What does the Boston Fed's sample actually measure?
The disclosed loan books of 168 business development companies, nearly 890,000 company-quarter loan observations, read from the schedules of investments in their SEC filings and tracked most closely from the start of 2022, when reporting became consistent enough to compare lenders against each other. BDCs are the exception in private credit: every other kind of private lender keeps its holdings to itself.
The sample is the middle market. From the first quarter of 2022 to the fourth quarter of 2025 the number of reporting BDCs rose from an average of 105 to 166 a quarter and the number of loans roughly doubled, with a modest decline in both during 2025 that matches the wider slowdown in leveraged lending. The borrowers are, in the authors' description, too large for small-business lending and too small for the bond market.
The books are concentrated and they differ. Internet and software companies are the largest sector in the median BDC's portfolio by loan count at about 20 percent, with manufacturing and industrial firms, professional services and health care each at 10 to 15 percent. The interquartile range for the software share runs from under 15 percent to nearly 30 percent, and some lenders hold more than a third of their book in technology. A borrower's lender is not the average lender; it is one of these.
Which industries' borrowers are paying in kind?
Nearly all of them, and construction most. The share of BDC loans on which the borrower is actively capitalizing interest rather than paying it in cash rose from approximately 6 percent in early 2022 to roughly 10 percent by early 2026, a 67 percent increase over three years. The measure is strict: it counts only loans whose end-of-period rate structure includes a PIK component, so an unexercised toggle does not appear.
The industry breakdown is the part of the paper a borrower should read twice. Construction shows the largest increase, from under 5 percent of loans in 2022 to nearly 20 percent by early 2026. Wholesale trade and transportation and warehousing both more than doubled. Accommodation and business services, where the baseline was already high, rose more modestly. The authors' point is that the rise is spread across diverse sectors rather than concentrated in a few troubled ones, which reads as broad pressure on middle-market cash flows: a company that serviced debt at SOFR plus 5 points when SOFR was near zero carries a substantially greater burden with SOFR above 4.
Our reading is narrower and more practical. A lender that holds construction, distribution or freight paper is now underwriting your refinancing against a cohort where one loan in five, or one in ten, has already stopped paying cash. That cohort is the lender's reference set for what happens to companies like yours, and it shapes the covenant package, the cash sweep and the amortization it asks for, before anyone has looked at your numbers. A founder in one of these sectors should walk in knowing what the cohort looks like, because the lender does.
| Measure | Early 2022 | Early 2026 |
|---|---|---|
| Share of BDC loans with active PIK, all industries | ~6% | ~10% (+67%) |
| Construction | under 5% | nearly 20% |
| Wholesale trade | more than doubled | |
| Transportation and warehousing | more than doubled | |
| Accommodation; business services | already elevated | modest increase |
| Median spread over SOFR, BDC loans | 4 to 5 pts | narrowed ~1 pt over two years |
| Spread over SOFR, large-bank high-yield term loans | ~2 pts | |
| Fair value to cost, BDC portfolios | ~1.0 | ~1.0 |
| S&P BDC Index vs S&P 500 | about -20% (early 2025 to mid-2026) | |
| Committed bank credit to BDCs | ~$10bn (2013) | >$50bn (2025); ~$35bn drawn |
Why did the spread fall while the deferrals rose?
Because more capital was chasing the same loans, on the paper's reading, and possibly because lenders were cutting the price of debt to keep borrowers current. The median BDC spread sits at 4 to 5 percentage points over SOFR, with an interquartile range of about 1.5 points that barely moves across industries with quite different risk. Against that, large banks subject to the Federal Reserve stress test charge roughly 2 points over SOFR on high-yield term loans, per their FR Y-14 filings. The premium is the price of lending to companies the banks will not.
Over the past two years that median spread has narrowed by approximately 1 percentage point. The authors call the combination a puzzle: if borrowers are increasingly unable to pay cash interest, lenders should be demanding more, and instead they are accepting less. They offer two explanations and do not choose between them. One is implicit restructuring, lowering the cost of debt to reduce the probability of default. The other is competition, with rates falling because the market for the loan got crowded.
For a borrower the two explanations point the same way. The price you are quoted on a refinancing today says more about how many lenders want the paper than about your credit, and it will not hold if the crowd thins. And when cash gets tight, the concession you are most likely to be offered is not a lower spread but a switch of some of the coupon into kind, which is the cheaper concession for the lender and the more expensive one for your equity. What that switch costs at the accreted balance is set out in our companion piece on payment-in-kind interest.
“A market where the spread narrows while the share paying in kind rises is a market where the lender has become cheaper and more lenient at the same time. Neither is a comment on the borrower. Both will reverse together, and the borrower who priced its plan on the lenient version will find out first.”
Does the lender's own mark tell a borrower anything?
Yes, and it is public. BDCs value their loans with internal models, and the ratio of reported fair value to cost stayed near 1.0 across the sample, with a slight dip into 2023 and a gradual recovery: in aggregate the lenders say the books are worth par despite the rise in deferrals. The dispersion between lenders is wide. Their shareholders disagree with them: the S&P BDC Index fell roughly 20 percent against the S&P 500 between early 2025 and mid-2026.
The marks still carry information. A one standard deviation fall in a BDC's fair value ratio in one quarter is associated with about 50 basis points of lower abnormal return on its shares the next, which means public investors read the disclosed mark and act on it. A middle-market borrower whose loan sits in a BDC book should know that its own line in the schedule of investments, cost and fair value side by side, is a number anyone can read each quarter, including its customers, its competitors and its next lender.
What is behind the lender, and is it tightening?
Bank credit, and yes. Committed bank credit to BDCs grew from approximately 10 billion dollars in 2013 to more than 50 billion by 2025, with about 35 billion drawn. The paper judges the exposure modest, at under 2 percent of large banks' Tier 1 capital and predominantly senior secured, so a bad stress in BDC books would not threaten bank solvency. It also notes the April 2026 Senior Loan Officer survey's special questions, in which large and regional banks reported tighter standards for business credit intermediaries and private equity funds on maximum loan size, maturity, risk premiums, covenants and collateral, citing the outlook, reduced risk tolerance and borrower credit risk, at the same time as demand from those vehicles strengthened.
The caveats are the authors' own and they matter. BDCs are a portion of private credit and, being listed, may lend differently from the fully private vehicles that publish nothing; the patterns may not generalize to insurance balance sheets or private equity credit arms. Their conclusion is that the concerning signs, rising deferrals, compressing spreads and concentrated exposures, likely exist in some form across the wider market. For a borrower, the practical version is three questions before signing: which industry cohort the lender holds you in, what share of that cohort is paying in kind, and what the lender's own bank line costs it now that the banks have tightened.
As of August 2026
Sources: Federal Reserve Bank of Boston, Current Policy Perspectives 26-6, Early Warning Signals in Private Credit? What BDC Portfolios Reveal about Emerging Risks, by José L. Fillat, Leslie Sheng Shen and J. Christina Wang, 5 August 2026, for the sample of 168 BDCs and nearly 890,000 company-quarter loan observations; the rise in reporting BDCs from an average of 105 to 166 a quarter and the roughly 100 percent cumulative growth in loan count from the first quarter of 2022 to the fourth quarter of 2025; the median portfolio shares of about 20 percent for internet and software with an interquartile range from under 15 to nearly 30 percent and 10 to 15 percent each for manufacturing and industrial, professional services and health care; the share of BDC loans with active PIK rising from approximately 6 percent to roughly 10 percent by early 2026, a 67 percent increase, with construction from under 5 percent to nearly 20 percent, wholesale trade and transportation and warehousing more than doubling, and accommodation and business services rising modestly from an elevated base; the median spread of 4 to 5 percentage points over SOFR with an interquartile range of about 1.5 points, roughly 2 points over SOFR on large-bank high-yield term loans per FR Y-14 data, and the narrowing of approximately 1 percentage point over two years; the two candidate explanations of implicit restructuring and competition; the fair value to cost ratio near 1.0 with a dip into 2023, the roughly 20 percent decline of the S&P BDC Index relative to the S&P 500 from early 2025 to mid-2026, and the approximately 50 basis point lower abnormal return associated with a one standard deviation decline in the fair value ratio; committed bank credit to BDCs of approximately 10 billion dollars in 2013 rising to more than 50 billion by 2025 with roughly 35 billion drawn, under 2 percent of large banks' Tier 1 capital; the April 2026 Senior Loan Officer Opinion Survey special questions as summarized in the paper; and the authors' caveats on generalizing from listed BDCs. The views in the paper are its authors' and not those of the Federal Reserve Bank of Boston or the Federal Reserve System. The reading of the industry cohort as a lender's reference set, of the spread and PIK movements as a lender becoming cheaper and more lenient at once, the observation on a borrower's mark being public, and the three questions before signing are ours. Companion articles on this site cover what payment-in-kind interest costs at the accreted balance, why lenders wrote loans down before anything defaulted, the discount at which the market prices listed BDCs, and who funds your lender.