Learning to Set Prices
75 Pages Posted: 9 Nov 2018 Last revised: 7 Oct 2019
Date Written: October 8, 2018
This paper empirically studies whether, and how quickly, entrants to new markets learn about demand and adapt towards fully-informed pricing. Focusing on the privatized Washington State liquor market, we document large price changes across a broad range of products and present novel evidence showing that these changes result from retailer learning about demand: prices change in response to realized demand shocks and adjust to better reflect demand primitives. To investigate the nature of initial pricing mistakes and the process by which firms correct them, we structurally estimate demand and costs primitives using a combination of aggregate-level data on quantity and price, micro-data on customer compositions, and direct cost data from liquor-control states. We find that retailers do learn and converge to fully-informed pricing by the end of the sample, but initial pricing mistakes are consequential and account for an average 9% loss in variable profits. In addition, prior experience and sales information affect the starting position and adjustment processes of retail prices. Finally, initial pricing mistakes are consistent with misinformation about Washington's unique customer composition by product category. Our results broadly support the view that firms are optimizing agents but face significant information constraints.
Suggested Citation: Suggested Citation