Immigration and Wages — Digest
TL;DR
The biggest empirical event of the year for this topic landed: the first major academic study of the 2021–24 migration surge's labor-market effects finds no detectable native employment loss and modest positive native wage effects. Meanwhile, the H-1B fee regime's first year produced a striking revealed-preference datum — roughly 700 payments, effectively a hiring prohibition — and the data infrastructure itself took a hit, with the 2025 ACS release delayed by a Commerce disclosure-avoidance order.
The post-pandemic surge literature arrives, and the headline result is native wages up, employment null
The long-awaited first academic replication of the surge's local labor-market effects landed as a Brookings Papers on Economic Activity conference draft, and it cuts against the substitution story. Jennifer Hunt, Pia Orrenius, and Madeline Zavodny measure the surge (~6.5 million entrants outside usual legal pathways, 2021–2024) using EOIR immigration-court data — because the ACS undercounts surge migrants — and identify effects with a shift-share instrument plus a novel "buslift" instrument exploiting the Texas/Arizona busing of ~125,000 migrants to six cities (post-pandemic-surge-labor-housing-market-impacts):
"it raised U.S. natives' wages 0.9% over the period" — Hunt, Orrenius & Zavodny, BPEA conference draft via post-pandemic-surge-labor-housing-market-impacts
The native employment effect is a precisely-estimated zero, and the paper's most distinctive contribution is the adjustment mechanism: in the average metro, 90% of surge migration was offset by reduced net domestic in-migration, so effects diffused across geographies and became undetectable in cross-metro comparisons by 2023–24. The two other working papers in this nascent literature mostly agree — Clemens, Nice & Rigol find native wages +1.3% with a null employment effect — but Wilson & Zhou (Dallas Fed) estimate −0.7% native wages, significant only at the 10% level, and Hunt et al. attribute the sign flip to weighting, instrument construction, and metro-vs-commuting-zone choices. That dispute is now the main unresolved question in the literature (see the open thread below).
What to watch: revised versions, response papers, or a direct replication that pins down which specification choice drives the Wilson–Zhou divergence.
The H-1B fee's first year is a revealed-preference result: at $100K, demand collapsed
The high-skilled side produced its own clean behavioral datum this cycle. The September 18, 2026 extension proclamation revealed that employers made the $100,000 payment for only about 700 individuals, and USCIS receipts for initial consular H-1B filings fell 91.2% year-over-year (high-skilled-immigration-h1b-lottery-and-firm-performance). For labor economics, that's effectively a demand curve: at a $100K price, firms stopped hiring rather than absorbing the cost or passing it into wages.
The fee also remains legally dead on paper-alive: vacated by the District of Massachusetts on June 8, 2026, with the First Circuit denying a stay on July 24, yet extended by proclamation through September 21, 2027 (high-skilled-immigration-h1b-lottery-and-firm-performance). The successor DHS rule at $103,265 is where the economics fight now lives — and DHS's revenue math is under fire:
"would raise not $8.8 billion but only $2.6 billion, 70% less than DHS claimed" — Forbes (Stuart Anderson) via high-skilled-immigration-h1b-lottery-and-firm-performance
NFAP's deeper objection is methodological and should sound familiar: DHS extrapolated from small-fee elasticity estimates to a fee more than 100× larger, and justified the fee by citing Borjas's finding that H-1B workers earn 16% less than comparable natives — a contested claim, with NFAP noting the forthcoming DOL prevailing-wage rule would raise required salaries enough to eliminate the alleged underpayment. A contested study claim doing load-bearing work in a federal rule's economic justification is exactly the generalization-risk pattern this topic exists to flag.
What to watch: the First Circuit's ruling on the underlying appeal and whether the $103,265 rule (comments closed September 24) survives its APA vulnerabilities.
The data infrastructure is degrading at exactly the wrong moment
The 2025 ACS 1-year estimates — the first full post-retrenchment data year, and the release this topic's measurement program was counting on — are in limbo. Census stated on August 6 that the release date is "being determined" while it assesses the Commerce disclosure-avoidance order (DAO 216-26, June 2026), and The Census Project's September 4 letter attributes the delay directly to that order (acs-2025-release-delay-disclosure-avoidance):
"The release of the 2025 1-year ACS estimates is being delayed as a direct result of a new order that the Department of Commerce issued in June" — The Census Project letter to Commerce via acs-2025-release-delay-disclosure-avoidance
This compounds a problem the BPEA draft documents independently: the ACS already undercounts surge migrants because Census revised net international migration upward without revising ACS weights — Venezuela shows roughly 800,000 CBP entries against ~255,000 ACS entrants (post-pandemic-surge-labor-housing-market-impacts). Neither development changes any substantive finding, but both mean 2025-vintage ACS-based claims — in either political direction — deserve extra caution until the release lands with documented methodology.
What to watch: whether Census publishes the 2025 1-year tables at all this year, and with what disclosure-avoidance methodology.
What surprised us
- The surge's short-run native wage effect was positive and then decayed to zero (+2.3% in 2021–22, ~0 by 2023–24) — the opposite of the standard static prediction, and consistent with the complementarity pattern the Korea sudden-stop paper showed last cycle (post-pandemic-surge-labor-housing-market-impacts
).
- The buslift instrument's negative late-surge estimate shouldn't be generalized. The −2.6% native wage LATE for 2023–24 comes from what the authors concede is "essentially a case study of Denver" — a good example of an instrument's local nature being honestly disclosed (post-pandemic-surge-labor-housing-market-impacts
- Rents absorbed part of the wage gain. Native rents rose 1.8% per point of surge population, so the net-of-rent native wage gain shrinks to +0.6% — the housing channel is not incidental to the labor question.
- The surge can account for about one-fifth of 2021–24 real GDP growth in the 164 major metros — a macro-scale stake that rarely surfaces in the wages fight (post-pandemic-surge-labor-housing-market-impacts
Open threads worth a vote
- [question] Reconciling the Wilson-Zhou vs. Clemens/Hunt native-wage estimates for the post-pandemic surge — the sign of the native wage response to a ~2%-of-population supply shock is the live dispute; steer which replication attempts and response papers to prioritize.