Quantitative Marketing and Economics Маркетинг

The asymmetric value of algorithmic information in two-sided platforms: Evidence from Zillow

Troncoso, Isamar, Malik, Nikhil, Proserpio, Davide, Fu, Runshan
Springer Nature · 2026-10-26 · DOI 10.1007/s11129-026-09316-y

Аннотация

Platforms increasingly deploy algorithms that provide information to both sides of a market, yet the value of this information may be unevenly distributed. We study this asymmetry in the context of Zillow, a leading online real estate platform that publishes “Zestimates,” algorithmic estimates of home values, for over 110 million U.S. residential properties. We leverage an exogenous negative shock to trust in the Zestimate that reduced reliance on the algorithm among both buyers and sellers. We develop a simple framework that captures how algorithmic information shapes the beliefs and pricing strategies of buyers and sellers, and test its predictions using secondary data on 27,908 properties listed for sale. We find that following the shock, sellers move away from the Zestimate, setting list prices that deviate more from the algorithmic estimate. The effect is robust to controls for housing composition and observed local market conditions and is not driven by contemporaneous changes in Zestimate values. Although we cannot rule out some delayed adjustment of the Zestimate to local market conditions, our analyses suggest that it is unlikely to fully explain the results. Despite higher list prices, homes sell faster and at higher prices. The shorter time on market is consistent with the Zestimate being more informative to buyers than to sellers, such that the negative shock disproportionately increases buyer uncertainty. These findings highlight how asymmetries in the value of algorithmic information can shape market outcomes, with implications for predicting market responses and designing platform policies.

Читать у издателя →