Alternative Approaches to Technical Efficiency Estimation: A Comparative Study

Abstract

<p>Pursuing efficiency is a fundamental characteristic of economic activity. correspondingly, efficiency measurement seems an eternal interest of production economists. The present dissertation is a comparative study of alternative technical efficiency estimation methods. Two recently developed methods based on different methodologies, namely, data envelopment analysis (DEA) and the stochastic frontier approach (SF) are studied. In this dissertation we review the production and efficiency structure defined by modern production theory. Based on earlier works of Afriat, we discuss a set of propositions underpinning the non-parametric programming approach (or DEA). Further, we demonstrate the relationship between non-parametric and parametric production frontiers as references for technical efficiency measurement. We also explore the corresponding relationships between various versions of the DEA model and their implications regarding returns to scale properties. On the side of the SF approach, we work out a conditional estimation model to extract technical efficiency from a composite error structure. The main empirical contribution is a simulation study that is carried out to examine the capabilities of both approaches under various circumstances. In the first set of experiments we examine the performances of the two methods under assorted efficiency profiles, by which we describe the industry's efficiency distribution. Then, in a second set of experiments we investigate the performance of the two methods when the experimental data has different returns to scale properties. Finally, we test the robustness of the two models in regards to varied magnitudes of random noise . Our results indicate that though the SF model often leads the competit ion by a small margin i n our experimental environment, both methods have reasonable performances.</p>