This article examines a dynamic adverse‐selection model that generates equilibrium employment cycles. In the model, firms hire workers from unemployment, observe workers' productivity through time, and (following the profit‐maximizing rule) eventually fire unproductive workers. If hiring costs are low, the dynamical system converges to a steady state in which the unemployment pool contains mostly low‐ability workers. However, if hiring costs are sufficiently large, this “lemons effect” would make firms unwilling to hire workers. In this case, the system converges to a cyclical equilibrium in which firms alternate between hiring and not hiring.

In recent years workplace smoking policies have become increasingly prevalent and restrictive. Using data from two large-scale national surveys, we investigate whether these policies reduce smoking. Our estimates suggest that workplace bans reduce smoking prevalence by 5 percentage points and daily consumption among smokers by 10 percent. Although workers with better health habits are more likely to work at firms with smoking bans, estimates from systems of equations indicate that these results are not subject to an omitted variables bias. The rapid increase in bans can explain all of the recent drop in smoking among workers relative to nonworkers.