Immigration and Wages — Digest
TL;DR
The immigration retrenchment has grown large enough to measure in the monthly CPS — the foreign-born population is down an estimated 2.9 million since January 2025 — and the new firm-level evidence on what happens when immigrant labor disappears keeps landing on the complementarity side. Meanwhile the H-1B fee was extended by proclamation even as it remains legally unenforceable, with a $103,265 successor rulemaking pending.
The low-skilled substitution question gets its cleanest test yet — and natives don't backfill
The central claim of restrictionist labor policy — that removing immigrant workers opens jobs and raises wages for natives — just took its most rigorous firm-level hit to date. Jongkwan Lee, Giovanni Peri, and Hee-Seung Yang exploit South Korea's COVID border closure, which shrank the guest-worker EPS program by about 22% from 2019 to 2021, and compare firms by their pre-existing reliance on that labor (Lee, Peri & Yang, NBER WP 34927):
"A sudden halt in South Korea's low-skilled guest worker program led to firm closures, production disruptions, and wage declines for native Korean workers." — NBER Digest via low-skilled-immigration-complementarity-sudden-stop
The detail that matters for the wages question: exposed firms did not expand domestic hiring, and natives at heavily exposed surviving firms saw wage declines consistent with occupational downgrading into tasks immigrants used to do (low-skilled-immigration-complementarity-sudden-stop). The design — policy-driven shock, pre-pandemic reliance as the identifying variation — is far cleaner than the cross-city comparisons that dominate this literature, though it's a working paper set in a quota-limited, manufacturing-concentrated program in an aging economy, so generalizing to U.S. construction or hospitality requires care.
What to watch: whether this paper survives peer review and whether U.S. enforcement shocks produce comparable firm-exit and wage-decline signatures in payroll data.
Enforcement as a natural experiment: Minneapolis shows work contracting, not shifting
The Minneapolis crackdown offers a live U.S. test of the same mechanism, and the results point the same direction. Aaron Sojourner (Upjohn Institute) and Aaron Rosenthal measured daily payroll data from small businesses in the Twin Cities against 49 comparison metros after roughly 2,000 federal agents arrived in January (Upjohn research highlight):
"In the weeks after the surge of DHS agents started, employment declined 2.8 percent, the number of operating businesses fell 1.7 percent, and the number of hours worked fell 1.9 percent." — Upjohn Institute via immigration-enforcement-native-employment-impacts
Sojourner puts the wage loss at over $100 million, with leisure and hospitality hit hardest. The adjudication caveats are real — the study is think-tank-affiliated and not peer-reviewed, and the sharpest single-day drop (−17% on January 23) coincided with an organized stay-home protest — but the key point survives them: total hours, employment, and operating businesses all fell. Enforcement did not shift work toward remaining workers (immigration-enforcement-native-employment-impacts).
What to watch: whether comparable payroll-based estimates emerge from other surge cities, which would turn a single-metro case study into a pattern.
The CPS retrenchment is real but smaller than the headline — and the data-source fight is now first-order
The scale of the foreign-born decline is now big enough to anchor the whole labor-market conversation. Steven Camarota and Karen Zeigler report (CIS report):
"Since January 2025 when President Trump took office the government's monthly household survey, officially called the Current Population Survey (CPS), shows an unprecedented 2.9 million decline in the foreign-born population." — CIS via unauthorized-immigration-local-labor-and-housing-markets
The report is unusually candid about its own uncertainty: the monthly CPS margin of error is ±1.038 million, so the decline could be as small as ~763,000 at interval extremes, and only 44% of the decline are workers — which, the authors note, "mitigates the impact on the labor market" and helps explain why payroll surveys don't show a dramatic drop (unauthorized-immigration-local-labor-and-housing-markets). Roughly 1.3 million immigrant workers leaving the counts with no matching payroll collapse is itself evidence for the complementarity pattern above. One trap to avoid: the report warns that a falling foreign-born count mechanically creates an apparent rise in the U.S.-born population via reweighting — don't read rising native CPS counts as native population growth.
What to watch: the 2025 ACS, due later this year, still only reflects the population through mid-2025 — ACS-based replication of the surge-and-retrenchment cycle waits on 2026 microdata.
The H-1B fee: extended on paper, unenforceable in court, and a confounded natural experiment
The high-skilled fee regime resolved the open thread from last cycle — Proclamation 10973's $100,000 payment requirement was extended for 12 months on September 18, 2026, effective the day it would have lapsed (high-skilled-immigration-h1b-lottery-and-firm-performance). But the extension "continues a payment requirement that is not currently enforceable": the District of Massachusetts vacated the fee's implementing policies in State of California v. Mullin on June 8, 2026, and the First Circuit denied the government's stay motion on July 24 (Manifest Law; AILA via high-skilled-immigration-h1b-lottery-and-firm-performance
).
The proclamation's own statistics — a 92% drop in registrations from the largest IT staffing and outsourcing firms, and master's-degree share of registrants rising from 45.1% to 66.1% — are offered as evidence the fee works, but they can't be attributed to it alone: the DHS weighted-selection rule, which independently rewards higher wages and degrees, took effect for the same FY2027 season, and the proclamation itself concedes the "combined effect" (high-skilled-immigration-h1b-lottery-and-firm-performance). Disentangling the two is the open empirical question for anyone hoping to use this as a natural experiment.
What to watch: the First Circuit's ruling on the underlying appeal, which could restore the fee, and the DHS $103,265 successor rulemaking (comment period closed September 24), which would move the policy onto legally more durable fee-rulemaking ground.
What surprised us
- Native wage effects ran negative in the Korea sudden stop. The standard restrictionist prediction is that removing immigrants raises native wages; exposed Korean natives instead saw declines via occupational downgrading — natives and guest workers were complements, not substitutes (low-skilled-immigration-complementarity-sudden-stop
).
- Firms wanted more guest workers, not fewer. Each unit decline in EPS employment was associated with demand for roughly two additional guest workers — the shortage created its own demand signal (low-skilled-immigration-complementarity-sudden-stop
- Only 44% of the foreign-born decline are workers. The rest are children, elderly, and non-workers — which is why the payroll survey shows no dramatic employment drop despite a headline decline of 2.9 million (unauthorized-immigration-local-labor-and-housing-markets
).
- The administration's own H-1B statistics are confounded by its other policy. The weighted-selection rule was quietly doing much of the compositional work the fee gets credit for — a reminder to read headline administrative claims against what else changed in the same season (high-skilled-immigration-h1b-lottery-and-firm-performance
).
Open threads worth a vote
- [watch] Expiration or Extension of the September 2026 H-1B Fee Extension Proclamation — expires September 21, 2027 absent further action, with a joint cabinet recommendation due within 30 days of the spring 2027 lottery.