SBA 504 Lending Statistics
The SBA 504 programme is the part of federal small-business lending that buys buildings and long-life equipment. A non-profit Certified Development Company issues a long-term, fixed-rate debenture, and a private lender takes a first mortgage on the same project. We analysed every loan-level 504 record SBA has released — 117,983 of them — to measure how much financing the programme anchors, how much private capital sits beside it, how deal sizes have moved, which industries use it, and how the oldest loans have actually performed.
Latest complete fiscal year analysed: FY2025. Data through the SBA snapshot of 30 June 2026. Historical series 2010–2025 (16 complete fiscal years). Last updated 2026-09-17.
The 504 programme moved more money through bigger deals in FY2025
SBA publishes 504 activity reports and lender rankings. It does not publish how much total project financing the programme anchors, how stable the private-capital share has been, how deal sizes have changed over 16 years, or how the oldest cohorts have performed by industry. Those are the figures below.
Every headline measure rose, and the private-capital ratio did not move
Both fiscal years are complete and are measured on the same population rule, so these comparisons are like for like. The ratio of third-party dollars to SBA debentures changed by -0.0003 — effectively flat while combined project financing moved by +17.1%, about a sixth.
| Measure | FY2025 | FY2024 | Change |
|---|---|---|---|
| 504 approvals | 6,762 | 5,993 | +12.8% |
| SBA 504 debentures approved | $7.80bn | $6.66bn | +17.1% |
| Third-party lender dollars | $10.90bn | $9.31bn | +17.1% |
| Combined project financing | $18.70bn | $15.97bn | +17.1% |
| Median debenture | $747,000 | $719,000 | +3.9% |
| Third-party dollars per $1 of debenture | $1.40 | $1.40 | -0.0003 |
| Jobs supported | 71,894 | 62,328 | +15.3% |
Percentages are computed from unrounded values, so a change may not reproduce exactly from the rounded figures displayed.
An SBA 504 debenture is not the size of the project
A 504 project is normally financed in three parts: a private lender takes a first mortgage for roughly half of the cost, the SBA-guaranteed debenture issued through a Certified Development Company covers roughly 40%, and the borrower contributes the rest — typically about 10%, and more for a start-up or a special-purpose property. Quoting only the debenture therefore understates the investment a 504 deal represents by well over half.
SBA’s loan-level file records the third-party lender dollars on the same project, which is what makes the combined figure on this page computable. It is present on 99.85% of FY2025 records; the handful of rows without it are left out rather than estimated, so the combined total is very slightly conservative. The borrower’s own contribution is not in the file at all. “Combined project financing” on this page means the SBA debenture plus the recorded third-party dollars, and nothing else — it is not total project cost, and the true project total is higher.
One further caution. The ratio below is an accounting relationship between two recorded amounts. It shows how 504 deals are structured. It is not evidence that the private money would have been absent without the programme, and nothing here should be read as a causal or “crowding-in” claim.
For every $1 of SBA 504 debenture, $1.40 of third-party lender financing accompanied the projects
In FY2025 the 6,762 approvals carried $7.80bn of SBA debentures and $10.90bn of recorded private third-party lender dollars — a ratio of $1.40 to $1. What is striking is how little it varies: over 16 fiscal years spanning a recovery, a pandemic and a rate cycle, the ratio has stayed between $1.29 (FY2021) and $1.41 (FY2016).
FOIA_504_FY2010_Present_asof_260630.csv, 117,983 records). Fiscal years run 1 October to 30 September.The median 504 debenture has nearly doubled since FY2010
The typical FY2025 debenture was $747,000, up +3.9% on FY2024 and 94.0% above FY2010’s $385,000. It is the largest median of the 16 years analysed. The mean was $1,154,122, well above the median — a minority of large projects pulls the average up, which is why medians are used throughout this page.
FOIA_504_FY2010_Present_asof_260630.csv, 117,983 records). Fiscal years run 1 October to 30 September.Who borrows through 504, and how far deal sizes diverge
19 NAICS sectors used the programme in FY2025. Accommodation and Food Services led on both counts and dollars, taking 16.9% of approvals (1,143 loans) and 24.1% of combined project financing. At the four-digit industry level, 31 industries had 50 or more approvals and 15 had 100 or more; the ten largest accounted for 40.8% of all FY2025 approvals. Among the industries below, median debenture size ranges 6.00× — from Hotels (except Casino Hotels) and Motels at $2,255,000 down to Beauty Salons at $376,000.
| Industry (NAICS-4) | Approvals | Median debenture | Combined financing | Jobs / loan | Third-party per $1 |
|---|---|---|---|---|---|
| Limited-Service Restaurants | 601 | $725,000 | $1.34bn | 16.5 | 1.31 |
| Hotels (except Casino Hotels) and Motels | 414 | $2,255,000 | $2.87bn | 12.5 | 1.71 |
| General Automotive Repair | 366 | $645,000 | $813m | 6.6 | 1.40 |
| Offices of Mental Health Practitioners (except Physicians) | 271 | $389,000 | $352m | 6.9 | 1.27 |
| Child Care Services | 213 | $1,110,000 | $667m | 14.8 | 1.31 |
| Fitness and Recreational Sports Centers | 211 | $1,179,000 | $845m | 12.3 | 1.44 |
| All Other Specialty Trade Contractors | 204 | $666,000 | $400m | 11.1 | 1.25 |
| Other Building Equipment Contractors | 176 | $650,000 | $387m | 11.4 | 1.31 |
| Offices of Physicians (except Mental Health Specialists) | 159 | $668,000 | $426m | 7.3 | 1.33 |
| All Other Personal Services | 143 | $581,000 | $266m | 7.0 | 1.36 |
| Offices of Lawyers | 124 | $583,500 | $209m | 5.3 | 1.25 |
| All Other Professional, Scientific, and Technical Services | 112 | $552,500 | $228m | 11.2 | 1.29 |
| Janitorial Services | 111 | $514,000 | $166m | 10.7 | 1.22 |
| Gasoline Stations with Convenience Stores | 107 | $1,411,000 | $482m | 7.7 | 1.64 |
| Beauty Salons | 105 | $376,000 | $112m | 5.7 | 1.26 |
15 industries with 100 or more FY2025 approvals. The full four-digit table is in the downloads below.
How the oldest 504 loans actually performed
504 debentures run 10, 20 or 25 years, so recent lending tells you nothing about losses. This section uses only FY2010–FY2015 approvals, now 10 to 16 years old. It says nothing about FY2025 loans, which are far too young to have an outcome.
Of those cohorts, 26,972 loans have a resolved status — either paid in full or charged off — and 2.80% of them were charged off. A further 11,351 loans from the same years are marked exempt in SBA’s file, meaning the status is withheld; they cannot be scored either way and are excluded from the denominator. If exempt loans differ systematically from resolved ones, these rates are biased by an unknown amount.
The dispersion is the finding. Accommodation and Food Services (5.39%) and Educational Services (5.29%) carry the highest rates — they are named together because the difference between them is not statistically significant at 95% — while Transportation and Warehousing is lowest at 1.12%, a spread of 4.82×. Note that the sector borrowing most heavily through 504 in FY2025, Accommodation and Food Services, is also the one with the highest seasoned charge-off rate.
FOIA_504_FY2010_Present_asof_260630.csv, 117,983 records). Fiscal years run 1 October to 30 September.SBA 504 lending, FY2010 to FY2025
16 complete fiscal years on one consistent population rule. The current partial fiscal year is excluded entirely; it will be added here once it closes.
| Fiscal year | Approvals | SBA debentures | Third-party dollars | Combined financing | Median debenture | Per $1 | Jobs supported |
|---|---|---|---|---|---|---|---|
| FY2010 | 7,833 | $4.43bn | $6.03bn | $10.46bn | $385,000 | 1.36 | 88,123 |
| FY2011 | 7,983 | $4.85bn | $6.52bn | $11.36bn | $387,000 | 1.34 | 91,045 |
| FY2012 | 9,471 | $6.71bn | $8.74bn | $15.45bn | $448,000 | 1.30 | 127,363 |
| FY2013 | 7,708 | $5.23bn | $7.05bn | $12.28bn | $429,000 | 1.35 | 92,413 |
| FY2014 | 5,885 | $4.20bn | $5.84bn | $10.04bn | $444,000 | 1.39 | 67,787 |
| FY2015 | 5,787 | $4.30bn | $5.89bn | $10.19bn | $464,000 | 1.37 | 62,781 |
| FY2016 | 5,938 | $4.74bn | $6.69bn | $11.43bn | $486,000 | 1.41 | 63,256 |
| FY2017 | 6,218 | $5.01bn | $6.95bn | $11.97bn | $494,000 | 1.39 | 60,579 |
| FY2018 | 5,874 | $4.75bn | $6.61bn | $11.36bn | $497,000 | 1.39 | 56,391 |
| FY2019 | 6,099 | $4.96bn | $6.79bn | $11.75bn | $516,000 | 1.37 | 53,347 |
| FY2020 | 7,119 | $5.83bn | $7.77bn | $13.60bn | $523,000 | 1.33 | 57,727 |
| FY2021 | 9,676 | $8.22bn | $10.59bn | $18.81bn | $547,000 | 1.29 | 84,989 |
| FY2022 | 9,254 | $9.21bn | $12.23bn | $21.44bn | $665,500 | 1.33 | 97,310 |
| FY2023 | 5,924 | $6.42bn | $8.97bn | $15.39bn | $691,000 | 1.40 | 64,195 |
| FY2024 | 5,993 | $6.66bn | $9.31bn | $15.97bn | $719,000 | 1.40 | 62,328 |
| FY2025 | 6,762 | $7.80bn | $10.90bn | $18.70bn | $747,000 | 1.40 | 71,894 |
Geography: 51 states and DC recorded at least one FY2025 approval. CA alone took 1,352 of them, 20.0% of the national programme, and the five largest states accounted for 44.2%. Among the 32 states with at least 50 approvals, third-party dollars per $1 of debenture ran from 1.60 in ND down to 1.23 in ID. 64 approvals in PR are reported separately and are not included in the state table. The full state table is in the downloads.
Download the tables
CSV, free to use with attribution to this page. No borrower-identifying field is published.
How these figures were produced
Source. SBA 504/CDC loan-level FOIA file, snapshot 30 June 2026 (FOIA_504_FY2010_Present_asof_260630.csv, 117,983 records). Fiscal years run 1 October to 30 September. The file is archived locally with a SHA-256 hash so these figures stay reproducible
after SBA replaces the snapshot. Period covered: FY2010 to the snapshot date.
Which rows are counted. The headline series is all approvals recorded for a fiscal year.
LoanStatus is deliberately not used to filter it. Two of its values are functions of how long ago a
loan was approved rather than properties of the loan: NOT FUNDED is a pending state that affects only
recent years, and CANCLD accumulates with age (3.43% of FY2025 rows).
Filtering on either imposes a time-varying bias — excluding NOT FUNDED reverses the direction of
the FY2025 change from +12.8% to a fall. Our pipeline asserts on every run that this
trap is still present in the data, so the rule cannot be dropped by accident.
Fiscal year. 1 October to 30 September, as recorded in SBA’s ApprovalFY field. The
current fiscal year is incomplete in this snapshot and is excluded from every trend, ranking and year-over-year
figure on this page.
Money. “SBA 504 debenture” is GrossApproval. “Third-party lender
dollars” is ThirdPartyDollars, present on 99.85% of FY2025 rows;
missing values are excluded, never imputed. “Combined project financing” is the sum of the two and is
not total project cost, because the borrower’s contribution is not in the file. Medians are used for
typical deal size because the distribution is right-skewed; means are shown only where labelled.
Industry. NAICS as recorded by the lender. Sector totals combine the split NAICS ranges — Manufacturing 31–33, Retail Trade 44–45 and Transportation and Warehousing 48–49 — into one sector each, and sectors are labelled with their official NAICS titles rather than the description of any single industry inside them.
Geography. State tables cover the 50 states and DC. Territory approvals are reported separately and are never folded into a national or state figure.
Jobs. JobsSupported is recorded by the lender at approval. It is a programme metric, not a
measured employment outcome, and is never described here as jobs created.
Seasoned performance. FY2010–FY2015 approvals only. The denominator is loans resolved to paid in full or charged off; exempt-status loans are excluded. Sectors are reported only at 300 or more resolved loans. Where two sectors cannot be separated at 95% by a two-proportion z-test they are named jointly rather than ranked.
Verification. Every figure on this page is recomputed from the raw file by an independent QA script that does not share code with the script that produced the page, and the QA itself is mutation-tested: deliberately corrupting any single figure must make it fail. Rounding is applied for display only; all percentages and ratios are computed from unrounded values.
Limitations. The file records lender-reported approvals, not disbursements, completed projects or business outcomes. It cannot separate purchase, construction, refinance and equipment inside a project. The private-capital ratio is an accounting relationship, not a causal claim.
Citing this analysis
These figures may be quoted or republished with attribution to Fundur and a link to this page. The underlying tables are downloadable above, and the method section states every population rule and limitation. If you need a cut that is not published here — a particular state, county, industry or fiscal year — ask and we will derive it from the same file.
Fundur, “SBA 504 Lending Statistics”, 2026-09-17. https://fundur.com/resources/sba-504-lending-statistics/
Where this comes from, and what sits next to it
- U.S. Small Business Administration — 7(a) & 504 FOIA loan-level data, snapshot 30 June 2026
- SBA — 504 loan programme overview
- SBA 7(a) lending by industry — the working-capital and acquisition side of SBA lending
- SBA 7(a) lending by state
- SBA 7(a) lending by metro area
- SBA loans — programme mechanics, eligibility and timelines
- Equipment financing — the non-SBA route for the equipment side of a project
- All Fundur research and guides