Working Paper: NBER ID: w30570
Authors: Zoe B. Cullen; Shengwu Li; Ricardo Perez-Truglia
Abstract: Firms are allowed to use aggregate data on market salaries to set pay, a practice known as salary benchmarking. Using national payroll data, we study firms that gain access to a tool that reveals market benchmarks for each job title. Using a difference-indifferences design, we find that the benchmark information reduces salary dispersion by 25%. Thus, salary dispersion must stem partly from aggregate uncertainty about the salaries offered by other firms. Our model formalizes how salary dispersion can arise even in competitive labor markets for identical workers when such uncertainty exists, and we discuss implications for an ongoing policy debate.
Keywords: salary benchmarking; employee compensation; labor markets; pay setting
JEL Codes: D83; J31; J38; M52
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| salary benchmarking (J31) | pay setting (J33) |
| salary benchmarking (J31) | salary dispersion (J31) |
| salary benchmarking (J31) | salary dispersion (low-skill positions) (J31) |
| salary benchmarking (J31) | salary dispersion (high-skill positions) (J31) |
| salary benchmarking (J31) | average salaries (low-skill positions) (J31) |
| salary benchmarking (J31) | retention rates (M51) |