The purpose of this paper is twofold. The primary purpose is to try and explain both the temporal and the spatial variation of poverty rates in terms of unemployment insurance indicators and structural/sociodemographic factors. Secondly, the paper aims to test the 'convergence hypothesis' of the poverty rate, i.e. whether or not poverty rates in modern welfare states have converged in recent decades. The analyses tentatively suggest that structural change in terms of the unemployment rate, the proportion of single-earner households, the female labour-force participation rate, and the proportion of families with children primarily explains the temporal variation in poverty rates. The unemployment insurance factors primarily explain the spatial variation, i.e. variation between countries. In part, however, the temporal variation is explained by the cuts in net replacements in unemployment insurance. This factor alone explains a larger share of the overall variance than all the structural factors put together. Where the convergence hypothesis is concerned, a random-effects model is proposed as a statistical tool for testing convergence (and dispersion). The results of this analysis suggest that for the 16 countries and the time frame under review the convergence hypothesis must be rejected, an outcome that is in line with some other analyses presented recently.