Estimating stochastic frontier systems with unbalanced panel data: the case of floor tile manufactories in Egypt

A generalized stochastic frontier system is fitted with unbalanced panel data of Egyptian floor tileries and estimated using OLS, `within', feasible GLS, and maximum likelihood. Hausman tests indicate `within' is the correct estimator for the data. Estimates of firm-level technical and allocative inefficiencies (and their costs) are presented for each of the estimators. Finally, sources of technical inefficiency are investigated.