This week's departmental seminar will be hosted by Sarah Smith and invites Suphanit Piyapromdee (UCL) to speak on her recent work.

Professor Piyapromdee is an Assistant Professor at UCL, as well as being a Resesearch Affiliate at CEPR and IZA. Her work focuses on labor economics, the intersection between economics and public policy, and development economics. She will be discussing "The Anatomy of Sorting: Evidence from Denmark"

 

Abstract

In this paper, we use the finite mixture approach of Bonhomme et al. (2019) andAbowd et al. (2019) to estimate a model of wages and employment mobility with two-sided heterogeneity that we particularize in two ways. First, we develop a new Classifi-cation Expectation-Maximization algorithm for estimation to improve the classification offirms into a finite number of groups. Second, we impose a parametric structure on workers’transition probabilities for improved precision and interpretability. The major contributionof this research is showing that the process by which workers match with firms is con-voluted and one cannot understand its sources by studying wages alone. Extending thestudy of sorting beyond the correlation of wage fixed effects, we also employ a broadernotion of sorting, the mutual information (MI), which can accurately represent the depen-dence between worker and firm latent types independently of their effect on wages. UsingDanish data from the years 1987−2013, we find that it is important to allow for flexibleinteractions between job tenure, labor experience, and worker and firm types in both meanwages and employment mobility. Sorting is moderately increasing over time wherein somesorting is positive on wages. Through the lens of MI, we find increased importance ofnon-wage amenities as workers age and select into long tenure relationships. The job offerarrival processes while employed and unemployed are key drivers of sorting during earlycareer stages. However, as workers age job preferences become the dominant determinantof matching and give rise to a positive age trend in sorting. The positive age trend in sortingshows in the MI index but is not noticeably apparent in the wage fixed effect correlation. Itis, hence, important to allow for both non-wage factors and non-linearity in matching whenmeasuring sorting.

 

The seminar will be held in the usual seminar room, 1B6