The buyer and the seller are the same firm

I tried to sell a car.

Two people came to look at it. One offered me about two thirds of what I asked. I said no and shook his hand and watched him drive off in his own car.

So I kept it. And in my head it stayed worth my price all summer.

As long as you don't sell, you never have to find out.

Private equity is doing that with 32,000 companies.

Bain counted them in February. Thirty-two thousand businesses sitting inside buyout funds, bought and not yet sold, carried on the books at about $3.8 trillion. The pile has grown three years running.

These funds are supposed to have a clock. Ten years, maybe twelve. Buy the companies, improve them, sell them, send the money back.

The money is not going back. Cash returned to investors has stayed under 15% of fund value for four years in a row. Bain calls that an industry record.

The average company now sits in a fund about seven years before it leaves. A decade ago it was five or six.

But the clock still runs. And a fund manager who cannot find a buyer has one option left that nobody talks about at dinner.

He becomes the buyer.

Here is how that works. The manager raises a new fund. The new fund buys the company out of the old fund. He runs both funds. Old investors who want out take cash. The company never changes hands in any way you would recognise… and the fees keep running.

The industry calls it a continuation vehicle. Investors call it liquidity. I call it selling the car to yourself and writing down the number you wanted.

In the first six months of this year, deals run by the manager passed deals run by outside investors for the first time on record.

I have no idea when this breaks. Prices only become facts when somebody who wants out meets somebody who wants in.

Right now the same firm is standing on both sides of the door.

This connects directly to the exit problem. When a buyout fund runs out of time and cannot find an outside buyer, the manager can move the company into a new fund he also controls.

He negotiates with himself. He agrees a price with himself. Then he keeps running the business and charging fees on it.

Single-company versions of this reached $34 billion in the first half of 2026, according to Evercore. That is nearly double the same period last year, and 53% of everything managers arranged.

Now look at the prices. Most of those single-company deals cleared at the manager's own stated value, and 14% of them cleared above it. Where several companies were bundled together, 71% priced below.

So the trophies fetch full price and the rest take a haircut. Nigel Dawn, who runs private capital at Evercore, uses that word for them.

Jefferies reckons nearly 80% of the hundred largest sponsors had done one of these by the end of last year. It is not a workaround any more. It is the plumbing.

And here's where it spreads.

On 7 August 2025, President Trump signed an executive order telling the Labor Department to open workplace retirement plans to private assets. The department published its proposed rule on 30 March this year. Comments closed on 1 June. Officials say they want it finished by the end of 2026.

Today about 4% of these plans offer private assets at all. They hold about a tenth of one percent of the money.

The department's own estimate is that the rule would move about $178 billion a year, touching 4.5 million savers. It expects the money to arrive through target-date funds, which is where 84% of participants already sit without choosing anything.

Meanwhile the Supreme Court took a case in January about whether savers can sue over private equity inside a target-date fund. That argument and this rule are running at the same time.

A fund that cannot sell a company can still sell a slice of it to somebody with a paycheque.

Nobody has to mis-sell anything here. The money arrives by default, from people who never opened the document.

Meanwhile, the same companies are being priced two different ways in two different places.

Investors spent this spring deciding that artificial intelligence will eat a chunk of ordinary business software. Listed software companies outside the AI names fell roughly 30%.

Buyout funds own a lot of the same kind of business. Software has been about a quarter of all private equity deal value over the past five years.

Through 31 March, those private software holdings were written down about 8%. In Europe it was 4.2%. That is MSCI data, read by Bain.

John Zito, who co-runs asset management at Apollo, told a room of UBS clients in March what he thought of that. I literally think all the marks are wrong, he said.

Buyers of these funds noticed. Software's share of manager-arranged deals fell eight percentage points in the first half. Interesting… they will still transact, at a price. Just not at that one.

Three more things worth keeping track of this week.

The Dutch pension switch. On 1 January, about €550 billion of Dutch retirement money moved into a new system that needs far less hedging at the very long end of the bond market. Traders spent last autumn bracing for a jam in 40- and 50-year euro swaps. January came and went and Rabobank's desk said activity rose, but nowhere near what people expected. Another €900 billion is due to move next year. The buyer who held down the far end of Europe's curve for two decades is walking away in instalments.

Jets worth more in pieces. Roughly one in three Airbus A320neos fitted with Pratt & Whitney's geared turbofan sat parked at the worst of it, waiting for engine slots. A shop visit that used to take 60 to 90 days has been running near 300. Teardown yards in Spain are now stripping aircraft under six years old, because a working engine sells for more than the plane it came off. Pratt says groundings have started falling. The queue has not cleared.

Cattle at a 75-year low. The US herd stood at 86.2 million head on 1 January, the smallest since 1951, with beef cows down to 27.6 million. Ranchers are getting record prices. The plants that buy from them are losing $200 to $300 an animal, and Tyson stopped killing cattle at its Joslin, Illinois plant on 13 August after 43 years. When the herd finally rebuilds, some of the capacity to process it will be gone. We'll see.

A bank sold shares in a fund.

It was December 1928, the bank was Goldman Sachs, and the fund was called the Goldman Sachs Trading Corporation. Ordinary people could buy in. Plenty did.

Then the fund started buying its own shares with the money. By the middle of March it had bought 560,724 of them and spent about $57 million doing it.

That pushed the price up. The rising price brought in new buyers. The new buyers' money bought more shares.

In July 1929 the fund launched a second fund, Shenandoah. Investors asked for seven times what was on offer. Shares went out at $17.50 and closed the first day at $36.

Twenty-five days later, Shenandoah launched a third fund called Blue Ridge.

Each fund owned most of the one below it. So each one's worth depended on the price of the one below it… and the sponsors set those prices among themselves.

Bernard Baruch was offered a piece of both new funds. He called the whole thing financial whoopee and passed.

Nobody in the chain had to sell to anybody outside it. So nobody found out what any of it was worth.

Until October, when they did.

The first fund had traded as high as $326 a share. By 1932 you could buy it for $1.75.

Fourteen leading investment trusts lost $172.5 million between them in that stretch. That one fund accounted for about 70% of the total.

The businesses inside did not get 99% worse in three years. Steel was still steel.

What changed is that somebody outside the circle finally had to be asked for a price.

That is always the moment. We'll see.