Stochastic Production Frontier and Technical Inefficiency : Evidence from Tunisian Manufacturing Firms ∗

In this paper, a stochastic frontier production model is estimated on panel data, and technical inefficiency indices are computed for Tunisian manufacturing Þrms over the 1983-1993 time period. The most commonly used one-sided distributions of the inefficiency error term are speciÞed, namely the truncated normal, the half-normal and the exponential distributions. A generalized version of the halfnormal, which does not embody the zero-mean restriction, is also explored. For each distribution, the likelihood function and the counterpart of Jondrow et al. (1982) estimator of technical efficiency are explicitly stated. Based on our data set, formal tests lead to a strong rejection of the zero-mean restriction embodied in the half normal distribution. Our main conclusion is that the degree of measured inefficiency is very sensitive to the postulated assumptions about the distribution of the one-sided error term. The estimated inefficiency