Open dataset · SBA 7(a) Loan Outcomes · doi:10.5281/zenodo.24177485

SBA 7(a) Loan Outcomes by Franchise Brand and Industry 2026: the open dataset behind the SBA Loan Outcome Check

Of 1,168,658 loans approved in fiscal 2000 to 2021 (at least five years old), 15.6% were charged off by the lender. Among the 602 franchise codes with 30 or more such loans, the charged-off share runs from 0% to 73%, median 10%. Counts of loans, not ratings of brands. Version 1.0, published 13 November 2026. Every figure is reproducible from the files below; try it on your own case in the SBA Loan Outcome Check.
Charged-off share of 7(a) loans approved in fiscal 2000 to 2021

All loans

15.6%

Median franchise code (602 with 30 or more loans)

10%

Highest franchise code

73%

What is in it?

1,416,244 disbursed SBA-guaranteed 7(a) loans aggregated by franchise brand (3,293 codes with 5 or more loans) and by six-digit NAICS industry (1,412), by approval fiscal year and status (paid in full, charged off, not yet resolved). Borrower names, addresses and lender names are never read into any file.

How should it be read?

A charge-off is a loan the lender wrote off; part may be recovered later, so the charge-off amount is not the final loss. Only SBA-guaranteed 7(a) loans are in the file. A franchisee's loan is not a measure of the franchisor. Recent approval years are shown by count and status only because a loan's outcome depends on its age.

What are the files?

The files are hosted on Zenodo with the DOI. Each has a SHA-256 hash, so a downloaded copy can be checked against it.

FileSizeSHA-256 (first 16)
README.md2 KB1d9c453f0138f5ee
groups.csv254 KBfad497f2676da95d
cohorts.csv3.4 MBdede15c98ef97513
data.json853 bytes87dabd4d517c0b22
control-result.json239 bytes46d48f9b4f29de66
PROTOCOL-SBA-LOAN-OUTCOMES-2026-11-12.md4 KB8e3661619a8e4a89
protocol.sha256107 bytese07ecb0fd030793d
code.zip25 KB61a5593926fba5f1
SHA256SUMS.txt737 bytes

How was it checked?

The method was frozen before any figure was published, on 12 November 2026; its SHA-256 hash begins 8e3661619a8e4a89. A separate reader recomputed figures from the raw files: a random sample of 300 recomputed, 0 mismatches. Read how the SBA Loan Outcome Check was checked, including what the check found and what it does not cover.

What is not claimed?

That a franchise brand or industry is safe, risky, good or bad; that a share predicts a new loan; any cause; anything about a borrower or lender.

Has it been corrected?

None yet. Control: 300 sampled groups, 0 mismatches against the raw CSV rows.

How do I cite it?

Couey, V. W. (2026). SBA 7(a) Loan Outcomes by Franchise Brand and Industry 2026: Charge-off Shares by Approval Year for 1,416,244 Loans (Version 1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.24177485

Cite all versions: doi:10.5281/zenodo.24177484. This version: doi:10.5281/zenodo.24177485.

Where does the data come from?

US Small Business Administration, 7(a) FOIA loan data, data.sba.gov, as of 30 September 2026, downloaded 2026-11-12. Compiled by Vincent Wesley Couey (ORCID 0009-0005-6869-308X). Tool: https://ceocult.com/tools/sba-loan-outcome-check

Questions about this dataset

What is in the SBA 7(a) Loan Outcomes dataset?

1,416,244 disbursed SBA-guaranteed 7(a) loans aggregated by franchise brand (3,293 codes with 5 or more loans) and by six-digit NAICS industry (1,412), by approval fiscal year and status (paid in full, charged off, not yet resolved). Borrower names, addresses and lender names are never read into any file.

Can I reuse it?

Yes. The compilation is licensed CC BY 4.0: reuse it with credit, citing the DOI (10.5281/zenodo.24177485). The underlying records keep the terms of their public source.

How was it checked?

The method was frozen and hashed before any figure was published (12 November 2026), and a separate reader recomputed figures from the raw files: a random sample of 300 recomputed, 0 mismatches. The full account is on the how-we-check page for the tool.

What does it not claim?

That a franchise brand or industry is safe, risky, good or bad; that a share predicts a new loan; any cause; anything about a borrower or lender.

Compiled by Vincent Wesley Couey (ORCID 0009-0005-6869-308X).

☆ Save