Fama-MacBeth two-pass regressions: Improving risk premia estimates

In this paper, we provide the asymptotic theory for the widely used Fama and MacBeth (1973) two-pass risk premia estimates in the usual case of a large number of assets. We demonstrate analytically and using simulations that the standard OLS and GLS estimators can contain large bias when the time series sample size is small, but our proposed OLS and GLS estimators can reduce the bias significantly.