Overview
This series explains why H-1B wage-gap estimates change when the comparison changes. It traces the LCA, I-129, and ACS evidence chain; separates raw pay differences from conditional associations, legal wage floors, and policy models; then examines worker and employer heterogeneity and visa-fee simulations. The evidence describes selected, cap-subject, for-profit new hires, not every H-1B worker or an individual career. Conflicting NBER and EIG estimates are best read as design-sensitive findings, not averaged into one verdict.
Learning outcomes
- An H-1B wage gap changes with the baseline, wage clock, geography, and institutional constraint.
- The revised NBER estimate is a conditional association for selected cap-subject, for-profit new hires, not a universal worker-level finding.
- The H-1B legal required wage is employer- and place-specific, so a regression gap does not prove underpayment.
- Visa-fee revenue and demand effects are model outputs shaped by hidden costs, excess demand, exemptions, and worker heterogeneity.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
Why can H-1B workers earn more on average but less in an adjusted comparison?
They can because the broad comparison reflects concentration in high-paying jobs and places, while the adjusted estimate compares petition wages with a constructed native baseline after selected controls.
Does a negative H-1B wage gap prove an employer violated wage rules?
No. The legal test compares an employer’s actual and prevailing wage for a particular job and place, not the regression’s population baseline.
Why do Borjas and EIG report different H-1B wage gaps?
They align petition and survey wages differently and use different geographic controls and sample filters; the supplied material does not independently settle which construction is preferable.
Can a $100,000 H-1B fee reduce visa use or change who gets hired?
It could change who employers hire and may reduce some demand, while excess demand can keep visa use near the cap in modeled scenarios. The effects depend on hidden costs, registrant behavior, and in-country exemptions.