The headline number in corporate credit is that nothing is happening. Investment-grade and high-yield spreads have spent the year in a narrow band. On the index, this is one of the quieter credit markets in a decade.
Underneath it, two separate pieces of work published in the past week point at the same thing from different angles.
The first: Bloomberg reports that the pile of debt trading at spreads unusually wide for its credit rating has more than doubled since the start of the year, to roughly $1 trillion — about $580 billion of US bonds and close to $400 billion in Europe, looking at non-financial high-grade paper with more than three years to run. These are bonds the agencies still call one thing and the market is pricing as another. An index average cannot show you that, because the tightening in the median name and the widening in the tail net out.
The second: the Federal Reserve Bank of Boston, examining business development company portfolios as a window into private credit, finds the share of BDC loans with payment-in-kind features has risen from around 6% to roughly 10% by early 2026 — a 67% increase over three years. Construction went from under 5% in 2022 to nearly 20%. Wholesale trade and transportation and warehousing more than doubled.
Why PIK is the tell
Payment-in-kind means the borrower stops paying interest in cash and starts paying it in more debt. There are benign reasons a loan is structured that way at origination — growth financings, sponsors preserving cash for capex. There is one dominant reason a loan converts to PIK later: the borrower could not make the cash payment and the lender preferred an amended loan to a defaulted one.
The Boston Fed’s useful observation is not the level but the spread of it. The increase is broad-based across diverse sectors rather than concentrated in two or three troubled industries. Concentrated distress is a sector story. Distress that shows up everywhere at once is a cost-of-capital story — middle-market borrowers who financed at one rate and are now servicing at another.
PitchBook’s work on the same universe adds the corroborating detail: non-accruals rising by both borrower count and cost basis, software marks falling faster than the rest of the book, and PIK loans taking the hardest valuation cuts of anything held. That last point matters. If PIK were simply a structuring choice, it would not be marked down harder than everything around it. The people holding it do not believe it either.
Our take: The index is not lying, it is averaging — and averaging is exactly the wrong lens for a market where the distribution has gone bimodal. What both datasets describe is dispersion: a large, healthy, tight-trading core and a growing tail that is being repriced quietly, name by name, without a spread-widening event to put a date on. Dispersion is how credit cycles start every time. The mistake is reading calm at the index level as calm in the book you actually own. It also connects to the story we ran this morning on hyperscaler bond issuance — when order books thin from five times covered to two, the marginal buyer is already being choosier than the headline spread suggests.
The part that has a date on it
Maturities in this universe build steadily through 2028, and a substantial amount of the software debt and the PIK debt comes due inside that window. PIK is a deferral instrument: it converts a cash-flow problem today into a larger principal balance at maturity. That works if the borrower’s cash flow recovers or if refinancing is available on reasonable terms. It works considerably less well if neither happens.
Which is why the funding side is worth watching alongside the asset side. Banks have been reporting tighter standards for lending to business credit intermediaries and private equity funds, citing a weaker economic outlook and higher perceived borrower risk. Tightening supply into rising demand is not, on its own, a crisis. It is the condition under which a refinancing window closes.
What to watch
- PIK as a share of investment income, not of loans. The percentage of a BDC’s income that is accrued rather than received in cash is the honest version of the metric. It appears in the filings.
- Non-accruals by cost basis. Counting borrowers understates it when the larger positions are the ones going bad.
- Dispersion within ratings buckets. The share of a rating cohort trading more than a set distance from its cohort median tells you more right now than the median does.
- BDC discounts to net asset value. Listed vehicles reprice daily against marks that reprice quarterly. The gap is a live opinion on the marks.
- The 2027–28 maturity wall. Specifically how much of it is refinanced early and at what cost. Early refinancings that clear easily are the all-clear. Extensions are not.
None of this is a forecast that something breaks. It is a note that the instrument most people use to check — the spread on the index — is currently the instrument least able to see it. This is reporting on market structure, not investment advice.