Private credit's marks go on trial

A private credit fund does not have a price. It has a valuation, produced quarterly, built from discounted cash flow models informed by the manager's own assessment of how its borrowers are doing. For most of the asset class's growth that distinction was academic.

The distinction is no longer academic. Practitioners describe the sector entering a live test of whether managers can prove the values inside their portfolios, which is a question that only becomes urgent when someone needs the answer.

The structural criticism is old and simple. Where publicly traded credit moves sharply on news, reported valuations in private vehicles move by a fraction of that, and the smoothness is a property of the valuation method rather than of the loans.

Investors bought that smoothness on purpose. A portfolio whose reported value does not swing is easier to hold, easier to report to a board, and easier to lever against, and for a decade the absence of volatility was treated as a characteristic of the asset rather than of its measurement.

Litigation is now testing the claim directly. Firms tracking the sector describe suits alleging misstated values and delayed loss recognition, which move the argument out of investment committee meetings and into a forum with discovery.

Discovery is the part that changes behavior. A valuation methodology that has never been examined by an adversary is a different document from one that has, and managers know that every quarterly mark is now a potential exhibit.

Supervisors have converged on disclosure and stress testing rather than on prescribing values. That is the appropriate instrument: the objection was never that a model is inherently wrong, but that a model whose inputs are unobservable and whose author is compensated on the output deserves scrutiny.

The mechanical problem is that private credit lends to companies that do not trade either. A manager marking a loan to a mid-sized borrower has no market price to reference, and reasonable people can build defensible models that differ by a lot.

That range is fine when nobody transacts. It becomes concrete when a fund permits redemptions, because a redeeming investor is paid the manager's number and the remaining investors keep whatever the loans turn out to be worth.

Managers argue, with justification, that the underlying loans have mostly performed and that dispersion in marks is not the same as error. The counterargument is not that the marks are wrong. It is that nobody outside the manager can currently tell.

That is what has changed, and it is a governance shift rather than a credit event. Investors are asking to see the workings of a number they previously accepted, and the managers who can produce them will be distinguishable from the ones who cannot.