An Investigation of the Risk and Return Relation at Long Horizons

This paper examines the relation between expected stock returns and their conditional volatility over different holding periods and across different states of the economy. Seminonparametric density estimation and Monte Carlo integration are used to obtain the expected returns and conditional volatility at various holding intervals. We uncover a significantly positive risk and return relation at long holding intervals, such as one and two years, which is nonexistent at short holding periods such as one month. We also show that the existing finding in the literature of a negative risk and return relation may be attributable to misspecification.