Firm exit and technical inefficiency

The purpose of the paper is to examine formally the fundamental implication that technical inefficiency (TI) is related to firm exit. Traditional stochastic frontier models allow for the measurement of TI but do not allow for a direct effect of TI on exit. We propose a model which allows for such effects and consists of a stochastic frontier model, and an additional equation that describes the probability of exit as a function of covariates and TI. Since TI is unobserved, econometric complications arise, and obtaining consistent estimates is non-trivial due to the presence of integrals in the likelihood function. We propose and implement maximum likelihood estimation one step, employing data for 3,404 manufacturing firms in Greece. We find significant positive effects from TI on the probability of exit. We also propose and provide measures of TI that respect the fact that unobserved TI affects the probability of exit and compare them to TI measures from the traditional stochastic frontier model.