Why the incidence of shirking varies across employers

Existing models of shirking are not consistent with the underlying behavior of employers and employees. In these models, either shirking does not occur in equilibrium or it may occur but the offending employee is immediately dismissed. These conclusions conflict with evidence that shirking is sometimes ignored by firms for many years. To resolve this conflict, the author models the interaction between employer and employees as a finite repeated game, and introduces the quasi-fixed costs of dismissing and replacing an employee. Shirking is distinguished from more serious malfeasant behavior by the criterion of imposing current period costs on the employer that are less than the cost of dismissing and replacing a shirking employee. When both the employer and the employees are rational and there are no informational asymmetries, shirking is not deterred by threat of dismissal. Introducing informational asymmetries leads to an equilibrium where shirking is deterred, to a great extent, by the dismissal threat. Comparative static analysis of the model yields testable implications on the incidence of shirking across firms, and contributes to our understanding of other labor market issues.