Analyzing the determinants of the regional digital divide in the European Union

Over the last 2-3 decades, European regional economic convergence has been extensively analyzed, because it represents a European Commission policy goal. Empirical evidence has indicated that despite progress on the road to convergence at the country level, regional divergence has increased over time, especially in the aftermath of the global economic crisis. These regional divergences exist in the digital area as well. Even for the most digitally developed countries in the EU, some internal gaps remain and need to be addressed (Cruz-Jesus et al., 2016). These findings, along with the availability of regional data collected at different levels of aggregation (e.g. the NUTS2 data provided by Eurostat), have led to a growing body of regional studies.

Despite the consistent, diverse and growing literature, the regional digital divide has seldom been investigated. To our knowledge, there are few papers focused on the regional digital divide within the EU, however their scope and approach are completely different from ours. The innovative contribution that this paper could bring to the literature is the explanation of the regional digital divide within the EU by both regional- and country level socioeconomic factors, and not only by regional factors (Vicente and López, 2011). Moreover, the study examines the dynamics of the regional digital divide over a time span of 16 years (2001-2016) rather than at a single point in time. In order to reveal the influence of both regional- and country-level factors on the regional digital divide from a dynamic perspective, a multilevel analysis is used as the econometric framework. Specifically, the three level random slope model is proposed to best fit our hierarchical data.

The methodological framework of multilevel analysis, i.e. the three level random slope model, was chosen not only for conceptual reasons related to the hierarchical structure of our data, but also because preliminary tests applied to our data confirmed the appropriateness of this model. We expect to find that the regional and country level factors play different roles in shaping the regional digital divide, so that only a mix of regional and national policies are effective in reducing the digital gap at this level.