Statistical Modeling of Structured Peer Group Dependence Arising from Salary Benchmarking Practices

While salary benchmarking is widely used to help set compensation, there has been little econometric modeling of peer-group identification and the impacts of benchmarking on the compensation decision. We adapt some empirical tools from spatial econometrics to analyze compensation decisions exhibiting peer-group dependence, and apply the methods to compensation of administrators in Texas nursing facilities. We find that average pay depends on compensation in other facilities having similar outlays on nursing services; there are statistically significant peer-group effects associated with the occupancy rate and the revenue received from Medicaid and from private-pay residents. Our peer-group model appears applicable to other areas of organizational, regulatory and behavioral research and can easily be implemented using publicly available software