CLUSTERING REGIONAL COMPETITIVENESS IN CENTRAL AND EASTERN EUROPE: INSIGHTS FROM THE K-MEANS METHOD

Abstract

Purpose: This paper investigates regional competitiveness of NUTS 2 regions in eleven post-transition EU Member States in Central and Eastern Europe (CEE) from 2011 to 2021. It applies Martin’s (2004) “Regional Competitiveness Hat” model to identify whether distinct regional profiles, knowledge hubs, production locations, and regions with growing yields, can be empirically validated using clustering techniques.
Methodology: The study utilises the k-means clustering method to classify 61 NUTS 2 regions based on three key indicators: GDP per capita, population density, and gross domestic expenditure on R&D. Data were standardised and tested for outliers using Mahalanobis distance. ANOVA and post-hoc Games-Howell tests were conducted to verify statistical significance and interpret the stability and movement of regions between clusters over time.
Results: The analysis produced three statistically robust and theoretically consistent clusters: knowledge hubs (e.g., Zagreb and Bucharest), production locations (regions with low GDP per capita and population density), and regions with growing yields (moderate GDP per capita and lower density). The results affirm the utility of Martin’s model in the CEE context and reveal stability among clusters, with notable mobility only between production locations and growing yield regions.
Conclusion: This study confirms the applicability of Martin’s framework to post-transition CEE regions and offers a dynamic, data-driven classification tool for regional development policy. It highlights GDP per capita, population density, and R&D investment as critical competitiveness indicators. The findings support targeted EU policy-making and suggest future inclusion of digitalisation and sustainability metrics.