Private Credit
Private Credit's Upside-Down Waterfall
A second lien cannot be worth more than the first. Two June 30 schedules say it is.
Once a first-lien loan is marked below 80 for credit reasons, every claim beneath it is an option on enterprise value. Marking those claims one at a time, by yield, overstates them by 40 to 55 points. Two marks in one capital structure are two statements about one enterprise value, and in this summer's public filings they do not agree.
Different securities in one company should carry different marks. Priority, coupon and maturity all differ. My claim is narrower. The marks have to come from one enterprise value, run through one waterfall. They can differ. They cannot contradict each other.
The Wall Street Journal reported that lenders have marked loans to Cornerstone OnDemand and Symplr down by more than 30%, and tied private equity's trouble to the Fed's quarter-point hike. A quarter point adds about $5 million a year of interest on a $2.0 billion floating-rate loan. Through the discount rate it takes about 3% off enterprise value. These loans lost thirty points. That is a statement about enterprise value, not rates.
What the June 30 schedules show
Ares Strategic Income Fund's 10-Q for June 30, 2026 holds both loans in each company. The marks below are fair value over principal.
| Company | First lien | Second lien |
|---|---|---|
| Cornerstone OnDemand | 63.2 | 64.0 |
| Symplr | 65.4 | 67.0 |
| Kaseya | 76.9 | 60.1 |
In Cornerstone and Symplr, the junior loan is marked above the senior loan. At those prices, coupons and maturities, the second liens yield less than the first liens: roughly 31% against 35% for Cornerstone, and 35% against 47% for Symplr. No market participant accepts less yield for more risk in the same company. Read as recovery prices, the marks say the second lien recovers as much as the first. Absolute priority says it cannot.
The schedule's own footnotes show where the break sits. Both second liens carry the footnote for Level 3, valued using unobservable inputs. Both first liens do not. Kaseya is the one name where the order is right, and neither of its loans carries that footnote.
Six months earlier, the same schedule had the Cornerstone first lien at 91.5 and the second lien at 91.0. Both fell 27 to 28 points. A waterfall does not do that. It takes the junior claim down first and hardest.
One enterprise value, run through the stack
Take a hypothetical stack: a $2.0 billion first lien, a $0.5 billion second lien, a $0.4 billion preferred, and 2.25 years to maturity. I use the option pricing method, with each claim's face as a breakpoint, and assign 3 points of each loan's discount to non-credit factors. At 38% enterprise value volatility, a first-lien mark of 63 implies enterprise value of $1.52 billion, or 0.76x the first lien's face. From that value the second lien is worth about 14 and the preferred about 7.
| EV volatility | Implied EV | Second lien from that EV | Filed | Gap |
|---|---|---|---|---|
| 33% | $1,448M | 10.1 | 64 | 54 |
| 38% | $1,517M | 13.7 | 64 | 50 |
| 49% | $1,729M | 20.6 | 64 | 43 |
The range is the interquartile range of de-levered two-year volatility for 24 public software companies at June 30, 2026. At the highest volatility in the set, 57%, the gap is still 39 points. Those companies are healthier than these borrowers, and higher volatility helps the junior claim. It does not rescue the mark.
Run it in reverse. A second lien at 64 requires enterprise value of $3.40 billion, 2.2 times what the first lien implies. At that value the first lien prices at 90, not 63.
One enterprise value, every claim
Set the first-lien mark and see what it leaves for the claims below it. Then move the second-lien mark and watch the two marks pull EV apart.
Ordering test fails. The junior loan is marked at or above the senior loan.
What the first-lien mark leaves for the rest
| EV implied by the first-lien mark | $1,517M |
| Implied EV to first-lien face | 0.76x |
| Second-lien mark from that EV | 13.7 |
| Preferred value per 100 of claim, same basis | 6.8 |
| Value left for common | $74M |
| Filed second lien above the one-EV mark | 50.3 points |
What the second-lien mark requires
| EV the filed second-lien mark requires | $3,400M |
| Ratio to the EV the first lien implies | 2.24x |
| First-lien mark the model gives at that EV | 90.1 |
| Gap to the filed first-lien mark | 27.1 points |
Assumptions
How this calculates
- The stack is hypothetical. First lien $2,000M, second lien set by the slider, preferred $400M, then common. Breakpoints are the running totals.
- Credit price of a loan = (mark + non-credit points) ÷ 100, capped at 0.99. The model treats the rest of the discount as credit.
- Call on EV at breakpoint K:
C(K) = EV × N(d1) − K × N(d2), withd1 = [ln(EV ÷ K) + σ²T ÷ 2] ÷ (σ√T)andd2 = d1 − σ√T.C(0) = EV. - Value of the claim between two breakpoints =
C(lower) − C(upper). Value per 100 of claim = that value ÷ claim size × 100. - The model assumes zero net drift in EV and a zero rate. Coupons are assumed to offset discounting. This is generous to junior claims.
- Forward: solve by bisection for the EV at which the first-lien claim value equals its credit price. Value the other claims at that EV. The second-lien and preferred figures subtract the same non-credit points, floored at zero.
- Reverse: solve for the EV at which the second-lien claim value equals its filed credit price. Price the first lien at that EV and subtract the non-credit points.
- Guard: when a mark plus the non-credit points exceeds 88, that direction reports nothing. A mark near par does not bound EV.
- Ordering test: fails when the filed second-lien mark is at or above the first-lien mark.
Two marks in one capital structure are two statements about one enterprise value. The gap shows how far apart they are.
Illustrative. Hypothetical capital structure. Not a valuation opinion and not reliance-grade.
The preferred tells the same story. Blue Owl Technology Finance carries Cornerstone's Series A preferred at 48% of cost and Kaseya's perpetual preferred at 57%. New Mountain Finance carries Symplr's Series A preferred at 71%. All are Level 3.
What this means for September 30
Third-quarter marks are being set now. If any first lien in a company sits below 80, value the company once. Treat the senior mark as an input to the junior model. ASC 820-10-35-24C requires a technique built on unobservable inputs to reflect observable market data at the measurement date, including the price of a similar asset. The first lien of the same borrower is the most similar asset there is.
Ask whoever signs your marks whether each position gets checked against the other positions in the same issuer. Third-party valuation work is commonly scoped position by position. Valuing the enterprise once and running every tranche off it is how MELD scopes private credit marks.
The counterargument
Each tranche is its own unit of account, with its own market participants. A thinly traded first lien can print at a distressed seller's price. The Level 3 model on the second lien may be the better read on enterprise value.
I accept all of that, and it does not help. If the second-lien model is right, enterprise value is near $3.40 billion and the first lien belongs near 90. If the first-lien mark is right, the second lien belongs near 14. I am not saying which mark is wrong. Both cannot stand in one schedule.
I am wrong if the second liens have support the waterfall ignores: separate collateral, a sponsor guarantee, or a pending transaction that pays them out of order. No split between credit and non-credit discount fixes it. With the same non-credit points on both loans, the junior never reaches the senior.
Once the first lien is impaired, there is one enterprise value, and every mark in the stack has to come from it.