French Carhart framework, and find that only the market and size factors work as with modern data. The value factor is statistically insignificant, and the momentum factor, while significant, has the opposite sign of the modern momentum factor.

Elaborado por: Miguel Cantillo Simon and Nick Wonder