Investor sentiments revisited: negligence of stock-level sentiments may be a mistake

Purpose The primary objective of this research is to provide evidence that there are two distinct layers of investor sentiments that can affect asset valuation models. The first is general market-wide sentiments, while the second is biased approaches toward specific assets. Design/methodology/approach To achieve the goal, the authors conducted a multi-step analysis of stock returns and constructed complex sentiment indices that reflect the optimism or pessimism of stock market participants. The authors used panel regression with fixed effects and a sample of the US stock market to improve the explanatory power of the three-factor models. Findings The analysis showed that both market-level and stock-level sentiments have significant contributions, although they are not equal. The impact of stock-level sentiments is more profound than market-level sentiments, suggesting that neglecting the stock-level sentiment proxies in asset valuation models may lead to severe deficiencies. Originality/value In contrast to previous studies, the authors propose that investor sentiments should be measured using a multi-level factor approach rather than a single-factor approach. The authors identified two distinct levels of investor sentiment: general market-wide sentiments and individual stock-specific sentiments.