EU structural funds and GDP per capita: Spatial VAR evidence for the European regions

This paper focuses on the impact of EU structural funds (SFs) on the GDP per capita of 183 European NUTS2 regions throughout the 1990-2015 period. To allow for the endogeneity of funds allocation to regions, we estimate a bivariate structural panel VAR model, allowing for unobserved heterogeneity through a rich menu of deterministic controls. Our main identifying restriction is rooted on the widely documented long lags affecting the implementation of EU’s Cohesion Policy. Through a spatial VAR specification, we also estimate spillovers from local SF expenditure on other areas. We find positive and highly significant multipliers measuring the local response of GDP to an exogenous shock in local SF expenditure, with a long-run value settling at 2.6. The spillover effects on GDP from an exogenous shock to SFs are also positive and significant, but much smaller (about one fifth of the within-region responses). When partitioning our sample according to features suggested by the literature (size, stage of development and EU funding regimes), we find that within-region multipliers are higher in regions with a larger population, located in countries supported by the Cohesion Fund, and interested by the Convergence objective. Spillovers are also heterogenous across different groups of regions, turning out to be negative in regions belonging to countries not supported by the Cohesion Fund. All this evidence is validated in qualitative terms by a robustness check concerned with the choice of spatial weights