Discrete Adjustment Costs, Investment Dynamics, and Productivity Growth: Evidence from Chilean Manufacturing Plants

Trustees Of Boston University, Boston

Investigators

Abstract

ABSTRACT PROPOSAL NO: 0351454 INSTITUTION: Boston University NSF PROGARM: ECONOMICS PI: Gilchrist, Simon TITLE: Discrete adjustment costs, investment dynamics, and productivity growth: Evidence from Chilean manufacturing plants. Standard models of investment predict gradual responses to policy intervention. However, in the face of large non-convexities of adjustment costs, the response of investment to policy intervention may be "lumpy", either because firms immediately alter capital outlays or because they wait to take any action. Non-convexities are particularly important for emerging markets, making emerging market economies respond differently to policy reforms than developed economies. Understanding these dissimilarities within the context of a structural econometric framework will provide useful information for reform-oriented policy makers in a variety of economic settings. For example, whether reforms are enacted quickly or gradually will depend on the nature of adjustment costs that firms face. This research uses Chilean data to contribute to recent research focusing on the non-convex adjustment costs in the plant-level investment process. It makes important methodological contribution by estimating a fully specified structural model of plant-level investment using maximum like-likelihood techniques. It differs from standard models of adjustment costs by allowing for fixed adjustment costs. This research characterizes both the decisions of when and how much to invest as functions of underlying observable state variables such as the current capital stock and the current productivity parameter that governs plant-level profitability. A contribution of this research project is to examine the joint productivity / investment dynamics for Chilean manufacturing plants over the 1979-2000 period. Existing studies of Chilean plant-level productivity dynamics use data from 1979-1986 but this research project will extend the existing data set on Chilean plants to cover the entire 1979-2000. This long data series will for better evaluation of Chilean trade reforms, reforms that were enacted over a relatively long period of time and the impact can take many years to express itself. Because the model is structural, it can be used to conduct policy experiments (such as changes in tariff or subsidy policy) or may be used as an input into calibrated general equilibrium models. In addition to policy relevance, the project will contribute to the infrastructure of academic research by extending the widely used data set on Chilean manufacturing plants to the year 2000. The research also makes methodological advances in estimating dynamic structural models with both continuous and discrete choice variables.