N°26-17: From Memes to Markets: How WallStreetBets Attention Drives Robinhood Trading—A Causal Machine Learning Approach

We investigate the causal effect of social media attention from the WallStreetBets subreddit on retail trading behavior on the Robinhood platform. Using the potential outcomes framework and a double machine learning approach, we isolate the causal impact of attention while flexibly controlling for high-dimensional confounders and remaining robust to model misspecification. Our results show that social media attention significantly increases the absolute number of retail investors holding an asset, with effects rising monotonically in the intensity of attention. Moreover, the effects are heterogeneous, with attention-driven trading being strongest for niche small-cap and highly visible large-cap assets. The impact of attention-driven trading on an asset is further amplified when that asset is accompanied by highly relevant news, included in the Robinhood Top Movers list, or approaching an earnings announcement. Finally, comparisons with OLS and IPW estimators highlight the risk of model misspecification and underscore the advantages of double machine learning.