What actually happened to loan prices last quarter?
The average stood still and the market underneath it did not. The weighted average bid on the US leveraged loan market ended the second quarter of 2026 at 94.96, essentially unchanged from about 95 at the end of March and modestly below its year-end 2025 level of 96.64.
That flat line is two movements cancelling. Excluding software, the average secondary bid firmed by roughly 0.4 points over the quarter. Software loans fell approximately 1.9 points over the same three months.
The result is a gap between software and the rest of the index of 9.6 points, which Carlyle describes as historically wide. Two loans of the same seniority, written to the same documentation standard, are now being bid nearly ten points apart on the basis of what the borrower sells.
The equity market recorded the same split in its own currency. Software stocks finished the first half of 2026 down 20 percent while traditional economy sectors led performance and small capitalization shares outperformed large by more than 1,000 basis points. Two markets that price the same companies through different instruments reached the same conclusion in the same six months.
| Measure | Reading | Date or period |
|---|---|---|
| Weighted average bid, US leveraged loan market | 96.64 | 31 December 2025 |
| Weighted average bid, US leveraged loan market | about 95 | 31 March 2026 |
| Weighted average bid, US leveraged loan market | 94.96 | 30 June 2026 |
| Average secondary bid, loans excluding software | about +0.4 points | Quarter to 30 June 2026 |
| Average secondary bid, software loans | about -1.9 points | Quarter to 30 June 2026 |
| Gap, software against the rest of the index | 9.6 points | 30 June 2026 |
| Leveraged loan defaults plus distressed exchanges | 2.77% | Quarter to 30 June 2026 |
| Software stocks, first half total return | -20% | Half year to 30 June 2026 |
Why is one sector moving against the index?
Because the market has stopped treating recurring software revenue as the safest cash flow in the loan universe, and it has not said what it now thinks the right price is. Carlyle notes that new issuance of collateralized loan obligations, the largest single buyer of broadly syndicated loans, moderated from last year's pace amid tight loan spreads and continued uncertainty around software and energy exposures.
That is a demand statement, not a credit statement. The buyer that sets the marginal bid for this paper slowed down, and it slowed down specifically where it is unsure. A loan does not need to miss a payment to fall two points. It only needs the person who would otherwise have bought it to wait.
Note what the split is not. It is not the whole market softening, because the other side of the index went up. It is not a default wave, because defaults fell. It is a repricing of one sector's assumed durability, carried out by buyers rather than by borrowers.
For an owner, the honest reading is that a sector label has become a price input in the loan market in a way it was not two years ago. That is uncomfortable if you sit on the wrong side of it and worth understanding either way, because the label is applied to you by someone else and it does not consult your numbers first.
“An index average is a convenience, and it stops being useful the moment its halves separate. When the composite barely moves and the two sides of it move a full two points in opposite directions, the average has stopped describing anybody in the market. The only number that matters to a borrower is where paper in their own sector is bid, and that is a question you have to ask specifically because no headline will answer it.”
If defaults are at a three year low, why does this matter?
Because price and default are different clocks, and the price clock runs first. Leveraged loan defaults plus distressed exchanges fell to their lowest level in over three years at 2.77 percent in the second quarter of 2026. On that measure the market is in better credit health than it has been since 2023.
Both readings are correct at once. Borrowers are meeting their obligations, and buyers have simultaneously decided they will pay less for a subset of those obligations. Credit performance describes what has already happened. A traded bid describes what the buyer expects to happen, and it moves years earlier.
The amendment data show the same forward posture. Repricing and extension activity held roughly steady from the first quarter, with a more than doubling of extensions largely offsetting a decline in refinancings. Borrowers are buying time in preference to resetting terms, which is what a market does when it prefers not to test pricing.
So the low default rate is genuinely good news about the recent past and tells you very little about the terms available on your next facility. An owner who reads only the default headline will be surprised by the quote.
What does the split mean for a borrower or an owner?
That the sector question now arrives before the credit question. A lender pricing a new facility looks at where comparable paper trades in the secondary market, and if that reference point has fallen two points while the index has not, the quote reflects the reference point rather than the index.
It changes what to ask for in a process. The useful request is not for the market rate. It is for the recent comparable transactions the lender is actually pricing against, and specifically whether those comparables sit inside or outside the sector the market has marked down.
It changes the value of evidence that cuts against a sector label. If buyers are discounting an entire category on assumed durability, then contracted revenue, demonstrated retention through a full renewal cycle, and customers who cannot leave cheaply are worth more than they were when the whole category was assumed durable. The proof has to be built before a financing, because during one it reads as advocacy.
It also changes timing for anyone on the firmer side of the split. Loans outside software firmed. A business in a traditional economy sector is being refinanced into a market that has improved for it specifically, and that is a narrower and more perishable advantage than a general view about interest rates.
What would tell you the split is closing?
Issuance of collateralized loan obligations returning to its prior pace, because that is the buyer whose retreat opened the gap. Watch the pace of new formation rather than the level of spreads, since a buyer coming back shows up in volume before it shows up in price.
After that, the direction of the discounted side rather than the level of the index. The gap narrows honestly when software bids recover and dishonestly when the rest of the market falls to meet them. Those two paths produce the same headline and very different conditions for a borrower.
Treat one quarter as one quarter. A 0.4 point move in one direction and a 1.9 point move in the other are small absolute numbers, and the composite index has travelled less than two points since year-end 2025. What makes this worth watching is the separation, not the magnitude.
The practical conclusion is about sequence. The loan market repriced a sector before anything in that sector defaulted, which means the information reached lenders before it reached borrowers' terms. The interval between those two events is the window in which a prepared company still negotiates on its own record rather than on its category's.
As of August 2026
Sources: The Carlyle Group Inc., quarterly report on Form 10-Q for the quarterly period ended 30 June 2026, filed with the US Securities and Exchange Commission on 10 August 2026 (accession 0001527166-26-000045), section Trends Affecting Our Business, for the statement that leveraged loan defaults plus distressed exchanges fell to their lowest level in over three years at 2.77 percent, that the weighted average bid on the US leveraged loan market ended the quarter at 94.96 against a year-end 2025 level of 96.64 and was essentially unchanged from 95 at the end of the first quarter, that amendment activity comprising repricings and extensions remained relatively steady from the first quarter as a more than doubling of extensions largely offset a decline in refinancings, that weakness remained concentrated rather than broad-based, that excluding software the average secondary bid firmed by roughly 0.4 points over the quarter while software loans fell approximately 1.9 points, that this widened the gap between software and the rest of the index to a historically wide 9.6 points, and that new issuance of collateralized loan obligations moderated from last year's pace amid tight loan spreads and continued uncertainty around software and energy exposures; and, from the same section, for the statement that software stocks finished the first half of 2026 down 20 percent while traditional economy sectors led performance and small capitalization shares outperformed large capitalization shares by over 1,000 basis points. The description of a bid as a price per 100 of face value is standard market convention and is our own explanation rather than a figure from the filing. A companion article on this site reads a single private credit fund's own valuation of its loans over the same quarter, another covers what the issuance of collateralized loan obligations signals about whether the loan market is open, and a third covers the sector rotation in deal value that this loan price split accompanies.