Structural Demand Models for Retailer Category Pricing

One critical aspect of retailer category management is pricing the brands within the category in order to maximize total category profits. Unlike the manufacturer, a retailer must carefully assess how changes in brand prices affect substitution between brands and overall category profits. In this paper, we demonstrate that structural demand models have inherent advantages over commonly used non-structural approaches, such as the linear and log-linear demand models. While numerous papers have utilized these non-structural demand models, we show that these models have unappealing theoretical properties for category pricing. We then show that a variant of the AIDS model provides for very flexible substitution effects and category expansion effects. In addition, the model specification allows for prices and merchandising (display/feature) to separately affect brand substitution and category expenditure. We compare this model to the logit model with an outside good, which can also been used for category pricing. We describe how the underlying assumptions and structure of each model yield different implications. Finally, we apply each model to a retail category pricing problem in three different categories. The results show how the AIDS model can be used to assist in retailer pricing decisions and illustrate its comparative advantages/limitations with respect to the logit model.