We study the effect of a firm’s new information disclosure on the information asymmetry between its informed and uninformed investors and its liquidity. To do this, we employ advanced natural language processing (NLP) methods and introduce a novel measure of the surprising language in a firms’ 10-K filing. Our findings show that more surprising language is associated with higher bid-ask spreads and lower trading volumes. An event study analysis further reveals that more surprising information triggers increased trading activity and abnormal returns.

We develop a rational expectations model to study the optimal precision of firm’s public disclosure when traders have heterogeneous preferences and the firm learns from market prices to guide its ESG investment decisions. We show that when the manager’s private information is imprecise, the investment opportunity is large, and prices are more informative about the value of the ESG component that the manager wants to learn about, greater precision of public disclosure may impair firm’s ability to learn from prices and may reduce firm value. Thus, the firms’ optimal disclosure policy balances the benefits of price efficiency against the costs of reduced market-based learning, generating a non-monotonic relationship between disclosure and firm value.