The collateral is a store on a corner

Another one closed.

The Subway on the corner had paper over the windows. A note on the door sent people to a store eleven miles away. Somebody had already pulled the sign off the wall, and you could still see the outline of it in the brick.

You have seen the same thing where you live. A Hardee's in one town, an Applebee's in the next. The lot sits empty for a few months. Then it becomes a vape shop or a dentist.

I used to read a closed store as a private loss. Somebody's savings. Somebody's job. A landlord with a problem.

But there is a bond attached to that store. And the bond is rated investment grade.

Subway does not operate a single restaurant. Every one of its 18,773 American locations belongs to a franchisee. The company earns its money by collecting a royalty on what those stores sell, either 8% or 10% depending on the agreement.

In 2024 the private equity firm Roark Capital bought Subway for $9.6 billion. To pay for it, Roark pledged that royalty stream to bondholders. By the end of last year that put $5.7 billion of debt on a company that carried almost none before.

The structure has a name. A whole business securitization. Subway set up its trust in June 2024, and the rating agency KBRA reports that 96.7% of the cash flowing into it comes from franchise royalties.

So the bonds are a claim on how many sandwiches get sold in American strip malls. Nothing else.

Meanwhile the stores keep closing. Subway lost a net 631 US locations in 2024 and another 729 in 2025. That was the tenth straight year of decline.

The collateral is a store on a corner. And the corners keep going dark.

You can see it in the money. Subway's royalty revenue has fallen 10% in two years. Total revenue held up better only because the company now takes far more from vendor rebates than it used to.

In 2021, rebates from suppliers brought in $44.7 million, about 5.4% of company revenue. Last year they brought in $136.5 million, or 14.8%.

In other words, the franchisor found a way to be paid more per store while the stores were disappearing. That is a fine trade for the owner. It does not put a single royalty back on the books.

The debt does not shrink. The base paying it does, roughly 4% a year, mostly because operators walk away when a lease or a franchise agreement runs out.

I have no idea when that becomes a problem. Bonds like these run for years on covenants and cash traps, and the senior holders sit at the front of the line by design…

But it only moves one way. It has moved that way for ten years.

And it is not one brand. That's what worries me.

This is the part that makes the whole arrangement work. A restaurant chain with a junk credit rating borrows through a trust, and the paper that comes out the other end is investment grade.

The uplift runs from two notches to eight, depending on the deal. Take Dunkin'. Its 2015 securitization raised $2.6 billion, and the two rated pieces came out at BBB. That was five notches above the company's own bank facility, which S&P rated B-plus and Moody's rated B1.

Same business. Same sandwiches and coffee. A different box around the cash, and the borrowing cost drops.

Borrowers noticed. Franchise deals raised about $2.3 billion in 2023, then $13.5 billion in 2024 and $11.0 billion in 2025. Four more deals priced in the first quarter of this year for $2.4 billion, and one of them came from a home services company rather than a restaurant at all.

New deals price at BBB-plus, BBB or BBB-minus, and they pay more than similar corporate bonds carrying the same letters. Investors treat that gap as a reward for illiquidity.

But a BBB on a franchise trust is not the same promise as a BBB on a corporate bond. One is a company. The other is a waterfall with a brand attached.

And here is where it spreads. The people who run these restaurants are failing faster than the brands above them.

By the middle of July, at least ten multi-unit restaurant operators had sought Chapter 7 or Chapter 11 protection this year. That is more than the whole of 2025, and it leaves out the small filings, like the two-store Checkers operator that went the same week.

Sailormen ran 136 Popeyes restaurants across Florida and Georgia. It filed in January with $342 million of liabilities and an $18.8 million operating loss on $233 million of sales. Twenty of its locations closed after the filing.

Superior Star, with 59 Hardee's, filed on July 9. A 65-store Carl's Jr. operator filed. So did a 53-store Applebee's group in Georgia, a 43-store Subway operator, a 12-store Farmer Boys group and a Domino's operator in California.

Several of them had turned to merchant cash advances first. One Subway operator's filing recorded rates of 59% and 94%.

Hardee's had 1,485 restaurants in January, down 86 in a year and down from 1,707 at the start of its 2024 fiscal year. The brand did not file for anything. Its operators did.

Meanwhile, two of these structures have already been tested, and both tests went badly.

TGI Fridays borrowed $375 million through a trust in 2017. On September 3, 2024, the party controlling those bonds declared a manager termination event and removed the company from running its own securitization. The trigger was a failure to deliver an auditor's report on time.

A British franchisee had been trying to buy the brand. It walked away. Two months later, on November 2, Fridays filed for Chapter 11.

KBRA said it was the first time since the financial crisis that a manager had been terminated and then filed for bankruptcy.

Fridays had 601 American restaurants in 2008. It had roughly 161 when it filed.

Hooters was next. It borrowed through the same kind of trust in 2021, owed $31 million of debt service in 2024, and told the court that the size of those payments had put heavy pressure on the business. It filed in March 2025, having already closed 48 restaurants.

The credit analysts at Octus wrote in May that these two cases challenge whether the structure really is bankruptcy remote, and called it a blunt instrument when things go wrong.

So the box around the cash is not a wall. It is a set of tripwires. And a chain under strain is exactly the kind of chain that misses a filing deadline.

Three more things worth keeping track of this week.

Copper for delivery today is worth much more than copper for delivery in three months. On Monday the gap on the London Metal Exchange reached as much as $545 a tonne, the widest since the squeeze of 2021. Exchange stocks have fallen for forty-two sessions running, to about 205,000 tonnes, and roughly half of what is left is already booked to leave the warehouse. Backwardation that steep is not a forecast. It means somebody needs metal now and cannot find it.

Euro area governments borrow at almost identical spreads while their economies behave nothing alike. The European Central Bank reports that yield dispersion across member states sits near historic lows, with the weighted ten-year around 3.6%. Yet headline inflation runs at 3.9% in Spain and 2.9% in Italy, and last quarter Germany grew 0.2% while Ireland posted 3.9%. Bond markets normally charge for that spread of outcomes. This one is charging for membership instead.

Investors keep buying stocks and keep hoarding cash at the same time. In the week to August 12 they put $18.62 billion into global equity funds, the twelfth straight week of inflows, and $18.01 billion into bond funds. Money market funds took $28.41 billion, more than either. People are staying in the market with one hand and holding the door with the other. We'll see.

The first bonds of this kind were sold on beer.

In the 1990s the British brewers were forced to let go of their pubs, thousands of them, all at once. A company called Punch Taverns bought them. Its founders worked out that you could raise the money by pledging the rent and the beer orders from the pubs themselves.

In 2003 Punch bought its larger rival Pubmaster for £168 million and took on £1 billion of its debt. That made it the biggest pub operator in Britain, with more than 7,000 houses. The borrowing sat in two securitisations, Punch A and Punch B, and eventually came to about £2.3 billion.

The people behind the bar were tenants. They paid Punch rent, they were tied to buying Punch's beer, and many of them had handed over a deposit to get the keys. In some cases £20,000.

Then Britain drank less in pubs. Income per house fell. Punch began selling pubs to service the debt, and roughly 1,300 went in three years.

In February 2014 the company put a restructuring to its bondholders. One group withdrew support and the plan died on Valentine's Day. The senior notes were trading at about 105 pence in the pound, because those holders would be repaid at face value if Punch defaulted. Default paid them better than any deal.

The bonds were fine. The pubs were not.

A revised version passed that autumn. By late August 2014, Punch owned 3,809 pubs. It had held more than 7,000 a decade before.

So the structure did its job. It put the noteholders first, took their cash before anything else, and kept paying while the estate it sat on halved. Legal & General, Prudential, Standard Life and Aviva all held the paper. They fought each other over the terms and they got paid.

Nobody wrote a covenant for the tenant.

That is the shape to hold on to. A financing built on the takings of thousands of small operators, engineered so the bondholders keep collecting as the operators thin out. It works, and it goes on working, right up to the point where there are not enough of them left to carry the payment.

We cannot know how long that takes. Punch had a decade of it.

The last one standing pays for everyone who left. We'll see.