Return Decomposition

Abstract

A crucial issue in asset pricing is to understand the relative importance of discount rate (DR) news and cash flow (CF) news in driving the time-series and cross-sectional variations of stock returns. Many studies directly estimate the DR news but back out the CF news as the residual. We argue that this approach has a serious limitation because the DR news cannot be accurately measured due to the small predictive power, and the CF news, as the residual, inherits the large misspecification error of the DR news. We apply this residual-based decomposition approach to Treasury bonds and equities, and find results that are either counter-intuitive or unrobust. Potential solutions, including modeling both DR news and CF news directly, the Bayesian model averaging approach, and the principal component analysis, are explored.