Working Papers

with Diego García, Alexandre Jeanneret, and Benjamin Loos

(Draft and slides by request)

Abstract: Aggregate measures of investor attention cannot distinguish a stable, returning audience from a constantly changing one. Using individual browsing records from China's largest investment platform, we measure attention stickiness, the fraction of a stock's viewers who return to it the next day, a dimension no existing proxy captures. When viewers are returning investors, an attention shock is associated with a same-day return above 110 basis points that fully reverses within forty trading days; when they are new, attention barely moves prices. The level of stickiness predicts lower future returns, cumulating to 160 basis points over forty days. In the cross-section, stocks both rarely viewed and rarely revisited earn 2.2% per month more than heavily monitored, high-stickiness stocks. The composition of retail attention, not merely its level, is a first-order predictor of short-horizon returns; the evidence is more consistent with behavioral reinforcement than with rational learning.

Presentations: CityUHK International Finance Conference (Hong Kong, 2026); Penn State University (2026); HEC Lausanne (2026); University of Colorado (2026); SBFC (Sydney, 2025); FIRN UQ Asset Management Meeting (Brisbane, 2025)