Estimation of a Dynamic Discrete Choice Model of Irreversible Investment

In this paper we propose and estimate a dynamic structural model of fixed capital investment at the firm level. Our dataset consists of an unbalanced panel of Spanish manufacturing firms. Two important features are present in this dataset. There are periods in which firms decide not to invest and periods of large investment episodes. These empirical evidence of infrequent and lumpy investment provides evidence in favour of irreversibilities and nonconvex capital adjustment costs. We consider a dynamic discrete choice model of irreversible investment with a general specification of adjustment costs including convex and nonconvex components. We use a two stage estimation procedure. In a first stage, we obtain GMM estimates of technological parameters. In the second stage, we obtain partial maximum likelihood estimates for the adjustment cost parameters. The estimation strategy builds on the representation of conditional value functions as a computable function of conditional choice probabilities. It is in the line of structural estimation techniques which avoid the solution of the dynamic programming problem.