Data Preferences in Firm Learning: Evidence from an Online Auction Platform
(A2026-134245)

AUTHORS

Xiaojie Li, Boston College

ABSTRACT

Firms rely on data to improve decisions, but when both their own and competitors’ data are available, do they use others’ data to accelerate learning? This paper studies used-car auction sellers on Ali Auction, who choose only the ending hour—an important decision because different hours attract different bidders and yield different payoffs. Although experienced offline, sellers face new demand patterns and competition online and can access both their own and others’ auction data. I develop a structural model that allows differential use of data sources. Results show that inexperienced sellers lean more on competitors’ data but increasingly prioritize their own as they gain experience. Counterfactuals indicate that full-information could raise profits by 6.4%, driven mainly by sellers’ preference for their own data. These findings suggest that data sharing alone is insufficient and that platforms can guide new sellers using others’ data to improve decisions.