Efficiency in Brazilian Federal Universities: a study with Data Envelopment Analysis (DEA)

Résumé

This paper aimed to analyze the level of technical efficiency of Brazilian federal universities, from 2012 to 2018, using the Data Envelopment Analysis (DEA) method, with an output-oriented DEA BCC model. Four management indicators instituted by TCU were used as inputs and two quality indicators, as outputs. For that, a sample of 56 federal universities was used. Considering the standard efficiency scores, the results indicated high levels of technical efficiency during the years under analysis. The average technical efficiency of federal universities was 93.7% in 2012, with 17 efficient universities, and 94.5% in 2018, with 18 efficient universities. The years 2015 and 2016 stood out, with 28 and 27 efficient universities respectively, whose national averages were 96.1%. However, due to the benevolence of the BCC model with very small or very large units, some universities may have been considered efficient by default. From 2012 to 2018, the Malmquist Index indicated an increase in the productivity of universities by 2.2%, which occurred in a greater proportion due to the catch-up effect (1.8%) and in a lower proportion by the frontier shift effect (0.4%). In addition, the increase of 1.8% in relative efficiency occurred in a greater proportion due to the increase in pure technical efficiency (1.1%) and in a lower proportion due to the increase in scale efficiency (0.7%). Yet, despite the increase in productivity, there is still room to achieve better quality results and to improvements in the management of resources, in order to reduce waste.