Speaker: Giacomo Calzolari (European University Institute)

Location: University of Mannheim, Room S031

Time: 1:45 p.m.

Abstract: We study how algorithmic recommendations affect product-market competition. We embed recommender systems, trained on sparse and noisy feedback, in a model of consumer search and price competition. Recommendations create personalized prominence: consumers start searching from products predicted to match their tastes. This improves matching and reduces search costs, but also raises equilibrium prices. Prices increase through front-loading of high-value consumers, reliance on informative recommendations, and reduced price salience under directed search. Due to the higher prices, consumer surplus may fall despite better matches. We also show that manipulated recommendations can intensify competition and limit the profitability of self-preferencing.