Fitch counted 11 private credit default events in January 2026. The 2025 monthly average was 5.9. That is not a trend bending toward containment. That is a trend accelerating while the rating agencies publish forecasts that assume it won't.
The baseline numbers from Moody's (3.8%) and S&P (4.0%) are reassuring in the same way a doctor telling you your average blood pressure is fine while ignoring the spike that just happened. Those forecasts describe a world where private credit stress stays quarantined, where leveraged borrowers refinance on schedule, where macro conditions hold. I think those assumptions are already breaking, and the market is not pricing the breakage.
The Footnote Nobody Reads
Moody's own risk range extends to 8.3% through October 2026. That upper bound is not a stress test fantasy. It is the agency's own model output for what happens when conditions deteriorate modestly. The distance between 3.8% and 8.3% is enormous in dollar terms. For a pension fund holding $500 million in speculative-grade credit, the difference between a 4% and an 8% default rate is roughly $20 million in realized losses, before recovery haircuts.
Recovery rates are the second problem nobody wants to discuss. Marathon Asset Management's Bruce Richards estimates $0 to $0.30 on the dollar for software direct loans. That means on a $100 million defaulted position, you might recover $30 million. Maybe zero. These are not hypothetical numbers from a bearish blogger. Richards runs $24 billion in assets and is describing loans his firm has seen up close, leveraged at 8 to 10 times annual earnings with no refinancing available at anything close to original terms.
The last time I saw leverage multiples that high paired with collapsing refinancing markets was 2007, in a different asset class. The loans were different. The math was the same.
Redemptions Turn Paper Losses Into Real Ones
Private credit's defining feature has been its ability to avoid mark-to-market discipline. Funds hold illiquid loans, value them internally, and report smooth returns. That works until investors want their money back. The $20 billion in redemption requests now hitting private credit funds changes the calculus entirely, because managers who need to return capital must sell or restructure positions they would otherwise extend and pretend.
Richards named the sequence explicitly: extend and pretend, then distress, then defaults, then massive losses. The redemptions compress that timeline. A fund facing 10% outflows cannot sit on a non-performing software loan for 3 years hoping the borrower finds new revenue. It has to mark it down or sell it at a discount that establishes a price. Once a price exists, every other fund holding similar paper has to acknowledge it.
This is how contagion works in credit markets. Not through some dramatic Lehman-style explosion, but through forced price discovery that reveals what everyone already suspected but nobody wanted to book.
The leveraged loan market's composition makes this worse. Charles Schwab has noted that the majority of leveraged loans carry B ratings or lower. These are borrowers one bad quarter away from distress. Retail, housing, autos, consumer goods: Moody's has already flagged these sectors as vulnerable. Trade policy uncertainty and cooling demand are not abstract risks. They are Q1 earnings calls happening right now.
I will concede that emerging market corporate defaults look genuinely contained, with Neuberger Berman projecting 2.7% for 2026, well below the long-term average. That is a real bright spot. It does not help the U.S. private credit investor staring at a portfolio of software loans that cannot refinance.
The consensus view treats the 3.8% baseline as the most likely outcome and the 8.3% tail as a remote possibility. I think the January data, the redemption pressure, and the leverage multiples have already moved us off the baseline. Investors positioning for containment are making a bet that every transmission mechanism fails simultaneously. In 2007, the same bet had a name. It was called the soft landing.
