Understanding European Integration with Bipartite Networks of Comparative Advantage

European employment network structure

Core objectives of the European integration are economic convergence and growth across member states, but these ambitious goals are challenged by competition and value chain asymmetries within the common market. A fundamental tension exists between fostering specialized industries for global competitiveness and bridging the productivity gap between more and less developed countries.

A difficult challenge for the EU is how to harmonize the specialization of industries and countries to achieve global competitiveness while simultaneously ensuring that less developed regions can catch up. The question is whether convergence requires similar specialization patterns across the EU, or whether a division of labor with complementary specializations can achieve the same goal more effectively.

A Network Approach to Revealed Comparative Advantage

We develop a novel bipartite network approach that traces the pairwise similarity of Revealed Comparative Advantage (RCA)—a widely-used measure of economic specialization—within and between EU15 (the older, more developed member states) and Central and Eastern European (CEE) countries from 2000 onwards. RCA indicates which industries a country is particularly good at relative to the global average, revealing its productive strengths.

This network-based approach offers several advantages over traditional methods. It can assess redundancies and divisions in the system as a whole, while also providing insights at the level of individual industries and countries. This granular view enables us to investigate how co-specialization across countries impacts economic growth—a question that has been difficult to address with conventional economic metrics.

Our analysis reveals a striking pattern: we find substantial overlap of RCA among CEE countries themselves, suggesting similar specialization patterns within this group. However, we observe a diverging RCA structure between EU15 and CEE countries, indicating that these two groups are specializing in different industries rather than converging toward similar economic profiles.

The Productivity Puzzle: Division Beats Imitation

Our econometric analysis provides a surprising and policy-relevant finding: productivity increases most significantly in those CEE industries that have co-specialized with other CEE countries after EU accession. In contrast, RCA similarity between CEE and EU15 countries shows a weaker relationship with productivity growth.

This result challenges the intuitive notion that catching up requires imitating the industrial structure of more developed countries. Instead, it suggests that productivity convergence can be achieved through a division of sectoral specialization, where different regions focus on complementary industries that fit their particular advantages and capabilities.

The findings have important implications for European policy. They suggest that efforts to promote convergence should not necessarily push CEE countries to replicate the industrial patterns of Western Europe. Rather, policies might be more effective if they facilitate the development of specialized clusters within CEE that can serve complementary roles in European and global value chains.

This division-of-labor approach could allow CEE countries to leverage their specific advantages—whether in labor costs, geographic position, institutional arrangements, or accumulated capabilities—while still achieving productivity improvements through deeper integration into European production networks. The key is coordination and complementarity rather than imitation.

Research Metrics

Altmetric Attention

Citations

References

Di Clemente, R., Lengyel, B., Andersson, L-F. & Eriksson, R.

Understanding European Integration with Bipartite Networks of Comparative Advantage

PNAS Nexus, 1(5), pgac262 (2022)

Python implementation of the Bipartite Configuration Model

BiCM Python Package - Statistical Null Model for Bipartite Networks

Multi-language implementation of bipartite network motifs analysis

bmotifs Package - Python, R and Matlab Implementation