Buying a franchise three different ways, and what happens to the loan

The finding

Franchise loans approved FY2018–2022, by what the loan was for:

A new-unit startup fails on its loan roughly three times as often as a resale, and the gap is not an artifact of how the rate is framed — it holds whether you measure against finished loans or against every loan booked.

It matters because startups are the majority of the market. 52% of franchise loans in this window funded a new unit — the path with three times the charge-off rate is also the most travelled one.

The part that surprised us

The three paths fail on roughly the same schedule. Median time from approval to charge-off:

Nobody fails fast. The common story — that a bad new unit reveals itself in year one — is not what the file shows. Failures cluster around years three and four across every path.

So it is not that startups fail sooner. They fail more often, on the same clock. Anyone underwriting their own ramp should plan for a runway measured in years rather than months, whichever way they buy in.

What a resale costs

Resales are not the cheap option:

A resale borrows about 2.5x what a new unit does. That is the trade the data describes: you pay considerably more, and you are buying a unit with a trading history instead of a projection. On this evidence the premium buys a real reduction in risk — though it does not follow that it is priced correctly, only that the outcomes differ.

How this was measured, and what it does not show

The whole point of a number like this is the denominator, so:

Source

SBA 7(a) FOIA disclosure files, FY2010–FY2019 and FY2020–present, as published 30 June 2026. Restricted to loans carrying a franchise name.

“Finished” means finished

A rate is computed only over loans that were repaid in full or charged off. Loans still running are neither a success nor a failure, and counting them as repaid flatters everything.

SBA withholds a lot

The agency publishes no outcome for a large share of loans — 40% to 50% depending on the path:

Those loans are excluded, not assumed good. The withheld share is similar across all four paths, so it is unlikely to be generating the gap — but it is a real limit and anyone quoting these figures should quote it too.

The window is five years, not fifteen

SBA only began populating the BusinessAge field in FY2018. Every one of the four paths can only be compared from that year forward, which is why this covers five approval cohorts rather than the full file.

“Expansion” is an inference

SBA’s label is “Existing or more than 2 years old”. Reading that as an existing operator adding a unit is our interpretation; the file does not say so.

Recent cohorts are not finished

A 2018 loan has had years to fail; a 2022 loan has not. Later vintages are weighted toward the failures that happen early, which cuts both ways — it can overstate risk for a brand that simply lent recently.

Why nobody reports this

The field is in the file, sitting next to the loan amount. It is not in the franchise disclosure document, it is not in any brand’s marketing, and it is not in the aggregate default rates usually quoted — which pool all three paths into a single number that describes none of them.

That single number, for franchise loans in this window, is 10.2%. It is also the least useful figure here: it is a blend of a 15% path and a 4% path, weighted by how many people took each.