Методы оценки показателя эффективности в моделях стохастической производственной границы.
This paper discusses the problems of modeling efficiency of firms. There are two the most popular methods to estimate efficiency of firms: DEA (data envelopment analysis) and SFA (stochastic frontier analysis), and popularity of the last one is fast growing. There are a lot of different SFA models, so most researches often choose in advance one or two models, which they are going to estimate. So survey of different SFA models is one of goals of this paper.
We discuss 15 popular SFA models. Also we discuss problems of SFA models and their prospects. In our paper we compare models, estimated by classical method of moments (MoM), and models, estimated by maximum likelihood approach (MML). Today there are no such papers, so we try to discuss pros and cons of using method of moments approach in SFA models. Interesting, that this method is very unpopular today, but its’ estimates are asymptotical normal and consistent.
Because there are no formal criteria to compare different SFA models, we investigate the estimation results from 9 SFA models on the concrete industry data. We use correlation analysis of estimates of efficiency ranks and also we try to find out the causes of the most serious differences between models.