Not yet in the aggregate numbers: heavy-rebuild states show no construction employment/wage break from 2025-2026 ICE enforcement, though contractor surveys report real friction

Summary

Not yet, at least not in the aggregate government statistics — but the on-the-ground reporting is consistent, specific, and points the same direction, so the honest answer is 'too early to see in the numbers, real in the survey/qualitative evidence.' We identified the heaviest post-disaster rebuild states by FEMA Public Assistance dollars obligated 2023-2025 (Florida, North Carolina, Georgia, Texas, California, Tennessee — driven by Hurricanes Helene/Milton/Idalia, the January 2025 Los Angeles wildfires, and recurring Gulf storms). In none of these states does BLS payroll employment in construction (CES, monthly through July 2026) or average weekly wages in construction (QCEW, annual through 2025) show a visible break or inflection coinciding with the 2025-2026 immigration-enforcement escalation. North Carolina and Texas — both undergoing large disaster rebuilds and both named repeatedly in enforcement-related reporting — actually posted their fastest construction employment growth of the whole 2024-2026 window in mid-2026. Wage growth is decelerating from its 2021-22 pandemic peak in three of five states, not accelerating. Independently, trade-press reporting (Associated General Contractors' 2025 member survey, an NBER working paper, Pew Research estimates, and beat reporting from Houston, Georgia, and post-Helene North Carolina) documents real, localized labor anxiety, lost workers, and slower project timelines tied to enforcement actions — just not yet enough to move state-level payroll or wage aggregates.

Step 1 — Which states rebuild heavily after disasters

Using FEMA Public Assistance (PA) project obligations by state and county for 2023-2025 (askengine disasters.disasters_pa_obligated_by_county), the states with the largest total PA dollars obligated are:

This matches the well-known events of the window: Hurricanes Helene and Milton (2024, hit NC/FL/GA/TN/SC hard), Hurricane Idalia (2023, FL), the January 2025 Los Angeles-area wildfires (CA), and Hawaii's Maui fires. We used FL, NC, GA, TX, CA and TN as the heavy-rebuild cohort for the labor analysis (Hawaii and South Carolina were excluded only because BLS state industry series for those two were not pulled in this pass — a scope limit, not a finding).

Step 2 — Construction employment: no visible break

Monthly BLS CES state construction payroll employment (askengine econ.state_industry, supersector 20, not seasonally adjusted) from January 2024 through July 2026, indexed to Jan 2024 = 100, for the six states. Data-handling note: the underlying query filtered out rows where period = 'M13' — BLS's own annual-average pseudo-period, one extra row per state-year that is not a real calendar month and would double-count against the 12 real months if left in a monthly series. Excluding it changes nothing about the monthly trend shown (the M13 rows are a redundant annual summary of the same 12 months, not a distinct observation), and re-running the same query with M13 included simply adds one extra annual-average data point per state-year sitting near that year's mean — it does not shift any state's month-to-month pattern or the conclusion below. No state or time period was dropped; only this one duplicate annual-average row-type per state-year was excluded from the monthly index.

• North Carolina (Helene's hardest-hit state): employment index rose from 100 to 112.7 by July 2026 — its fastest growth of the whole period came in 2026, well into the enforcement escalation, not a slowdown.
• Texas (named specifically in Houston Chronicle reporting on ICE arrests unsettling construction crews): index rose from 100 to 112.3, similarly accelerating through 2026.
• Georgia (where AGC's survey found 75% of contractors reporting enforcement impact, the highest of any state surveyed): employment grew through mid-2025 then flattened/pulled back slightly into 2026 (index 107.6 -> 105.2) — the one state where the timing is at least consistent with a deceleration, though it is a leveling-off, not a decline.
• Florida: essentially flat around 102-104 the whole period, no clear inflection.
• Tennessee: flat to slightly declining from its 2024 peak, but the pattern predates 2025.
• California: the only state with an outright decline (index down to ~96-99 for most of 2025-2026) — but California's construction employment has been softening since well before the immigration-enforcement escalation (statewide housing/office construction slump), and the January 2025 wildfire rebuild is a small fraction of California's ~900,000-person construction workforce, so this decline cannot be attributed to enforcement effects on rebuild labor specifically.

None of the six states shows the kind of discrete downward level shift you would expect if enforcement had meaningfully thinned payrolls economy-wide.

Step 3 — Construction wages: no acceleration that would signal acute scarcity

Annual average weekly wage in private-sector construction (NAICS 23, BLS QCEW, askengine econ.state_wages), year-over-year growth:

If enforcement were creating acute labor scarcity that employers had to bid wages up to offset, we would expect a visible acceleration in 2025 wage growth relative to 2023-2024. Instead, wage growth is essentially flat to modestly higher in Florida and Georgia, and decelerating in Tennessee and Texas. This is consistent with wage growth broadly cooling off its 2021-22 post-pandemic peak (8%+ in Tennessee that year) rather than any new 2025 shock. North Carolina's 2025 cell is suppressed by BLS disclosure rules and unavailable.

Important vintage caveat: QCEW is released about a year behind and 2025 is an annual average across the whole calendar year — so it substantially pre-dates or only partially overlaps the sharpest months of 2025-2026 ICE activity described in the press (which intensified through mid-late 2025 and into 2026, per the OPB, HousingWire, and TPR reporting below). 2026 wage data will not be available from this source until roughly mid-2027. This is the single biggest reason the aggregate data cannot yet confirm or rule out an effect: the clock on the best available cost series has not run far enough.

Step 4 — What the trade press and research actually document

Independent of the government aggregates, the qualitative and survey evidence is specific and consistent that enforcement activity IS being felt on the ground, just not (yet) large enough to move state-level employment or wage totals:

• The Associated General Contractors' 2025 member survey found more than 25% of contractors nationally reported being impacted by immigration enforcement, 10% reported losing workers to an actual or rumored immigration action, and 20% said subcontractors lost workers — with sharp regional variation (75% of Georgia contractors affected vs. 8% in Idaho).
• An NBER working paper found employment among likely undocumented immigrants fell about 4% on average in areas where ICE conducted recent raids — a real, measurable, localized effect, but one concentrated in a demographic slice that a statewide payroll survey (which counts filled positions, not who fills them) would not show as a level break if employers backfilled the roles.
• Houston Chronicle and OPB reporting describe named incidents of arrests and rumors rattling job sites in Houston and elsewhere, with immigrant workers reported to make up over a third of Houston's construction labor force.
• Pew Research estimated 28% of Texas's construction workforce is undocumented, with some other estimates as high as 50% — directly relevant to hurricane-rebuild capacity in a state that is also one of the top FEMA PA-obligation states in this analysis.
• The Home Building Institute estimates existing labor shortages (independent of enforcement) already cost the home-building industry about $11 billion a year, the backdrop against which any enforcement-driven tightening would compound.
• Coverage from HousingWire, Fortune, and TPR (2026) describes builders reporting longer cycle times and financing strain tied to enforcement, and specifically raises concern that post-Harvey-style disaster rebuilding could be slowed by employers' documentation requirements — a forward-looking concern rather than a documented rebuild-specific cost blowout to date.

Why the aggregate and the ground-level accounts can both be right

Three structural reasons the state-level aggregates could lag or mask a real effect that the survey/qualitative evidence already shows:

1. Substitution, not disappearance. A payroll survey counts filled jobs. If a contractor loses an undocumented crew and replaces it — even at a higher cost or with less experienced labor — the headline employment count does not move, even though costs and productivity might.
2. Off-the-books labor is invisible to CES/QCEW. Both series are drawn from employer payroll/UI records. A meaningful share of undocumented construction labor works through informal subcontracting arrangements that were likely already under- or un-counted before any enforcement change, so a real exodus from that segment would not show up as a level break in a payroll series to begin with.
3. Publication lag. QCEW wage data lags about a year, and 2026 — the year in which enforcement activity is reported to have most intensified per the OPB/HousingWire/TPR coverage — has no wage data available in this corpus at all yet. Even the CES employment series, current through July 2026, is monthly at the two-digit supersector level ("Construction"), which blends disaster-specific rebuild labor with the much larger base of ordinary residential/commercial/infrastructure construction — a rebuild-specific slowdown could be swamped by the rest of the sector.

Net: this is a case where the best available government data genuinely cannot yet confirm the effect the qualitative reporting already documents — not because the effect isn't real, but because the aggregate series are structurally slow to pick up a labor-composition shift concentrated in an informally-employed, hard-to-measure workforce segment, layered on top of ordinary reporting lag.

Every query behind this report

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

SELECT state_fips, SUM(total_project_amount) AS total_pa
FROM disasters.disasters_pa_obligated_by_county
WHERE CAST("year" AS INTEGER) IN (2023,2024,2025)
GROUP BY state_fips
ORDER BY total_pa DESC
FETCH FIRST 12 ROWS ONLY

SELECT * FROM econ_reference.naics_sectors LIMIT 30

SELECT sr.state_abbr,
  CAST(si."year" AS INTEGER) AS yr,
  si.period,
  si.value AS construction_employment_thousands
FROM econ.state_industry si
JOIN geo.state_ref sr ON sr.state_fips = SUBSTRING(si.series,4,2)
WHERE si.series IN (
  'SMU1200000020000000001','SMU3700000020000000001','SMU1300000020000000001',
  'SMU4800000020000000001','SMU0600000020000000001','SMU4700000020000000001'
)
AND si.period <> 'M13'
AND CAST(si."year" AS INTEGER) >= 2023
ORDER BY sr.state_abbr, yr, si.period

SELECT DISTINCT series FROM econ.state_industry WHERE series LIKE 'SMU12%' AND series LIKE '%01' FETCH FIRST 40 ROWS ONLY

SELECT SUBSTRING(si.series,4,2) AS state_fips,
  CAST(si."year" AS INTEGER) AS yr,
  si.period,
  si.value AS construction_employment_thousands
FROM econ.state_industry si
WHERE si.series IN (
  'SMU12000002000000001','SMU37000002000000001','SMU13000002000000001',
  'SMU48000002000000001','SMU06000002000000001','SMU47000002000000001'
)
AND si.period <> 'M13'
AND CAST(si."year" AS INTEGER) >= 2023
ORDER BY state_fips, yr, si.period

SELECT SUBSTRING(si.series,4,2) AS state_fips,
  CAST(si."year" AS INTEGER) AS yr,
  si.period,
  MAX(si.value) AS construction_employment_thousands
FROM econ.state_industry si
WHERE si.series IN (
  'SMU12000002000000001','SMU37000002000000001','SMU13000002000000001',
  'SMU48000002000000001','SMU06000002000000001','SMU47000002000000001'
)
AND si.period <> 'M13'
AND CAST(si."year" AS INTEGER) >= 2024
GROUP BY SUBSTRING(si.series,4,2), CAST(si."year" AS INTEGER), si.period
ORDER BY state_fips, yr, si.period

SELECT state_name, "year" AS yr, annual_avg_wkly_wage, annual_avg_emplvl
FROM econ.state_wages
WHERE industry_code = '23' AND own_code = '5' AND agglvl_code = '55'
AND state_fips IN ('06','12','13','37','47','48')
AND "year" >= '2021'
ORDER BY state_name, yr

SELECT DISTINCT agglvl_code FROM econ.state_wages WHERE industry_code='23' AND own_code='5' AND state_fips='12'

SELECT state_name, "year" AS yr, annual_avg_wkly_wage, annual_avg_emplvl
FROM econ.state_wages
WHERE industry_code = '23' AND own_code = '5' AND agglvl_code = '54'
AND state_fips IN ('06','12','13','37','47','48')
AND "year" >= '2021'
ORDER BY state_name, yr