Optimal Taxation with On-the-Job Search

We provide a theoretical and empirical study of the optimal taxation of labor income in the presence of search frictions and unobservable amenities using comprehensive Danish matched employer-employee data. Heterogeneous workers undertake costly search off- and on-the-job in order to locate more productive jobs that pay higher wages. More productive workers search harder, resulting in equilibrium sorting where low-type workers are overrepresented in low-wage jobs while high-type workers are overrepresented in high-wage jobs. Absent taxes, worker search effort is efficient, because the social and private gains from search coincide. The optimal tax system balance efficiency and equity concerns at the margin. Equity concerns make it desirable to levy low taxes on (or indeed, subsidize) low-wage jobs including unemployment, and levy high taxes on high-wage jobs. Efficiency concerns limit how much taxes an optimal tax system levy on high-paid jobs, as high taxes distort the workers' incentives to search. Using detailed micro data on wages, labor market transitions, and income tax filings we estimate the structural model and compare the the actual Danish tax regime with the optimal one. Preliminary results suggest the optimal tax schedule exhibits less progressivity than the actual tax system in place. The model allows us to quantify the welfare gains from adopting an optimal income tax schedule.