How it actually works
A bank picks a reference portfolio — corporate loans, project finance, SME debt, whatever it wants capital relief on — and leaves it exactly where it is. The loans stay on the balance sheet, the borrower relationship stays intact, the borrower is generally none the wiser. What moves is the credit risk, sliced into tranches: a first-loss piece that eats the initial defaults, mezzanine above it, and a senior tranche that only gets hit once everything below has been wiped out.
The protection comes in two flavours. Funded protection means investors buy credit-linked notes, usually issued through a , and their cash sits as collateral; if losses land in their tranche, the note principal is written down and the collateral pays the bank. Unfunded protection means a guarantee or credit default swap, where nobody posts the notional up front and the bank is instead exposed to whether the protection seller is good for the money.
The capital relief works because junior securitisation positions are punished ferociously. Under the Basel framework as implemented in the EU's Capital Requirements Regulation, first-loss and mezzanine positions can carry risk weights of up to 1,250% or be deducted straight from Common Equity Tier 1. Get them off your books and Article 244 lets you exclude the underlying exposures from your risk-weighted assets altogether, holding capital only against what you retained.
"Significant risk transfer" is not a marketing phrase. It is the formal test. Under Article 244(2), risk is deemed transferred if the originator's mezzanine positions account for no more than 50% of the mezzanine risk-weighted amounts in the deal, or — where there is no mezzanine — if the first-loss tranche exceeds a reasoned estimate of expected loss by a substantial margin and the bank keeps no more than 20% of it. Article 244(4) offers a slower alternative: the supervisor grants permission case by case, on evidence that the capital saved is justified by a commensurate transfer of risk. The ECB published a guide on SRT notifications in 2026 with a fast-track process, requiring significant institutions to notify at least three months before closing.
Where it came from
JPMorgan's BISTRO deal in 1997 is the prototype: credit default swaps transferring the default risk of a corporate loan portfolio to investors through an while the loans stayed put. Basel II's recognition of credit risk mitigation encouraged the model, synthetic CDOs boomed in the 2000s, and 2008 destroyed both the issuance and the confidence.
The BIS account of the revival attributes it to a combination of Basel III's higher capital requirements, depressed bank equity valuations, supervisory changes that supported the trade, and the growth of non-bank investors hunting yield. Post-crisis SRTs are simpler and more scrutinised than the pre-crisis credit-risk-transfer market, which is the industry's standard and largely fair rejoinder to anyone who says "synthetic CDO" in a disapproving tone. In 2021 the EU extended simple, transparent and standardised treatment to synthetic transactions, which market participants described as a game changer.
The 2024–2025 issuance boom
The American unlock came on 28 September 2023, when the Federal Reserve published FAQs on Regulation Q addressing credit-linked notes. Mayer Brown called it one step forward for US banks; the practical effect was that banks began applying for reduced risk weights on directly issued CLNs rather than assuming the answer was no. Credit Benchmark describes the result as a modest surge in US issuance, with the US share rising to around 30% of global SRT issuance.
The volumes: market data for 2024 put global issuance at roughly $23 billion of tranches, against about $865 billion of loan portfolios already covered and around $280 billion referenced in new transactions that year. For 2025, an IACPM survey cited by GARP found banks globally issued €30 billion of new tranches on €378 billion of loans, up 35% and 21% respectively on 2024. The FT, analysing 56 European and UK banks, found 35 of them using synthetic SRTs as of 2025, with €579 billion of assets in underlying portfolios against €460 billion in 2024. Oxane Partners, citing BIS data, puts issuance up nearly fivefold since 2016, with close to €800 billion of referenced loans by end-2024 supported by about €72 billion — 9% — of transferred risk.
Individual deals give the flavour. Bank of Montreal completed $5 billion of synthetic risk transfers in 2025 across two corporate loan portfolios under its Muskoka and Algonquin programmes, with first-loss tranches of more than 7% and over 6% respectively on $2.5 billion portfolios. ING lined up roughly $10 billion of risk transfers in 2025 across US and European project finance and a portfolio of Dutch SMEs, with Apollo Global Management expected to participate. In 2026, Deutsche Bank weighed an SRT tied to $1.7 billion of German SME debt.
One number cuts against the celebration. Oxane reports that margins on first-loss and mezzanine protection tightened by almost three percentage points between 2023 and early 2026. More money is chasing the same risk, and being paid less for it.
Why it matters to this crash
The official case for SRTs is genuinely good. The IMF's October 2025 working paper frames them as enabling capital relief that supports additional lending, and regulators approved the framework precisely to credit risk out of leveraged deposit-taking institutions and into investors who can afford to lose money. Every word of that is defensible.
The problem is where the risk goes. It goes into funds, insurers and alternative asset managers operating in bilateral markets, and the disclosure stops there. Who bought the first-loss tranche on BMO's Muskoka portfolio, what they paid, and whether they borrowed to do it are not public. One 2026 research note put indicative yields on SRT credit protection at 10–15% APY, but the figure is unattributed and not specific to first-loss tranches, which tells you how thin the public record is. The market-level statistics are decent. The counterparty-level ones do not exist.
We wrote on 20 August 2026 about the Basel Committee flagging the growth of significant risk transfers as warranting continued monitoring, on the same day the BIS reported that non-banks now account for more than 40% of non-centrally cleared euro-denominated cash-borrowing , up from under 30% at the end of 2020. Those are the two halves of the same migration: the risk moving out of banks, and the funding of the entities taking it moving into markets nobody can see whole.
What would make this dangerous
The circularity is the thing to watch. PIMCO's credit market commentary of 27 July 2026, The Return of Financial Engineering: Not 2008, But Not Nothing, notes that banks have increasingly provided leverage and liquidity to alternative asset managers through subscription lines, and warehouse financing, while simultaneously relying on SRTs to shift credit risk to institutional investors — and that in some cases the ultimate investors exposed to the underlying risk are affiliated with private equity firms that appear elsewhere in the same financing chain as borrowers or sponsors. If a bank sells the first loss on its loan book to a fund and then lends that fund the money to hold the position, the risk has taken a scenic route back to where it started, arriving with an extra margin call attached.
What is not established: no public source we have seen documents a specific transaction where bank X sold a first-loss tranche to fund Y and also extended fund Y a NAV loan. Nobody has quantified how much first-loss SRT exposure is leverage-financed. That is not evidence of absence; it is an absence of evidence, and the market structure is bilateral enough that the data may simply not exist anywhere outside the two firms involved.
Four things would turn this from a curiosity into a problem. First, actual write-downs on the first-loss tranche of a large, named deal, which would test whether the protection sellers are good for it and reveal who they were. Second, SRTs written on fund finance itself. MUFG was reportedly in talks in 2026 on a $2 billion SRT on credit lines and JPMorgan on a $4 billion-plus SRT on , which would be the first at scale on NAV facilities — banks buying protection on loans made against private equity portfolios whose sponsors may also be selling that protection. Third, further drift into riskier collateral: the FT has reported a Morgan Stanley–Blackstone deal as evidence the credit risk transfer market is expanding to riskier assets, and each step away from plain corporate loans widens the gap between the modelled expected loss and the actual one. Fourth, a supervisor refusing SRT recognition on a completed deal, which would put the risk-weighted assets straight back on the issuing bank's balance sheet at the worst possible moment.
And underneath all of it, the compression: three percentage points of margin gone between 2023 and early 2026 means the first-loss buyers are being paid materially less to stand in front of exactly the same defaults.