Detrending Persistent Predictors for Forecasting Stock Returns
34 Pages Posted: 12 May 2011
Date Written: May 1, 2011
Researchers in finance very often rely on highly persistent – nearly integrated – explanatory variables to predict returns. However, statistical inference in predictive regressions depends critically upon the stochastic properties of the posited explanatory variable, and in particular, of its persistence. This paper proposes to stand up to the usual problem of persistent regressor bias, by detrending the highly autocorrelated predictors. We find that some evidence of predictability at short horizons based on financial ratios.
Keywords: forecasting, persistence, wavelet, expected returns
JEL Classification: C14, C58, G17
Suggested Citation: Suggested Citation