Association for Public Policy Analysis & Management

This paper studies the effect of payroll subsidies on employment by workers of heterogeneous skill levels using evidence from the 2003 introduction and 2010 elimination of a set of large payroll subsidies affecting nursing homes in Maine. The amount of subsidies received by a nursing home was tied to the number of their residents on Medicaid, allowing this paper to identify the effect of the subsidy using variation in treatment intensity across firms. Estimates using control states constructed according to a modification of Doudchenko and Imbens’ generalized synthetic control estimator are also presented. I find that when the subsidies are first introduced, the increase in direct care worker staffing levels are 15 minutes per resident per day larger at the top of the Medicaid intensity interquartile range than at the bottom, a difference in excess of 6% of the mean Maine direct care worker staffing level. Strikingly, this effect is driven nearly half and half by low skill nursing assistants and high skill registered nurses, suggesting that nursing homes may have spent a disproportionately large amount of their subsidy dollars on high skill labor. A collection of payroll subsidy cuts in 2010 have opposite, though smaller, effects on staffing levels.