Why are wages in cities like New York or Paris higher than in others? This paper uses firm mobility to separate the role of “location effects” (e.g., local geography, infrastructure, and agglomeration) from the spatial sorting of workers and firms. Using French administrative records and U.S. commercial data, we first document that firm mobility is widespread: 4% of establishments relocate annually. Establishments retain their main activity and structure as they move, but adjust their workforce and wages. Combining firm and worker mobility, we then decompose wage disparities across French commuting zones. We find that spatial wage differences are largely driven by the sorting and co-location of workers and firms: location effects account for only 2–5% of disparities, while differences in the composition of workers and establishments account for around 30% and 15%, respectively. The remaining half is accounted for by the co-location of high-wage workers and firms, especially in cities with high location effects. Revisiting the elasticity of local wages to population density, we find a significant coefficient of 0.007—two to three times lower than estimates not controlling for firm composition.

How do the employer and the worker interact during a dismissal? This paper tests whether they cooperate to minimize costs, or instead engage in conflict—i.e., deliberately amplify costs. We leverage a unique feature of the French labor market: an employer and a worker can jointly opt to replace a costly dismissal by a cheaper and more flexible “separation by mutual agreement” (SMA). Introduced in 2008, SMAs eliminate red tape costs, enable severance pay bargaining, and preclude litigation. However, we find that only 12% of dismissals are resolved through SMAs—far below the efficient level predicted by standard bargaining models. Surveying HR directors, we identify three drivers of conflict that hinder cost minimization: (i) hostility between the employer and the employee, (ii) employers using dismissals as a “discipline device” to maintain incentives, and (iii) asymmetric beliefs about subsequent labor court outcomes. Using counterfactual scenarios in the survey, we find that removing these three drivers of conflict would increase SMA adoption from 12% to 67% of dismissals. We confirm that less conflictual dismissals—due to either better employer-employee relationships or workers benefiting from early retirement—end more often as SMAs.