Investor Sentiment, Regimes and Stock Returns

Abstract

In this paper, we empirically examine the relationship between return predictability and investor sentiment when the stock fundamentals exhibit regime shifts. This study is motivated by the fact that the predictive power of sentiment may be weakened if we do not separately identify the price change as a correction of a mispricing due to sentiment and/or an adjustment dynamic in relation to the regime shift. We propose a simple way to explore this issue within the conventional predictive regression framework and a testing procedure to tackle the potential econometric problems. Our main empirical findings are: (1) the effects of sentiment on predicting the cross-section of future stock returns are significant only under a certain regime (bullish regime); (2) dividend- and earning-oriented portfolios show strong conditional predictability patterns only after conditioning on sentiment and regime; (3) the appearance of the size and value effects is associated with sentiment and the state of regime; (4) the cross-sectional predictability patterns associated with sentiment reflect the mispricing, not the compensation for systematic risk.

Keywords: Sentiment, Return Predictability,Mispricing, Markov-Switching Vector Autoregressive Model, Bootstrap

JEL Classification: G10, G12