Using matched employer‐employee data from the state of Georgia, this paper investigates how employment of undocumented workers varies along the business cycle and how it differs from the adjustment in employment of documented workers. The cyclical component of undocumented employment is found to be significantly more volatile than the cyclical component of documented employment. Simulation results indicate that complementarities between documented workers and capital account for almost 90% of the difference in measured volatility between documented and undocumented employment. (JEL E24, J21, J15)