The historical magnitude of the equity premium in the US has been the object of intense study over the last decade. Since the seminal work of Mehra and Prescott (1985), many authors have modified the basic theoretical model to account for the wide discrepancy between the time series generated by a complete markets Arrow-Debreu economy and the data. In Discussion Paper No. 1119, Research Fellow Fabio Canova and Gianni De Nicolo document the time series properties of the equity premium/risk free rate pair for seven countries (Canada, Germany, France, Italy, Japan, the UK and the US) for three samples (1973–91, 1973–81 and 1982–91) and for three holding maturities (3, 6 and 12 months). They show the existence of sub-sample instabilities, of some cross-country differences and of inconsistencies with the expectations theory of the term structure.

The second part of their paper attempts to confront a standard Arrow-Debreu model with the cross-country, cross-sample and cross-maturity evidence they have discovered. The basic results of their simulations suggest that simulated and actual data differ substantially: in most cases, the distribution of the moments of the actual and simulated equity premium/risk free rate pair hardly overlap. Moreover, regardless of the country, the model generates values for the mean and the standard deviation of the equity premium that are too low to be consistent with the cross-sectional post-Bretton Woods experience.

The Equity Premium and the Risk Free Rate: A Cross -Country, Cross-Maturity Examination
Fabio Canova and Gianni De Nicolo

Discussion Paper No. 1119, January 1995 (IM)