Estimation of Dynamic Decision Models with Corner Solutions: A Model of Price and Inventory Decisions

This paper proposes and applies a method of moments to estimate dynamic decision models with corner solutions. The method extends previous results by Hotz and Miller (1993) and Pakes (1994), and it allows for unobserved state variables affecting both the continuous choice (interior solution) and the discrete choice (interior solution / corner solution). The method is applied to estimate a model of price and inventory decisions using data of individual goods from a supermarket chain. The estimation shows that lump-sum ordering costs and lump-sum price adjustment costs are significant, but ordering costs are quantitatively more important. Numerical solutions of the model show that, in that context, the interaction between price and inventory decisions can explain the high frequency of nominal price reductions observed in the data.