Accounting for health expenditures, migration, and CO₂ emissions in the EU: a bidirectional analysis.

BACKGROUND: Migration, health expenditures, and carbon emissions are critical policy concerns in the European Union. However, limited evidence exists on how these factors interrelate across diverse member states, and most existing studies examine only one pairwise linkage at a time, rely on non-EU settings, or employ methods that do not accommodate cross-sectional dependence and heterogeneous country dynamics. This study tests for bidirectional Granger-causal relationships among all three variable pairs across the EU-27.

METHODS: We compiled national-level data from 27 EU member states spanning 2000-2020. Using the Emirmahmutoğlu and Köse heterogeneous panel Granger causality test, which accommodates mixed integration orders and country-specific lag structures, with bootstrap critical values to address cross-sectional dependence, we tested for pairwise bidirectional predictive relationships among CO₂ (Carbon dioxide) emissions per capita, current health expenditure per capita, and net migration at both the panel and country levels. A Cross-Sectionally Augmented Autoregressive Distributed Lag (CS-ARDL) model was estimated as a robustness check to assess the long-run magnitude and sign of the identified relationships.

RESULTS: At the panel level, the Fisher test statistics reject the null of no Granger causality in both directions for all three variable pairs, CO₂ and net migration, health expenditure and net migration, and health expenditure and CO₂ (all p < 0.01), confirming panel-level bidirectional Granger causality. At the country level, however, the patterns are markedly heterogeneous: only Italy exhibits bidirectional causality between CO₂ and migration; four countries (Germany, Sweden, Croatia, Poland) show bidirectional health expenditure-CO₂ feedback; and Portugal and Slovenia show bidirectional migration-health expenditure linkages. Unidirectional results emerge in a further 17 countries, while Ireland, Luxembourg, and the Netherlands show no significant linkages. The CS-ARDL robustness analysis, estimated for the CO₂ equation only, confirms a significant negative long-run association between health expenditure and CO₂ and a significant positive long-run association between net migration and CO₂. Structural interpretation of the country-level heterogeneity identifies four broad regime types, integrated nexus countries with feedback dynamics, environment-sensitive migration regimes, demographic-pressure-driven systems, and structurally decoupled systems, though several countries exhibit multi-linkage profiles that span more than one category.

CONCLUSION: The EU-wide panel results mask several distinct national pathways shaped by differences in energy mix, health-system design, and migration exposure. Policy responses should be regime-specific: decarbonising migrant-absorbing infrastructure in demographic-pressure systems, coupling pollution remediation with place-based investment in environment-sensitive regimes, strengthening migrant-inclusive healthcare in integrated nexus countries, and pairing pollution control with healthcare-sector decarbonisation in feedback systems.