This study investigates spatial variation in fertility in Europe. We analyse spatial variation in total fertility rates using small-scale geographical data from 21 European countries for 2010 and investigate the role economic, sociocultural, and spatial factors play in regional fertility levels. We compare the performance of conventional ordinary least squares regression and multilevel modelling with that of different spatial regression models and show that the spatial approach is superior for modelling regional fertility variation. The analysis shows that fertility levels in a region are strongly related to gross domestic product per capita and the share of divorced individuals in the region, and fertility levels in neighbouring regions, supporting that all three realms of fertility determinants-economic, sociocultural, and spatial-are relevant for understanding modern fertility variation. K E Y W O R D SEurope, NUTS 3, spatial lag model, spatial regression, spatial variation, total fertility rate

Since 2010, the Nordic countries have experienced substantial fertility decline. This was unexpected, as these countries have well-established systems of family support policies and they did not experience a fertility shock in the immediate aftermath of the 2008 economic recession like many other European countries. Previous studies paid little attention to the spatial dimension of this unexpected fertility change, despite evidence of large spatial variation in fertility. This paper aims to close this gap through a spatial perspective and deepen our understanding of how Nordic fertility change varies by economic and social contexts. We apply advanced spatial panel models on data for 1,099 municipalities covering Denmark, Finland, Norway, and Sweden. Next to analyzing total fertility rates, we also compare the relevance of economic and social contexts for spatial fertility variation at younger or older ages. We note distinct differences by levels of urbanization and differing trends between the age groups. We also find that measures of unemployment and union stability are associated with lower levels of fertility across space and time, while support for conservative parties is related to higher fertility. Age-specific analyses highlighted that economic conditions are more relevant for fertility variation under age thirty than over age thirty. Overall, our analysis provides support for the view that both economic and social factors are highly relevant for understanding spatial variation in the Nordic fertility decline. Given the strong spatial component in Nordic fertility change, policy initiatives would also benefit from a spatial dimension to effectively address fertility change.

No abstract