Development Of Data Envelopment Analysis Research Articles

FairDEA—Removing disparate impact from efficiency scores

Achieving fairness in algorithmic decision-making tools is an issue constantly gaining in need and popularity. Today, unfair decisions made by such tools can even be subject to legal consequences. We propose a new constraint that integrates fairness into data envelopment analysis (DEA). This allows the calculation of relative efficiency scores of decision-making units (DMUs) with fairness included. The proposed fairness constraint restricts disparate impact to occur in efficiency scores, and enables the creation of a single data envelopment analysis for both privileged and unprivileged groups of DMUs simultaneously. We show that the proposed method - FairDEA - produces an interpretable model that was tested on a synthetic dataset and two real-world examples, namely the ranking between hybrid and conventional car designs, and the Latin American and Caribbean economies. We provide the interpretation of the FairDEA method by comparing it to the basic DEA and the balanced fairness and efficiency method (BFE DEA). Along with calculating the disparate impact of the model, we performed a Wilcoxon rank-sum test to inspect for fairness in rankings. The results show that the FairDEA method achieves similar efficiency scores as other methods, but without disparate impact. Statistical analysis indicates that the differences in ranking between the groups are not statistically different, which means that the ranking is fair. This method contributes both to the development of data envelopment analysis, and the inclusion of fairness in efficiency analysis.

Benefit-of-the-doubt approach to workload indicators: Simplifying the use of case weights in court evaluations

In practice, the calculation of workloads in different entities and the allocation of resources between such entities is often done using simple task weighting models. Different tasks are assigned weights and the total workload is the weighted sum of the tasks performed. Traditional approaches for establishing the weights, including the using of Delphi methods or detailed time-studies, are however expensive, and the resulting weights are often challenged by the evaluated entities. In this paper, we discuss how to mitigate these problems.We study the problem within the context of the judiciary system. To ensure an efficient judiciary, it is usually considered necessary to have a reliable case weighting system (CWS). Different types of court cases, e.g. criminal versus civil cases, have different resource needs, and to get a relevant aggregate measure of the tasks at hand, it is therefore necessary to weight the different court cases, i.e. to construct a case mix corrected workload measure.We suggest a “benefit-of-the-doubt” (BoD) approach inspired by recent developments in Data Envelopment Analysis (DEA). We allow for uncertain weights based on only partial information about the “true” weights, and we evaluate individual courts with the weights that put them in their most favorable light. In addition to making the weight setting easier and less disputable in applications, our approach dispenses with several limitations of a traditional weighted caseload approach, including the implicit assumptions of constant returns to scale and a constant rate of substitution between caseloads. Moreover, most of the applications of detailed case weights are still available using our BoD approach. We can continue to evaluate the efficiency of individual courts and device sound and fair resource allocation procedures that are robust to the remaining uncertainty about weights.We illustrate our approach by several real-world applications and discuss relevant extensions and applications in other areas, including regulation.

A two stage data envelopment analysis model with undesirable output

The dependent relationship among the decision making units (DMU) is usually assumed to be non-existent in the development of Data Envelopment Analysis (DEA) model. The dependency can be represented by the multi-stage DEA model, where the outputs from the precedent stage will be the inputs for the latter stage. The multi-stage DEA model evaluate both the efficiency score for each stages and the overall efficiency of the whole process. The existing multi stage DEA models do not focus on the integration with the undesirable output, in which the higher input will generate lower output unlike the normal desirable output. This research attempts to address the inclusion of such undesirable output and investigate the theoretical implication and potential application towards the development of multi-stage DEA model.

Novel theory and methodology developments in data envelopment analysis

This is a post-peer-review, pre-copyedit version of an article published in Annals of Operations Research. The final authenticated version is available online at: http://dx.doi.org/10.1007/s10479-017-2413-7

An approach to evaluating relative effectiveness in non-profit institutions

We present, in this study, a method for comparing the relative effectiveness of different non-profit institutions with similar objectives. In addition, we show how this measure of relative effectiveness is related theoretically to their relative efficiency. Relative effectiveness is shown to be a product of the efficacy with which potentially utilizable resources can be converted into usable inputs, and the efficiency with which the inputs are converted to outputs or outcomes. Finally, drawing on developments in data envelopment analysis, we illustrate the new methodology using data from 109 institutions of higher education.

An Approach to Evaluating Relative Effectiveness in Non-Profit Institutions

We present in this study a method for comparing the relative effectiveness of different non-profit institutions with similar objectives. In addition, we show how this measure of relative effectiveness is related theoretically to their relative efficiency. Finally, drawing on developments in data envelopment analysis, we illustrate the new methodology using data from 107 institutions of higher education.

Profitability and Marketability of the Top 55 U.S. Commercial Banks

Utilizing recent developments in data envelopment analysis (DEA), this paper examines the performance of the top 55 U.S. commercial banks via a two-stage production process that separates profitability and marketability. Substantial performance inefficiency is uncovered in both dimensions. Relatively large banks exhibit better performance on profitability, whereas smaller banks tend to perform better with respect to marketability. New context-dependent performance measures are defined for profitability and marketability which employ a DEA stratification model and a DEA attractiveness measure. When combined with the original DEA measure, the context-dependent performance measure better characterizes the profitability and marketability of 55 U.S. commercial banks. The new approach identifies areas for improved bank performance over the two-stage production process. The effect of acquisition on efficiency and attractiveness is also examined.

Evaluating the information content of a measure of plant output: An application to high-technology manufacturing

Commonly used measures of plant output have been criticized for their inability to provide information required to manage the dynamic operations of high-technology manufacturing plants. In this paper, we propose tests to evaluate the information content of a measure of plant output that is specifically directed at these issues. These tests are based on recent developments in DATA Envelopment Analysis (DEA), namely the Cone Ratio Envelopments. In this new application of DEA models, we shift the focus from Decision Making Units (DMUs) being evaluated to the DMUs that are being used to effect evaluations. The proposed tests are applied to evaluate the information contnet of a complexity adjusted measure of plant output, which we refer to as Total Net Die Equivalent (TNDE). Developed recently in the context of a high-technology manufacturing plant—a wafer fabrication plant of a merchant semiconductor manufacturing company-TNDE reflects the ongoing changes in product and process technologies, process flow characteristics, and volume of production. Evaluating the information content on joint criteria of “recency” and “efficiency”, the results of our tests, conducted over a 28-month period in the wafer fabrication plant,show that TNDE as a single aggregate (scalar) measure of plant output outperforms the two outputs from which it is synthesized. Thus, TNDE as a single measure of output can be used to provide an improved basis for planning a plant's future operations. En route to the development and application of the proposed tests, we illustrate how DEA concepts and models provide a rigorous and systematic basis for conducting ex post technology evaluation to guide continuous improvements in a plant's current operations.

An Application of Frontier Analysis: Handicapping Running Races

Economists have used frontier analysis for some time now to measure the efficiency of the firm. Recently, particularly since the development of data envelopment analysis, management scientists have applied efficient frontier analysis to a variety of problems. We used linear programming to estimate a frontier function that describes where the age records for a particular running race should be, given the current records. The frontier function is used to handicap running races based on age and sex. The frontier method is an improvement over current methods in that it avoids the problem of soft records for some ages.