Working Paper: NBER ID: w27410
Authors: Jacob Boudoukh; Ronen Israel; Matthew P. Richardson
Abstract: Analogous to Stambaugh (1999), this paper derives the small sample bias of estimators in J-horizon predictive regressions, providing a plug-in adjustment for these estimators. A number of surprising results emerge, including (i) a higher bias for overlapping than nonoverlapping regressions despite the greater number of observations, and (ii) particularly higher bias for an alternative long-horizon predictive regression commonly advocated for in the literature. For large J, the bias is linear in (J/T) with a slope that depends on the predictive variable’s persistence. The bias adjustment substantially reduces the existing magnitude of long-horizon estimates of predictability.
Keywords: predictive regressions; small sample bias; asset pricing; long-horizon forecasts
JEL Codes: C01; C22; C53; C58; G12; G17
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| small sample bias (overlapping regressions) (C24) | small sample bias (non-overlapping regressions) (C24) |
| correlation between return innovations and predictive variables (C29) | small sample bias (C83) |
| autocorrelation of the predictive variable (C22) | small sample bias (C83) |
| sample size (C83) | small sample bias (C83) |
| horizon (j) (F01) | small sample bias (C83) |
| small sample bias (C83) | estimated predictability (C53) |
| small sample bias (C83) | evidence of predictability (C53) |