Stepwise selection of variables in data envelopment analysis: Procedures and managerial perspectives

Insolvency of Brazilian Electricity Distributors: A Dea Bootstrap Approach

“…Ali and Seiford (1990) and Pastor (1997) showed that some models are not translation-invariant when using inputs and outputs with negative values -an aspect that was ignored in the Cielen et al (2004) analysis of insolvency. In addition, there is the classic difficulty of selecting input and outputs in DEA (Adler, Yazhemsky 2010;Jenkins, Anderson 2003;Kittelsen 1993;Titko et al 2014;Wagner, Shimshak 2007).…”

“…Section 2 is divided into three subsections, devoted respectively to presenting the methodological foundations for using DEA, the stepwise procedure of Wagner and Shimshak (2007) and the bootstrap truncated regression of Simar and Wilson (2007). Section 3 presents the results of the article and the respective robustness analysis in two subsections.…”

“…This is performed using DEA to estimate an inefficiency frontier (Premachandra et al 2011;Paradi et al 2004;Sueyoshi 2005;Cielen et al 2004) through the selection of indicators using the stepwise procedure of Wagner and Shimshak (2007). A second stage analysis is then undertaken, relating firms' in(efficiency) scores to different contextual variables.…”

Performance Evaluation of Property and Real Estate Companies Listed on Indonesia Stock Exchange Using Data Envelopment Analysis

“…The DEA model forms a relative efficiency score by converting the multiple-input or multiple-output variables to a single measure of performance for each DMU (Horta et al, 2010). This happens by establishing an empirically based "bestpractice" or efficient frontier as a result of classifying a set of efficient DMUs which lies on the frontier and inefficient DMUs which do not lie on the frontier (Wagner & Shimshak, 2007).…”

Evaluating efficiency in the Brazilian trucking industry

“…Techniques proposed include i) judgmental screening by experts, in order to indicate the most relevant variables for the DEA model (GOLANY; ROLL, 1989); ii) regression analysis, in order to indicate highly correlated variables as redundant (LEWIN;MOREY;COOK, 1982); iii) the application of DEA to reduced models, in order to rank the effect of variables on efficiency scores (WAGNER; SHIMSHAK, 2007); iv) the use of multi-criteria approaches for weighting variables (MIRANDA; ALMEIDA, 2004); and v) the addition of a virtual target into the sample to identify changes in the adherence of a given DMU to the frontier (BERECHMAN; ADLER, 1999).…”