What If the Silver Tsunami Depends More on Lenders Than Retiring Owners?
redactedNear Minneapolis a business changes hands with an SBA loan 5.6 times as often as near New York. 15 tests of why, and what the closest one shows.
Part five of five.
This is the last one, and it is the least settled.
Among the 40 largest metro areas, a fast market has 2.1 to 2.3 times the SBA-financed sales of a slow one, for its size.
I tried 13 explanations, and then two more. I show every one, including the ones that went nowhere.
The closest was not on my list. A reviewer raised it: which lenders work where. Add how much they lend, and it lines up with the map better than anything I had tried. It changed what I think the map shows.
I still do not have the answer. I would rather end on that than on a tidy one I cannot support.
Among the 40 largest metro areas in the United States, a fast market has 2.1 to 2.3 times the SBA-financed sales of a slow one, for its size. Chance alone would give 1.2. We tried 13 explanations, and then two more. One at a time, none accounts for the gap. The closest is which lenders work where, and how much they lend. So this may be a map of SBA lending as much as a map of sales. This piece lays out the map and every test.
1. The same business, a different market
Four pieces have asked what is out there and what changes hands with an SBA loan. This one asks where. It is the least settled question in the series, so it is laid out in full: the map first, then every explanation we tried.
Start with the two ends, as piece 1 did. The five boroughs of New York City hold 120,421 in-scope business locations. Those are locations in industries that clear our licence screen, leaving out 25 industries in finance, property and gambling. In the six fiscal years 2020 to 2025, 100 of them changed hands with an SBA 7(a) loan, and 2 more through 504.
Marinette County, Wisconsin, is the other end. It has 564 such locations, and 17 changed hands. That is 36 times the rate. Counted once per borrower, the counts are 93 and 14. That is 32 times the rate.
Neither end is typical, and small places bounce. Marinette's 17 sales are a small number. So the fair comparison is among big markets, where chance has less room.
Among big markets, the gap in SBA-financed sales is wider than chance alone would give. Take the 40 metro areas with at least 20,000 in-scope locations. They hold 54% of all such locations in the United States. A fast one has 2.3 times the SBA-financed sales of a slow one, for its size. If all 40 sold at the same rate and only chance differed, the gap would be 1.2 times.
That 2.3 is the high reading. Some sales take two loans. Count each borrower once, and the gap is 2.1 times. Move the size bar anywhere from 10,000 to 30,000 locations, and it runs from 1.9 to 2.4. So the size of the gap depends on how we count.
At the two ends, businesses in Minneapolis–St Paul change hands with a 7(a) loan 5.6 times as often as in the New York metro area. At the national rate, the New York area would have had about 1,410 such sales in six years. It had 588. Minneapolis–St Paul would have had 256. It had 597. Both areas had almost the same number of such sales, though the national rate would give the New York area far more.
The New York area had 588 SBA-financed sales where the national rate would give 1,410. Minneapolis had 597 where it would give 256.
The gap is not fixed. In fiscal years 2020 to 2022 it was 2.7 times, and Minneapolis ran at 6.6 times New York. In 2023 to 2025 it was 2.2 times, and Minneapolis ran at 4.8 times New York. Chance alone would give 1.3 in three years. So the gap was narrower in the later years, and still wider than chance alone would give. The New York area had 157 such sales in fiscal 2025, against 80 the year before.
In short
Among the 40 largest metro areas, a fast market has 2.3 times the SBA-financed sales of a slow one, for its size. Counted once per borrower it is 2.1, and chance alone would give 1.2. Minneapolis–St Paul runs at 5.6 times the New York area. The rest of this piece is the map, and 15 attempts to explain it.
What these numbers count. Every sale in this series is an SBA-financed sale. The Small Business Administration (SBA) releases loan-by-loan records for its two lending programmes, 7(a) and 504. They are the only free, public, deal-level record of business sales that we know of. Our best reading of Census data is that the SBA finances about a tenth of business sales. So every count of sales here is a floor, and "sells" means "sells with an SBA loan". Industries lean on SBA lending to different degrees, and nothing public that we know of shows what share of each industry's sales it finances. Rates for places in this piece count 7(a) loans only, as in piece 1. Adding the smaller 504 programme barely changes the order of places. A low rate for a place may mean few sales, or few sales that use an SBA loan. This piece cannot tell those apart.
2. Metro areas and states
Here are the 40 largest metro areas, in order.
redacted14 of the 40 had clearly more SBA-financed sales than the national rate would give, and 14 clearly fewer. The other 12 were too close to call. So most of these metro areas sit clearly above the national rate or clearly below it. Minneapolis, Denver, Milwaukee, Indianapolis and Seattle lead. New York, Miami, San Jose, Nashville and Washington trail.
Size does not decide it. New York, Los Angeles and Miami hold 620,488 in-scope locations between them, which is 27% of the 40's. They had 1,717 sales where the national rate would give 2,994. But Dallas–Fort Worth and Atlanta, two of the six largest, had clearly more than the national rate would give. Chicago was level with it. So big markets fall on both sides of the national rate.
Across all 251 metro areas with at least 2,000 locations, size and the SBA-financed sale rate do not line up. The link is −0.03, on a scale where 0 means no connection and 1 means they move in lockstep. Among the 40 it is −0.18, which is within chance. So the size of a market says little about how often its businesses change hands with an SBA loan.
Do the very largest markets have fewer SBA-financed sales for their size? Among the 10, 15 and 20 largest the link is −0.31, −0.52 and −0.40. Of those three, only the middle one is past chance. The 14 and 15 largest give the strongest link of any cut from 8 to 40. The link moves with where the cut is made. So the very largest may run slower, but this cannot settle it.
By state the order is similar.
redactedMinnesota, Colorado and Wisconsin lead, at 1.62, 1.59 and 1.47 SBA-financed sales per 1,000 locations a year. Hawaii and New York trail, at 0.29 andredactedThe median state isredactedSo the states that lead are far above the median state, and those that trail are far below it.
States hide a good deal, though. Between 11% and 22% of the differences between counties lie between states. The figure depends on how counties with no sale are treated. Chance alone would give about 4%. So states matter, but most of the difference between two counties lies inside a state. A good part of that is chance, because counties are small. Not all of it is, as the smaller metro areas show.
redactedAmong these 251, a fast metro area has 3.0 times the SBA-financed sales of a slow one, for its size. Chance alone would give 1.7 times. The gap is wider than that. Grand Junction, Colorado, had 35 sales where the national rate would give 12. McAllen, Texas, had 2 where it would give 30.
McAllen is one of four metro areas on the Texas border near the bottom of the list. The others are Brownsville, Laredo and El Paso. Together they hold 20,301 in-scope locations. In six years they had 18 sales with a 7(a) loan. The national rate would give 98. Texas as a whole sits at the median. So the state's average hides them.
In short
14 of the 40 largest metro areas had clearly more SBA-financed sales than the national rate would give, 14 clearly fewer, and 12 were too close to call. Size does not tell you which. Four metro areas on the Texas border had 18 sales where the national rate would give 98, while Texas as a whole sits at the median. State averages hide a good deal of the map.
3. Small places bounce
One level down, the spread looks dramatic. Much of it is chance.
Among counties, a fast one sells with a 7(a) loan 6.5 times as often as a slow one. But chance alone would give 3.5 times. That is across 1,253 counties with at least 400 in-scope locations. 67 of those counties had no sale with a 7(a) loan in six years.
So we do not rank counties. Here are the eight fastest on the numbers. Every one rests on fewer than 20 sales. Their location counts are estimates too, because Census withholds small cells. Read the table as a warning, not a league.
redactedWhat counties do show is that a state that is slow for SBA-financed sales is slow in most of its counties. Half of all counties are below the national county median for these sales, by definition. In New York state it is 38 of 53, or 72%. In Tennessee it is 86%, and in Virginia 74%. In Minnesota it is 9%, and in Wisconsin 19%. Of the 12 slowest states with five or more such counties, 11 have most of their counties below that median.
In short
A fast county sells with a 7(a) loan 6.5 times as often as a slow one. Chance alone would give a gap of 3.5 times, so we do not rank counties. What counties do show is that slow states are slow in most of their counties: 72% of New York's are below the national county median, against 9% of Minnesota's.
4. Where the turnkey businesses are
Piece 1 counted turnkey locations by state: business locations with 20 or more staff in the 34 industries that pass on the screen as built. Set them beside the sale rates.
redactedCalifornia holds 15,894 turnkey locations, the most of any state. Its in-scope locations of every size change hands with a 7(a) loan at 0.69 per 1,000 a year. New York holds 6,261, and its rate isredactedThe five fastest states hold fewer than 3,200 each.
So the biggest pools are not the fastest markets for SBA-financed sales. But they are not the slowest either. Five of the six largest sit within 15% of the median state, which isredactedOnly New York is far below. It is 50th of the 51.
Count only the 34 industries, and the picture is much the same. We set 7(a) sales in those industries against their own locations, state by state. The order of states matches the order for all in-scope industries at +0.81.
Some states move. Florida is 25th on all in-scope industries and 15th on the 34. New Hampshire goes the other way, from 4th to 21st. On the 34 alone California is 34th, Texas 22nd, Illinois 23rd, Pennsylvania 28th and New York 49th. So on this count too, the biggest pools do not lead.
Do states with bigger pools have SBA-financed sales more often, or less often? Line up all 51 by their number of turnkey locations, and again by the sale rate for all in-scope industries. The two orders match at +0.25, which is within chance.
Against the rate for the 34 industries alone, the match is stronger. But it comes from the small end. It is +0.37, and chance gives that about 1 time in 130. Nine of the ten states with the smallest pools are below the median state on that rate. Each had fewer than 30 such sales in six years, so their rates are rough. Among the other 41 the link is +0.02. There, pool size tells you nothing.
So a big pool is neither a fast market nor a slow one. The number of turnkey locations in a state says little about how often its businesses change hands with an SBA loan.
6,261 turnkey locations are in New York, the fourth-largest pool of any state. Its in-scope locations change hands with a 7(a) loan at 0.36 per 1,000 a year, 50th of the 51. It is the one big pool that is clearly slow.
In short
Five of the six largest pools of turnkey locations sit within 15% of the median state, and only New York is far below it. On the 34 industries alone, nine of the ten states with the smallest pools are below the median. Among the other 41, pool size tells you nothing. A big pool is neither a fast market nor a slow one.
5. Each one taken in turn
Why would a business in Minneapolis change hands with an SBA loan so much more often than one in New York? We tried 13 explanations. Here is every one, with what it came back as. Most were tested across states, because that is where the data are. Section 7 runs those we could on the 40 metro areas.
The 13 explanations, and what each came back as
• 1. Industry mix differs from place to place: Explains 0.6% of the differences between states. Mix alone would put the fastest state at 1.36 times the slowest. It is 5.6 times.
• 2. Credit is harder to get in slow places: Cannot be tested directly. We see only loans that were made, and SBA rules set their length and guarantee.
• 3. Lenders are missing from slow places: In the median state, 64.4% of loans carry a lender's address in another state. Fast states have slightly fewer: a faint link of −0.28. But the file holds one address for each bank.
• 4. Lending follows an old map: A place's share of SBA lending in the 1990s still matches its share today (+redactedBut shares mostly measure size. Lending per location in the 1990s has no clear link with today's sale rate (+0.08).
• 5. People start businesses instead of buying: No clear link between applications to start a business with planned wages and sale rates (−0.10).
• 6. Costs are higher in slow places: No clear link with the cost of living (−0.07).
• 7. Slow places have smaller businesses: Count only locations with five or more staff and the states rank almost the same (+0.95).
• 8. Older owners mean more sales: The opposite. States with more owners aged 65 or over have fewer SBA-financed sales (−redactedAcross all counties it is −0.07, and in the 182 largest counties −0.29.
• 9. Places with more non-citizen owners sell less: A faint link across states (−redactedAmong the 40 largest metro areas it is −0.56, the strongest of these 13. We do not know why.
• 10. Deals are built differently: Little sign of it. The median loan is $699,900 in the slowest quarter of states and $615,000 in the fastest. 22% and 20% report 20 or more jobs.
• 11. High prices show buyers competing: Places with larger loans have fewer SBA-financed sales (−0.30, 61 metros). That fits competition, or its opposite.
• 12. Too many buyers chase too few businesses: Two indirect tests, nothing clear in either (−0.02 and +0.09).
• 13. Licensing business brokers helps: The median of 15 states that a public list names as licensing brokers has 1.28 times the sale rate of the median of the other 36. The gap is about the same whether or not the state licenses the sale of a business as such: 1.28 andredactedBut 7 states against 8 cannot settle it.
How to read the links. With 51 states, chance alone gives a link of 0.28 or stronger about 1 time in 21. Read anything fainter across states as nothing. Even that is generous, because neighbouring states resemble each other: the nine Census divisions alone hold 36% of the differences between states, where nine random groups would hold 16%. Across 1,251 counties even −0.07 is more than chance would give, but it is small.
Six of those deserve a paragraph.
Credit
We cannot test whether credit is harder to get where SBA-financed sales are slow. The file holds only loans that were made. So we cannot see a loan that was turned down.
The loans' length and guarantee say nothing either way, because SBA rules set both. As a rule, a 7(a) loan over $150,000 can carry a guarantee of 75% at most. A loan that pays for no property or long-lived equipment can run ten years at most. 71% of these loans sit at exactly 75%. Of those not since charged off, 64% run exactly 120 months.
That count leaves out the failed loans. The file rewrites a failed loan's length, as piece 4 explains. Leave out loans the March file marks as purchased or in liquidation as well, and it is 65%. That reading of the file's two status codes is ours, as piece 4 says. Either way the figure barely moves.
Interest rates differ a little. The median rate was the same in the slowest and the fastest quarter of states in fiscal 2020, 2022 andredactedIn the other three years it was up to 0.5 points lower in the fastest. So in the years when the rates differed, borrowing cost a little less in the fastest group of states.
Lenders
The money seems to travel. 1,324 lenders made these loans, and one bank made 9.0% of them. In the median state, 64.4% of loans carry a lender's address in another state.
But the file holds one address for each bank. A loan from a branch down the road counts as made from elsewhere when that address is in another state. So this test says little about whether a local lender was there. Section 8 asks a better question: which lenders.
Owner age
This is the explanation the wave story needs, and it runs the wrong way. States with a larger share of owners aged 65 or over have fewer SBA-financed sales, not more. The link is −0.45, and it accounts for 24.5% of the differences between states.
Across counties and metro areas the link is fainter. Across all 1,251 counties it is only −0.07. But county figures come from a sample survey, and are rough for small places. In the 182 counties with 5,000 or more locations it is −0.29. Across the 251 metro areas it is −0.23. None of these links runs the way the wave story needs.
Loan size
Places where loans run larger have fewer SBA-financed sales. That could be buyers bidding up a short supply. It could be costlier businesses proving harder to sell. Loan data cannot tell those apart.
Broker licensing
Piece 1 said brokers "publish lists of 15 states that require a licence to sell a business".
The list we can trace is on Wikipedia's page on business brokers. It names 16 states: these 15, and Illinois, marked "registration only". The median state among the 15 has 1.28 times the sale rate of the median of the other 36. With Illinois counted as listed, 16 states against 35, it isredactedSo counting Illinois barely changes the gap.
We read the 15 laws. By our reading, 7 require a licence to sell a business as such. The other 8 require one only where real estate or a lease is part of the deal. Set each group against the other 36 states, and the gap is about the same: 1.28 and 1.27 times. Piece 1 set each group against every other state, and printed 1.25 and 1.24.
That does not settle it. It is 7 states against 8, and the split is our reading. Arizona's law defines a business broker by deals "where a lease or sale of real property is either a direct or incidental part of the transaction". So Arizona could sit in either group. Move it, and the two figures part: 1.17 andredactedWhere that single state sits changes the result.
Nor are the 36 a clean comparison. Illinois is among them, though the Wikipedia page lists it. Washington is not on the list. Yet its real estate licensing law defines a "business opportunity": a business sold "when the transaction or business includes an interest in real property". That is close to how we read the 8. We did not read the other 35.
Buyers
We could not count buyers. So we looked at where issuers that offered securities in connection with a merger or acquisition are based. An issuer that sells securities privately files a notice with the SEC. The notice asks whether the offering is "in connection with a business combination transaction", such as a merger, an acquisition or an exchange offer.
In eight quarters, 1,408 new notices from issuers based in the 50 states or DC said yes. 14% were from investment funds, and the rest from issuers of other kinds. 37% were based in New York, California and Texas, which hold 27% of in-scope locations. That is more than their share of locations.
Neither of two tests found a link with SBA-financed sale rates. One caution: a notice records where the issuer is based, not where it buys.
In short
13 explanations, taken one at a time, and none accounts for the gap alone. Industry mix explains 0.6% of the differences between states. Older owners go with fewer SBA-financed sales, not more. Credit cannot be tested from loans that were made, and 15 states cannot settle the broker-licensing test.
6. Is it just where the SBA lends?
The second piece in this series found that much of the gap in sale rates between industries tracks how much each industry uses SBA loans at all. So the same has to be asked of places. Is a slow market just a place where the SBA lends little of anything?
Partly. More SBA lending of other kinds goes with more SBA-financed sales. We counted every other 7(a) loan in the same industries and years: 161,607 of them, for any purpose but buying a business. Places with more of that lending per business location do have more SBA-financed sales. Rank places by each, and the two orders line up at +0.45 across the 51 states, and at +0.47 across the 40 largest metro areas. That accounts for 21% and 22% of the differences. So this lending goes with part of the differences, not most of them.
But those other loans are not like loans to buy a business. 50% of them are for $150,000 or less, against 12% of the loans to buy a business. 42% are SBA Express loans, against 14%. 31% are revolving lines, against 10%. So we counted again, five more ways, keeping the loans most like loans to buy a business.
Other SBA lending, like for like: loans a year per 1,000 in-scope locations
• Every other loan: New York area 6.49, Median of theredacted, New York, from the bottom of the 40 16th, Share of the differences among the 40 it accounts for 22%
• Not a revolving line: New York area 3.77, Median of theredacted, New York, from the bottom of the 40 4th, Share of the differences among the 40 it accounts for 34%
• Not SBA Express: New York area 2.96, Median of theredacted, New York, from the bottom of the 40 6th, Share of the differences among the 40 it accounts for 29%
• Over $150,000: New York area 2.85, Median of theredacted, New York, from the bottom of the 40 6th, Share of the differences among the 40 it accounts for 41%
• Over $350,000: New York area 1.57, Median of theredacted, New York, from the bottom of the 40 5th, Share of the differences among the 40 it accounts for 30%
• Over $350,000, not Express, not a revolving line: New York area 1.41, Median of theredacted, New York, from the bottom of the 40 4th, Share of the differences among the 40 it accounts for 21%
On every count but the first, the New York area is short of SBA lending. On loans over $350,000 it has 1.57 a year per 1,000 locations, against 2.21 for the median of the 40. That is fifth from the bottom.
And the like loans account for more. Loans over $150,000 account for 41% of the differences among the 40, and 46% between states. Across the six counts it runs from 21% to 41% among the 40, and from 21% to 46% between states.
But the gap survives every count. Take sales per 100 other loans, not per location. Among the 40, a fast market still runs 2.1 to 2.5 times a slow one, whichever loans are counted. Chance alone would give 1.2. The gap is still wider than that.
And the New York area is last of the 40 for sales per loan on every one of those counts. It has 5.2 sales for every 100 other loans of any kind. The median is 11.1, and Minneapolis–St Paul has 18.0.
The four Texas border metro areas are low on both counts. They have 3.9 other loans per 1,000 locations a year, against 6.6 across all 251 metro areas. And they have 3.8 sales per 100 other loans, againstredactedSo they have less SBA lending of other kinds for their size, and fewer SBA-financed sales for each of those loans.
5.2 sales for every 100 other SBA loans in the New York area, the lowest of the 40 largest metro areas. 56% of those loans are for $150,000 or less. On loans over $350,000 New York is fifth from the bottom for lending, and still last for sales per loan.
In short
Places with more SBA lending of other kinds have more SBA-financed sales. That accounts for 21% to 46% of the differences, depending on which loans are counted. Counted per 100 other SBA loans, a fast large metro area still has 2.1 to 2.5 times the sales of a slow one. The New York area is short of the larger SBA loans, and last for sales per loan on every count.
7. The same tests where the headline is
The headline is about 40 metro areas, and most of those tests were run on states. So we ran every one we could on the 40. Then we added them up.
redactedChance is counted by dealing one of the two orders at random 4,000,000 times. Figures over 100 are rounded to the nearest 50, and over 1,000 to the nearest 1,000. The share of the differences is measured on a log scale of the sale rate. Industry mix has no link figure: the test asks what rate each metro area's mix of industries would give.
Three things stand out. First, the citizenship link is far stronger among the 40 than across states. It is −0.56, and chance gives that about 1 time in 5,000. In the New York area 7.3% of owners are not US citizens, and in Minneapolis–St Paul 1.6%. The link holds without San Jose and Miami, the two with the most such owners, at −0.49. We do not know what is behind it. The next section shows it weaken once the lender test of that section is held constant.
Second, slower metro areas paid slightly more for these loans (−0.39), and larger loans go with fewer sales (−0.32).
Third, the brokerage link is not there among the 40. It is +0.13, well within chance.
Now add them up. Across the 51 states, five measures each line up with the sale rate at 0.28 or more. They are other SBA lending, owners aged 65 or over, brokerage offices, owners who are not citizens, and the lender's address. Taken together they account for 58% of the differences between states.
That flatters them. Five measures fitted to 51 states will always account for something. A fairer sum predicts each state from the other 50, and gives 42%. The best five of 12 unrelated measures would give about 18% by luck on the first sum, and nothing on the second. So the fairer sum gives more than luck would.
Among the 40, four kinds of measure pass the same sort of bar: chance gives each link less than 1 time in 20. They are citizenship, other lending, the interest rate and loan size. With owners as the citizenship measure, the four account for 41%. Predicting each metro area from the other 39 gives 23%.
So "none of them explains it" was true one at a time, and too quick for states. On the fairer sum the five account for 42% of the differences between states, where the best single measure gives 17%. Two of the five are not among the 13 explanations: other SBA lending and brokerage offices. The three that are give 23%. The 15 listed states pass the same line as a group, and with them it is 26%. All six give 40%.
Among the 40 the four give 23%, and citizenship alone gives 25%. So together they give no more than citizenship alone. None of this says why. These are places that differ in many ways at once.
In short
Run on the 40 largest metro areas, the strongest of the first 14 tests is the share of owners who are not US citizens, at −0.56. Other SBA lending is +0.47, and brokerage offices +0.13. On the fairer sum, five measures together account for 42% of the differences between states. Among the 40, four together give 23%, no more than citizenship alone. One at a time, none accounts for the gap.
8. Test 15: which lenders work where, and how much
One explanation was not on our list. A reviewer of this piece raised it. Lenders differ. Some finance the purchase of a business, and many hardly ever do. So which lenders are active in a place could shape how many of its sales carry an SBA loan.
One caution on the word lender. The file names the bank that holds each loan now, which may not be the bank that made it.
The differences between lenders are large. Across the United States, about 11% of in-scope 7(a) loans buy a business. For the largest acquisition lender it is 52%. For the bank that made the most other loans in the New York area it is 0.7%.
The New York area shows what that means. One bank made 27% of its other SBA loans. Three banks made 46%. Across the United States, 0.7%, 5.0% and 0.1% of those three banks' loans buy a business.
Minneapolis–St Paul is served differently. 95 lenders made its 597 loans to buy a business, and 110 made the New York area's 588. The numbers are alike.
What differs is who makes each area's other SBA loans. From their records elsewhere, those lenders would make 8.2 loans to buy a business for every 100 other loans in the New York area. That is the second-lowest of the 40. In Minneapolis they would make 13.8, and the median isredactedOn the eight versions of the test below, New York's figure is lowest or second-lowest each time. Minneapolis's runs from 11.4 to 21.0.
So we built a test. For each metro area, take each lender's other SBA loans there. Multiply by that lender's loans to buy a business per other loan in every other place. Add up. That is the number of SBA-financed sales the metro area would have if its lenders behaved there as they do elsewhere. The metro area's own sales are left out of each lender's record.
That expected rate lines up with the real one at +0.72 across the 40. Call it the lender sum. It accounts for 58% of the differences. On the fairer sum of section 7, which predicts each metro area from the other 39, it is 53%. That is more than any other test here.
The lender sum has two parts. The volume of other lending alone accounts for 22%. The lenders' habits alone account for 21%. Together they give the 58%. Counting only other loans over $150,000, the two parts are 41% and 27%, and the sum 62%. So the sum is as much about how much SBA lending a place has as about which lenders do it.
redactedWe ran it eight ways, and the link runs from +0.64 to +0.83. The share of the differences runs from 48% to 69%. The link holds each time.
It does not rest on the two largest acquisition lenders. Take both out of every count, and the gap among the 40 barely changes. The test then gives +0.65 and 58%.
Seven other builds give 34% to 62%. The lowest sets the lenders of fiscal 2023 to 2025 against the sales of 2020 toredactedUse a lender's own record only where it has 50 other loans elsewhere, not 20, and that build gives 27%. At 100 loans it gives 24%. Against those early sales, owners' citizenship alone gives 27%. So in both cases that build gives no more than citizenship alone.
The lender sum takes most of the other links with it. Hold the lender sum constant, and the citizenship link falls from −0.56 to −0.26 for owners. For residents it falls to −0.39, which is still more than chance. The interest-rate link falls from −0.39 to −0.03, and loan size to −0.11. Other lending falls to −0.02, as it must: it is inside the sum.
The fall in the owners' link takes both parts of the sum. With only the lenders' habits held constant, the owners' link is −0.53.
Add the four measures of section 7 to the lender sum, and the fairer figure gets worse, not better. It is 48%, against 53% for the lender sum alone.
The lender sum does not close the gap. The New York area's lenders would give it 0.53 sales a year per 1,000 locations, the third-lowest of the 40. It hadredactedMinneapolis–St Paul's would give it 1.44, and it hadredactedSo the New York area came in below what its lenders would give, and Minneapolis–St Paul above.
The lender sum alone would put the two 2.7 times apart: 1.6 from the volume of other lending, times 1.7 from the lenders' habits. They are 5.6 times apart. Across the 40, a gap of 1.6 times is left, where chance would give 1.2.
What this test can and cannot show
It cannot say which way the cause runs. Lenders that finance purchases may simply go where businesses are sold. Or a place may have fewer SBA-financed sales because the lenders that serve it do not offer them. Both would give this result.
Four more tests follow. The first two ask whether a lender's habit is its own. The last two bear on which way the cause runs.
First, a lender's habit travels. Take each lender with 20 or more other loans in one of the 40 and 20 or more elsewhere. Its loans to buy a business per other loan there line up with its figure everywhere else at +0.76, across 838 such pairs.
Second, loan by loan, the lender tells you far more than the place. Take the 106,122 in-scope loans in the 40. Differences between lenders hold 19.2% of the differences in whether a loan buys a business. Differences between the 40 metro areas hold 1.2%. Even with lenders put in 40 groups it is 11.8%.
Third, the place still shows inside a lender. 87 lenders have 50 or more in-scope loans in both the faster 20 and the slower 20 of the 40. For 55 of them a larger share of loans buys a business in the faster half, and for 30 a smaller share. Chance gives a split that lopsided about 1 time in 110. Weight each lender by all its loans in the 40 on both sides, and 9.3% of loans buy a business in the faster half, against 7.9% in the slower.
Fourth, one lender made loans to buy a business in all 40. It is the largest, with 10% of the 40's sales. If one local market for sales lay behind the map, that lender and everyone else should be busy in the same places. They are not. The two line up at −0.01.
That is not proof either. A local lender may take deals a national one would have made. In Minneapolis–St Paul, lenders with fewer than 20 other loans anywhere else made 30% of the loans to buy a business. The median of the 40 is 8%.
The two ends show two different things. With that lender the New York area is third-lowest of the 40, and Minneapolis–St Paul is second-lowest. With all the other lenders taken together New York is the lowest, at 41% of the median, and Minneapolis the highest, at 263%. So New York is short with both, and Minneapolis's lead comes from the other lenders.
Seven lenders made loans to buy a business in 30 or more of the 40, and made 24% of the 40's SBA-financed sales. Two of them made 17%. The second of those has its address in Ohio, and its five busiest of the 40, for their size, are Cincinnati, Cleveland, Columbus, Detroit and Indianapolis. So the seven together say less than the one in all 40. Their sales line up with everyone else's at +0.10, within chance.
Nor is it only a matter of region. The four Census regions hold 35% of the differences among the 40, and the Midwest leads. Four random groups would hold 8%. Inside regions, the lender test still lines up at +0.60.
We also asked whether the sum lines up for other kinds of loan. We ran it on loans to start a business, where no local market for sales is in play. It lined up just as well: +0.85, and 60% of the differences among the 40. For loans to existing businesses it was weaker, at +0.49 and 28%. So the lender sum goes with more of a place's SBA lending than purchases alone.
And loans to start a business are spread more unevenly than loans to buy one. For its size, a fast large metro area has 3.1 times the start-up loans of a slow one. For loans to buy a business it is 2.3 times.
That is the point to hold on to. We set out to map where businesses change hands. What the file maps is where SBA loans buy them. And that goes closely with who is lending, and how much.
53% to 58% of the differences in SBA-financed sale rates among the 40 largest metro areas go with which lenders work there, and how much they lend. No other single test accounts for as much. This may be a map of SBA lending as much as a map of sales.
In short
Lenders differ: 52% of one lender's loans buy a business, and 0.7% of another's. The rate a metro area's lenders would give it lines up with its real rate at +0.72, and accounts for 53% to 58% of the differences among the 40. The volume of other lending alone accounts for 22%, and the lenders' habits alone for 21%. The sum is the closest of 15 tests, and it cannot say which way the cause runs. The one lender that made loans to buy a business in all 40 is not busy in the same places as the rest. The same sum fits loans to start a business just as well. For loans to buy a business, a gap of 1.6 times is left.
9. A faint signal, and how far it goes
One measure of the people in the middle lines up with sale rates too, faintly. Piece 1 leaned on it. It carries less than we gave it.
Census counts offices of real estate agents and brokers by county. We have no count of business brokers across states or counties, so this is the nearest measure.
Across the 51 states, more brokerage offices per business location goes with more SBA-financed sales: +0.31. Chance gives that about 1 time in 36.
Across the 1,243 counties with a published count it is +0.11. Among the 40 largest metro areas it is +0.13, which is within chance.
redactedHold the size of the market constant and the county link survives, at +0.10. Hold brokerage density constant, and market size fades to +0.03. Each of those is uncertain by about 0.03 either way.
The link shows mostly in the largest quarter of counties, at +0.29, give or takeredactedIn the other three quarters it is +0.03 to +0.05, give or takeredactedAt the national rate, the counties in those three quarters would have 2 to 13 such sales in six years. So that may be measurement as much as markets.
Five things cut against reading much into it.
• The industry is real estate brokerage, and mostly residential. It stands in for business brokers. It does not count them.
• New Hampshire and Wisconsin have high SBA-financed sale rates with thin layers. Florida has an ordinary rate with a thick one.
• Brokers may follow sales rather than cause them.
• Investment-advice offices and insurance agencies show the same pattern in big counties, at +0.26 and +0.25. Law offices run the other way, at −0.18. So this may measure how financially busy a place is.
• Among the 40 largest metro areas, where the headline is, the link is not there.
In short
Brokerage offices per business location line up with SBA-financed sale rates at +0.31 across states and +0.11 across counties. Among the 40 largest metro areas the link is +0.13, within chance. It is faint, it shows mostly in large counties, and insurance agencies show the same pattern. A count of business brokers by place would test it.
10. What the map corrects
1. "Go where the businesses are"
The size of a market says little about how often its businesses change hands with an SBA loan. Across 251 metro areas the link is −0.03. New York, Los Angeles and Miami had clearly fewer such sales than the national rate would give. Dallas–Fort Worth and Atlanta had clearly more.
2. "It is a big-city problem"
Minneapolis–St Paul and Denver are large, and they are the two fastest of the 40 for SBA-financed sales. And 38 of New York state's 53 counties are below the national county median. 28 of those are outside the New York metro area.
3. "A slow market is one where businesses do not sell"
It is one with few SBA-financed sales. That may be few sales, or few that use the SBA.
In the New York area, three banks make 46% of the other SBA loans. Across the United States, few of those banks' loans buy a business. 56% of the area's other SBA loans are for $150,000 or less. The one lender that made loans to buy a business in all 40 is short there too. For its size, New York is third-lowest of the 40 with it, and lowest with all the other lenders taken together. Whether New York's owners sell less often, or sell without the SBA, the file cannot say.
11. If you are choosing where to look
If you are picking a market
• Ask how often businesses change hands there, not how many there are. For SBA-financed sales the two do not line up.
• Use metro areas, not states. Four Texas border metro areas are near the bottom of the national list, and Texas sits at the median.
• Do not trust a county ranking. Chance alone would give a gap of 3.5 times between a fast county and a slow one, against 6.5.
• A slow market on this map is not proof of a slow market. These are SBA-financed sales, and we cannot see the rest.
• If you plan to borrow from the SBA, look at how much SBA lenders lend there, and how often those lenders finance purchases elsewhere. Together, that lines up with the map more closely than anything else we tested.
If you are selling in a slow market
• In the New York area, 521 buyers used a 7(a) loan in six years, among 292,192 in-scope locations. Do not count on an SBA-financed buyer.
• We cannot see buyers who pay another way. A slow market on this map may still have them.
If you are lending or advising
• Do not size a local market from owner age. States with more owners aged 65 or over have fewer SBA-financed sales, not more.
• A place with few SBA-financed sales may be one whose SBA lenders finance purchases less often wherever they lend. The file cannot say which came first.
• If you know of a count of business brokers by market, or of sales made without an SBA loan, either would test this map.
12. Assumptions, gaps and limits
Assumptions
• That SBA-financed sales track all sales from place to place. This is the big one, and the evidence here gives reason to doubt it. The lender sum lines up with the map more closely than anything else. If cash and seller-financed sales are commoner in New York and Miami than in Minneapolis, the map is about financing rather than selling. Nothing public that we know of can settle that.
• That six years is long enough to average out local cycles. The gap was 2.7 times in fiscal 2020 to 2022, and 2.2 in 2023 toredactedChance alone would give 1.3 in three years.
• That a metro area is the right unit. A buyer willing to move faces a different map.
• That a lender's habits elsewhere say what it would do here. Across 838 lender-and-metro pairs the two line up at +0.76. That is close, not the same.
Data gaps
• No count of business brokers by place. Real estate brokerage offices stand in for brokers here. We have not used any state register of brokers.
• No count of sales made without an SBA loan.
• No listings, so we cannot tell "nothing was offered" from "nothing sold".
• No count of buyers. SEC notices record where an issuer is based, not where it buys, and 14% are from funds.
• No count of credit turned down. The file holds only loans that were made.
• We read the broker laws of the 15 listed states, and Washington's. We did not read the other 35.
Limitations
• Rates for places count 7(a) loans only. Adding the 504 programme lifts the median state from 0.73 to 0.78 and barely changes the order.
• The map counts locations, not firms. Each branch of a chain counts once.
• Each loan counts as one sale, and some sales take two loans. Count each borrower once, and a fast large metro area runs 2.1 times a slow one, not 2.3. Minneapolis–St Paul then runs at 5.3 times the New York area, not 5.6. The order of the 40 barely changes.
• The bar at 20,000 locations is ours. From 10,000 to 30,000 the gap runs from 1.9 to 2.4 times.
• Some loans may fund one partner buying out another. The file cannot separate those.
• 436 of the 20,738 loans, or 2.1%, were approved but not yet paid out when the file was drawn. 380 of those were approved in fiscal 2024 and 2025.
• Census withholds any county-and-industry cell with fewer than three locations. We put those back from industry-group totals, which is an estimate.
• Connecticut is left out of the county and metro figures, because it changed from counties to planning regions.
• Small places bounce. We give chance baselines for that reason, and we do not rank counties.
• Every test here is of places that differ in many ways at once, and neighbouring places resemble each other. So no link here is proof of a cause.
• The lender test was run last. We tried eight versions and report their range, not the best. Seven other builds give 34% to 62%. The lowest gives 24% to 27% when a lender's own record is used only from 50 or 100 other loans elsewhere, or on records in other states only.
• 15 tests are not all the tests there are. The answer may be something we did not think to try.
13. What five pieces found
Five pieces, and the shape of it is this.
The wave is ageing, not arriving. The share of owners aged 55 or over moved from 50.8% to 51.3% in seven readings. The share past 65 rose from 20.1% to 23.1%. Older owners are already leaving, at about 8% a year.
What is out there is far smaller than the usual numbers. 30,427,808 businesses have no employees. 6,294,604 firms have some. Counted by industry, and leaving out membership bodies, holding companies and funds, that is 6,077,257 businesses. 92,986 pass every test we set.
That count moves with how it is made. As separate, for-profit firms, sized in March 2021, that is nearer 80,000. On a fuller matching of industries, which piece 3 sets out, 36 industries pass, holding 96,744. And the list moves with how a failed loan is counted. On five ways of counting, 32 to 37 industries pass, as piece 4 shows.
26% of businesses with employees are in industries our licence screen flags for a licensed profession or trade. That is 1,601,343 businesses. On our reading, an owner must hold the licence in 44% of them. Their SBA loans go bad less often, and the firms are smaller.
Passing every test says little about how often a business changes hands with an SBA loan. In a year, about 200 of the 92,986 do. That is 1 in 457, or 1 in 529 counting each borrower once. Counting businesses of every size, the rate among the 34 industries runs from 1 in 153 to 1 in 5,060.
And where a business is matters to whether an SBA loan buys it. Among large metro areas, a fast market has 2.1 to 2.3 times the SBA-financed sales of a slow one, for its size, depending on the count. Which lenders work there, and how much they lend, goes with 53% to 58% of the differences on the build we print. Across the 18 builds listed in the methodology it is 24% to 69%, and 16% to 65% on the fairer sum. We cannot say whether any of it is fewer sales or fewer SBA loans.
Of the 92,986 that pass every test, about 1 in 457 changes hands with an SBA loan in a year, or 1 in 529 counting each borrower once. Where a business is matters too, and it goes with who lends there and how much they lend.
We would rather end on that than on a tidy answer we cannot support. Two counts would test it. One is a count of business sales made without an SBA loan, by place. The other is a count of business brokers by place. If somebody has either, we would like to see it.
The full version is on acquidex.com. It carries all 50 states, the 40 largest metro areas, every test and the complete method.
This is the last of five parts. If you know of a count of business brokers by market, or of business sales made without an SBA loan, I would like to see it.
________________________________________
Sources
Every count here is public, free to download and recomputable. The SBA figures are from the agency's own loan-by-loan releases.
• SBA 7(a) and 504 loan-level FOIA releases, 7(a) as of 30 June 2026; 504 as of 31 MarchredactedNot linked, because the download path moves between releases
• Census County Business Patterns, 2017 to 2023: census.gov/programs-surveys/cbp.html
• Census Annual Business Survey, owner characteristics, reference years 2017 to 2023: census.gov/programs-surveys/abs.html
• Census Business Formation Statistics: census.gov/econ/bfs/
• Census American Community Survey: census.gov/programs-surveys/acs
• BEA regional price parities: bea.gov/data/prices-inflation/regional-price-parities-state-and-metro-area
• SEC Form D data sets, mid-2024 to mid-2026: sec.gov/data-research/sec-markets-data/form-d-data-sets