Document Type

Closed Project

Publication Date

Summer 2004


Timothy Anderson

Course Title

Productivity Analysis

Course Number

EMGT 537/637


Oregon University System, Portland State University -- Management -- Analysis, Data envelopment analysis, Production management


This paper uses Data Envelopment Analysis with information gleaned from the website of the Oregon University System,, in an attempt to gain a better understanding of state's higher education system, and in particular, of course, of Portland State University. Two different approaches are used, as a function of data availability: the Malmquist Productivity Index for one set of data and a more traditional envelopment model for the other. In this paper we will also discuss issues of model building and the use of Professor Joe Zhu's DEA Solver software.

In general we found this exercise interesting but frustrating. The difficulty involved in collecting meaningful data (in particular financial information) left us feeling that we had only scratched the surface of a very hard problem. We have ongoing questions about the application of the rule surrounding the ratio of DMUs to the sum of inputs and outputs, especially in a Malmquist context. In addition, we have questions about using Malmquist when the frontier of "best" performance moves backwards as well as forwards, something not usually encountered in technology. Even if the technique is capable of handling such a situation, in a mathematical sense, we had a hard time making intuitive sense of it. And apart from showing a list of numbers, how might a case of rising and falling efficiency (due, perhaps, to bad economic conditions and tight budgets) be represented graphically. In the end, we feel that while DEA may present an excellent way to generate "red flags," or indications that a subject needs to be studied more closely, we are not at all sure that we would want to base any serious decisions on DEA alone. Is this, in fact, the way the technique should be used? Are we talking about something rather more exploratory than analytical, at least in education?

Note: The presentation associated with this report is included here as a supplemental file.


This project is only available to students, staff, and faculty of Portland State University

Persistent Identifier