Key points are not available for this paper at this time.
Abstract In this paper, we propose a novel model-based approach to measure labour market polarization. The methodology, grounded in Generalized Joint Regression Models, allows for flexible specification and is particularly well-suited to regional labour market analyses with hierarchical data structures. We employ a trivariate model to jointly capture the growth rates of job advertisement shares for high-skill occupations (Managers, Professionals, and Associate Professionals) and middle-skill occupations, along with their correlation—a conceptually natural measure of job polarization. As an empirical application, we investigate the degree of labour market polarization across EU regions from the demand side during the post-COVID-19 period. Our analysis draws on institutional data from the Cedefop European Centre, which contains job advertisements published on online portals across EU regions, disaggregated by region and main ISCO-08 occupational groups. To identify potential determinants of the three outcomes, we incorporate a set of structural regional factors sourced from Eurostat Regional Statistics. Issues of representativeness in vacancy data are addressed by benchmarking against new hires recorded in LFS microdata. Empirically, we found a highly heterogeneous pattern of labour demand across the EU, with some countries exhibiting pronounced polarization processes while others show signs of depolarization. Among determinants, labour productivity acts as a depolarizer, tertiary education as a strong polarizer, while the technology level in a region increases the shares of both skill groups. To our knowledge, this study provides one of the first model-based analyses of demand-side job polarization integrating supply- and demand-side official data within an EU regional framework.
Pietro Giorgio Lovaglio (Tue,) studied this question.