Abstract
This study examines technical efficiency in European higher education (HE) institutions. To measure efficiency, we consider the capacity of each HE institution, on one hand, to provide competencies to graduates and, on the other hand, to match competencies provided during education to competencies required in the job. We use a large sample of young graduates interviewed three years after graduation from 209 HE institutions among eight European countries. A non‐parametric approach (Data Envelopment Analysis) is used to evaluate efficiency of converting multiple inputs into multiple outputs. Objectives selected are consistent as the same types of institution are found to be efficient in different specifications.
Acknowledgements
The authors would like to thank Tim Coelli and Sergio Perelman for their answers to questions concerning the DEA methodology, Christoph Meng for his help in providing the data, and participants of Human Capital Workshop (Maastricht, December 2003) for helpful comments.
Notes
Magazines and newspapers are reluctant to choose those performance indicators that are more associated with their readership.
See Nunamaker (Citation1985) for an overview on efficiency measurement of non‐profit organizations and Ruggiero (Citation2004) for applications to education performance.
Detailed presentation of the methodology can be found in Coelli et al. (Citation1998) and Coelli (Citation1996).
Efficiency of a unit consists of two components: technical efficiency, which reflects the ability of an unit to obtain maximal outputs from a given set of inputs, and allocative efficiency, which reflects the ability of an unit to use the inputs in optimal proportions, given their respective prices. These two measures are combined to provide measures of total economic efficiency.
A summary of mathematical DEA is included in Appendix 1.
The input and output orientated models will estimate exactly the same frontier and therefore identify the same DMUs as being efficient. It is only the efficiency measure associated with inefficient DMUs that may differ.
For a discussion concerning output slack see Coelli (Citation1996, p. 23).
According to Coelli (Citation1996), the CRS assumption is only appropriate when all DMUs are operating at an optimal scale (in case of perfect competition, no financial constraints, etc.).
Appendix 2 presents countries and institutions included in the sample.
See Appendix 3 for details of the hierarchical clustering method.
The internal consistency of the two clusters is supported by Cronbach alpha findings of 0.77.
To evaluate the importance of each institution in determining the frontier, we check how often an efficient institution is a peer for other institutions. As those other institutions all have at least two other peers, excluding efficient institutions from the sample does not modify the ranking.
Those competencies are linked to social capital as access to networks, knowledge of rules, communication skills, and so forth.
The number of efficient universities increases when two types of output are used in the model because universities that more efficiently produce one type of output will place a higher weight on that type of output. Based on this, it would seem that a university that efficiently produces either competencies outputs or match outputs can choose weights that will produce an efficiency score of 1 when both types of outputs are considered in the same model.
DEA treats the observed inputs and outputs as given constants.