Private credit's 0.8% default rate is a size-weighting artifact

Twelve per cent of private credit loans to borrowers with $10m to $20m of EBITDA now sit below 90 per cent of par, against roughly 1 per cent in 2023. Houlihan Lokey's Q2 2026 Private Credit DataBank Market Trends & Insights report, as reported by Alternative Credit Investor, shows the same pressure easing as size rises: 6 per cent of loans to borrowers with $20m to $100m of EBITDA are below that threshold, the highest level in three years, and 3 per cent of loans to borrowers above $100m are.

Step down one band and the sub-90 share doubles — 3 per cent above $100m of EBITDA, 6 per cent between $20m and $100m, 12 per cent below $20m — a sharp rise across three years that the report describes as twelve times the 2023 reading.

A sub-90 mark is a valuation failure, a default a payment failure, and this cycle the two have separated along exactly the line the report draws. Houlihan Lokey puts second-quarter defaults among borrowers with less than $100m of EBITDA at 3 per cent on a size-weighted basis and 3.6 per cent by borrower count. Across the whole private credit market, defaults accounted for 0.8 per cent of outstanding loan principal and 2.5 per cent of borrowers by count.

Allocators quote the 0.8, which is arithmetically correct and close to useless as a base rate for a mid-market book, because size-weighting lets the largest loans dominate the denominator — plausibly the credits with the deepest sponsor support and the most room to amend — while the population that contains the 12 per cent cohort defaults at 3.6 per cent by count. The distance between 12 per cent marked below 90 and 3.6 per cent in default is not an inconsistency. It is an inventory: loans already revalued and not yet restructured.

Cindy Ma, managing director and global head of portfolio valuation and fund advisory services at Houlihan Lokey, describes the increase as "concentrated, not broad," and attributes the sub-1 per cent market-wide figure to the largest borrowers continuing to perform. Her framing carries the point that matters for allocation: sub-$100m EBITDA "is where much of the direct lending market operates," and she expects "this divide by borrower size to define the market through the balance of the year."

That puts the report's least comfortable number in the middle band rather than the bottom one: six per cent of loans to $20m-to-$100m EBITDA borrowers below 90 of par is a three-year high, and sub-$100m EBITDA is, on Houlihan Lokey's own description, where much of the direct lending market operates. The 12 per cent at the bottom of that range is the louder figure; the 6 per cent at its core is the one that says the repricing is moving up in size rather than sitting at the micro-cap fringe.

Healthcare is the only sector showing elevated stress on both measures, with 4.2 per cent of borrowers in default by count and 2.7 per cent size-weighted, which makes it the one place in the report where the count-based and dollar-based readings agree. A sector that is stressed by borrower and by dollar has the least room left before marks move against fund valuations, and it is the likeliest origin of third-quarter amendments.

The payment-in-kind data complicates the loudest version of the bear case: in the second quarter, 11.8 per cent of loans on a size-weighted basis elected to pay at least some interest in kind, and those elections represented 6.3 per cent of total interest dollars. Amended PIK, added after origination and treated by Houlihan Lokey as the closer proxy for borrower stress, accounted for 1.6 per cent of interest dollars. Timothy Kang, a managing director in the same Houlihan Lokey practice, argues the distinction is routinely lost: "A PIK option is a structuring feature, and not necessarily a distress signal, and the two get conflated." Option availability, he says, is at a record high while election stays modest.

On those numbers Kang has the better of it: a market where amended PIK is 1.6 per cent of interest dollars is not one in which sponsors are papering over cash-interest shortfalls at scale. The read that does more work is of a market with a valuation problem and no matching payment problem yet, where the arithmetic for both sponsor and lender favours an amendment over an enforcement, and where the marks that have already moved are the ones that would move first in that process. That is the shape of a workout pipeline forming, and it is why the size split in this report is a better guide to the next two quarters of marks than the default rate is.

Uniform unitranche pricing is over; manager selection is now the only durable alpha in direct lending. The DataBank split suggests the selection test has moved axes: borrower size, rather than sector or vintage, separates the 12 per cent cohort from the 3 per cent one, and it is the variable a fund's sector label will not tell you. An allocator who can see how much of a manager's book sits in the $10m-to-$20m EBITDA band has a better predictor of fourth-quarter marks than any healthcare or software thesis. Direct lending spent a decade pricing small enterprises at spreads that assumed scarcity rather than survival, the same inversion this publication flagged in European football finance, where relegation rather than scarcity sets the price. Houlihan Lokey has now put the correction on the page, band by band. A fund's weighted-average borrower size is the disclosure that would settle the question, and the managers who publish it first will be the ones whose books can take it.

The third quarter's test is conversion: twelve per cent of the smallest loans are marked below 90 of par today, and the question for the next two sets of marks is whether that cohort holds as a plateau or becomes the waypoint. Amended PIK is the number to watch: at 1.6 per cent of interest dollars it is small enough that any sustained rise carries information, and specific enough to distinguish a cash-flow problem from a structuring preference.