What about regions in regional science? An exercise of convergence using different geographic scales of European Union

Although the choice of the spatial scale of analysis is a problematic issue in applied research (specifically, in the literature of regional economic growth), it is evident that regional scientists have been very slow at demonstrating the implications of aggregation problems on empirical economic growth studies. The aim of this paper is to examine the alterations in the empirical results caused by the use of different spatial scales in the analysis of convergence process of European Union (EU15) regions by systematically repeating a method originally developed to examine this phenomenon at a single scale across multiple scales (e.g., NUTS1, NUTS2 and NUTS3). Despite the analysis of the robustness of growth regression to various aspects of research design conducted by several studies, it is still necessary to incorporate in this important line of research a deeper investigation of changes in spatial scale of analysis which is usually known by Modifiable Areal Unit Problem (MAUP). Thus, convergence processes among EU15 are examined across three spatial scales often used in applied work, namely, NUTS1, NUTS2 and NUTS3, between 2000 and 2008. The paper suggests that convergence pattern of EU regions depends on the spatial scale of analysis. In sum, the results for NUTS 1 and 2 do not support the hypothesis of b-convergence when country dummies are included in the OLS regressions. On the other hand, results for NUTS 3 controlling for country specific characteristics suggest conditional b-convergence. This might be an indication that convergence occurs within countries at finer scales (NUTS3 level) rather than between EU countries. Furthermore, these findings hold when we correct our econometric specifications for spatial dependence.