ICE Enforcement Surge Has Cut Local Employment for Both Immigrants and U.S.-Born Workers — Not Raised Wages

Summary

Yes, but not the shift the deportation-frees-up-jobs theory predicts. The first national, quasi-experimental study of the 2025 ICE surge (NBER Working Paper 35129, Cox & East, revised June 2026) compares 58 metro areas that saw a sudden spike in ICE arrests against otherwise-similar areas that did not, using monthly Current Population Survey data January–October 2025. In surge areas: employment among likely-undocumented immigrants who remained fell 1.3 percentage points (-1.4%) overall and 1.8 points (-2%) in high-immigrant-share sectors (agriculture, construction, manufacturing, wholesale, leisure/hospitality) — a “chilling effect” on people who were never arrested. Employment of U.S.-born men with a high-school degree or less also fell, by about 1.3 points nationally and roughly 3% in construction specifically — twice the average effect. Critically, the study finds no evidence that wages rose for U.S.-born workers to attract them into jobs vacated by immigrants; if anything the point estimate on weekly earnings is slightly negative. There is essentially no job-openings surge and no wage premium — the labor market in affected industries is shrinking, not reallocating.

The primary evidence: NBER Working Paper 35129

Economists Elizabeth Cox and Chloe East (University of Colorado Boulder) built the first national causal estimate by comparing regions with a sudden, large increase in monthly ICE arrests (Jan–Oct 2025) against control regions with no such spike, using a staggered difference-in-differences design on CPS microdata (~59,000 households/month) plus geocoded ICE arrest counts. Key figures, read directly from the paper's Results and Magnitudes sections:

• First stage: treated areas saw roughly 200 more daily arrests/month than control areas — a 114% increase over the pre-period baseline arrest rate.
• Immigrant employment: employment rate of likely-undocumented immigrants fell 1.3 pp (1.4%) overall, 1.8 pp (2%) in high-impact sectors; effects concentrated among men (-2.8 pp overall, -3.5 pp in high-impact sectors); effects on their pay were statistically insignificant (Appendix Table A1).
• Sector heterogeneity: the largest negative effect on immigrant employment is in construction (13% undocumented share); effects are smaller in transportation, finance and information (each <5% undocumented).
• U.S.-born workers: null effect on employment for the full sample (rules out gains larger than 0.3 pp); negative and significant for men with a high-school degree or less, especially in construction (~3% employment-rate decline, twice the average effect) and in non-tradable, immigrant-reliant sectors (leisure/hospitality, other services). Weekly earnings effects are negative, not positive, and hourly wage effects for U.S.-born workers are statistically null (Appendix Table A1).
• Magnitude: in an average treated area with 1,200 ICE arrests, the study implies about 7,574 fewer undocumented immigrants and roughly 1,200 fewer U.S.-born men with a high school degree or less employed — i.e., for every six male undocumented workers lost, about one U.S.-born male job is also lost, not gained. This chilling-effect ratio (7 immigrant workers stopped working per arrest) is larger than in the Obama-era enforcement surges studied by East et al. (2023), where the ratio was 2.3.
• Placebo test using 2015-2017 CPS data around the first Trump inauguration shows no similar effects, supporting a causal (not merely correlational) reading.

The paper explicitly notes it captures local, cross-area effects and does not extrapolate to a national aggregate, and that it cannot fully distinguish workers voluntarily staying home from employers cutting production — both plausibly operate together.

Cross-check: BLS QCEW wages and employment in four large enforcement-active counties

Because NBER w35129's CPS-based estimates end in October 2025 and use metro-area anonymized regions rather than named counties, we independently pulled BLS Quarterly Census of Employment and Wages (QCEW) county-level annual data (askamerica econ.county_qcew, private ownership) for four counties that saw well-documented, high-profile 2025 ICE operations — Los Angeles County CA, Cook County IL (Chicago), Mecklenburg County NC (Charlotte), and Harris County TX (Houston) — in the three industries the NBER paper flags as most exposed: agriculture (NAICS 11), construction (NAICS 23), and food services (NAICS 722).

This is a simple year-over-year comparison, not a causal design — it has no control group and is confounded by each metro's broader 2025 economic conditions (including 2025 tariff effects, which the NBER paper controls for and we cannot here). Read as description, not attribution:

• Employment fell in agriculture and construction in Los Angeles and Chicago (agriculture down 7.8% and 8.0%; LA construction down 3.3%) — consistent with the chilling-effect/reduced-production story for the two metros with the most intense, most publicized 2025 ICE activity. Charlotte and Houston, by contrast, show employment growth in the same industries, illustrating that local demand and industry mix can swamp an enforcement effect at this coarse, non-causal grain.
• Nowhere do wages show the sharp acceleration a labor-shortage story would predict. Nominal average weekly wage growth across the three industries ranged from -1.3% to +8.3%, but CPI-U rose about 2.6% over the same period, so most of these gains are modest-to-flat in real terms — matching the NBER paper's finding of no significant wage response for either immigrant or U.S.-born workers.

Job-openings data at the state-by-industry grain (BLS JOLTS) does not yet cover 2025 in this corpus (econ.jolts_state and econ.jolts_industry are complete only through 2024), so we could not independently verify a job-openings shift; this is a genuine data-vintage gap, not a finding of “no change.”

Corroborating context

Independent reporting and a regional case study point the same direction:

• Minneapolis officials estimated the city's economy lost about $203 million in February 2026 from falling restaurant, hotel and retail revenue plus lost wages tied to enforcement-driven pullback — a demand-side channel consistent with East's finding that employers cut production/hours rather than raise pay.
• An Oxnard, California case study (agriculture-heavy Ventura County) estimated a 20–40% reduction in the local agricultural workforce following 2025 ICE raids, with associated crop losses and produce-price increases — a supply-shock story for one specific, highly exposed local labor market, though this is a single-region estimate, not a national one.

Bottom line and caveats

Measurable shift: yes, but it is a contraction, not a reallocation. Areas with concentrated 2025-26 ICE enforcement activity show a real, statistically significant drop in employment among both immigrant workers who remained and U.S.-born men in the same high-exposure industries (construction most of all), with no offsetting wage increase for U.S.-born workers and, if anything, a slight negative pull on earnings. The “deportations free up jobs for Americans at better pay” hypothesis is not supported by the best available national causal estimate.

Caveats: (1) NBER w35129 is a working paper, not yet peer-reviewed. (2) Its window ends October 2025; enforcement intensity and adaptation could look different a year later. (3) Our own QCEW county comparison is illustrative and non-causal, included to extend the paper's window into 2025 annual data and check directional consistency, not as independent proof. (4) National job-openings-by-industry data for 2025 was not available in this corpus at query time.

What This Report Does Not Answer

• Job openings: BLS JOLTS by state/industry (the standard job-openings series) is only loaded through 2024 in this corpus, so 2025-26 job-openings-rate figures could not be pulled directly; we substituted QCEW employment-level changes (hires/separations proxy, not openings) and noted the gap explicitly rather than presenting it as a null finding.

Every query behind this report

2 warehouse calls ran in this session, in order. Each is reproducible against the same snapshot.

SELECT area_fips, industry_code, "year", annual_avg_emplvl, annual_avg_wkly_wage, disclosure_code
FROM econ.county_qcew
WHERE area_fips IN ('06037','17031','37119','48201') AND own_code='5'
AND industry_code IN ('10','11','23','722') AND agglvl_code = '78'
AND "year" IN ('2023','2024','2025')
ORDER BY area_fips, industry_code, "year"

SELECT area_fips, industry_code, agglvl_code, "year", annual_avg_emplvl, annual_avg_wkly_wage, disclosure_code
FROM econ.county_qcew
WHERE area_fips IN ('06037','17031','37119','48201') AND own_code='5'
AND industry_code IN ('10','11','23','722')
AND "year" IN ('2023','2024','2025')
ORDER BY area_fips, industry_code, "year"
LIMIT 100