| Summary: | We use payroll data on over 1 million workers at 80,000 small firms to construct county-month measures of employment, hours, and wages that correct for dynamic changes in sample composition in response to business cycle fluctuations. We use this to estimate the response of small firms' employment, hours and wages following tighter local labor market conditions. We find that employment and hours per worker fall and wages rise. This is consistent with the predictions of the response to a demand shock in the well-known “jobs ladder” model of labor markets. To check this interpretation, we show our results hold when instrumenting for local demand using county-level Department of Defense contract spending. Correction for dynamic sample bias is important -- without it, the hours fall by only one third as much and wages increase by double
|
|---|