Abstract
In a one-factor model, such as the CAPM, it takes two assets to pin down the expected rates of return on every other asset in the economy. The risk-free rate is one. The market is traditionally the second. However, this numeraire can be switched to other assets. If we use long-term Treasuries, and believe earlier empirical evidence in Fama-Bliss (1987) that the long-term yield spread is entirely due to a risk premium and not at all due to a forecast of future yields, then the Treasury yield spread and the market-beta of the long-term Treasury pin down an estimate of the equity premium. For the period from 1962 to 2007, it was 8 er year.However, this method also fails often, specifically in times when the market beta estimate of the long-term Treasury is close to zero. Moreover, given that the CAPM performs so poorly, the author recommends using the resulting estimate only as an internally consistent estimate within the context of the model itself.