The influence of consumer, manager, and investor sentiment on US stock market returns

This study examines how consumer, investor, and manager sentiment explain US stock excess returns over 23 years. Its novelty resides in integrating the sentiments of three different types of economic and financial agents. It also performs a segmented temporal analysis using rolling window techniques, to assess sentiment’s impact across different time horizons. The empirical analysis utilizes the Paris-Winsten and Newey-West estimators, along with the ARMAX model to address autocorrelation and heteroscedasticity in linear regression, providing robust standard errors and reliable statistical inferences. The autoregressive moving average models estimate excess return based on the past values, shocks, and external variables. Combining the Fama-French five-factor model with the sentiment factor enriches the analysis. The study’s findings indicate that higher consumer optimism negatively impacts excess returns, as investors may anticipate a future decline in the stock market due to an existing overheated economy. Investor sentiment exhibits mixed behavior, where higher uncertainty may increase stock returns due to previous oversold markets creating opportunities for investors or due to the closing of short positions, which will also increase stock demand. It is also related to decreased stock returns depending on the proxy used. As for managers’ sentiment, this work did not demonstrate a relevant relationship between this sentiment and stock returns. The study also reveals that the importance of sentiment determinants of those three agents changes over time. The findings support behavioral models of asset pricing, which incorporate both market fundamentals and the psychological characteristics (sentiment) of different market participants. AcknowledgmentsThis work is funded by National Funds through the FCT – Foundation for Science and Technology, I.P., within the scope of the project Ref. UIDB/05583/2020. Furthermore, we would like to thank the Research Centre in Digital Services (CISeD) and the Instituto Politécnico de Viseu for their support.