Most of the studies on convergence in regional GDP per capita estimated (beta) convergence among EU regions, at both EU-15 and EU-27 level. However, the speed of convergence is not constant either in time or between regions. Results for other variables, like health variables, are inconclusive. Our objective in this paper is to analyse the health convergence of the EU-27 regions from 1995 to 2009, using (cause-specific) mortality and life expectancy at birth. Three are the contributions of our paper. First, we analysed regions instead of countries. Second, we explicitly allowed that convergence rate, if any, could vary in time and space. Third, we specified a model to control also explicitly, problems such as endogeneity of the explanatory variables of health convergence as well as the unobserved spatiotemporal extra-variability. Our results indicate that, in terms of health, although there was a catching-up process between the EU-27 regions between 1995 and 2009, there was also significant heterogeneity not only in the speed of that beta-convergence but even in itself, both among countries and over time. Furthermore, using the coefficient of variation as a summary measure of sigma-convergence, we were unable to estimate a reduction in disparities between regions in EU-27 over those fifteen years. In summary, if the concept of sigma-convergence is more revealing of the reality of convergence, we could not conclude that there is convergence in health among the regions of the EU-27 between 1995 and 2009.
