Estimation of a dynamic structural model of irreversible investment

In this paper we propose and estimate a dynamic structural model of fixed capital investment using firm-level Spanish data. We first characterize the optimal behavior of a firm facing uncertainty in demand and input prices, as well as kinked and lump-sum adjustment costs. This nonconvex adjustment cost structure is suggested by the observed evidence of inaction and lumpiness in microeconomic investment data. The optimal decision rule we derive is then used to specify a microeconometric discrete choice model. Our estimation method consists on a nested algorithm. The outer algorithm maximizes the likelihood function in each iteration in the search of the structural parameters. The inner algorithm computes the difference in conditional value functions from the nonparametric estimation of conditional choice probabilities. Thus, our method fits into the literature on structural estimation that avoids the solution of the dynamic programming problem. This estimation exercise is implemented on a dataset consisting on an unbalanced panel of 3060 Spanish manufacturing firms for the period 1990-1997.