ABSTRACT
This paper examines the effects of borders on the performance of metropolitan areas in Europe. An innovative multidimensional conceptualization of border effects into four factors (separation, contact, differentiation and affirmation) is elaborated on and empirically tested. Estimation results confirm the ambivalent effects of the differentiation factor: economic differentials display positive effects, while cultural differences have negative effects on metropolitan performance. For the other factors, the estimations are of the expected sign, but their signal is weak and calls for further research. Ultimately, this exploratory analysis represents a promising attempt to disentangle the intrinsic dimensions of borders and consider their ambivalent effects.
ACKNOWLEDGEMENTS
A previous version of this paper was presented at the Regional Studies Association (RSA) annual conference in Graz, Austria, 4–6 April 2016, where it was granted the Award for Best International Conference Paper. The authors particularly thank three anonymous referees for providing valuable comments and suggestions for improvements to this paper. They also thank Vincent Dautel and Antoine Paccoud for their insightful feedback; Rupert Kawka for data on European metropolitan areas; and Timothée Cuignet for data on functional urban areas. The usual disclaimer applies.
DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.
SUPPLEMENTAL DATA
Supplemental data for this article can be accessed here. https://doi.org/10.1080/00343404.2017.1410537.
ORCID
Christophe Sohn http://orcid.org/0000-0002-2448-288X
Julien Licheron http://orcid.org/0000-0002-6181-3561
Notes
1. For an example of the impact of US gun laws on homicides in Mexican border municipios, see Dube, Dube, and García-Ponce (Citation2013).
2. The border or cross-border dimension of metropolitan areas is assessed from a geographical point of view and, therefore, is not indicative of the potential bi-national character of the cities concerned (i.e., the sense of belonging of their inhabitants to a single community; cf. Buursink, Citation2001).
3. For a detailed description of the 30 indicators, see Table B1 in Appendix B in the supplemental data online.
4. Language differences were accessed in 2008, but as they typically display strong inertia, this time lag is considered not relevant.
5. Potential heteroskedasticity in the residuals of the Tobit model is also tested and accounted for using White’s correction.
6. The stepwise procedure is an automatic procedure for the choice of relevant explanatory variables. A backward-selection approach is used, starting with all explanatory variables in a first model, and then dropping the least significant variable and re-estimating the model, until the remaining explanatory variables are all significant at the 10% level.