Agricultural lending is concentrated in a small set of fruit/vegetable/dairy counties with the heaviest reliance on immigrant labor — a labor shock there is a loan-book shock, not a diversified risk

Summary

Agricultural lending exposure to a sudden farm-labor shortfall is highly concentrated, not diversified: the crops that depend most on immigrant (and disproportionately unauthorized) labor — fruit, vegetables, horticultural/nursery products, and dairy — are grown in a small number of counties that carry an outsized share of national crop-insurance liability, a standard proxy for the collateral value lenders are underwriting against. Ten counties, led by Fresno, Tulare, and Kern (CA), Grant and Yakima (WA), and Miami-Dade and De Soto (FL), account for a composite exposure score far above the national median, with 60-100% of their crop-insurance liability sitting in labor-intensive commodities that cannot be mechanically harvested at scale today. Nationally, labor-intensive crops carry $16.9B of $158.7B in total 2024 RMA liability (10.6%) but that liability is not spread evenly — it clusters in exactly the counties documented (NAWS, CRS, AEI) as most reliant on unauthorized and H-2A labor. A further, genuinely surprising finding from this corpus: FDIC's own "Agricultural Specialization" bank designation is concentrated almost entirely in Midwest/Plains commodity-grain counties, not in these labor-intensive counties — only 1 of 914 FDIC ag-specialized banks nationwide is headquartered in the top 25 exposed counties identified here. That means the lenders actually underwriting labor-intensive agriculture are predominantly the Farm Credit System (not covered in this corpus) and large regional/international commercial banks, not the small rural ag-specialty banks that dominate coverage of farm lending generally.

Why 'immigrant-dependent farm labor' cannot be measured directly here

No table in this corpus carries farmworker nativity or immigration status — confirmed via a catalog search where 'immigrant' matched nothing in any schema, table, or column description. This is a genuine data gap, not an oversight in this analysis: the only federal survey that asks farmworkers their legal status, USDA's National Agricultural Workers Survey (NAWS), is not warehoused here and is not published at county grain even by USDA itself (results are pooled into multi-state regions for confidentiality).

The published NAWS and CRS findings that anchor the proxy used below: in the 2020-22 NAWS wave, 42% of hired U.S. crop farmworkers held no work authorization, 19% were other authorized immigrants (mostly green-card holders), 7% were naturalized citizens, and only 32% were U.S.-born (USDA ERS Farm Labor; CRS IF12979). Unauthorized immigrants made up an estimated 4.8% of the overall U.S. workforce but over a third of crop workers (AEI, Charlton 2025). Critically, this dependence is not spread evenly across agriculture: it concentrates in fruit, vegetable, and horticultural (FVH) crops and in dairy — sectors requiring delicate, time-sensitive hand labor that has proven far harder to mechanize than grain and oilseed harvesting, which is already almost fully automated and relies on relatively little hired seasonal labor. NAWS itself excludes dairy, poultry, and other livestock workers, but a Texas A&M study found roughly half of the U.S. dairy workforce (76,968 of 150,418 workers in 2013) were immigrants, and dairy/poultry operations are categorically ineligible for the H-2A guest-worker visa, pushing them toward undocumented labor specifically.

The proxy used: crop-insurance liability in labor-intensive commodities

Since no direct labor-dependence table exists at county grain, this analysis uses ag.rma_crop_insurance — USDA RMA's Summary of Business, which reports insured liability (essentially the insured value of a county's crop production) by county and commodity for 2024, the most recent complete year. A commodity list of ~55 labor-intensive crop/tree/nursery categories (fruits, tree nuts, vegetables, berries, nursery/greenhouse) was built from the literature above and summed by county, then compared to each county's total RMA liability across all commodities to compute a labor-intensive-crop share.

Crop insurance liability is a reasonable, widely used proxy for agricultural lending exposure specifically because agricultural lenders (both Farm Credit System associations and commercial banks) routinely require or strongly favor borrowers carrying federal crop insurance as a condition of operating loans — RMA's own liability figure approximates the insured production value that underlies loan collateral and repayment capacity in a given county. It is not itself a loan-book figure (see limitations below).

A composite 0-1 exposure score was built per the standard normalize-then-combine method: for each county, min-max normalize (a) the absolute dollar liability in labor-intensive crops and (b) the labor-intensive share of that county's total liability, then average the two equally weighted components, so a county has to be both large in absolute labor-intensive collateral value and structurally dependent on it (not just large overall) to rank at the top.

These are, independently, exactly the geographies the immigration/farm-labor literature names as most exposed: California's Central Valley (Fresno, Tulare, Kern — the nation's top fruit/vegetable/nut producing counties, per American Immigration Council, which notes California produces 40%+ of U.S. fruits/vegetables/nuts and 'relies the most heavily on undocumented labor'); central Washington's Yakima and Columbia Basin (Grant, Yakima — apple/hop/vegetable country with among the highest H-2A certification counts nationally); and Florida's citrus/vegetable belt (Miami-Dade, De Soto, Polk, Highlands, Hendry, Hardee — Florida had the most H-2A-certified workers of any state, 51,000 in 2022, per the American Immigration Council).

Who actually holds the exposure: a genuine finding

Cross-referencing the top-25 exposed counties above against FDIC's own bank-specialization classification (banking.institutions.specialization_group = 'Agricultural Specialization') surfaces a real structural mismatch: nationally there are 914 active FDIC-insured banks classified as agriculturally specialized, holding $314.2B in combined assets — but only one is headquartered in any of the top 25 labor-intensive-exposed counties (a single $5.8B bank in San Joaquin County, CA). FDIC's ag-specialty designation is concentrated in the commodity-grain Midwest and Plains, where farming is heavily mechanized and hired seasonal/immigrant labor is comparatively marginal, not in the FVH/dairy counties that actually carry the labor-shock risk.

Total FDIC-insured bank assets headquartered in the top exposed counties (all specializations, not just ag) do exist — e.g., $119B in Miami-Dade, $71.5B in Polk County, FL — but these are large metro/regional bank totals dominated by non-agricultural commercial and international banking, not a measure of farm-loan exposure; this figure is reported here only to show that the lending capacity physically present in these counties is not itself concentrated in agriculture. The practical implication, consistent with the farm-finance literature (USDA ERS): the Farm Credit System — which holds roughly 46% of national farm debt (about $269B in 2024, split $188B real estate / $81B non-real-estate) and is not represented in this corpus — is very likely the dominant lender actually carrying labor-shock risk in these specific counties, alongside a handful of large regional commercial banks, rather than the small rural ag-specialty community banks that anchor most farm-lending coverage nationally.

The 2025 labor shock, in the literature

This is not a hypothetical scenario as of this writing. Multiple 2025 accounts document an actual, ongoing labor-supply shock from intensified interior immigration enforcement: national agricultural employment fell 6.5% between March and July 2025, some farms reported operating with up to 60% fewer laborers due to worker no-shows, and California farmers reported as many as 70% of workers missing in the wake of immigration raids (Fortune, Oct 2025; Christian Science Monitor, Oct 2025). The U.S. Labor Department itself has acknowledged the shortage risks food-supply disruption and higher production costs (Pro Farmer). The mechanism into lending risk is direct and well established in agricultural-economics literature even before 2025: county-level 287(g) immigration-enforcement agreements in the 2000s measurably reduced vegetable acreage, dairy production value, and the number of dairies in affected counties, while Arizona's 2008 mandatory E-Verify law reduced vegetable and melon acreage without any offsetting increase in mechanization sufficient to compensate (AEI, Charlton 2025, citing peer-reviewed studies of both policies). Timothy Richards' modeling cited in that same report estimates that removing all unauthorized immigrants from California's agricultural sector would raise farm wages roughly 42%, a margin compression large enough to push many operations out of business — the loan-performance channel a lender would watch directly. AgAmerica's own 2025 year-in-review for agricultural lenders makes the credit-risk channel explicit: labor predictability and cost are factors lenders weigh directly in approving and pricing farm loans, and labor shortages destabilize exactly those factors (AgAmerica, 2025 Year in Review).

Bottom line and limitations

Agricultural lending exposure to a farm-labor shock is geographically concentrated in a short list of Central Valley CA, central WA, and Florida counties where 60-100% of crop-insurance-backed production value sits in fruit, vegetable, and horticultural commodities that are the same commodities the labor-economics and immigration literature identifies as most dependent on immigrant — and substantially unauthorized — labor. That concentration means the financial risk from a sudden labor-supply disruption (deportation surge, enforcement crackdown, or a visa-program disruption) is not diversified across the farm-lending system; it falls disproportionately on whichever lenders — chiefly Farm Credit System associations and a small number of large regional banks, based on this corpus's bank data — hold agricultural paper in those specific counties, not on the broad population of small ag-specialty community banks that dominate national farm-lending coverage.

Limitations, stated plainly: (1) No table anywhere in this corpus carries farmworker immigration status or a county-level immigrant-labor-share figure; the ranking above is a crop-mix proxy validated against, not derived from, an immigrant-labor dataset. (2) Crop insurance liability approximates insured production value, not actual outstanding farm debt; no Farm Credit System data (the single largest ag lender, ~46% of the market) exists in this corpus at all, so the loan-exposure side of this analysis is necessarily incomplete without it. (3) Dairy — one of the sectors the literature flags as most dependent on undocumented labor — is under-represented here because RMA crop insurance does not cover most dairy production; this likely understates exposure in dairy-heavy counties (e.g., parts of WI, NY, CA's Central Valley, and Aroostook County, ME, which appears in the wider ranking on potato/produce liability but not dairy). (4) FDIC bank-asset totals by county reflect where institutions are headquartered, not where their agricultural loan books are concentrated, and can be dominated by unrelated commercial banking activity (Miami-Dade's $119B figure is a clear example).

Every query behind this report

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

SELECT DISTINCT statisticcat_desc, unit_desc FROM ag.nass_crop_production WHERE commodity_desc IN ('VEGETABLES','FRUIT & TREE NUTS','BERRIES') LIMIT 30

SELECT DISTINCT commodity_name FROM ag.rma_crop_insurance WHERE "year" = 2024 ORDER BY commodity_name

WITH labor_intensive AS (
  SELECT county_fips, county_name, state_abbr, SUM(liability_amount) AS li_liability
  FROM ag.rma_crop_insurance
  WHERE "year" = 2024
    AND commodity_name IN (
      'Apples','Avocados','Avocado Trees','Cherries','Grapes','Table Grapes','Raisins',
      'Fresh Freestone Peaches','Processing Cling Peaches','Processing Freestone','Peaches',
      'Pears','Plums','Prunes','Strawberries','Blueberries','Caneberries','Cranberries','Figs',
      'Kiwifruit','Olives','Oranges','Orange Trees','Grapefruit','Grapefruit Trees','Lemons','Lemon Trees',
      'Mandarins/Tangerines','Mandarin/Tangerine Trees','Tangelos','Tangerine Trees','Tangors',
      'Pomegranates','Papaya','Papaya Tree','Mango Trees','Macadamia Trees','Pecans','Pistachios','Walnuts',
      'Cabbage','Cucumbers','Chile Peppers','Peppers','Onions','Potatoes','Sweet Potatoes',
      'Fresh Market Tomatoes','Tomatoes','Pumpkins','Fresh Market Sweet Corn','Sweet Corn',
      'Fresh Market Beans','Processing Beans','Green Peas','Nursery (FG&C)','Nursery (NVS)','Fresh Apricots','Processing Apricots'
    )
  GROUP BY county_fips, county_name, state_abbr
),
total AS (
  SELECT county_fips, SUM(liability_amount) AS total_liability
  FROM ag.rma_crop_insurance
  WHERE "year" = 2024
  GROUP BY county_fips
)
SELECT l.county_fips, l.county_name, l.state_abbr, l.li_liability, t.total_liability,
       l.li_liability / NULLIF(t.total_liability,0) AS li_share
FROM labor_intensive l
JOIN total t ON l.county_fips = t.county_fips
WHERE t.total_liability > 5000000
ORDER BY l.li_liability DESC
FETCH FIRST 25 ROWS ONLY

SELECT
  SUM(CASE WHEN commodity_name IN (
      'Apples','Avocados','Avocado Trees','Cherries','Grapes','Table Grapes','Raisins',
      'Fresh Freestone Peaches','Processing Cling Peaches','Processing Freestone','Peaches',
      'Pears','Plums','Prunes','Strawberries','Blueberries','Caneberries','Cranberries','Figs',
      'Kiwifruit','Olives','Oranges','Orange Trees','Grapefruit','Grapefruit Trees','Lemons','Lemon Trees',
      'Mandarins/Tangerines','Mandarin/Tangerine Trees','Tangelos','Tangerine Trees','Tangors',
      'Pomegranates','Papaya','Papaya Tree','Mango Trees','Macadamia Trees','Pecans','Pistachios','Walnuts',
      'Cabbage','Cucumbers','Chile Peppers','Peppers','Onions','Potatoes','Sweet Potatoes',
      'Fresh Market Tomatoes','Tomatoes','Pumpkins','Fresh Market Sweet Corn','Sweet Corn',
      'Fresh Market Beans','Processing Beans','Green Peas','Nursery (FG&C)','Nursery (NVS)','Fresh Apricots','Processing Apricots'
    ) THEN liability_amount ELSE 0 END) AS li_liability_national,
  SUM(liability_amount) AS total_liability_national
FROM ag.rma_crop_insurance WHERE "year" = 2024

SELECT DISTINCT specialization_group FROM banking.institutions

SELECT i.county_fips, COUNT(*) AS n_ag_banks, SUM(i.total_assets_thousands)/1000.0 AS total_assets_millions
FROM banking.institutions i
WHERE i.active = true AND i.specialization_group ILIKE '%agricultur%'
  AND i.county_fips IN ('06019','06029','06107','53025','53077','12086','06055','06077','12027','06097','12105','06031','12055','53005','06111','06039','53021','12051','12049','06113','53047','06047','48215','53071','12021')
GROUP BY i.county_fips
ORDER BY total_assets_millions DESC

SELECT COUNT(*) AS n_ag_banks_national, SUM(total_assets_thousands)/1000.0 AS total_assets_millions_national
FROM banking.institutions WHERE active = true AND specialization_group ILIKE '%agricultur%'

SELECT county_fips, COUNT(*) AS n_banks, SUM(total_assets_thousands)/1000.0 AS total_assets_millions
FROM banking.institutions
WHERE active = true
  AND county_fips IN ('06019','06029','06107','53025','53077','12086','06055','06077','12027','06097','12105','06031','12055','53005','06111','06039','53021','12051','12049','06113','53047','06047','48215','53071','12021')
GROUP BY county_fips
ORDER BY total_assets_millions DESC

WITH labor_intensive AS (
  SELECT county_fips, county_name, state_abbr, SUM(liability_amount) AS li_liability
  FROM ag.rma_crop_insurance
  WHERE "year" = 2024
    AND commodity_name IN (
      'Apples','Avocados','Avocado Trees','Cherries','Grapes','Table Grapes','Raisins',
      'Fresh Freestone Peaches','Processing Cling Peaches','Processing Freestone','Peaches',
      'Pears','Plums','Prunes','Strawberries','Blueberries','Caneberries','Cranberries','Figs',
      'Kiwifruit','Olives','Oranges','Orange Trees','Grapefruit','Grapefruit Trees','Lemons','Lemon Trees',
      'Mandarins/Tangerines','Mandarin/Tangerine Trees','Tangelos','Tangerine Trees','Tangors',
      'Pomegranates','Papaya','Papaya Tree','Mango Trees','Macadamia Trees','Pecans','Pistachios','Walnuts',
      'Cabbage','Cucumbers','Chile Peppers','Peppers','Onions','Potatoes','Sweet Potatoes',
      'Fresh Market Tomatoes','Tomatoes','Pumpkins','Fresh Market Sweet Corn','Sweet Corn',
      'Fresh Market Beans','Processing Beans','Green Peas','Nursery (FG&C)','Nursery (NVS)','Fresh Apricots','Processing Apricots'
    )
  GROUP BY county_fips, county_name, state_abbr
),
total AS (
  SELECT county_fips, SUM(liability_amount) AS total_liability
  FROM ag.rma_crop_insurance
  WHERE "year" = 2024
  GROUP BY county_fips
),
joined AS (
  SELECT l.county_fips, l.county_name, l.state_abbr, l.li_liability, t.total_liability,
         l.li_liability / NULLIF(t.total_liability,0) AS li_share
  FROM labor_intensive l JOIN total t ON l.county_fips = t.county_fips
  WHERE t.total_liability > 5000000
),
scaled AS (
  SELECT *,
    (li_liability - MIN(li_liability) OVER()) / NULLIF(MAX(li_liability) OVER() - MIN(li_liability) OVER(),0) AS scale_norm,
    (li_share - MIN(li_share) OVER()) / NULLIF(MAX(li_share) OVER() - MIN(li_share) OVER(),0) AS share_norm
  FROM joined
)
SELECT county_name, state_abbr, li_liability, li_share,
       ROUND((0.5*scale_norm + 0.5*share_norm)::numeric,4) AS exposure_score
FROM scaled
ORDER BY exposure_score DESC
FETCH FIRST 15 ROWS ONLY

WITH labor_intensive AS (
  SELECT county_fips, county_name, state_abbr, SUM(liability_amount) AS li_liability
  FROM ag.rma_crop_insurance
  WHERE "year" = 2024
    AND commodity_name IN (
      'Apples','Avocados','Avocado Trees','Cherries','Grapes','Table Grapes','Raisins',
      'Fresh Freestone Peaches','Processing Cling Peaches','Processing Freestone','Peaches',
      'Pears','Plums','Prunes','Strawberries','Blueberries','Caneberries','Cranberries','Figs',
      'Kiwifruit','Olives','Oranges','Orange Trees','Grapefruit','Grapefruit Trees','Lemons','Lemon Trees',
      'Mandarins/Tangerines','Mandarin/Tangerine Trees','Tangelos','Tangerine Trees','Tangors',
      'Pomegranates','Papaya','Papaya Tree','Mango Trees','Macadamia Trees','Pecans','Pistachios','Walnuts',
      'Cabbage','Cucumbers','Chile Peppers','Peppers','Onions','Potatoes','Sweet Potatoes',
      'Fresh Market Tomatoes','Tomatoes','Pumpkins','Fresh Market Sweet Corn','Sweet Corn',
      'Fresh Market Beans','Processing Beans','Green Peas','Nursery (FG&C)','Nursery (NVS)','Fresh Apricots','Processing Apricots'
    )
  GROUP BY county_fips, county_name, state_abbr
),
total AS (
  SELECT county_fips, SUM(liability_amount) AS total_liability
  FROM ag.rma_crop_insurance
  WHERE "year" = 2024
  GROUP BY county_fips
),
joined AS (
  SELECT l.county_fips, l.county_name, l.state_abbr, l.li_liability, t.total_liability,
         l.li_liability / NULLIF(t.total_liability,0) AS li_share
  FROM labor_intensive l JOIN total t ON l.county_fips = t.county_fips
  WHERE t.total_liability > 5000000
),
bounds AS (
  SELECT MIN(li_liability) AS mn_l, MAX(li_liability) AS mx_l, MIN(li_share) AS mn_s, MAX(li_share) AS mx_s FROM joined
)
SELECT j.county_name, j.state_abbr, j.li_liability, j.li_share,
  (j.li_liability - b.mn_l)/(b.mx_l-b.mn_l) AS scale_norm,
  (j.li_share - b.mn_s)/(b.mx_s-b.mn_s) AS share_norm,
  0.5*((j.li_liability - b.mn_l)/(b.mx_l-b.mn_l)) + 0.5*((j.li_share - b.mn_s)/(b.mx_s-b.mn_s)) AS exposure_score
FROM joined j CROSS JOIN bounds b
ORDER BY exposure_score DESC
FETCH FIRST 15 ROWS ONLY