A Note on Nonsense Predictive Regressions Arising from Structural Breaks

Preprint
Publisher Website
Google Scholar
Abstract
In this short note we respond to the argument advanced by Baker, Taliaferro, and Wurgler (2006) that our criticism of the market timing literature is simply a reinterpretation of Stambaugh's (1999) small sample bias. We show analytically how structural breaks in an economic time-series may result in spurious, or nonsense, predictive regressions, whether or not there is any small-sample bias at play. We also provide a simple example showing that the magnitude of this bias could explain the predictive power of certain variables.