The Market Wasn't Hot. The Clock Was.
redactedSeptember was the biggest month for SBA acquisition loans since the SBA started flagging them in its public loan data, in late 2017.
The SBA approved 1,270 loans to buy a business.
The obvious read is demand.
One press release last week ran the headline “SBA 7(a) Lending Hits Five-Year High as Growth Shifts Toward Larger Loans and Business Acquisitions.”
We celebrate. The Silver Tsunami has arrived.
Maybe.
But I read it a little bit differently. September wasn’t a boom.
It was actually a deadline. And the most interesting thing about it isn’t the SBA.
It’s us.
What changed on October 1
As most of you already know, on October 1, the SBA’s new rulebook took effect. For an acquisition with a business purchase price of $3M or more, the lender now has to obtain a Quality of Earnings report before it lends. The same rulebook raised the minimum debt-service coverage for initial acquisitions to 1.25x, and that applies at every size.
The rule keys off the date an application gets its SBA loan number. Get the number in September, and the old rules apply.
So everybody tried to get the number in September.
What the data shows
Now look only at the big loans: $2.7M and up, roughly where a $3M purchase lands with 10% down.
In September the SBA approved 234 of them. For reference, a typical month is about 70.
redactedThe month before each rule change, big SBA acquisition loans spiked. Source: SBA 7(a) loan-level data (FOIA), as of 30 SeptemberredactedBusiness acquisitions only; approvals, not closings. October 2025 is blank because of the federal shutdown.
Every deal size roughly doubled, which fits the changes that hit every size.
The loans the QoE rule would cover? They tripled.
But there's a caveat. The QoE rule counts the business purchase price, not the loan, and it leaves out owner-occupied real estate. Unfortunately, the public data shows neither. But what it does show is the loan’s term, and a 20- to 25-year term means real estate is in the deal.
So if we split it that way, the big loans with no real estate in them, the closest match to the deals the rule covers, ran 3.1 times their summer pace. The big loans with real estate ran 3.6 times.
So this wasn’t only the QoE.
Every deal size roughly doubled, and the real-estate deals tripled too.
The whole rulebook changed on 1 October, the coverage minimum included, and anyone with a deal in motion had a reason to beat it. The deals closest to the QoE line just bunched hardest at the end: 59% of them came in the last two weeks before the deadline, against 50% of the real-estate deals.
The calendar doesn’t record demand. It schedules it.
September is the last month of the SBA’s fiscal year, and it is often the busiest. Since 2010 it has been the SBA’s busiest month in 7 of 16 fiscal years. For acquisition loans, it has been the busiest in 5 of the last 8, counting this one.
But year-end on its own doesn’t explain much. For acquisition loans, an ordinary September runs about 1.1 to 1.8 times a normal month. What we found is that the Septembers that ran about twice a normal month all had a deadline in them:
SeptemberredactedUnder the CARES Act, the SBA paid six months of payments on new loans approved between 27 March and 27 SeptemberredactedA new rulebook followed on 1 October.
SeptemberredactedThe SBA’s fee waiver and 90% guarantee ran to 30 September 2021, or until the money ran out.
SeptemberredactedThe new rulebook, on 1 October.
Each ran about 2.1 times a normal month.
September 2020 settles it, because that deadline was the 27th, not the 30th. In the two weeks before it, the SBA approved about 260 to 360 loans a business day. On Monday 28 September, the first business day after, it approved 127.
Still September. Still fiscal year-end.
The rush stopped on the deadline, not on the calendar.
The biggest months outside September tell the same story.
December 2010 is the biggest month in the file: 9,480 loans, about 2.7 times the pace of the six months before, as the Small Business Jobs Act’s fee waivers and 90% guarantee came up on their year-end expiry.
July 2015 is second, as lenders raced to get loans in before the program hit its $18.75 billion annual cap. It did, and approvals stopped on 23 July. And in May 2025, the month before the last rulebook took effect, big acquisition loans doubled against April.
Rules change behavior before they take effect. Every time.
Why we wait
Here is the part I find most interesting. The SBA announced the new rulebook in a notice dated 14 August. That gave buyers, sellers and lenders about seven weeks.
In SBA time, that isn’t long. Lenders commonly quote 60 to 90 days from a complete application to closing, and the SBA loan number only comes once the lender has finished its own approval.
A deal that hadn’t started by mid-August was never going to make it. The rush was the pipeline: deals already in motion, pushed to the front. And for anyone mid-deal, racing was the rational move. Get the loan number in September and the deal went ahead under the old rules.
Of the 278 big acquisition loans approved between 14 August and 30 September, 53% came in the final two weeks, which held 24% of the business days. Week by week from mid-August, big loans ran 20, 18, 23, 27, 39, 72, and then 76 in the last three days. In the same weeks of 2024, with no deadline, they ran between 6 and 23 a week.
Some of that is just pipelines finishing. But a longer runway doesn’t change the shape.
The last rulebook was announced in late April 2025 and took effect on 1 June: a third of the big loans in that window came in its final week. The Jobs Act gave three months of warning in 2010: more than a third of the large loans came in the last two weeks.
However long the warning, the work lands at the end.
Psychologists have a name for every piece of this.
We think we have more time than we do.
In a classic 1994 study, psychology students predicted they would finish their honours theses in 33.9 days on average.
They took 55.5.
Fewer than a third, 29.7%, finished by the date they predicted.
Daniel Kahneman and Amos Tversky called it the planning fallacy: we plan from the best case of our own project and ignore how long projects like it actually take.
Every deal has a seller who is slow with the bank statements, or a landlord who won’t sign the lease assignment, or a franchisor with its own approval queue.
Everyone knows that. But then again, most people may not plan for it on their own deal.
I’ve seen this firsthand building Acquidex. I expected us to be further along by now, but our initial timeline underestimated the development, testing, and validation required. It’s a reminder that even with direct visibility into a project, we can systematically underestimate its complexity.
Kahneman’s fix is a boring one: forecast from how long things like yours have actually taken, not from your own plan.
The cost of starting is today. The cost of skipping is later. Economists call it present bias. People put off tasks whose costs come now and whose payoff comes later, and when the new date arrives, they put them off again. Diligence is the textbook case, here. The fee, the document requests and the awkward conversation with the seller are this week. The surprise in the add-backs is two years out.
Deadlines are where deals get done. In bargaining experiments, agreements pile up just before the deadline, whatever the rules of the game. Union contract talks show the same thing in the field. That isn’t just procrastination. A deal has four or five parties, each on its own clock. A deadline is the one clock they all share. It is what makes the seller find the documents and the lender make the call.
Under pressure, we check less. When time is short, decision-makers speed up, focus on the few things that seem to matter most and drop the rest. In diligence, that means revenue, margins and add-backs get checked. The payroll tax deposits, the full-time crew on 1099s, the customer deposits, the card-sale advances and the vacation owed at closing don’t.
It isn’t just SBA lending
The federal government has a September problem too. Its fiscal year also ends on 30 September. Federal agencies spend about 4.9 times as much in the last week of the year as in an average week, and the IT projects bought in that rush were rated lower in quality. The one agency allowed to carry unused IT money into the next year had neither the spike nor the drop.
Small business sales did it inredactedWith tax rises coming on 1 January 2013, brokers reported 43.4% more closed sales in the last three weeks of December than in the same weeks of 2011, according to BizBuySell. The quarter as a whole was flat. Plenty of the deals that raced the date slipped into January anyway.
Home buyers did it inredactedThe $8,000 first-time buyer credit needed a signed contract by 30 April. In July, existing-home sales fell 27.2%. The National Association of Realtors’ chief economist said consumers had “rationally jumped into the market before the deadline”.
Car buyers did it inredactedCash for Clunkers paid for about 678,000 new cars that summer. Economists Atif Mian and Amir Sufi found that almost all of the extra purchases were pulled forward from the months just after. Within about seven months, the extra sales had disappeared.
And deals done to a date do worse. In private equity, one study looked at funds buying companies from other funds. Deals done late in the buying fund’s investment period, when the money has to be spent, returned 88 cents for every dollar the stock market would have made. The same kind of deal without that pressure did about as well as other buyouts.
In public company M&A, a shorter gap between agreeing a deal and closing it, a proxy for less diligence, goes with lower profits afterwards and more goodwill write-downs.
So who pays for the rush?
Well, it's not the bank.
I checked whether the loans that came in during the last days of a rush do worse than the ones approved earlier in the same window.
And they don’t.
In 2010, 2015, 2020 and 2021, the last-minute loans charged off at the same rate as the earlier ones or less, and late-month loans look the same in ordinary years. Whole rush months don’t do consistently worse than the months around them either.
So this September’s 231 big loans that are still live, after three cancellations, will probably behave like the loans before them. Run them at the rates comparable loans have shown and you get somewhere between 4 and 10 defaults, and $11M to $20M charged off over the life of the loans. The SBA booked at least $752M of charge-offs in fiscal 2026 alone. The rush won’t dent the portfolio.
Every one of the 234 is flagged as collateralized.
The rush is made of the same businesses every time: hotels, gas stations, car washes, liquor stores. Property-heavy, quick to underwrite. More than four in ten of the big loans include real estate, and every one carries the SBA guarantee.
The deadline doesn’t hurt the bank.
But it may hurt the buyer.
Take an illustrative buyer who overpaid by $400K: $100K of add-backs that don’t survive, at four times earnings. That buyer still makes the loan payment. They just make it out of earnings that were never there, for ten years. That never shows up as a default.
It shows up as a business that works harder than it should.
The market pays a little too. After the fee waiver ended in September 2021, acquisition loans fell by about half for the next three months. After May 2025 they fell by about a fifth.
A rush doesn’t create deals. It borrows them.
Isn’t the answer just to plan ahead?
Yes. And let's be honest, it is harder than it sounds, for every reason above. Everyone believes their own deal will close on time. The cost of starting is today.
The deadline is the only clock all the parties share.
Telling a buyer to plan ahead is like telling a thesis student to start earlier. True, and not much use.
What works is changing the structure, not the intention.
One principle I took from prior work experience was the discipline of operating cadence. Monthly business reviews, quarterly planning, annual operating plans, and regular scrutiny of performance against budget.
None of it was particularly sophisticated in isolation. But the overall value came from consistency: identifying deviations early, challenging assumptions, and making adjustments before small problems became material.
But disciplined execution doesn't eliminate uncertainty. Projects encounter delays, financial discrepancies emerge, and counterparties become unresponsive.
Individually, these events may be difficult to predict. Collectively, some level of disruption is almost inevitable.
The thesis study illustrates this well. Even participants' pessimistic estimates tended to run short: 48.6 days on average, against an actualredactedThe authors note that gap was small enough to be chance, but it pointed the same way. The lesson isn't simply to build more contingency into a forecast. It's to establish a process that continuously tests the assumptions behind it.
A good operating cadence doesn't prevent surprises. It reduces the cost of discovering them too late.
Here are 5 useful disciplines that I would include in standard in buy-side advisory:
1. Manage the critical path, not just the closing date.
Establish a weekly review of outstanding diligence, dependencies, accountable parties, and emerging risks to completion. The objective isn't more reporting. It's identifying slippage early enough to intervene, rather than discovering in the final week that a critical issue has been unresolved for a month.
2. Build timelines around historical outcomes, not expectations.
Check the actual LOI-to-close periods across comparable transactions. Account for financing, financial reporting quality, legal complexity, and third-party approvals. Historical completion times already show some of the friction that optimistic forecasts tend to ignore. A seller's preferred closing date is not necessarily an execution plan.
3. Challenge the earnings before the valuation becomes an anchor.
Before signing an LOI, pressure-test the financial assumptions supporting the purchase price. Examine the earnings quality, material add-backs, customer concentration, and obvious inconsistencies in the financial statements. A comprehensive QoE review may follow during exclusivity, but fundamental weaknesses in the earnings story should be identified as early as access permits.
4. Where timing is constrained, address the residual risk explicitly.
Not every uncertainty can be resolved before close. Identify and quantify outstanding exposures, then determine whether purchase price adjustments, escrows, indemnities, or specific closing conditions can provide appropriate protection. Contractual safeguards should address defined risks, not compensate for diligence that was never performed.
5. Understand whose deadline is driving the transaction.
A seller targeting a December 31 close, a lender approaching a credit approval deadline, or a buyer facing a regulatory change may all have legitimate reasons for urgency. But deadlines influence negotiating leverage, decision-making, and willingness to accept uncertainty. Understanding those incentives is part of assessing the transaction itself.
As year-end approaches, more transactions will face pressure to close before December 31.
The question isn't whether a deal can close on time. It's whether the assumptions supporting the transaction have survived sufficient scrutiny before the money changes hands.
What this data cannot tell you
These are approvals, not closings. Some rushed approvals will cancel. Three of the 234 already have.
I can’t separate the QoE rule from the rest of the rulebook. The coverage minimum and the equity rules changed on the same day. Big loans jumped more than small ones, which fits the QoE rule. But the big real-estate deals jumped most of all, so I can’t pin it on the QoE.
$2.7M is a proxy. Some loans include working capital and fees; some $3M deals borrow less. And the rule’s $3M test leaves out owner-occupied real estate, which the file doesn’t separate. The loan term is the best stand-in it offers.
Charge-offs measure the lender’s loss, not the buyer’s. Overpaying never shows up in this file. That part of the argument rests on the research above, not on SBA data.
Not every spike is a deadline. January 2026 ran high too, after the shutdown, and I haven’t tied it to anything.
The full method and every count are in the site version of this piece: the method, the monthly numbers and the daily data [add the site post link at publish].
The test is coming. If September was timing, not demand, big acquisition loans should run below the 70 pace from October to December. The 31 December loan file, expected around late January, will show it. We’ll publish the answer either way.
The verdict
The best month ever was a deadline.
The deadline wasn’t the problem. Leaving the work until it arrived was.
Sources: SBA 7(a) loan-level data (FOIA), fiscal 2010 to 2019 and fiscal 2020 to present files, as of 30 September 2026, published 30 October 2026, analysed 9 NovemberredactedSBA SOPredactedand Information NoticeredactedAugustredactedSOPredactedeffective 1 JuneredactedCARES Act sectionredactedCongressional Research Service R41146 (Small Business Jobs Act). SBA 7(a) program cap, JulyredactedBuehler, Griffin and Ross, Journal of Personality and Social PsychologyredactedKahneman and TverskyredactedO’Donoghue and Rabin, American Economic ReviewredactedRoth, Murnighan and Schoumaker, American Economic ReviewredactedPayne, Bettman and Johnson, Journal of Experimental PsychologyredactedLiebman and Mahoney, American Economic ReviewredactedBizBuySell Insight Report, Q4redactedNational Association of Realtors, 24 AugustredactedMian and Sufi, Quarterly Journal of EconomicsredactedDegeorge, Martin and Phalippou, Journal of Financial EconomicsredactedWangerin, Contemporary Accounting Research (2019).