Why segregation matters more than scale.

The structured credit industry, like most corners of finance, has a gravitational pull toward scale. Bigger books, bigger balance sheets, bigger AUM figures on the tombstone. Scale is marketable. It photographs well in pitch decks and reads well in league tables. It implies permanence. And for the managers who achieve it, scale becomes the organizing principle of the franchise — the thing that determines compensation, strategy, and, eventually, risk.

We think this is the wrong organizing principle. The platforms that survive credit cycles are rarely the largest. They are the ones whose architecture prevents one book's trouble from traveling into another. The discipline is not glamorous. It does not expand the addressable market. It imposes real costs — more legal entities, more operational overhead, more friction at the moment a portfolio manager wants to cross-collateralize or net exposures. But it is, in our view, the single choice that most cleanly separates platforms that last from platforms that go spectacularly wrong.

The historical record is unambiguous on this point. The monoline insurers of the pre-2008 era — MBIA, Ambac, FGIC — were, until very late, regarded as disciplined credit shops. What destroyed them was not the quality of any single underwriting decision but the structure of the enterprise around those decisions. A guarantee written on a municipal bond sat on the same balance sheet as a guarantee written on a structured CDO. When the CDO book soured, it took the municipal franchise with it, not because muni credit had deteriorated but because the holding company's capital was fungible across lines. One book's trouble traveled. The pattern repeated, in miniature, with Archegos in 2021, and again with several family-office and multi-strategy vehicles whose prime-brokerage relationships bled across strategies that were supposedly independent. The lesson is always the same, and it is always relearned.

The shape of segregation

Segregation, as we practice it, is not a policy. Policies are promises, and promises degrade under stress. Segregation at River is a construction — a set of hard architectural choices that a future risk committee, under duress, cannot quietly unwind. Every book on the platform is issued from its own bankruptcy-remote special-purpose vehicle, with its own facility agreement, its own waterfall, its own trustee, and its own independent directors. The SPVs do not cross-guarantee. The parent does not backstop. There is no omnibus credit line against which a struggling book can draw to buy time.

• Each origination vertical — LatAm consumer receivables, APAC card portfolios, European SME trade finance — sits in a separately incorporated, ring-fenced facility with its own capital stack.
• Trigger-based wind-downs are enforced in code, not in committee. When a facility breaches its stated covenants, the waterfall redirects automatically; there is no discretionary forbearance at the platform level.
• No asset, cash account, or servicing contract is shared between facilities. Operational providers are contracted per-SPV, with severability built into the master services agreements.
• The parent holds equity interests in each SPV but no recourse obligations. A total loss in one facility is bounded by the equity committed to that facility.

The point of this construction is not elegance. It is that when — not if — one of our facilities experiences stress, the stress is contained by the legal and technical perimeter of that facility. A default in the LatAm consumer book cannot reach into the APAC cards facility, because there is no pipe between them. The code enforces this. The structure enforces this. A well-intentioned risk officer, three years from now, trying to smooth a quarterly result, cannot enforce otherwise — because the rails they would need do not exist.

We are aware that this is the harder story to tell. Scale is a narrative with a clear upward slope; a prospective LP can underwrite it in an afternoon. Segregation is an argument about what will not happen, about losses that will not compound, about contagion that will not occur. It is, in the language of options, short gamma on marketing and long gamma on survival. Howard Marks has written, repeatedly, that the investors who outlast cycles are those who refuse the trades that look obviously correct at the peak. The institutional equivalent, at the platform level, is refusing the architecture that looks obviously correct at the peak.

Scale is easier to sell. Segregation is harder to sell, and — this is the part that matters — harder to back out of once you have built it. The legal entities exist. The trustees are named. The code is deployed. You cannot quietly consolidate your way out of the discipline when the cycle turns and the temptation arrives. River chose the harder one, on purpose, at the founding, because we intend to be here in 2046.