Predictability Hidden by Anomalous Observations

TL;DR

This paper introduces a new testing framework for predictive regressions that remains reliable despite small violations of assumptions, revealing hidden market return predictability in the presence of anomalous data.

Contribution

The authors develop a robust testing method compatible with nearly integrated regressors and approximate models, improving inference in complex predictive regression settings.

Findings

The new approach significantly outperforms existing tests in simulations.

Empirical analysis uncovers strong evidence of market return predictability.

Predictive variables like dividend yield and volatility risk premium are effective in revealing hidden signals.

Abstract

Testing procedures for predictive regressions with lagged autoregressive variables imply a suboptimal inference in presence of small violations of ideal assumptions. We propose a novel testing framework resistant to such violations, which is consistent with nearly integrated regressors and applicable to multi-predictor settings, when the data may only approximately follow a predictive regression model. The Monte Carlo evidence demonstrates large improvements of our approach, while the empirical analysis produces a strong robust evidence of market return predictability hidden by anomalous observations, both in- and out-of-sample, using predictive variables such as the dividend yield or the volatility risk premium.