BDC non-accruals climb as stress stays contained

KBRA's second-quarter review of 35 rated BDCs puts the median non-perpetual-life vehicle at 2.75% of investments at cost on non-accrual, up from 1.81% in the first quarter, a jump the agency frames as the visible edge of headwinds that have become harder to ignore. The survey covers BDCs with published and unpublished ratings, and the deterioration it describes sits with a relatively limited number of borrowers and managers, reading more like a normalization after an unusually benign stretch than broad deterioration, even as selective credit troubles became more common in the quarter while worries about a wider credit-quality slide and AI-related disruption to software moderated.

A market-wide non-accrual jump forced by macro pressure would argue for cutting exposure; an increase that is real but narrow argues for finding which credits and underwriters are producing it. The KBRA data points to the second, and the median misleads by making the sector look as though it moves as one body when the actual stress sits with a few managers.

Income pressure in the same period is not credit loss, but it is still a BDC problem, which KBRA traces to interest rates lower than they were in 2025, tighter loan spreads, and slower M&A and refinancing activity, not to a broad credit downturn. For shareholders that distinction is cold comfort, because a market-driven income gap can persist for several quarters even when the credit book is stable.

Non-perpetual-life BDCs kept leverage broadly stable, with a few increases tied to weaker asset quality, while perpetual-life vehicles took on modestly more debt because capital deployment—and for some issuers redemption requests—ran ahead of debt repayment. Managers that need extra leverage while also carrying non-accruals are carrying both sides of the story.

Redemptions at BDCs spilled into the second quarter, following the earlier run of concerns about private credit lending standards and the fear that AI could batter the software sector, where the asset class carries significant exposure. The report does not blend the redemption and non-accrual data into a single stress number, but the two lines are related for any BDC forced to raise cash while some loans have stopped paying.

Direct lending, as this publication has argued, has become a manager-picking market; the KBRA data shows what that looks like on a BDC balance sheet, with a rising median driven by a concentrated base of distressed borrowers.

For owners of BDCs the practical motion is from category to book: the 2.75% median summarizes how the asset class is doing, not how any one manager is doing. The spread that matters is the one between the BDCs contributing to that number and the ones that are not, and the report points allocators to exactly that distinction.

Next quarter the least informative question will be whether non-accruals rose or fell. The most useful question is whether the managers that drove this quarter's increase are the same managers whose earnings are under the most pressure; in a market this differentiated, the median is the last number an investor should lead with.