Large Sample Estimators of the Stochastic Discount Factor

Abstract We propose estimators of the stochastic discount factor (SDF) using large cross-sections of individual stocks. Our small-sample bias corrections allow us to exploit unbalanced panels of individual stock returns. Our estimators can accommodate prespecified traded and non-traded factors, and latent factors. The estimators perform well in simulations. We apply our estimators to return data for U.S. individual stocks over a 50-year sample period and identify those factors in popular asset pricing models that command significant premia. A number of proposed non-traded factors have insignificant risk premia. Contrary to many studies, we find the market factor has a significant premium, as do profitability, value, and momentum factors