Efficiency and Ranking Measurement of Vendors by Data Envelopment Analysis

Abstract

The research area of this thesis is efficiency measurement in the context of multiple input and multiple output production. Its major objective is to contribute to the development of non-parametric techniques for the measurement of technical efficiency, an important component of the global efficiency of an organization. More particularly, it is concerned with extensions of Data Envelopment Analysis (DEA).1. The first contribution of this thesis is toward a unified theory of efficiency. Based on a unified axiomatic framework, the thesis suggests a way of synthesizing production possibility sets and non-parametric efficiency measures in a generic form. A generic formulation of DEA for the convex cases is then proposed.2. The second contribution consists of a Common Set of Weights (CSW) formulation which provides a CSW for a selected subset of Decision Making Units (DMUs) and also indicates whether a priori formed group is homogeneous and distinctive enough from the rest of the observation set. Computational experiments indicate that the solution algorithm is efficient. The suggested homogeneity index opens new avenues for classifying and analyzing the performances of DMUs.3. The third contribution concerns two new DEA formulations in order to deal with shortcomings of the envelopment procedure of conventional DEA. One formulation defines an Inefficiency Frontier (IF) for the purpose of identifying those DMUs who are considered as "best practices" and but are also on "inefficiency frontier". The other formulation, called Maximal Envelopment Analysis (MEA), is a two-stage optimization procedure. MEA yields lower bound on efficiency via face extension with a controlled degree of envelopment. The analytical superiority of MEA in terms of envelopment degree and of greatest lower bound for a given degree of envelopment over Constrained Facet Analysis (CFA) is demonstrated.4. Finally, conventional formulations as well as the new models proposed in this dissertation are applied in a real-life setting. The empirical experience gained during the evaluation of commercial bank branches suggests that such models are not only complementary to traditionally used financial ratios but also a useful bank management tool in reallocating resources between bank branches in order to achieve higher service efficiency and profitability levels.