An analysis of the equity risk premium

Abstract

The authors analyse the equity risk premium over 1926–2000, extending the results of Clinebell et al. (1994) who use simple regression to test for autoregression, random variation and random walk. The present authors find that Clinebell et al.'s significant autocorrelation in the first half is restricted to the behaviour of July 1932, and it is insignificant otherwise. The slope co-efficient for October is insignificant over 1926–1950 and indicates negative (positive) autocorrelation over 1951–1975 (1976–2000). A small autoregressive component is detected for the remaining months over 1926–1975 which the authors relate to a lead-lag pattern consistent with market efficiency.