Abstract
In some contexts data envelopment analysis (DEA) gives poor discrimination on the performance of units. While this may reflect genuine uniformity of performance between units, it may also reflect lack of sufficient observations or other factors limiting discrimination on performance between units. In this paper, we present an overview of the main approaches that can be used to improve the discrimination of DEA. This includes simple methods such as the aggregation of inputs or outputs, the use of longitudinal data, more advanced methods such as the use of weight restrictions, production trade-offs and unobserved units, and a relatively new method based on the use of selective proportionality between the inputs and outputs.
Similar content being viewed by others
Notes
For example, in the assessment of school performance, the overall number of students may be highly correlated with the number of students achieving good grades in exams. However, omitting one of these indicators from the model may change not only the efficiency measures but also the ‘flavour’ of the model used in the analysis.
The assumption that the trade-offs should be applicable globally in the entire technology means that they cannot equate directly to the marginal rates of substitution. Indeed, the latter are generally different over the efficient frontier and reflect different production patterns exhibited by efficient DMUs. (For example, the marginal rates of substitution between teaching staff and students, or students and publications will generally be different in a high-ranked research-driven and average-ranked university department.) In non-CRS technologies, the marginal rates of substitution also depend on the type of local returns to scale. These observations mean that the production trade-offs should be sufficiently relaxed and non-demanding (Podinovski 2004c, 2007b).
The HRS technology is a subset of the CRS technology if all inputs are included in the selective proportionality assumption (Podinovski 2004b).
We cannot specify by how much the publications will increase. Leaving this unchanged is a safe option.
References
Andersen P, Petersen NC (1993) A procedure for ranking efficient units in data envelopment analysis. Manage Sci 39:1261–1264
Allen R, Athanassopoulos A, Dyson RG, Thanassoulis E (1997) Weights restrictions and value judgements in data envelopment analysis: evolution, development and future directions. Annl Operat Res 73:13–34
Allen A, Thanassoulis E (2004) Improving envelopment in data envelopment analysis. Euro J Operat Res 154:363–379
Angulo-Meza L, Lins MPE (2002) Review of methods for increasing discrimination in data envelopment analysis. Annl Operat Res 116:225–242
Charnes A, Cooper WW, Rhodes E (1978) Measuring the efficiency of decision making units. Euro J Operat Res 2:429–444
Cooper WW, Seiford LM, Tone K (2000) Data envelopment analysis. Kluwer Academic Publishers, Boston
Cooper WW, Thompson RG, Thrall RM (1996) Extensions and new developments in DEA. Annl Operat Res 66:3–45
Despotis DK (2002) Improving the discriminating power of DEA: focus on globally efficient units. J Operat Res Soc 53:314–323
Dyson RG, Allen R, Camanho AS, Podinovski VV, Sarrico C, Shale EA (2001) Pitfalls and protocols in DEA. Euro J Operat Res 132:245–259
Dyson RG, Thanassoulis E (1988) Reducing weight flexibility in data envelopment analysis. J Operat Res Soc 39:563–576
Färe R, Lovell CAK (1978) Measuring the technical efficiency of production. J Econ Theory 19/1:150–162
Podinovski VV (1999) Side effects of absolute weight bounds in DEA models. Euro J Operat Res 115:583–595
Podinovski VV (2001) Validating absolute weight bounds in data envelopment analysis (DEA) models. J Operat Res Soc 52:221–225
Podinovski VV (2004a) Suitability and redundancy of non-homogeneous weight restrictions for measuring the relative efficiency in DEA. Euro J Operat Res 154:380–395
Podinovski VV (2004b) Bridging the gap between the constant and variable returns-to-scale models: selective proportionality in data envelopment analysis. J Operat Res Soc 55:265–276
Podinovski VV (2004c) Production trade-offs and weight restrictions in data envelopment analysis. J Operat Res Soc 55:1311–1322
Podinovski VV (2005) The explicit role of weight bounds in models of data envelopment analysis. J Operat Res Soc 56:1408–1418
Podinovski VV (2007a) Improving data envelopment analysis by the use of production trade-offs. J Operat Res Soc, Advance online publication, September 6, 2006; doi:10.1057/palgrave.jors.2602302
Podinovski VV (2007b) Computation of efficient targets in DEA models with production trade-offs and weight restrictions. Euro J Operat Res, doi:10.1016/j.ejor.2006.06.041
Sarrico CS, Dyson RG (2004) Restricting virtual weights in data envelopment analysis. Euro J Operat Res 159:17–34
Thanassoulis E (1995) Assessing police forces in England and Wales using data envelopment analysis. Euro J Operat Res 87:641–657
Thanassoulis E, Allen R (1998) Simulating weights restrictions in data envelopment analysis by means of unobserved DMUs. Manage Sci 44:586–594
Thanassoulis E (2001) Introduction to the theory and application of data envelopment analysis. Kluwer Academic Publishers, Dordrecht
Thanassoulis E, Portela MCS, Allen R (2004) Incorporating value judgments in DEA. In: Cooper WW, Seiford LM, Zhu J (eds) Handbook on data envelopment analysis. Kluwer Academic Publishers, Boston
Thanassoulis E, Portela MCS, Despić O (2007) DEA—the mathematical programming approach to efficiency analysis. In: Fried H, Lovell K, Schmidt S (eds) The measurement of productive efficiency and productivity growth. Oxford University Press
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Podinovski, V.V., Thanassoulis, E. Improving discrimination in data envelopment analysis: some practical suggestions. J Prod Anal 28, 117–126 (2007). https://doi.org/10.1007/s11123-007-0042-x
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11123-007-0042-x
Keywords
- Efficiency
- Data envelopment analysis
- Productivity
- Weight restrictions
- Unobserved DMUs
- Selective proportionality