Overvaluing Algorithmic Advice: Evidence from a Stock Price Forecasting Experiment

This study investigates willingness to pay (WTP) for stock forecasting advice from algorithms, financial experts, and peers. In two incentivized forecasting experiments, participants purchased advice using an incentive-compatible mechanism and then decided how much to incorporate it into their forecasts. Participants assigned the highest WTP to algorithmic advice and relied on it as heavily as expert advice, despite its forecasting performance being no better than alternative sources. Consequently, participants overpaid for advice, especially algorithmic advice, whose realized benefits were insufficient to offset its cost. A second experiment shows that overpayment persists even after repeated opportunities to revise WTP with detailed feedback on advice quality and realized net benefits. The results suggest that individuals place excessive value on algorithmic advice perceived as sophisticated or credible, even when its realized economic value is limited. These findings highlight the importance of tools and disclosure policies that help individuals better assess the economic value of algorithmic advice.