The $400 Million That Wasn't a Default Wave

On 7 July the *Financial Times*, with confirmation from Bloomberg, reported that HSBC has told clients it will stop renewing certain credit lines to private-credit funds and will no longer provide them 'back leverageback leverageWhen a bank lends to a private-credit fund (rather than to the company) so the fund can make more loans than its own capital would allow, an amplifier bolted onto the fund's lending.' where the returns are not seen as justifying the risk. Coming from the largest bank in Europe, the move landed as a verdict on an entire asset class: record defaults are here, the reasoning went, so the smart money is heading for the exits of what the market frames as a $3.5 trillion sector.

That is a clean story. It is also two different stories welded together, and the weld is where the interesting question lives.

**Private creditPrivate creditLending done by investment funds instead of banks, direct loans to companies that bypass both the bank and the public bond market.** is lending by investment funds instead of banks, direct loans to companies, cut out of the bond market. The sector gets sized anywhere from about $1.5 trillion (direct corporate lending) to $3.5 trillion (everything defined broadly). That $3.5 trillion is the number the flight narrative reaches for, and we use it here only to describe the story being told, not as our own accounting. The ambiguity is itself part of why the narrative is so slippery.

**Back leverage** is the amplifier. A bank does not lend to the company; it lends to the *fund* that lends to the company, so the fund can make more loans than its own capital allows. The bank earns a steady spread and sits one step removed from the borrower, which also means one step removed from seeing what the borrower is really doing.

**Payment-in-kind (PIK)** is when a borrower, instead of paying interest in cash, pays it by issuing more debt. It keeps a loan technically current while the borrower's cash position quietly rots. Hold that one; it matters for the 'record default' number.

On 5 May, in its first-quarter results, HSBC disclosed a surprise expected-credit-loss charge of roughly **$400 million**, the item that pushed group expected credit losses to about **$1.3 billion** (the total also included a separate ~$300m Middle East reserve) and pre-tax profit below consensus, at about $9.4 billion. The shares fell more than **5% intraday** in London (and about 4.9% in Hong Kong) on the day. Within days JPMorgan, Goldman Sachs and Morgan Stanley all argued the sell-off was an overreaction, with JPMorgan reaffirming its 'overweight' rating and calling the weakness a buying opportunity. CFO Pam Kaur described the charge as 'idiosyncraticidiosyncraticSpecific to one name or situation rather than part of a broad, system-wide trend.,' telling analysts the bank had 'completed a review of the highest areas of risk' in its portfolio and had 'not identified any comparable fraud concerns.'

Trace the charge and it does not fan out across hundreds of struggling mid-market borrowers. It runs down a single wire: **HSBC** lent into a structure connected to **Apollo's Atlas SP** asset-backed-lending unit, which financed a **securitizationsecuritizationBundling loans together and selling the package as a financial instrument, so the risk and cash flows can be financed and traded.** tied to **Market Financial Solutions (MFS)**, a Mayfair bridging lenderbridging lenderA firm that makes short-term, high-interest property loans meant to be repaid quickly once the borrower sells or refinances. that **collapsed into administrationadministrationA UK insolvency process where an outside administrator takes control of a failed company to try to rescue it or repay its creditors.** on 25 February 2026. On the public record so far, the charge is dominated by this one item rather than a diversified spread of bad loans, which is why we say the *bulk* of it, not all of it, traces to MFS.

MFS had built a loan book over roughly two decades of short-term UK property finance. Creditors Zircon Bridging and Amber Bridging forced it into administration in late February, alleging **double-pledgingdouble-pledgingAn alleged fraud in which the same asset is promised as collateral to several lenders at once, so more than one lender believes it holds the security.**, the same property pledged as collateralcollateralAn asset a borrower pledges to a lender that the lender can seize and sell if the loan is not repaid. against loans from multiple lenders at once. Court filings put verifiable collateral at only about **£230 million** against roughly **£1.2 billion of debts**, a shortfall of more than 80%. This is not a credit cycle grinding a borrower down. It is an alleged fraud, and so far the losses cluster by *counterparty*, not by *sector*.

The clinching detail is who else is on the list. Barclays took a reported **£228 million** provisionprovisionMoney a bank sets aside on its books to cover a loss it now expects on a specific exposure. in its own Q1 results, tied, it said, to the *same* borrower, MFS, and is the single largest creditor, owed roughly **£500 million** (ahead of Apollo's Atlas SP at about £400m and Elliott Management at about £200m). Two banks, two headline 'private-credit losses,' one alleged fraud. But the two exposures are not identical: Barclays lent to MFS more directly, while HSBC sat further out on the chain as *secondary* securitization exposure through Atlas SP, possibly with some first-loss protection that, in hindsight, may have given a false sense of insulation from the underlying collateral. The public reporting has centred the loss on HSBC and Barclays rather than on Atlas SP absorbing it, so the reader should not assume the loss allocation across the whole chain is fully disclosed.

The single visual point: the 'private-credit sector losses' powering the flight narrative are, so far, concentrated in one alleged fraud shared by two lenders, not spread across a loan book.

The backdrop everyone cites is Fitch's US private-credit default ratedefault rateThe share of borrowers, by number or value, that have failed to meet their loan obligations over a set period., which hit a **record 6.0%** for the twelve months to 30 April 2026, up from 5.7% in March. Records make headlines. But two caveats sit right next to the word 'record.' First, the series is young, Fitch only began tracking this metric in **August 2024**, so a 'record' is a record over a short window, closer to a definitional artifact than a multi-cycle peak. Second, the 6.0% is not the cash-missed-a-payment number most readers picture.

Look at what the rate actually counts. Over the period there were **99 default events**, of which 81 were first-time defaulters. And, in Fitch's own words, **more than half** of those defaults were driven by interest-payment deferrals and the switch to PIK, the borrower paying interest in more debt rather than cash. In other words, most of the 'defaults' behind the record are the soft kind: a loan restructured to *avoid* a hard, missed-cash-payment default, not a borrower who simply stopped paying.

There is a second seam inside the headline. Fitch's rate is a blend of two tracks: a **model-implied** rate (machine-estimated across a broad universe) and an **analyst-monitored** rate (hand-scored on the names Fitch watches closely, which skew riskier). In the most recent snapshot with published component figures, the twelve months to 31 January 2026, when the blended rate read 5.8%, the model track sat near **4.7%** while the monitored track ran near **9.4%**. The headline is an average of two very different populations, so where it lands depends partly on the mix, not only on how borrowers are actually doing.

None of this makes the number fake. Consumer-facing borrowers really are deteriorating, Fitch flags consumer-products companies at an 11.1% default rate, up from 5.9% a year earlier. The point is narrower: a 6.0% blended, mostly-soft, two-year-old series is not the same object as 'a wave of borrowers missing cash payments,' and it is that stronger claim the flight narrative quietly borrows.

Size the exposure the way HSBC itself laid it out for analysts, as a set of nested layers rather than one scary headline. The bank's total private-markets exposure is about **$111 billion**. Inside that, its private-credit-related exposure is about **$22 billion**, which Kaur was at pains to note is roughly **2% of the group loan bookloan bookThe total pool of loans a bank has outstanding; an exposure is often quoted as a share of this book.**, a level she said the bank is comfortable staying within. Inside *that*, the securitization-financing layer where MFS lived is about **$3 billion**. And the charge itself is about **$0.4 billion**.

Read down the column and the loss is a sliver inside a sliver: roughly an eighth of the securitization layer, which is about a seventh of private credit, which is about a fiftieth of the loan book. That geometry is what lets a genuinely painful, genuinely embarrassing fraud coexist with a portfolio that is not, on the disclosed numbers, hemorrhaging.

Here is where the two welded stories come apart. Cutting back leverage is not the same as calling in loans to companies. HSBC is not marching into mid-market borrowers and demanding its money back. It is declining to *keep amplifying* certain funds, refusing to renew the facilities that let a fund lend more than its own capital would allow. The company-level loan can still exist; what changes is who funds the fund.

Cutting back leverage removes an amplifier from a fund, not a loan from a company. But an amplifier that two of the biggest suppliers stop selling does not become free.

That is the case for calling this a repricing, not a credit contraction. In principle the fund can replace HSBC's line with another bank's, or with a non-bank lender, and the underlying loans roll on untouched. If that migration happens smoothly, the real economy feels nothing; only the plumbing changes hands.

But 'in principle' is doing work there, and this is where we part company with the fully benign reading. HSBC is not alone. Barclays' CEO has said the bank is constraining lending to structured-finance counterparties that cannot satisfy it on the 'quality and independence of their financial controls,' explicitly linking MFS to an earlier loss. When two of the largest suppliers of fund leverage retreat from the *same* riskier segment at the *same* time, the price of that leverage rises for the funds that relied on it, even if the volume eventually migrates to someone else. Someone else will lend, but on stiffer terms. That is a real cost borne at the financing layer, not a null one, and it is exactly the kind of quiet tightening that does not show up as a 'default' anywhere.

The honest position is that the two readings are not yet fully separable, and there is a clean test coming. Both HSBC and Barclays report first-half results at the end of July. That print is the adjudicator.

• A *second, unrelated* credit provision, a different borrower, a different structure, would be the first genuine evidence for the default-cycle reading, because it would show the problem is not one name.
• A still-single MFS item, with recoveries tracking toward the >50% both banks have signalled, would keep the idiosyncratic reading intact and make the 'flight' look like disciplined repricing of one bad wire.
• Watch the *terms*, not just the losses: any sign that funds are paying visibly more for leverage, or shrinking, is the repricing cost showing up even with zero new defaults.

Kaur has already staked HSBC's credibility on the first outcome not happening, she told analysts the bank had reviewed its highest-risk areas and found 'no comparable fraud concerns.' If the mid-year numbers contradict her, the market's original, blunter story gets its evidence. If they don't, the record-default framing will have done a lot of narrative work on the back of one Mayfair bridging lender.

There is a last twist worth naming, because it is why this matters beyond two banks' balance sheets. Private credit's stability rests partly on belief. If enough lenders decide the sector is fragile and pull leverage at once, funds are forced to sell assets or refuse to roll loans, and the fragility becomes real. That is **reflexivityreflexivityWhen market beliefs change the very reality they describe, here, fearing sector-wide stress can help create it.**: the fear can help manufacture the thing feared. A retreat that starts as prudent repricing of a single fraud can, if it is read as a verdict and copied, tip into the very default wave it was mistaken for.

That is the whole reason to get the diagnosis right. 'One alleged fraud, priced and mostly recoverable' and 'the leading edge of a default cycle' call for opposite responses, and only one of them, repeated widely enough, can make itself come true.