| Article ID: | iaor19891161 |
| Country: | United Kingdom |
| Volume: | 17 |
| Start Page Number: | 123 |
| End Page Number: | 132 |
| Publication Date: | Jul 1990 |
| Journal: | Computers and Operations Research |
| Authors: | Sengupta Jati K. |
| Keywords: | programming: linear, education, statistics: data envelopment analysis |
Canonical correlation theory is utilized here to modify the set of efficiency measures in data envelopment analysis (DEA). This involves (1) the mean square error criterion, (2) the canonical correlation test, and (3) the set of redundancy tests. An empirical application to educational production frontier data shows the usefulness of these tests and how they can be statistically tested as a two-stage procedure. This may help provide the bridge between the parametric and the nonparametric approaches to the estimation of production frontiers.