Marginal Jobs and Job Surplus : A Test of the Efficiency of Separations

By the influential "Coasean" theory of employment relationships, job separations occur only once the worker and the employer have exhausted all remaining gains from trade through flexible bargaining and unrestricted contracting, with joint job surplus hence having turned negative. Our strategy to study this empirically elusive view is to track jobs longitudinally over the course of the introduction and sudden abolition of a policy that subsidized nonemployment and hence lowered job surplus: an age-and-region-specific extension of the maximum duration of unemployment benefits from one to four years in Austria. We document that this program destroyed 10.9ppt of jobs (a 27% increase in the separation rate). By the Coasean theory, these separations must have extracted marginal (low-surplus) jobs - a property not directly measurable. The testable prediction we instead investigate is that after the program abolition, the jobs having "survived" the treatment should be more resilient to any subsequent shocks (for lack of marginal i.e. low-surplus jobs), compared to their control peers. Strikingly, in the data, the two groups exhibit identical post-abolition separation behavior. The Coasean view can rationalize our findings only under narrow conditions: if surplus exhibits no persistence whatsoever. One non-Coasean model candidate fully accounts for our findings, building on wage rigidity and a constellation of large worker surplus and small firm-surplus, with the latter driving post-abolition separations