The predictability of industry portfolio returns

Abstract

This article studies the predictability of stock returns from industry portfolios. Consistent with the habit formation framework of Campbell and Cochrane (1999, 2000), we find that reasonably large portions of predictability of long-horizon industry portfolio returns are explained by the ratio of aggregate consumption in surplus of habit or its instrument, the consumption–wealth ratio. The time-varying βs and, more importantly, time-varying market risk premium associated with either the surplus consumption ratio or the consumption–wealth ratio help explain the predictable variation of long-horizon expected returns on over half of the industry portfolios. The conditional Capital Asset Pricing Model (CAPM) with βs varying with the proposed conditioning variable performs better than the static CAPM, but not as well as the Fama–French (1993, 1997) three-factor model in explaining the time-series variability of returns.