Smart money often trades actively during times of large corporate events. We document in the context of mergers and acquisitions (M&A) that, during the public bid negotiation period, institutional investors increase their holdings of acquirers in deals that generate positive value and decrease their holdings in those that generate negative value. The resulting trading profits create a significant gap between the return to the acquiring firm and the return to these investors, and this gap renders firm return a misleading measure of investors’ incentives in pursuing mergers. On average, institutional investors of acquiring firms earn 2.4% from M&A while the return to passive acquirer shareholders is only -0.9%. In deals that deliver volatile returns to acquiring firms, the gap widens to 6.3%. We further show how the trading motive impacts the ex ante holdings of institutional investors and how the trading decision and the resulting gap are impacted by deal characteristics such as merger size and stock liquidity as well as institutions’ characteristics such as initial holdings, portfolio weight, and trading skills. Institutions that earn a high return gap are associated with weak governance in preempting and correcting value-destroying mergers. Our study highlights that the group of investors who have influence over corporate actions do not necessarily bear the full consequences of such events, and therefore accounting for the dynamics of shareholder composition is critical in measuring investors’ incentives correctly.

We conduct the first analysis on the impact of democratized AI (ChatGPT) on the trading activities of investors by leveraging a dataset of long textual information spanning 19 years of earnings calls. We have three key findings. First, AI-sentiment generated by ChatGPT strongly predicts returns for up to 12 months, while traditional human-dictionary-based sentiment yields little predictability. Second, before the arrival of ChatGPT, short sellers traded in alignment with AI-sentiment within two weeks following earnings calls, while retail traders did not. Following the widespread deployment of ChatGPT, there was a significant 65-fold increase in retail-trader alignment with AI-sentiment, whereas the alignment of short-sellers with AI-sentiment may have weakened. Third, stock experiencing increased retail traders’ alignment with AI-sentiment also witnessed a significant decrease in bid-ask spreads. An exogenous variation in AI availability due to ChatGPT outages led to notable reduction in retail-AI alignment and reversal of narrowed bid-ask spreads, further supporting the causal role of AI. Overall, the study suggests that democratizing AI has the potential to level the playing field and bridge the information gap between privileged and ordinary investors.