When Is a Mark Wrong, and When Is It Just Different?

Sooner or later, someone sets two numbers next to each other and asks the fund to explain the difference. A consultant running LP diligence, an auditor pulling a sample, a co-investor comparing notes: here is your mark, here is someone else's on the same borrower, and they don't match. At the end of September 2023 the mismatch was worth real money. Six holders of Thrasio's debt reported it anywhere from 59 cents on the dollar to 84, Goldman Sachs Asset Management at the bottom and Monroe Capital at the top, four more scattered in between. Same borrower, the same quarter-end, twenty-five cents between the high and the low. The fund holding any one of those marks still has to say something when the question comes, and the two answers it reaches for first are both wrong.

The two easy answers are both wrong

The first answer is to close the gap. Move your mark toward the other number, because a visible difference invites questions and matching numbers don't. But your mark might have been the better one, and moving off it just to look consistent trades away the more accurate figure for a quieter file.

The second answer is to wave the gap off as valuation judgment. Sometimes that is exactly right. Sometimes it is the cover a stale or aggressive mark was counting on, because once every difference is just judgment, no difference ever has to be examined. Force the marks to agree and you have treated every gap as an error. Dismiss them as judgment and you have treated every gap as fine. Both skip the only question worth asking: what kind of difference is this?

A wide spread proves nothing

A spread as wide as Thrasio's tempts a third easy answer, quieter than the first two: that valuation is simply subjective, so every holder is entitled to its own number. It is the most comfortable read, and it is still wrong. Some of those marks may have been current and carefully built. Others may have been stale, or aggressive, or resting on a position that only looked comparable. A wide range does not make everything inside it defensible.

The standards do not bless a number just because it came out of the right process. They keep the measurement objective, the technique, and the inputs separate, and even within a single valuation, when more than one technique yields a range, ASC 820 makes the reporting entity choose the point that best represents fair value. The range is where the work starts. It is not where the work is allowed to stop.

Three kinds of different

If neither easy answer holds, the fund has to do the thing both of them dodge: work out which of three things the gap actually is. The first is that the marks were never comparable. Two numbers only conflict if they measure the same thing: the same instrument and tranche, since a first-lien term loan is not the second-lien piece beneath it, the same unit of account, the same measurement date, the same objective. Miss any of those and the two numbers were never arguing; they were answering different questions in the same units. Principal market is subtler. Under ASC 820 the accessible market can differ from one holder to the next, so two holders each using the market appropriate to them, or the most advantageous market where none exists, can land on marks that were never directly comparable. A holder that used the wrong market has produced a defect instead.

The second kind of gap is real and still legitimate. Two holders can measure the same thing, each using the relevant information reasonably available to it, and land on different but supportable reads of the credit: the spread the risk demands, the expected cash flows, the odds of default, the recovery if it does. A covenant breach or a missed forecast does not set the mark by itself; it is evidence, and careful people weigh evidence differently. What it cannot be is ignored when it is material and reasonably in hand. A holder that sets aside evidence it had has not made a different choice, it has made a worse one. This is the case worth naming for what it is: reasonable estimation uncertainty, two defensible numbers on the same asset.

The third kind is the one the first two exist to expose. The marks are comparable, the gap is not explained by any supportable difference in judgment, and it traces back to something broken: information that never caught up to the current quarter, a technique wrong for the instrument, assumptions nothing supports, a method quietly changed from last period for no stated reason. Changing technique is fine when the new one measures fair value better; changing it without support is the defect. A defective mark gets corrected, not averaged toward or deferred to.

Go back to the six Thrasio marks. The spread alone cannot tell you whether you are looking at answers to slightly different questions, a set of supportable reads on a genuinely murky credit, or a mark that would not survive the diagnosis. All three produce a gap. Only the diagnosis tells them apart.

So when the consultant or the auditor finally sets your mark beside someone else's, neither reflex answers the question. "Ours is right" might hold up, but only the diagnosis can show the mark is supportable; "valuation is subjective" explains nothing. What survives is knowing which of the three you are holding, and being able to show it. That is the line between a fund that can sign its number and one that is guessing, however confidently it guesses.