Unemployment clusters across Europe's regions and countries

High unemployment and regional inequalities are major concerns for European policy-makers, but so far connections between policies dealing with unemployment and regional inequalities have been few and weak. We think that this should change. This paper documents a regional and transnational dimension to unemployment -- i.e., geographical unemployment clusters that do not respect national boundaries. Since the mid 1980s, regions with high or low initial unemployment rates saw little change, while regions with intermediate unemployment moved towards extreme values. During this polarization, nearby regions tended to share similar outcomes due, we argue, to spatially related changes in labour demand. These spatially correlated demand shifts were due in part to initial clustering of low-skilled regions and badly performing industries, but a significant neighbour effect remains even after controlling for these, and the effect is as strong within as it is between nations. We believe this reflects agglomeration effects of economic integration. The new economic geography literature shows how integration fosters employment clusters that need not respect national borders. If regional labour forces do not adjust, regional unemployment polarization with neighbour effects can result. To account for these 'neighbour effects' a cross-regional and transnational dimension should be added to national anti-unemployment policies. Nations should consider policies that encourage regional wage setting, and short distance mobility, and the EU should consider including transnational considerations in its regional policy, since neighbour effects on unemployment mean that an anti-unemployment policy paid for by one region will benefit neighbouring regions. Since local politicians gain no votes or tax revenues from these 'spillovers', they are likely to underestimate the true benefit of the policy and thus tend to undertake too little of it.