Matching in segmented labor markets: An analytical proposal based on high-dimensional contingency tables

Abstract

The Spanish economy has a very problematic labor market characterized by high and persistent levels of unemployment, elevated long-term unemployment, strong segmentation and low regional mobility, among other drawbacks. This paper uses labor matching data from a large database of administrative microdata (Continuous Sample of Working Lives, MCVL) and structures them into a contingency table which cross-classifies the information of workers and jobs at provincial and occupational levels. The association analysis performed allows us to identify a more precise vision of the structure of the labor market, and a better design regarding active labor market policies. Our results demonstrate, for example, that highly isolated markets and those that influence the entire national territory coexist in the Spanish labor market. Finally, we also propose a new smoothing method in order to deal with typical statistical problems in sparse contingency tables, such as the existence of non-structural zero frequencies or sparsity.