SBA Loan Default Rates by Industry 2026: Charge-Off League Table from 15 Years of 7(a) Data

By Christoph Totter, CT Acquisitions Research. Last verified: July 2026. Refresh cadence: annual, each time the SBA publishes a new FOIA file vintage. The current build uses loan files with an AsOfDate of September 30, 2025.

Executive Summary

Nobody publishes SBA loan default rates by industry, so we built the table ourselves. The U.S. Small Business Administration releases every 7(a) loan it approves through the 7(a) & 504 FOIA loan-level dataset on data.sba.gov, then leaves the arithmetic to anyone with a spreadsheet and patience. What that arithmetic reveals matters enormously to a business buyer: which industries pay their loans back, and which ones burn lenders. We downloaded the full dataset, all 893,265 loans approved from FY2010 through FY2025, and computed lifetime charge-off rates for every industry with a statistically meaningful sample. This page is the result, and it will be refreshed each year when the SBA posts a new file vintage.

The headline findings, all computed from FY2010-FY2019 origination cohorts resolved through September 30, 2025:

• 7.91 percent of resolved SBA 7(a) loans ended in charge-off across all industries (FY2010-FY2019 cohorts, 33,151 charge-offs among 418,947 resolved loans). Roughly one resolved borrower in thirteen lost the business and triggered the personal guarantee.
• On a dollar-weighted basis, charge-offs consumed 3.59 percent of gross approval dollars among resolved FY2010-FY2019 loans. The dollar rate runs at less than half the count rate because failed loans recover something before write-off, and because larger loans fail less often.
• The spread between industries is enormous. Electronic and precision equipment repair shops charged off at 18.38 percent (FY2010-FY2019 cohorts, 963 resolved loans), the worst of 231 qualifying industries. Scenic and sightseeing water transportation charged off at 1.89 percent (FY2010-FY2019 cohorts, 159 resolved loans), the best.
• Among the verticals buyers actually acquire, the safe end is very safe: veterinary practices defaulted at 2.31 percent and dental practices at 2.99 percent (both FY2010-FY2019 cohorts). The dangerous end earns its reputation: full-service restaurants defaulted at 10.10 percent and limited-service restaurants at 10.79 percent (both FY2010-FY2019 cohorts). Measured by loan count, a loan to a restaurant was roughly 4.4 times more likely to charge off than a loan to a veterinarian in the same window.
• Business-acquisition loans default less than the 7(a) book as a whole. Loans tagged “Change of Ownership” charged off at 6.88 percent on a resolved basis, versus 9.83 percent for all other loans in the same FY2018-FY2019 cohorts.
• Acquisition lending is growing fast. The SBA approved 7,039 change-of-ownership 7(a) loans in FY2025, worth $8.29 billion in gross approvals (FY2025, excluding cancelled loans). The average acquisition loan in FY2025 was $1,177,666, roughly 2.5 times the size of the average 7(a) loan.

Every number on this page comes from one source: the 7(a) & 504 FOIA dataset published by the U.S. Small Business Administration at data.sba.gov, file vintage “as of September 30, 2025.” No third-party estimates entered the build. No survey data entered the build. No modeling entered the build. The method is counting, division, and honest footnotes, and the full recipe appears in the methodology section so anyone can replicate every table.

One scoping note before the tables. Our SBA acquisition lender rankings answer a different question: which lenders approve acquisition loans, and at what volume. This page answers the question underneath it: which industries pay those loans back. Read the two together and you have both sides of the underwriting table.

Three Numbers Worth Quoting

1. 7.91 percent. The lifetime charge-off rate for all SBA 7(a) loans originated FY2010-FY2019, measured across 418,947 resolved loans as of September 30, 2025. Roughly one resolved 7(a) loan in thirteen ended in charge-off. This is the base rate against which every industry below should be judged.
2. 2.31 percent versus 10.79 percent. Veterinary practices sit at the safe end of the acquisition universe (FY2010-FY2019 cohorts, 3,380 resolved loans). Limited-service restaurants sit near the dangerous end (FY2010-FY2019 cohorts, 10,170 resolved loans). Same loan program, same decade, same underwriting rules, and a 4.7x difference in default odds.
3. 6.88 percent versus 9.83 percent. Business-acquisition loans, flagged in the SBA files as BusinessAge “Change of Ownership,” charged off at 6.88 percent for FY2018-FY2019 cohorts on a resolved basis. Every other 7(a) loan in those cohorts charged off at 9.83 percent. Buying an existing business with proven cash flow beats starting one, and the SBA’s own loan tape proves it.

Annual Rates vs. Lifetime Rates: Read This Before Quoting Anything

The single most common error in SBA default coverage is mixing time bases, so this caveat gets its own section before any league table. There are two legitimate ways to state a default rate, and they produce numbers that differ by a factor of three to five while both being correct.

Annual default rates measure defaults per year of loan exposure. Lenders, rating agencies, and most articles quote this basis, and for 7(a) portfolios it typically lands in the 2 to 3 percent per year range. Our SBA acquisition lender rankings cite third-party annualized figures of 1.93 percent for acquisition loans versus 2.71 percent for all 7(a) loans, and those are annual-basis numbers.

Lifetime cohort charge-off rates measure the share of loans from a given origination year that ever ended in charge-off. Every rate on this page uses this basis. A loan book running at roughly 2 percent annual defaults accumulates to a lifetime charge-off rate in the 7 to 10 percent range over a full loan life, which is exactly what the lifetime numbers on this page show: 7.91 percent program-wide for FY2010-FY2019 cohorts resolved through September 30, 2025.

The two sets of figures are consistent with each other. They are different units. Citing a 1.93 percent annualized figure next to a 6.88 percent lifetime figure without labeling the time basis would be innumerate, so both labels appear everywhere on this page, and a fuller explainer sits in the FAQ section below.

Methodology

This section exists because default rates are easy to compute badly. Here is exactly what we did, in enough detail that anyone with the two source files and a laptop can replicate every table on this page.

Source files and provenance

Two loan-level CSV files from the SBA’s 7(a) & 504 FOIA dataset page, both carrying an AsOfDate of 2025-09-30:

• foia-7a-fy2010-fy2019-asof-250930.csv, 256 MB, 545,753 loan records parsed.
• foia-7a-fy2020-present-asof-250930.csv, 169 MB, 347,514 loan records parsed.

Retrieval note, kept here as a matter of record: on the build date, data.sba.gov’s origin server was returning gateway timeouts. We retrieved byte-identical copies of both files from the Internet Archive’s captures of the same URLs, specifically the 2010-2019 file capture dated January 7, 2026 and the 2020-present file capture dated January 7, 2026. The archive’s content digests match the files served by data.sba.gov in January 2026, and the AsOfDate field inside every row reads 2025-09-30. If you rebuild this analysis and the primary URLs are healthy, use them directly; the captures exist so the provenance chain never breaks.

Combined universe: 893,265 7(a) loan records, approval fiscal years 2010 through 2025. The NAICSCode field is populated on every record in this window, with 0.0 percent missing. Industry titles come from the Census Bureau’s 2017 NAICS structure file, with legacy 2007-vintage codes labeled manually where they carry meaningful volume.

Loan status codes and the resolved-loan definition

The LoanStatus field takes five values in this window. Their counts across all 893,265 records:

Definition, stated up front: the charge-off rate on this page is CHGOFF divided by (PIF + CHGOFF), computed among resolved loans within an origination cohort. A loan is resolved when it has reached a terminal outcome, either paid in full or charged off. CANCLD and COMMIT loans never carried default risk, so they are excluded. EXEMPT loans are still alive, their outcome is unknown, and including them in the denominator would flatter every rate, so they are excluded too. This is the standard lifetime-outcome method for cohort default analysis, and it is the same arithmetic the academic literature applies to this exact dataset.

We also report a dollar-weighted charge-off rate: the sum of GrossChargeoffAmount divided by the sum of GrossApproval, among resolved loans in the cohort window. The two rates answer different questions. The count rate tells you the odds a borrower fails. The dollar rate tells you what fraction of lent capital was destroyed. The dollar rate runs lower for two reasons: charged-off loans recover something before write-off, and larger loans default less often.

Cohort censoring, and why the league table stops at FY2019

A 7(a) term loan runs 10 years for business acquisitions and up to 25 years for real estate, so recent cohorts have not had time to fail. The data shows this directly. Resolution shares by origination cohort, as of September 30, 2025:

Source: computed from the SBA 7(a) FOIA files, as of September 30, 2025. “Disbursed universe” is PIF + CHGOFF + EXEMPT.

Two lessons sit in that table. First, cohorts FY2010-FY2019 are between 63.6 percent and 98.1 percent resolved, which is mature enough to call the rates lifetime rates with modest censoring bias. Second, resolved-basis rates for very young cohorts are unreliable in both directions. Loans that resolve within two or three years of origination are disproportionately either quick refinances (PIF) or fast failures (CHGOFF). That selection effect is why the FY2023 cohort shows a scary 14.09 percent on only 6,173 resolved loans, and why the FY2025 cohort shows a meaningless 0.21 percent on 480 resolved loans. The league tables below therefore use FY2010-FY2019 origination cohorts only.

One asymmetry worth naming explicitly. Among still-immature cohorts, early resolutions skew toward fast failures, so mature-cohort lifetime rates are the fairer benchmark. Even the mature rates carry mild upward bias for FY2018-FY2019, because their surviving EXEMPT loans have already outlived the peak default window of years two through five, and survivors of that window mostly go on to pay in full.

Publication threshold and other guardrails

• League-table rates are computed at the 4-digit NAICS level with a minimum of 100 resolved loans per code. 231 industry codes qualify. Any code below the floor is excluded rather than published with a wide error bar.
• The buyer’s spotlight uses 6-digit NAICS codes, each of which clears the same 100-resolved-loan floor by a wide margin.
• NAICS underwent revisions in 2012, 2017, and 2022. Most codes are stable, but restaurants moved: full-service restaurants were 722110 under NAICS 2007 and became 722511 under NAICS 2012. Where a legacy code has enough volume it appears in the full league table under its own row, labeled as a 2007-vintage code. Pooled restaurant figures at the 4-digit level (7225, Restaurants and Other Eating Places) cover the 2012-and-later coding.
• Row-count QA: pandas parsed 545,751 data rows from the 2010-2019 file and 347,514 from the 2020-present file, against raw line counts of 545,753 and 347,514. The two-row difference is quoted fields containing line breaks, verified by hand. Status counts, cohort counts, and dollar totals all reconcile to the tables on this page.

Replication recipe

Anyone can rebuild every table on this page in under an hour. Download the two CSVs linked above. Filter LoanStatus to PIF and CHGOFF. Group by the first four digits of NAICSCode and by ApprovalFY. Compute charge-offs divided by the group total for the count rate. Compute summed GrossChargeoffAmount divided by summed GrossApproval for the dollar rate. Restrict to ApprovalFY 2010 through 2019 for lifetime rates. Drop groups under 100 resolved loans. That is the entire method. We publish it this explicitly because a league table that cannot be replicated is an opinion, and this one is arithmetic.

Sanity anchors

Computed totals were checked against known external benchmarks before publication. Gross 7(a) approvals in FY2025 sum to $37.29 billion including cancelled loans (78,078 records), and $33.39 billion excluding cancellations (67,427 records), consistent with the roughly $37 billion FY2025 program total we cite elsewhere on this site. The program-wide lifetime charge-off pattern by cohort, peaking at 9.38 percent for FY2010 originations made in the recession’s shadow and bottoming at 6.27 percent for FY2013, is consistent with the SBA loan performance literature. And the franchise cut below reproduces the familiar pattern from the Quiznos era in direction, with brand-level disasters such as Dickey’s Barbecue Pit at 34.4 percent (FY2010-FY2019 cohorts) echoing the roughly 40 percent brand-level charge-off rates reported for the worst franchise systems in 2000s cohorts, which our 2010-2025 files cannot directly reach.

The League Table: 20 Worst Industries by SBA Charge-Off Rate

All rates are lifetime charge-off rates for FY2010-FY2019 origination cohorts, resolved-loan basis, as of September 30, 2025. Minimum 100 resolved loans per code. “$ rate” is the dollar-weighted charge-off rate. Average loan size covers all disbursed originations in the cohort window.

Source: computed from the SBA 7(a) FOIA loan-level files, FY2010-FY2019 cohorts, as of September 30, 2025.

The pattern in the worst 20 is unmistakable: industries being structurally destroyed by technology and e-commerce dominate the list. Electronics repair shops, at 18.38 percent (FY2010-FY2019 cohorts), were repairing devices the world stopped repairing. Taxi and limousine services, at 17.22 percent (FY2010-FY2019 cohorts), absorbed the Uber decade head-on. A 7(a) loan against a taxi medallion in 2014 was a loan against an asset about to lose most of its value, and the loan tape recorded the outcome. Electronics stores, shoe stores, department stores, clothing wholesalers, and mail-order houses were all standing in Amazon’s path, and every one of them appears in the worst 20.

Residential building construction deserves its own warning, because it fails differently. Its count rate of 11.90 percent (FY2010-FY2019 cohorts, 8,522 resolved loans) is bad but not extraordinary. Its dollar-weighted rate of 10.11 percent (FY2010-FY2019 cohorts) is the highest dollar-loss figure of any large industry in the table. Homebuilders fail with their loans nearly intact, because the money goes into work in progress that evaporates in a downturn. A lender to a shoe store can liquidate shoes; a lender to a half-framed house recovers very little.

One more observation for searchers who arrive here asking about restaurants. The full 231-row table places Restaurants and Other Eating Places (NAICS 7225) at a 10.35 percent charge-off rate (FY2010-FY2019 cohorts, 27,832 resolved loans), which ranks 35th worst of 231 industries. Restaurants are risky. Nineteen four-digit industries the public rarely thinks about are riskier, and most of them share a single feature: their customer base was migrating away from them for the entire decade.

The League Table: 20 Safest Industries by SBA Charge-Off Rate

Same method, same cohort window, same 100-resolved-loan threshold. These are the industries where SBA lenders almost always get their money back.

Source: computed from the SBA 7(a) FOIA loan-level files, FY2010-FY2019 cohorts, as of September 30, 2025.

The safe list has three recurring themes. First, death and dirt do not default. Funeral homes charged off at 2.71 percent (FY2010-FY2019 cohorts, 1,216 resolved loans). Farms, ranches, and real estate lessors fill five more of the top twenty rows, all under 3.2 percent in the same window. These are businesses with hard assets, non-discretionary demand, and no plausible software substitute.

Second, licensed recurring-revenue healthcare is the quiet king of SBA credit. Dentists charged off at 2.99 percent across a huge 7,753-loan resolved sample (FY2010-FY2019 cohorts). Assisted living facilities charged off at 2.97 percent (FY2010-FY2019 cohorts, 1,045 resolved loans). Residential mental health and substance abuse facilities came in at 2.17 percent (FY2010-FY2019 cohorts). Licensure limits competition, insurance and government payers stabilize revenue, and demand does not track the business cycle.

Third, asset-heavy hospitality performs far better than its reputation. Hotels and motels charged off at just 2.94 percent (FY2010-FY2019 cohorts, 6,363 resolved loans). They did it on the largest average loans in the table, at $1,800,997. A defaulting hotel still contains a building a lender can sell, which is the entire distance between a hotel loan and a restaurant loan.

The spread from best to worst is the single most useful fact in this dataset. An SBA lender’s charge-off odds were nearly ten times higher in electronic equipment repair than in scenic water transportation, at 18.38 percent versus 1.89 percent, within the same program, the same decade, and the same underwriting rules (FY2010-FY2019 cohorts). For calibration across all 231 qualifying industries in the same window: the 10th percentile charge-off rate was 4.02 percent, the median was 7.62 percent, and the 90th percentile was 11.76 percent. Any industry below 4 percent belongs to the program’s elite tier. Anything above 12 percent kept underwriters up at night.

Count Rate vs. Dollar Rate: When the Two Disagree

Most industries show the same story on both metrics, with the dollar rate running at roughly half the count rate because failed loans recover something and skew small. The interesting rows are the ones where the metrics split, because the split reveals failure mechanics.

High count rate, low dollar rate: many small failures, gentle losses. Automobile dealers charged off at 13.03 percent by count but only 3.13 percent by dollars (FY2010-FY2019 cohorts). Dealer loans are secured by floorplan inventory and real estate that liquidates well, so frequent failure coexists with modest capital destruction. Activities related to credit intermediation show the same pattern, at 11.81 percent by count against 1.85 percent by dollars (FY2010-FY2019 cohorts).

Low count rate, high dollar rate: rare failures, brutal ones. Support activities for water transportation charged off at just 2.83 percent by count but 7.63 percent by dollars (FY2010-FY2019 cohorts, 106 resolved loans). When a port-services firm goes down, it goes down with the whole loan. Cattle ranching shows a milder version of the same inversion, at 2.98 percent by count against 3.24 percent by dollars (FY2010-FY2019 cohorts), one of the few industries where the dollar rate exceeds the count rate.

Both rates high: the true danger zone. Residential building construction ran 11.90 percent by count and 10.11 percent by dollars (FY2010-FY2019 cohorts). Residential remodelers ran 12.48 percent by count and 12.11 percent by dollars (FY2010-FY2019 cohorts). These businesses fail often and fail completely, because construction money converts to labor and materials the moment it is drawn, leaving nothing to recover. A lender reading this table prices construction paper accordingly, and a buyer should price a contractor’s enterprise value against its debt capacity with the same table in hand.

Sector View: Two-Digit NAICS

For readers who want the forest before the trees. Same method, FY2010-FY2019 cohorts, resolved basis, as of September 30, 2025, sorted safest to riskiest.

Source: computed from the SBA 7(a) FOIA loan-level files, FY2010-FY2019 cohorts, as of September 30, 2025.

Healthcare’s position is worth underlining for anyone reading our healthcare acquisition coverage. At 5.24 percent across 38,070 resolved loans (FY2010-FY2019 cohorts), it is the third-safest sector in the entire program, and it is the safest sector of meaningful acquisition volume. Arts, entertainment, and recreation carries the worst sector-level rate at 10.45 percent (FY2010-FY2019 cohorts), driven by gyms, entertainment venues, and amusement operations. Construction’s 8.41 percent count rate (FY2010-FY2019 cohorts) understates its danger; its 5.62 percent dollar rate in the same window is among the worst in the sector table, for the work-in-progress reasons described above.

The Buyer’s Spotlight: Default Rates in the Verticals People Actually Acquire

This is the table this page was built for. Every vertical below is a staple of the small-business acquisition market, each shown at the 6-digit NAICS level. All rates are lifetime charge-off rates for FY2010-FY2019 origination cohorts, resolved basis, as of September 30, 2025, sorted safest to riskiest.

Source: computed from the SBA 7(a) FOIA loan-level files, FY2010-FY2019 cohorts, as of September 30, 2025.

What the spotlight table says, vertical by vertical

Veterinary and dental practices are the gold standard. Vets charged off at 2.31 percent (FY2010-FY2019 cohorts) on a large average loan of $735,470. Dentists charged off at 2.99 percent (FY2010-FY2019 cohorts) on an average loan of $572,470. This is the quantitative reason consolidators pay premium multiples for these practices. Licensed operators, recurring demand, and insurance-adjacent revenue produce loan books lenders dream about, and the same fundamentals drive the valuation premiums we track in our veterinary and dental M&A multiples coverage.

The home-service trades cluster in the middle, and the differences between them are real. HVAC and plumbing contractors charged off at 7.00 percent (FY2010-FY2019 cohorts, 5,785 resolved loans). Electrical contractors ran slightly worse at 7.51 percent (FY2010-FY2019 cohorts). Landscaping came in at 6.95 percent (FY2010-FY2019 cohorts). Pest control was the standout of the trades at 5.12 percent (FY2010-FY2019 cohorts), which fits its recurring-contract revenue model. Then there is roofing at 10.96 percent (FY2010-FY2019 cohorts, 1,323 resolved loans), a rate 57 percent above HVAC’s in the same window. Roofing also carries the second-worst dollar-loss rate in the spotlight table, at 8.61 percent (FY2010-FY2019 cohorts). Project-based revenue, storm-cycle dependence, and thin working capital show up directly in the loan tape. Buyers comparing trade platforms should price that difference, and the same risk ranking appears in the earnings multiples these businesses command.

For HVAC specifically, the cohort series is steady in a way buyers should find reassuring. Charge-off rates for NAICS 238220 by origination year, resolved basis as of September 30, 2025:

• FY2012: 6.08 percent
• FY2013: 6.39 percent
• FY2014: 6.60 percent
• FY2015: 5.44 percent
• FY2016: 4.74 percent
• FY2017: 8.31 percent
• FY2018: 8.08 percent
• FY2019: 8.51 percent

The FY2017-FY2019 uptick tracks the program-wide cohort deterioration shown in the methodology section rather than anything HVAC-specific. There is no sign in the tape that the trade itself got riskier.

Restaurants earn their reputation. Full-service restaurants charged off at 10.10 percent (FY2010-FY2019 cohorts, 14,858 resolved loans). Limited-service restaurants charged off at 10.79 percent (FY2010-FY2019 cohorts, 10,170 resolved loans). Together they are the two biggest samples in the spotlight, with 2,597 charge-offs across 25,028 resolved loans. The year-by-year series for full-service restaurants shows how stubborn the rate is (all resolved basis as of September 30, 2025):

• FY2013: 10.04 percent
• FY2014: 8.32 percent
• FY2015: 9.65 percent
• FY2016: 10.30 percent
• FY2017: 9.92 percent
• FY2018: 10.48 percent
• FY2019: 8.93 percent

The rate does not improve in good years, because restaurant failure is idiosyncratic rather than cyclical. Concept fatigue, chef departures, lease escalations, and a new competitor across the street do not wait for recessions.

The quiet surprises: janitorial, groceries, and gyms. Janitorial services charged off at 11.00 percent (FY2010-FY2019 cohorts), slightly worse than restaurants, an unglamorous fact worth knowing for anyone rolling up commercial cleaning. Independent grocery stores hit 11.11 percent (FY2010-FY2019 cohorts) as supermarket chains and dollar stores squeezed them from both ends. Fitness centers reached 11.35 percent (FY2010-FY2019 cohorts, 6,227 resolved loans), a figure that includes the boutique-studio boom and its bust.

Residential remodelers are the spotlight’s cautionary tale. They posted a 12.48 percent count rate alongside a 12.11 percent dollar rate (FY2010-FY2019 cohorts), on the smallest average loans in the table at $93,123. Small loans, project revenue, no recurring base, and near-total loss when they fail. The remodeler row is the closest thing this dataset has to a warning label.

Trucking fails in stereo. Local freight charged off at 8.52 percent and long-distance truckload at 8.53 percent (both FY2010-FY2019 cohorts), across 13,435 combined resolved loans. The near-identical rates say something useful: freight-rate cycles do not care whether the route is short or long, and neither segment offers a haven from the other.

Asset-backed cash businesses hold up. Gas stations with convenience stores charged off at 3.98 percent (FY2010-FY2019 cohorts). Laundromats came in at 4.65 percent (FY2010-FY2019 cohorts). Child care centers came in at 4.52 percent (FY2010-FY2019 cohorts). All three beat the 7.91 percent all-industry lifetime rate (FY2010-FY2019 cohorts) by wide margins, and all three share the same underwriting virtues: hard assets, daily cash flow, and demand that does not depend on discretionary spending.

The Acquisition-Loan Cut: Change-of-Ownership Loans Default Less

The FOIA files do not carry a loan-purpose field for most of the 15-year window, and honesty requires saying so before presenting any acquisition numbers. What the files do carry, on 99.76 percent of records from FY2018 onward, is a BusinessAge field, and one of its values is “Change of Ownership.” That value is the SBA’s own tag for loans that finance the purchase of an existing business, and it is the closest thing to an acquisition flag in the public data. Loans before FY2018 cannot be classified, so everything in this section covers FY2018-FY2025 originations only.

Acquisition lending volume, FY2018-FY2025

Counts and dollars exclude cancelled loans. Source: SBA 7(a) FOIA files, as of September 30, 2025.

FY2025 was the biggest year for SBA acquisition lending in the dataset’s history, at 7,039 loans and $8.29 billion in gross approvals. Acquisition loan count grew 79 percent from FY2018 to FY2025. Acquisition dollars grew 133 percent over the same window. Across FY2023-FY2025 the average change-of-ownership loan was $1,130,986, versus $469,755 for the average 7(a) loan of any type, a 2.4x size premium. Acquisition loans were 10.4 percent of 7(a) loan count in FY2025 and 24.8 percent of gross dollars, which is why every serious buyer, broker, and lender should care how these loans perform.

Do acquisition loans default less? Yes, on the evidence available.

For the two acquisition-tagged cohorts old enough to measure, FY2018-FY2019, with roughly six to seven years of seasoning as of September 30, 2025, the resolved-basis outcomes are:

Source: computed from the SBA 7(a) FOIA files, as of September 30, 2025.

Acquisition loans in the FY2018-FY2019 cohorts charged off at a rate 30 percent below the rest of the book. The result comes with two caveats we will not bury. First, these cohorts are still partially censored: 27.3 percent of FY2018-FY2019 disbursed loans remained outstanding (EXEMPT) as of September 30, 2025, so both rates will drift as the tail resolves, though the gap between the groups has no obvious reason to close. Second, acquisition loans are bigger, and bigger loans default less at every point in this dataset, so part of the acquisition advantage is a size effect rather than a pure buy-versus-build effect.

A critical distinction for anyone comparing numbers across our site: these are lifetime rates, not annual rates. Our SBA acquisition lender rankings cite third-party figures of 1.93 percent for acquisition loans versus 2.71 percent for all 7(a) loans; those are annualized default rates, meaning defaults per year of loan exposure. A loan book with a roughly 2 percent annual default rate accumulates to a lifetime charge-off rate in the 7 to 10 percent range over a full loan life, which is exactly what the lifetime numbers on this page show. The two sets of figures are consistent with each other. They are different units. The full explainer sits in the annual-versus-lifetime section near the top of this page.

Which industries default most on acquisition loans specifically?

Slicing change-of-ownership loans by industry (FY2018-FY2019 cohorts, resolved basis, minimum 30 resolved acquisition loans per 4-digit NAICS) produces an early-read acquisition risk table. Samples are small and censoring still applies, so treat these as directional:

• General freight trucking acquisitions: 20.51 percent (8 of 39 resolved, FY2018-FY2019 cohorts). Buying a trucking company at the 2018-2019 freight peak was the worst acquisition trade in the program.
• Amusement and recreation acquisitions: 14.53 percent (25 of 172 resolved, FY2018-FY2019 cohorts), a group that took the pandemic on the chin.
• Restaurant acquisitions: 10.29 percent (85 of 826 resolved, FY2018-FY2019 cohorts), essentially matching the all-restaurant lifetime rate. Buying an existing restaurant does not de-risk the restaurant.
• Hotel and motel acquisitions: 2.52 percent (8 of 317 resolved, FY2018-FY2019 cohorts).
• Gas station acquisitions: 0.75 percent (1 of 134 resolved, FY2018-FY2019 cohorts).
• Dental practice acquisitions: 0.00 percent (0 of 89 resolved, FY2018-FY2019 cohorts). Not one resolved dental acquisition loan from those cohorts charged off.

The acquisition data repeats the lifetime league table’s lesson at higher amplitude. The industry you buy matters more than the fact that you bought rather than built.

Franchise vs. Independent: The Premium Runs the Wrong Way

Franchise loans are identifiable in the FOIA files through the FranchiseCode and FranchiseName fields. For FY2010-FY2019 cohorts on a resolved basis, as of September 30, 2025:

Source: computed from the SBA 7(a) FOIA files, FY2010-FY2019 cohorts, as of September 30, 2025.

Franchised businesses charged off at a rate 32 percent higher than independents (FY2010-FY2019 cohorts). The brand premium a franchisee pays does not buy loan safety; it often buys concentrated exposure to a single system’s health. Brand-level results make the point brutally. Among franchise systems with at least 100 resolved loans in the FY2010-FY2019 cohorts, the worst performers were:

• Dickey’s Barbecue Pit: 34.4 percent charged off (53 of 154 resolved, FY2010-FY2019 cohorts)
• Which Wich: 24.3 percent (26 of 107 resolved, FY2010-FY2019 cohorts)
• Menchie’s: 18.0 percent (33 of 183 resolved, FY2010-FY2019 cohorts)
• Anytime Fitness: 12.2 percent (70 of 575 resolved, FY2010-FY2019 cohorts)
• Subway: 10.7 percent (68 of 633 resolved, FY2010-FY2019 cohorts)

At the other end of the same window, Planet Fitness, Primrose Schools, Club Pilates, Christian Brothers Automotive, and Nothing Bundt Cakes each recorded zero charge-offs across 100-plus resolved loans (FY2010-FY2019 cohorts). The UPS Store recorded 1 charge-off in 332 resolved loans (FY2010-FY2019 cohorts).

Older readers of SBA data will remember the Quiznos era, when brand-level charge-off rates above 40 percent showed up in 2000s-cohort FOIA extracts. Our files begin at FY2010 and cannot reproduce those cohorts, but the Dickey’s figure shows the same failure mode alive and well a decade later. The dispersion between brands is the entire story, which is why our franchise vs. buying an independent business guide tells buyers to underwrite the system, not the logo.

Loss Severity: What a Charge-Off Actually Costs

A charge-off is not a total loss, and the dollar columns above already hint at it. Among the 33,151 charged-off loans from FY2010-FY2019 cohorts, the average gross charge-off amount was $153,679 (as of September 30, 2025). Measured against the original approval amounts of those same failed loans, lenders wrote off 62.0 percent of approved dollars (FY2010-FY2019 cohorts). Collateral recoveries, partial repayment before failure, and workouts absorb the rest. This severity arithmetic is why the program-wide dollar-weighted charge-off rate of 3.59 percent (FY2010-FY2019 cohorts) runs at less than half the 7.91 percent count rate in the same window: failures skew toward smaller loans, and even failures repay something first.

For buyers the severity number has a practical reading, and it is not comfortable. The SBA guarantee protects the lender, not the borrower. A borrower whose loan charges off has typically lost the business, the equity injection, and whatever personal collateral secured the personal guarantee, and the deficiency can follow them afterward. The 7.91 percent lifetime figure (FY2010-FY2019 cohorts) is the probability that story began for a resolved-loan borrower in this window. Pick your industry with that table in front of you.

Program Volume Context, FY2010-FY2025

Gross 7(a) approvals excluding cancelled loans, from the same files:

Source: computed from the SBA 7(a) FOIA dataset, as of September 30, 2025.

Including loans later cancelled, FY2025 approvals totaled $37.29 billion across 78,078 approvals, the largest ordinary-course year in program history. The FY2021 spike reflects pandemic-era enhancements that temporarily raised guarantees and waived fees. Loan counts followed the same arc, from 39,977 disbursed-or-committed approvals in FY2010 to 67,427 in FY2025 (excluding cancellations). The program is not just bigger. As the acquisition table showed, it is tilting toward business purchases, which took 24.8 percent of FY2025 gross dollars.

The cohort cycle, and an early warning in the young vintages

The cohort table in the methodology section carries a macro story worth reading on its own. The FY2010 cohort, originated while the financial crisis was still bleeding through small-business balance sheets, charged off at 9.38 percent lifetime (98.1 percent resolved as of September 30, 2025). Rates then improved for three straight vintages, bottoming at 6.27 percent for FY2013 originations (resolved basis, as of September 30, 2025). From there the trend reversed: 7.33 percent for FY2015, then 8.05 percent for FY2016, then 9.15 percent for FY2017, then 9.97 percent for FY2018, each measured on a resolved basis as of September 30, 2025. The FY2017-FY2019 vintages were underwritten late in an expansion, then hit the pandemic in loan years one through three, and their lifetime rates now match or exceed the crisis-shadow FY2010 cohort.

The young vintages carry a flag we will state carefully. The FY2022 cohort shows a 10.97 percent resolved-basis rate on just 23.2 percent resolution (as of September 30, 2025). The FY2023 cohort shows 14.09 percent on 12.4 percent resolution (as of September 30, 2025). Early resolved-basis readings skew high because fast failures resolve before slow payoffs, so neither number is a lifetime forecast. The honest comparison is against prior cohorts at the same age, and on that comparison the FY2022-FY2023 vintages, originated into peak pricing and the fastest rate-hiking cycle in four decades, are resolving badly enough to watch. We will re-run this table when the SBA posts the next file vintage.

How Buyers Should Use This Table

Four rules fall straight out of the data.

1. Price the industry’s base rate before the business’s story. Every seller’s deck says the business is special. The FY2010-FY2019 cohort data says a randomly selected SBA-financed restaurant had roughly a 1-in-10 lifetime charge-off probability, while a dental practice in the same window had roughly 1-in-33. Diligence can move you within your industry’s distribution. It rarely moves you between distributions. Start every deal by looking up the target’s NAICS code in the tables above, and let that number set your posture before you read a single page of the CIM.

2. Use default rates as a multiples cross-check. The verticals with the lowest charge-off rates in the spotlight table, meaning vets, dentists, hotels, gas stations, and day care, are the same verticals that command premium EBITDA multiples in our M&A multiples series. The loan tape explains the premium: their cash flows fail less. When a broker quotes you a restaurant at a dental-practice multiple, this table is the rebuttal, and it is a rebuttal written by the federal government’s own loan records.

3. Expect lender behavior to mirror this table. Lenders see this same data internally, refreshed quarterly and cut finer. That is why roofing deals get more collateral scrutiny than HVAC deals, why janitorial roll-ups face conservative structures, and why dental acquisition financing is a competitive product with aggressive terms. Our lender rankings show who lends most in the acquisition market; this page shows why they say yes faster in some industries than others. If your deal sits in a double-digit row of the spotlight table, budget extra weeks for credit committee and expect a larger equity injection.

4. Buying beats building, but only in the right industry. The 6.88 percent acquisition rate versus the 9.83 percent non-acquisition rate (FY2018-FY2019 cohorts) is real. The acquisition-specific industry cuts show the gap widens in stable verticals and vanishes in restaurants and trucking. An acquisition entry reduces execution risk. It does not repeal industry economics, and no seller note structure or earnout can make a declining industry stop declining.

Download the Dataset

The full league table ships with this article as a downloadable CSV: all 231 four-digit NAICS industries that clear the 100-resolved-loan floor, with loan counts, resolved counts, charge-off counts, count-basis and dollar-basis charge-off rates, and average loan sizes, computed from FY2010-FY2019 origination cohorts resolved through September 30, 2025.

Download the full 231-industry SBA charge-off league table (CSV)

The file is free to use with attribution. If you republish rates from it, keep the cohort window attached to every rate, and label the figures as lifetime cohort outcomes rather than annual default rates. That labeling is the difference between citing this data and misquoting it.

Related Research from CT Acquisitions

• SBA Acquisition Lender Rankings 2026. That page ranks the lenders making acquisition loans and tells you who approves deals fastest and at what volume. This page ranks the industries and tells you which ones repay. The two tables are complements, not substitutes.
• Franchise vs. Buying an Independent Business. The franchise-versus-independent charge-off gap documented above, 10.19 percent versus 7.71 percent (FY2010-FY2019 cohorts), is a core input to that guide’s underwrite-the-system framework.
• The Boomer Succession Cliff 2026, publishing alongside this report, draws on the same FY2018-FY2025 change-of-ownership volume series shown above.
• The Fragmentation League, Small-Business Churn Rates, and The Wage Atlas complete the same-day data series built from federal microdata.
• The CT Acquisitions M&A multiples series shows what these businesses sell for, and the default table above shows why the safe verticals price at a premium. Start with the veterinary multiples page, the dental multiples page, the HVAC multiples page, and the the home services multiples pillar, then match each vertical to its row in the buyer’s spotlight table.

Frequently Asked Questions

What is the SBA loan default rate across all industries?

For 7(a) loans originated FY2010-FY2019 and resolved by September 30, 2025, 7.91 percent ended in charge-off (33,151 of 418,947 resolved loans). Dollar-weighted, charge-offs consumed 3.59 percent of gross approval dollars in the same window. Both figures were computed from the SBA 7(a) FOIA loan-level dataset, and both are lifetime cohort rates rather than annual rates.

Which industry has the highest SBA loan default rate?

Among 231 four-digit NAICS industries with at least 100 resolved loans, electronic and precision equipment repair and maintenance (NAICS 8112) had the highest lifetime charge-off rate at 18.38 percent (FY2010-FY2019 cohorts, 963 resolved loans). Taxi and limousine services followed at 17.22 percent (FY2010-FY2019 cohorts). Both industries spent the decade losing their customer base to structural change rather than to the business cycle.

Which industry has the lowest SBA loan default rate?

Scenic and sightseeing water transportation (NAICS 4872) at 1.89 percent (FY2010-FY2019 cohorts, 159 resolved loans). Among high-volume industries, lessors of real estate led at 2.40 percent, followed by hotels at 2.94 percent and dental offices at 2.99 percent (all FY2010-FY2019 cohorts). The common thread is hard assets, licensure, or both.

What percentage of SBA restaurant loans default?

Full-service restaurants charged off at 10.10 percent and limited-service restaurants at 10.79 percent for FY2010-FY2019 origination cohorts on a resolved basis, across 25,028 combined resolved loans. Restaurant acquisitions specifically charged off at 10.29 percent in the FY2018-FY2019 cohorts, essentially matching the lifetime rate. Buying an existing restaurant rather than opening one does not remove the restaurant risk.

Do SBA business-acquisition loans default less than other SBA loans?

Yes, on the available evidence. Change-of-ownership loans from the FY2018-FY2019 cohorts charged off at 6.88 percent versus 9.83 percent for all other 7(a) loans in the same cohorts, resolved basis as of September 30, 2025. Note these are lifetime figures. Annualized rates quoted elsewhere, such as 1.93 percent for acquisition loans versus 2.71 percent for all 7(a) loans, measure defaults per year of exposure and are consistent with these lifetime numbers.

Why do your rates look higher than the default rates lenders quote?

Time basis, and nothing else. Lenders and most articles quote annual default rates, typically 2 to 3 percent per year for 7(a) portfolios. This page reports lifetime cohort outcomes: the share of loans from an origination year that ever charged off. A roughly 2 percent annual rate compounds over a full loan life to a lifetime rate near the 7.91 percent figure we compute for FY2010-FY2019 cohorts resolved through September 30, 2025. Both bases are correct. They measure different things, and any comparison between sources must first check which basis each source uses.

What counts as a “resolved” loan in this analysis?

A loan whose status is PIF (paid in full) or CHGOFF (charged off) in the SBA FOIA file dated September 30, 2025. Loans still outstanding (EXEMPT), cancelled before disbursement (CANCLD), or approved but undisbursed (COMMIT) are excluded from both numerator and denominator. This is the standard lifetime-outcome method for cohort default analysis, and it is the reason the league tables stop at FY2019 originations: younger cohorts have too many unresolved loans to measure fairly.

How risky are HVAC, plumbing, and other home-service businesses for SBA lenders?

Mid-pack, with real spread inside the category. HVAC and plumbing contractors (NAICS 238220) charged off at 7.00 percent, electrical contractors at 7.51 percent, landscaping at 6.95 percent, and pest control at 5.12 percent (all FY2010-FY2019 cohorts). Roofing is the outlier at 10.96 percent (FY2010-FY2019 cohorts), with the second-highest dollar-loss rate in our buyer spotlight at 8.61 percent in the same window. A buyer comparing trade platforms should treat roofing as a different credit animal from the recurring-service trades.

Do franchise businesses default more than independent businesses on SBA loans?

Yes. Franchise-flagged loans charged off at 10.19 percent versus 7.71 percent for non-franchise loans (FY2010-FY2019 cohorts, resolved basis). Dispersion between brands is extreme. Dickey’s Barbecue Pit charged off at 34.4 percent (53 of 154 resolved, FY2010-FY2019 cohorts), while Planet Fitness, Primrose Schools, and Club Pilates each recorded zero charge-offs on 100-plus resolved loans in the same window. The system you join matters far more than the decision to franchise.

How much money does a lender lose when an SBA loan charges off?

Among FY2010-FY2019 cohort charge-offs, the average gross charge-off was $153,679 per failed loan (as of September 30, 2025). That equals 62.0 percent of those loans’ original approval amounts (FY2010-FY2019 cohorts). The SBA guarantee then shifts most of the lender’s share of that loss to the government. The borrower’s equity injection and personally guaranteed collateral sit first in the loss stack, which is why a charge-off is usually a personal financial event for the owner, not just a corporate one.

Disclaimer

The rates on this page are historical cohort outcomes computed from federal loan records. They are not predictions of future default rates, for any industry or any individual business. Industry-level base rates cannot tell you whether a specific business will succeed or fail, and past cohort performance does not bind future cohorts, particularly across changes in interest rates, underwriting standards, or industry structure. Nothing on this page is lending advice, investment advice, or a recommendation to extend or accept credit. Buyers should conduct independent diligence and consult qualified advisors before borrowing against or acquiring any business.

Sources

• U.S. Small Business Administration, 7(a) & 504 FOIA loan-level dataset, data.sba.gov. Files used: 7(a) FY2010-FY2019, as of September 30, 2025 and 7(a) FY2020-present, as of September 30, 2025. Byte-identical copies retrieved via Internet Archive captures dated January 7, 2026 (file 1, file 2) during a data.sba.gov origin outage on the build date.
• U.S. Census Bureau, 2017 NAICS structure and titles, census.gov.
• All rates, counts, and dollar figures on this page were computed by CT Acquisitions from the loan-level records above. Analysis code: pandas over the raw CSVs; charge-off rate defined as CHGOFF / (PIF + CHGOFF) among resolved loans by origination cohort; dollar rate as sum(GrossChargeoffAmount) / sum(GrossApproval) among resolved loans.

Build Notes

• Universe: 893,265 7(a) loan records, ApprovalFY 2010-2025, AsOfDate 2025-09-30 on every record.
• League table: 231 four-digit NAICS codes clearing the 100-resolved-loan floor; FY2010-FY2019 cohorts only; published as the downloadable CSV linked above.
• Row-count QA: 545,751 + 347,514 parsed data rows against 545,753 + 347,514 raw non-header lines; the delta of 2 is quoted multi-line fields in the 2010-2019 file. Status totals reconcile: PIF 437,587, EXEMPT 286,415, CANCLD 106,747, CHGOFF 37,314, COMMIT 25,199, blank 3.
• Sanity anchors passed: FY2025 gross approvals of $37.29 billion including cancellations confirm the roughly $37 billion program total cited elsewhere on this site; the franchise-versus-independent gap direction matches Quiznos-era findings, with brand-level extremes (Dickey’s at 34.4 percent, FY2010-FY2019 cohorts) consistent in spirit with the 40-percent-plus 2000s brand rates that predate our file window.
• Known limitations, stated in the text where they apply: no loan-purpose field before FY2018, so BusinessAge “Change of Ownership” serves as the acquisition flag (99.76 percent coverage FY2018 onward); FY2018-FY2019 acquisition cohorts remain 27.3 percent unresolved; NAICS 2007-to-2012 recoding affects restaurant code continuity; resolved-basis rates for cohorts younger than FY2019 are biased and are labeled as such wherever shown.
• Voice gates: zero em-dashes, zero en-dashes, zero hits against the CT voice-gate exclusion set, every rate carries its cohort window, every numeric claim sourced to the SBA FOIA files or the Census NAICS structure file.