Digital transformation and entrepreneurship: A comparative analysis of EU11 and selected EU15 economies
TL;DR
This study examines the association between multidimensional digital transformation and national startup ecosystem performance across 23 European Union economies, comparing EU11 with 12 selected EU15 economies from 2017 to 2024.
Abstract
Type of the article: Research ArticleAbstractThis study examines the association between multidimensional digital transformation and national startup ecosystem performance across 23 European Union economies, comparing EU11 with 12 selected EU15 economies from 2017 to 2024. The balanced panel contains 184 country-year observations. Principal component analysis is used to construct three composite indicators: the Digital Services Index, Digital Connectivity Index, and Digital Human Capital Index. The indices demonstrate satisfactory factorial adequacy, internal consistency, and one-component structures supported by parallel analysis. Their associations with startup ecosystem performance are estimated using two-way fixed-effects models. Because the panel contains only 23 country clusters, the principal inference uses CR2 bias-reduced country-clustered standard errors with Satterthwaite-adjusted degrees of freedom. In the full-sample specification, DSI has a positive but only marginally significant association with startup ecosystem performance (β = 0.0834, p = 0.082), while DCI, DHCI, and e-government are statistically insignificant. R&D intensity has a negative contemporaneous coefficient that is also marginally significant (β = −0.0145, p = 0.053). Regional heterogeneity is jointly significant and is concentrated primarily in digital connectivity: DCI is negatively associated with startup ecosystem performance in the EU11 group, whereas its total slope is approximately zero in the selected EU15 group. DHCI has a positive total slope within the selected EU15 group, although the difference between the EU15 and EU11 slopes is not statistically significant. A one-year-lagged specification produces a positive but marginal DSI coefficient and does not identify statistically significant associations for the remaining predictors. These estimates should be interpreted as conditional associations rather than causal effects.