Conditional confidence intervals for the equity premium and other rates

Abstract

Almost all of the published estimates of the equity premium and of other rates, are point estimates. The original point of this article is to compute 95% confidence intervals for these parameters conditional on a theoretical dividend model. The monthly samples are considered to have a break after 1981, as deemed in the literature and this turns out to be appropriate. The main result is that these confidence intervals include all estimates of the parameters in the literature, making all of them probable. Moreover and contrary to the opinions held in the literature, the unexpected capital gains after 1981 were not due to an unexpected fall in discount rates but due to an unexpected fall in the difference between the discount rate and the growth rate.