The Woman Who Invented Synthetic CDOs Is Back at JPMorgan

You remember 2008.

It comes back to you as a feeling more than a date. The slow drip of finding out that Wall Street had engineered the safe things in your retirement account to look safer than they were. The dread on the days the index fell five percent. The longer dread afterward, when it became clear nobody had been watching.

The products that did the worst damage didn't go away when the crisis ended. Most of the bankers who built them stayed in the industry. They went into something like exile.

On September 28, 2023, the Federal Reserve put out three new FAQs on its website. The FAQs said the structures the bankers had built were okay again.

The product is called a synthetic risk transfer. SRT for short.

It works simply. A bank has a loan portfolio. Regulators tell the bank to hold capital against it — money set aside in case the loans go bad. The bank doesn't want to hold that capital. It would rather lend the money out and earn fees.

So the bank pays a hedge fund to absorb the first losses. The hedge fund gets a coupon. The bank still owns the loans. But on the bank's books, the regulators agree the capital requirement falls by half or more.

The risk didn't disappear. It moved.

European banks have been doing this for years. American banks couldn't — the rules were unclear and the Fed hadn't blessed them. That changed in September 2023.

Two and a half years on, U.S. issuance has grown so fast that Pemberton — the London asset manager that buys SRT tranches — published a paper this winter titled SRT Volumes: The Road to $200 Billion.

That's the road. The market today protects about $800 billion in bank loans worldwide. Ten years ago, the number was a tenth as large.

What worries me is who's on the other side. The buyers of the first-loss tranches — the people the banks pay to absorb the credit risk — are private credit funds. Apollo. Ares. KKR. Blackstone. Blue Owl.

The same private credit funds I've been writing about all spring. Their retail BDCs gated redemptions in the first quarter. Their lending arms book interest their borrowers can't pay in cash.

The banks have moved roughly $800 billion of credit risk into the part of the system everyone is already worried about.

The regulators say it's fine.

This connects directly to the lead story.

For two decades, synthetic risk transfers were a European product. European banks had clearer rules. U.S. regulators kept their distance. American banks couldn't easily get capital relief from the structures.

Then the Fed published its September 2023 FAQs, and the U.S. market started catching up.

U.S. Bank issued $2.5 billion of credit-linked notes against an auto loan pool — its first ever such deal. JPMorgan Chase followed with a $1.8 billion auto-loan CLN. Morgan Stanley got Fed approval for $250 million of credit-linked notes tied to first-loss risk on its bank's commercial loan book.

The big banks went first. The regional banks followed. Ares Management noted in a paper last fall that U.S. regional banks are using SRTs to manage commercial real estate exposure — the same exposure regulators have been warning about since the 2023 regional bank failures.

The banks most exposed to the credit problem the market already knows about are the ones most aggressively shedding that exposure to private credit through structures American regulators have only just approved.

And here's where it spreads.

The mezzanine and first-loss tranches of SRT deals — the parts that take the hits if the loans go bad — go to a small, concentrated set of buyers. Specialist credit funds. A handful of insurance companies. A few pension plans.

The buyer list reads as a who's who of private credit. The European specialists — Pemberton, Chorus Capital, PineBridge. And the American giants you already know — Apollo, Ares, KKR, Blackstone, Blue Owl — through their structured credit arms.

The Basel Committee on Banking Supervision flagged this concentration in its February report. The protection market, the committee wrote, depends on a relatively small number of private credit funds being willing to buy mezzanine paper at renewal. If those funds step back, banks lose the capital relief and have to find equity in a hurry.

The dependency runs both ways. The same private credit funds buying SRT tranches are underwriting middle-market loans, running non-traded BDCs, and selling preferred-stock structures into retail wealth channels. They are stretched. Their flagship vehicles have been gating redemptions. Their business development companies are booking PIK income at record shares of total income.

These are the firms that now sit between American banks and the loans the banks made.

Meanwhile, the math underneath the structures has a problem the regulators are starting to discuss openly.

A typical SRT contract is short-dated. It runs a few years. The underlying loans — corporate revolvers, mortgages, auto loans, commercial real estate — run longer than that. Often much longer.

When the SRT matures and the bank wants to renew the protection, it has to find buyers willing to take the next tranche. If markets are calm, those buyers are there at a reasonable spread. If markets are not calm, the spread widens or the buyers walk away.

A bank that loses its SRT protection has to take the capital charge back onto its balance sheet. Suddenly. In the worst possible quarter to need new capital.

The Bank for International Settlements report from February calls this rollover risk and notes that the SRT market has never been tested in a stressed credit environment. The product has only ever existed during periods of strong demand from non-bank investors.

That period is the entire history of the product. We don't know what the other side looks like.

Three more things worth keeping track of …

Thirty-year U.K. gilt yields hit 5.86 percent on May 15. That was the highest level since 1998. The trigger was political — Prime Minister Keir Starmer's Labour Party lost badly in local elections, and Manchester mayor Andy Burnham, seen as more aggressive on public spending, lined up to challenge for the leadership. But the move was bigger than the politics. U.K. debt is near 94 percent of GDP. The Office for Budget Responsibility gives the government only a 59 percent probability of meeting its current fiscal rule by 2029. Starmer announced his resignation in late June. Burnham is the consensus favorite to take over by mid-July. The 30-year gilt has come off the May peak but stays above 5 percent, a level the U.K. market has not lived with for a generation.

Same-day SPX options now account for around 59 percent of all S&P 500 options volume. A decade ago the number was closer to five percent. Citadel Securities reported in March that retail options activity is running 47 percent above its 2020–2025 average, and that 40 percent of total retail options volume on its platform now sits in zero-day contracts. The risk is not the volume itself. The risk is what happens on a fast tape when the flow gets one-sided. Dealer hedging on negative-gamma days amplifies moves rather than dampening them. The S&P 500 has not had its real test yet.

The New York Fed's ACM model puts the 10-year Treasury term premium at its highest reading since 2014. For the entire post-financial-crisis decade the premium was negative — investors paid up for the safety of long-dated paper. Now they demand extra yield to own it. The reasons are familiar one at a time and new in combination. Foreign demand at long-bond auctions has plateaued. The Treasury's quarterly refunding is heavy on the long end. Inflation expectations are stickier than the Fed expected. The practical effect is that when the Fed cuts rates, the long end doesn't necessarily follow. The old playbook for forecasting ten-year yields from the Fed dots has stopped working. We'll see.

In December 1997, JPMorgan launched a $700 million bond.

The deal name was the Broad Index Secured Trust Offering. BISTRO.

A special-purpose vehicle the bank set up entered into a credit default swap with the bank itself. The swap covered a $9.8 billion portfolio — 307 commercial loans, corporate bonds, and municipal credits, across Europe and North America. The vehicle then sold $700 million of notes to investors. The investors got the credit risk. The bank kept the loans and the income.

The reason had nothing to do with selling investors a new product. The bank wanted capital relief. It had $9.8 billion of credit exposure that the Fed required it to hold capital against. After BISTRO, the bank no longer held that capital. The risk lived somewhere else.

The Fed approved the structure. A credit derivatives team at JPMorgan structured the deal. The team had been building since 1994 under a 28-year-old managing director named Blythe Masters.

Other banks copied BISTRO over the next decade. The structure evolved. Between 2005 and 2007, dealers issued at least $108 billion of synthetic CDOs, the public descendants of BISTRO. The journalist Gregory Zuckerman has estimated that the actual figure was several times larger, because most synthetic CDO trades were never reported to any exchange.

JPMorgan dodged most of what came next. AIG's London office, and a great many other banks, did not.

The product that did the worst damage in the 2008 crisis was the synthetic CDO. Regulators did not ban it. Dealers simply stopped issuing them. The Federal Reserve's September 2023 FAQs reactivated the modern version of the same structure, under a new acronym.

The team that built the original at JPMorgan stayed in the industry. Blythe Masters left JPMorgan in 2014, ran a digital-asset startup for a decade, and returned to JPMorgan in February of 2025 as chair of its securities unit in London.

The original BISTRO deal protected $9.8 billion of loans. Today's synthetic risk transfer market is roughly a hundred times larger. The Guardian once called Masters a destroyer of worlds for her role in inventing the structure. Her students are running the new version, with regulatory approval, at the exact moment the only people willing to buy the first-loss risk — the big private credit funds — are gating their own retail investors.

The structure rhymes. The verdict does not have to.

We'll see.