A Structural Model of Assortment Choice: Theory and Empirical Analysis

Many markets are characterized by the purchase of assortments with multiple units/varieties. There are two underlying mechanisms in assortment shopping, namely state dependence (i.e., utility reinforcement or satiation) and consumption flexibility. Previous studies have not sufficiently investigated how these two factors may interactively influence optimal assortment shopping and subsequent consumptions. We plan to meet this challenge by examining three related issues in the proposed project. First, we will fully solve the general dynamic programing problem for any assortment. We will consider a random-utility framework in which a forward-looking shopper decides on the optimal assortment and subsequent consumptions. We will derive closedform solution for optimal consumption policy and maximum expected value of selecting any assortment. We will also investigate how the optimal assortment may vary with the decision parameters and evolve dynamically over time. Second, we will develop a structure model of grocery shopping for assortments and estimate it on panel data. This will allow us to demonstrate how our model can be embedded into a larger framework to construct structural models of assortment choice, which may yield distinctive advantages over extant reduced-form models. Third, we will develop algorithms to identify the optimal assortment by examining and exploiting its structural properties. This will allow us to efficiently simulate optimal assortment choices and conduct policy analysis to evaluate potential impacts of marketing decisions. The algorithms can also be used to recommend assortments to individual consumers.