Research

We study how institutional ownership concentration affects price informativeness, measured by the ability of stock prices to predict future earnings. Using fund-level holdings, we measure stock-level concentration separately among active and passive funds. Active-fund concentration is associated with less informative prices, whereas passive-fund concentration has no stable relation with informativeness. Mergers between active investors provide causal evidence that increased concentration reduces price informativeness. The results hold in U.S. and international samples.

This paper quantifies agency conflicts at the firm-year level using an offline deep reinforcement learning (ODRL) framework. Whereas structural estimation offers transparent economic interpretation, the model-free ODRL framework accommodates high-dimensional inputs and their flexible interactions, better capturing the complexity of managerial decision making. The framework estimates optimal management actions based on the objectives of maximizing compensation versus market cap. Actual actions that deviate from market-cap-maximizing actions and align with compensation-maximizing actions indicate high agency conflicts. Applied to a data valuation setting, the measure reveals that investors place less value on management earnings guidance from firms with high agency conflicts due to its lower quality, shifting reliance toward analyst forecasts.