The book comprises 18 chapters written by 29 contributors. In chapter 1, authored by the editors, the background and history of DEA are revisited and the various models and methods for treating allocative and overall efficiency are covered. The CCR model is presented in detail as well as extensions to deal with non-discretionary and categorical inputs and outputs, incorporation of judgments and a priori knowledge, and window analysis. A new additive model is also presented aimed at dealing with allocative and overall efficiency, which can be used whenever the usual ratio form of the efficiency measure gives unsatisfactory or misleading results. This model requires unit prices (associated with output slacks) and unit costs (associated with input slacks) to assess “profit efficiency”.

The topic addressed in chapter 2, by Banker, Cooper, Seiford and Zhu, is returns to scale (RTS) in DEA models. The discussion is centred on the relationships between DEA models and methods and the qualitative RTS characterizations they produce (such as whether RTS is identified as increasing, decreasing or constant). The RTS approaches within BCC and CCR models are revisited, extending them to models other than radial measure models.

In chapter 3, by Cooper, Li, Seiford and Zhu, methods for studying the sensitivity of DEA results to changes in the data are presented. The focus is the sensitivity of DEA efficiency evaluation, namely regarding the stability of the classification of the DMU status into efficient and inefficient. Global data changes are considered, in the sense that the stability of results is assessed when inputs and outputs change simultaneously for all DMUs.

The incorporation of value judgments in DEA models is addressed in chapter 4, by Thanassoulis, Portela and Allen. The important issue of the role and meaning of weights is revisited, and the main reasons for including value judgments in DEA models are described (generally motivated by real-world applications). Methods for incorporating value judgments and reducing the flexibility of DMUs in choosing their “value system” are presented, which are categorized in two broad classes: weight restrictions (absolute restrictions, assurance regions, and restrictions on virtual inputs and outputs) and changing the data set (transforming the data and adding new DMUs). The authors also discuss the changes that the incorporation of value judgments may introduce on the efficient frontier and its RTS characteristics.

Chapter 5, by Fare, Grosskopf and Whittaker, deals with distance functions and their duality relations. DEA estimations of various distance functions are presented and it is shown that their support functions (profit, revenue and cost) also may be estimated via DEA.

In chapter 6, Cook discusses the treatment of qualitative data in DEA models. The radial projection DEA model in the presence of ordinal data is examined and then applied to the efficiency evaluation of R&D projects and a telephone office. It is shown that by introducing the concept of rank position data within the DEA structure, the resulting model can be transformed into a VRS type model. The linking of the ordinal DEA model to multiple criteria decision making with ordinal data and criteria importance is also mentioned.

Cooper, Deng, Seiford and Zhu, in chapter 7, present various approaches to identify and manage congestion (understood as a form of technical inefficiency) with DEA models. It is illustrated how the DEA models can be used to determine the effects of congestion, namely the amount of congesting inputs, the output reduction due to congestion and the points where technical inefficiency gives way to congestion.

Chapter 8, by Tone, is a comprehensive study of the Malmquist productivity index to evaluate the productivity change of a DMU between time periods. This index includes a catch-up (recovery) term and a frontier-shift (innovation) term to capture the effects of both efficiency and technology changes. Three different approaches for the measurement of the Malmquist index are presented: radial, non-radial and non-radial and non-oriented. It is shown that the oriented radial models suffer from the neglect of slacks and infeasibility.

Chapters 9, 10 and 11 are devoted to probabilistic and statistical characterizations of the main efficiency evaluation models. In chapter 9, Cooper, Huang and Li deal with chance constrained programming extensions of the deterministic DEA formulations, thus making it possible to use characterizations such as “probably efficient” and “probably not efficient”. Expected value formulations are used to discuss DEA efficiency and its relationship with sensitivity analysis in stochastic situations. Other types of chance constrained programming models incorporate Simon’s satisficing concepts to extend the potential uses of DEA to cases in which full efficiency can be replaced by the attainment of aspiration levels of performance.

Chapter 10, by Simar and Wilson, presents bootstrap methods for statistical inference within non-parametric efficiency estimation. It is shown, via Monte Carlo experiments, that the iterated bootstrap offers a convenient approach for evaluating the performance of the bootstrap and providing corrections in a given applied context.

Chapter 11, by Banker and Natarajan, is devoted to statistical tests based on DEA efficiency scores, showing that the DEA estimator of the production frontier has desirable statistical properties enabling to develop a wide range of formal statistical tests. These can be used to test hypotheses of interest and relevance in the application of DEA such as existence of scale economies, separability and substitutability of inputs in production system, comparison of efficiency of groups of DMUs, etc.

Six DEA application chapters follow. In chapter 12, Ruggiero deals with DEA applications in education, one of the public sector activities in which a large amount of funds is invested and has witnessed a growing trend towards the consideration of accountability and efficiency issues. A discussion of the treatment of non-discretionary variables is also provided.

Chapter 13, by Paradi, Vela and Yang, discusses the DEA application to banking and provides a comprehensive review of the literature on bank branch performance DEA models (which describes variables, sample, type of RTS and model-orientation).

Chapter 14, by Triantis, presents the issues that researchers face when applying DEA to engineering problems, and proposes an approach for the design of an integrated DEA based performance measurement system. Moreover, it summarizes studies that have focused on engineering applications of DEA, and suggests systems thinking concepts that are appropriate for future DEA research in engineering. A bibliography of DEA applications in engineering is also provided.

Chapter 15, by Anderson, shows how DEA can be used to assess the player who had the most dominant baseball batting season, using the concept of super-efficiency. The adjusting capability of DEA to the changing circumstances of the game was recognized as one of its main strengths.

Chapter 16, by Athanassopoulos, provides a discussion on the performance of for-profit retail service industries. The study focus on the development of a unified methodological framework for assessing the operating efficiency of real networks (512 retail outlets in banking, sales forces, restaurants and betting shops). The monitoring of marketing and cost efficiency of service chains contributes to enhance the accountability of the marketing function and the decision making ability to improve the performance of individual branches.

Chapter 17, by Chilingerian and Sherman, focuses on health care applications of DEA. It offers a brief history of case studies in health sector (from hospitals to physicians) and discusses some of the models and motivations behind the applications. An eight-step procedure for DEA health application is provided with emphasis on the need for including quality measures of the services delivered.

Chapter 18, by Barr, presents a critical survey of DEA software packages, both commercial and non-commercial. Besides descriptions of eight individual packages, comparisons of their features and capabilities are provided, as well as links to further information on each of them.

The handbook is intended for researchers, students and practitioners. It aims at reflecting the state-of-the-art as well as representing a milestone in DEA advancing. I found this handbook a valuable reference for researchers, graduate students, and consultant analysts. However, it requires a relatively important degree of familiarity with the main DEA models and extensions to be used as an introductory door to this field. For this purpose (for instance, for classroom use in undergraduate classes) other references by the same authors are more appropriate (Charnes et al., 1994 and Cooper et al., 2000), mainly because the topics unfolding in more comprehensive, self-contained and written in a didactic way.

In this scope, I found all the chapters on the first part of the book, covering methodological issues, quite interesting and useful, in particular those devoted to the incorporation of value judgments and sensitivity analysis in DEA models. The chapters exploring the links with statistics, devoted to the performance of bootstrap techniques and statistical tests based on efficiency scores, also unveil important research directions. However, these chapters require from the reader a level of expertise on DEA models (as well as other topics), which cannot be acquired in the handbook itself. The chapter dealing with the consideration of qualitative data is the only one where the links between DEA and multiple criteria decision making are briefly explored. This is a relevant research and application topic and it would have been useful to have a whole chapter devoted to it.

The second part of this handbook, regarding application studies, left me with a sense of “incompleteness”. Of course, it will be impossible to include chapters, or even mention, all the areas in which DEA applications have been reported in the literature. Therefore, I believe a more judicious selection of material to be included should have been done to reflect state-of-the-art and relevance in DEA applications. The chapters included in the handbook are interesting and indeed lessons can be learned therein that can be replicated in studies in other areas. However, the handbook would have benefited from the inclusion of chapters describing studies in other (perhaps more relevant) areas, such as, for instance, energy, agriculture, environment, or telecommunications. Also, there is some imbalance in the treatment of applications in the chapters in the second part of the handbook. Some chapters enter into the details of model description (input and output factors, type of RTS, etc.) whereas other chapters merely do a survey of the literature. It goes without saying that both types can be useful for researchers and practitioners, but a higher consistency on this specific issue could have been pursued.

Nevertheless, this handbook provides an important value-added regarding DEA monographs, even though I think the editors could have organized it more under the perspective of a valuable complement to the other two books already mentioned above. This handbook constitutes, namely regarding some methodological chapters in the first part, an encouraging research agenda for further developments and uses of DEA. In this scope it is a valuable tool for researchers, graduate students and experienced practitioners. Moreover, up-to-date references are provided in most chapters that enable the reader to develop further his/her own specific interests in this continuously advancing area.