Abstract

We evaluated the efficiency of 21 Higher Education Federal Institutions in human capital qualification in Brazil from 2007 to 2010. This original model considers the qualification increment measured by external tests given to first and last year undergraduate students as educational output. It considers teacher qualification, perceptions of pedagogic quality, and infrastructure adequacy as educational resources in the productive process. Data Envelopment Analysis (DEA) is a well-known technique used in evaluating the efficiency of institutions. An institution is regarded as a productive unit that requires input to generate output. Its relative performance in the process is evaluated and ranked among the units under evaluation. This technique differs from other alternatives by allowing several inputs and outputs in a nonparametric fashion without requiring functional specifications. The rankings produced can be examined in static or dynamic ways, and they can be used to suggest improvements in inefficient units. Static analysis indicates a relatively high efficiency level, suggesting an average improvement of 21%. Dynamic analysis decomposes productivity variation in two different effects. The first, measuring frontier variation, indicates little average improvement. The second, measuring the frontier effect, indicates contraction of the efficient frontier of about 5% per year. Although each university can improve its efficiency by acting on indicated variables, it cannot control movements of the visible part of its frontier alone unless it is defined.