Rigidity, Dispersion and Discreteness in Chain Prices

Abstract

This paper studies price setting within a chain of grocery stores, using a scanner database that contains observations of retail prices for 435 products within 75 stores over 121 weeks. We find price dispersion within the chain. Although price dispersion is pervasive 75% of the prices are equal to the modal price. The mode changes frequently: 35% of the modes change in an average week. This suggests that the distribution of prices may react relatively fast to aggregate shocks. Stores differentiate themselves by the prices of relatively few items. Typically most prices in the store are at the mode of the cross sectional price distribution, some are above the mode and some are below the mode. The probability of a price change is 3.6% when the price is at the mode and 76.2% when the price is not at the mode. We explain the apparent attraction to the mode in terms of a model in which price discreteness plays an important role but there is no inertia. We also find that the probability of a price change is higher when the deviation from the mean of the cross sectional price distribution is large. But unlike conventional wisdom, the probability of a price change is higher for young prices.