2372. A centroid method for neutral cross efficiency
Invited abstract in session TA-60: DEA methodological developments II, stream Data Envelopment Analysis and its applications.
Tuesday, 8:30-10:00Room: Western LT
Authors (first author is the speaker)
| 1. | John Lamb
|
| Business School, University of Aberdeen |
Abstract
Cross efficiency extends data envelopment analysis. In it we compute a cross efficiency for every decision-making unit (DMU) and for every peer. Then we compute the cross-efficiency score as an average efficiency over all peers. Cross efficiency increases our ability to discriminate among DMUs. But it presents a challenge. The cross efficiencies are not uniquely defined. Doyle and Green (1993, 1994) suggest aggressive and benevolent methods to resolve this. But the solutions can be unusual. Wang and Chin (2010) and Shi et al (2021) suggest neutral cross-efficiency methods based on secondary goals. We show there is another method, which we call the centroid method, that does not require secondary goals and gives a unique, average, cross efficiency. We also show how existing neutral methods can be extended to ensure unique solutions and provide some comparisons among methods.
Keywords
- Data Envelopment Analysis
- Convex Optimization
Status: accepted
Back to the list of papers