Defaults and markdowns among private credit borrowers are rising, driven by stress among the smallest borrowers, according to an analysis by investment bank Houlihan Lokey.
As of Q2, roughly 12% of borrowers with less than $20 million in EBITDA saw their loans marked below 90 cents on the dollar, up from 1% in 2023, according to the Sept. 10 report. Among borrowers with $20 million to $100 million in EBITDA, 6% of loans were marked below 90 cents; for larger borrowers, just 3% were marked below 90 cents.
Defaults — which include both technical and payment defaults — affected 2.5% of borrowers by count but only 0.8% of loan principal.
“[D]efault risk in private credit remains primarily a function of borrower scale, with the largest companies, for the most part, continuing to perform while smaller and core middle-market borrowers underperform at a meaningfully higher rate,” the firm said.
The figures were drawn from Houlihan Lokey’s DataBank, a dataset tracking the performance of private credit loans sourced from the firm’s third-party portfolio valuation business. The dataset includes more than 60,000 loans as of its launch in November 2025.
Healthcare was the only sector with “elevated” defaults measured both by count (4.2%) and on a size-weighted basis (2.7%), the firm said. The consumer sector, which includes a range of household, medical and automotive businesses, saw defaults of 3.6% by count and 0.7% after weighting by size.
Notably, software borrowers posted “among the lowest default rates of any industry,” according to Houlihan Lokey. Valuations for software borrowers were “conservative,” the firm said, following scrutiny of the sector in the first half of 2026, triggered by fears of AI disruption and a public-market sell-off of securities linked to enterprise software companies.
“Operating performance has continued to build since these loans were originated: Median EBITDA is 20% higher than it was when the loans first closed,” the firm said of the software borrowers in its dataset.
After weighting for loan size, 11.8% of loans in the dataset included a payment-in-kind (PIK) election, representing 6.3% of interest paid. PIK features introduced after origination (sometimes called “bad PIK”) represented 1.6% of interest paid.
Nora Carol Photography/Getty Images
Learn about our editorial standards.