The non-linear and linear impact of investor sentiment on stock returns: An empirical analysis of the US market

Abstract

This study analyses the existence and both the linear and non-linear causal relationships between investor sentiment and stock returns on the US market. Our findings show that both linear and nonlinear links exist between sentiment and subsequent stock returns. Firstly, we find a significant positive causality from sentiment to future returns, especially from high sentiment, which is robust across quantiles of conditional distribution of returns. We also unveil a negative causal impact of positive lagged sentiment volatility to returns, albeit limited to a narrower range of quantiles and, hence, less robust. In addition, return volatility is found to be driven by the volatility of positive lagged sentiment, but again, this relationship is not robust as it only exists in a narrow range of quantiles. In light of the DSSW (1990) model, these findings suggest that those effects which help irrational investors to obtain higher expected returns and survive (the hold more and the create space effects) seem to dominate those effects hypothesised to drive them out of the market by inflicting systematic losses and driving them into bankruptcy (the price pressure and the Friedman effects).