Working Paper: NBER ID: w2647
Authors: Shalumit Kahn; Kevin Lang
Abstract: Almost all labor supply models are estimated under the assumption that workers are free to choose their hours. However, theory, casual empiricism and survey data suggest that many workers are not free to vary the hours within a job. Consequently, labor supply estimates based on actual hours of work may be biased. Using Canadian data on desired hours of work, we find that using actual hours causes labor supply estimates to be biased upwards but the bias is small.
Keywords: labor supply; hours constraints; wage elasticity; desired hours; actual hours
JEL Codes: J22; J23
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| lower-wage workers tend to desire more hours than higher-wage workers (J38) | positive correlation between actual hours worked and wages (J31) |
| bias from using actual hours (J38) | modest and consistent with previous literature (C20) |
| using actual hours instead of desired hours (J22) | upward bias in labor supply estimates (J20) |