- Time: 14:00-15:00
We propose an aggregate measure of employee sentiment based on millions of employee online reviews and we test whether big employee data embedded in expert financial models can improve stock return predictability. In line with behavioural finance theory, our results document that the collective employee sentiment is a strong predictor of stock market returns with lower future returns following high employee sentiment. This predictive power is more pronounced when the employee sentiment index is constructed using the expectations of employees about the near-term business outlook of their employer. Our market-wide sentiment measure has superior performance compared to existing proxies of investor sentiment and commonly-studied macroeconomic variables. The forward-looking property of this data is also evident in predicting industry returns or portfolio returns sorted on characteristics, such as size, age, risk, profitability, dividend payout, tangibility, financial constraints and growth opportunities. Importantly, market-wide employee sentiment has relative power in predicting future asset returns after controlling for firm-level employee sentiment. The predictive power of aggregate employee online data is explained by investors’ biased beliefs about expected cash flows and volatility. These results indicate that financial models can be enriched with sentiment factors derived from various big data sources and stakeholders, providing insights into mispriced assets and assisting investment decisions.
Panagiotis Stamolampros is a lecturer in Business Analytics at the Centre for Decision Research at Leeds University Business School. His work is at the intersection of the fields of Marketing, Operations, and Finance. His current research explores user-generated content as a source that reduces information asymmetries. He also examines the value relevance and predictive ability of that form of information for investors.