Evaluating the performance of higher education institutions in Europe: a nonparametric efficiency analysis of 944 institutions

ABSTRACT

Although a long tradition exists of studying the economics of education, performance comparisons of different kinds have traditionally been difficult to undertake. An important impediment has been the lack of comparable data. To a large extent, this study remedies the shortcomings of earlier studies. Our contribution is twofold. First, the current higher education policy of contraction and economizing, common to most European countries, make knowledge of potential efficiency gains crucial. Second, a unique database allows us to study a broad diversity of organizations as well as organizations located in a number of different European countries. The study suggests evaluating higher education institution (HEI) performance in a production theory context, applying the well-known data envelopment analysis (DEA) method to a cross section of 944 HEIs in 17 European countries. The DEA approach is particularly suitable in this context where little is known about production technologies and economic behaviour of the HEIs. On average, provision of education is found to be most efficient in the Slovak Republic followed by Belgium and Latvia, while Denmark and Norway display the lowest efficiency. The study also indicates a positive relation between efficiency and HEI size and efficiency and research intensity. Furthermore, the study points to the importance of continued data collection.

KEYWORDS:

• Higher education institutions
• OECD countries
• technical efficiency
• nonparametric analysis
• bootstrapping
• Eumida database

JEL CLASSIFICATION:

• D22
• D24
• I21

Acknowledgements

The authors are grateful to Hans Bjurek for helpful comments and to Florin Maican and Jianhua Zhang for computational assistance. We thank the participants of the Efficiency in Education Workshop, London, held on 19–20 September 2014 and Finn R. Førsund in particular for constructive remarks. We also appreciate helpful comments from the anonymous referees. This article has been written in context of the research programme: Radical Innovations for the Enhancement of the Swedish Economy, running at the Department of Economy and Society, Institute of Innovation and Entrepreneurship, University of Gothenburg, Sweden.

Notes

¹ European University Data Collection Consortium.

² Based upon a survey of 196 research universities in 14 European countries.

³ Should prices exist, it would be possible to calculate costs directly and compare these with performance.

⁴ For a more detailed presentation of different Farrell-type efficiency measures and their application to DEA, see, for example, Hjalmarsson and Veiderpass (1992).

⁵ Assuming constant returns to scale, the corresponding measure is calculated excluding the weight u₀ from the LP problems.

⁶ The Eumida dataset contains information on HEIs in 29 European countries. Due to missing observations on one or more of our variables of interest, our study focuses on 17 of those countries.

⁷ For these reasons, the project had to exclude France.

⁸ Number of students admitted and number of degrees awarded are strongly correlated when dropout rates are accounted for.

⁹ Neither do we have explicit information on the quality of the academic staff. To a certain extent, this may be reflected by the basic institutional descriptors.

¹⁰ Due to space restrictions, it is not possible to present all 944 individual efficiency values. These are, however, available from the authors upon request.

¹¹ The Programme for International Student Assessment.

¹² 5% display a first decimal difference (i.e. a difference on ‘percentage level’) while 8 of the 944 HEIs display an integer difference (i.e. difference on 10% level). Most of the latter units are frontier units.

¹³ This type of diagram, then based on the input coefficient in Salter (1960), was first introduced in Førsund and Hjalmarsson (1979a, 1979b).

Additional information

Funding

We acknowledge and thank the Sten A Olsson Foundation for Research and Culture for supporting this research.