ABSTRACT
This paper studies regional productive performance differentials among 243 NUTS-2 European regions for the period 2003–11. Within the last two decades the technology gap between European regions has increased considerably. Building on previous studies that have used data envelopment analysis (DEA) and which have neglected national production structures and pre-existing technological heterogeneity, we examine differences in regional productive performance by using a meta-frontier framework. Our findings confirm significant differences in productive performance across European regions and a large North–South technology gap. A panel vector autoregression (PVAR) shows that this regional technology gap can be attributed to differences in human capital and innovation activity.
ACKNOWLEDGEMENTS
The authors express their sincere thanks to the journal editor and two anonymous referees for their valuable comments.
DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.
Notes
1. In this paper the estimated measure of regional productive performance is the concept traditionally called technical (in)efficiency.
2. Hence, this paper relaxes the assumption of technological isolation (Tsekouras et al., Citation2016, Citation2017).
3. Given a standard assumption on S, the input distance function is not increasing in x, not decreasing, homogeneous of degree +1, and convex in y.
4. The term Θ measures the radial distance between the input–output observation and the technology frontier within the production set.
5. For a detailed presentation of the countries and regions involved in our research, see Table S1 in the supplemental data online.
6. A FEAR package (Simar & Wilson, Citation2007) was used to carry out our estimations.
7. We also estimated a stochastic meta-frontier production function following a referee’s suggestion (see the supplemental data online).
8. Table S2 in the supplemental data online shows the productive efficiency and meta-technical ratio scores for European countries for the period 2003–11.
9. To save space and avoid making the analysis more complex, we chose to report the results at the country level and to not report the confidence intervals.
10. The average TE for this year is 0.768.
11. Relying on previous results from empirical studies that indicated patents as a more representative and reliable measure of innovations (Acs et al., Citation2002) or as an ‘upstream indicator’ (Faber & Hesen, Citation2004) that can better generate productivity gains, we use patents instead of R&D.
12. The supplemental data online presents the results from a robustness check regarding our PVAR estimations.