Primary Submission Category: Machine Learning and Causal Inference
Prediction Markets as Mechanisms for Price Discovery and Market Efficiency
Authors: Kiet Le, Khoa Le, Nghi Le,
Presenting Author: Kiet Le*
Prediction markets have recently experienced rapid growth, yet their role in financial information production and market efficiency remains underexplored in the finance literature. This paper studies whether prediction markets improve price discovery for publicly traded firms and how they interact with traditional information intermediaries. We develop a framework in which informed agents have incentives to trade on prediction markets to monetize private information, potentially reallocating informed trading away from equity markets. Exploiting the quasi-exogenous listing of firm-specific corporate events on major prediction market platforms such as Polymarket and Kalshi, we implement a difference-in-differences, event-study design and AIPW method to identify causal effects. Using data on prediction market events linked to S&P 500 firms, combined with CRSP/Compustat stock data and IBES analyst forecasts, we examine changes in analyst coverage, forecast dispersion, abnormal returns around earnings announcements, and post-event price dynamics. Our results shed light on whether prediction markets crowd out or crowd in analyst activity, reduce insider trading incentives in equity markets, and enhance overall price efficiency. The findings contribute to the literature on information aggregation, market design, and the evolving role of alternative trading venues in modern financial markets.