Price Dispersion and Informational Frictions: Evidence from Supermarket Purchases

Working Paper: CEPR ID: DP10906

Authors: Pierre Dubois; Helena Perrone

Abstract: Traditional demand models assume that consumers are perfectly informed about product characteristics, including price. However, this assumption may be too strong. Unannounced sales are a common supermarket practice. As we show, retailers frequently change position in the price rankings, thus making it unlikely that consumers are aware of all deals offered in each period. Further empirical evidence on consumer behavior is also consistent with a model with price information frictions. We develop such a model for horizontally differentiated products and structurally estimate the search cost distribution. The results show that in equilibrium, consumers observe a very limited number of prices before making a purchase decision, which implies that imperfect information is indeed important and that local market power is potentially high. We also show that a full information demand model yields severely biased price elasticities.

Keywords: consumer behavior; demand estimation; imperfect information; price dispersion; price elasticities; product differentiation; sales; search costs

JEL Codes: D4; D83; L11; L66


Causal Claims Network Graph

Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.


Causal Claims

CauseEffect
Consumers are not perfectly informed about prices (D89)Significant price dispersion in the French food market (D49)
Consumers incur costs to determine prices (D40)Biased demand parameter estimates when these costs are ignored (C51)
Higher opportunity cost of time (J29)Search less and pay higher prices (D49)
Limited number of prices observed (E30)Limited search behavior implies that imperfect information is crucial in understanding consumer choices (D83)
Demand model with imperfect information (D83)Biased estimates of price elasticities compared to a model assuming perfect information (D11)

Back to index