Private Credit's Reckoning Will Land on Insurers, Not Banks

What's New

Nick Nemeth of Mispriced Assets argues in a conversation with Jack Farley on Monetary Matters that the standard defense of private credit, that the banks are not holding the paper, is technically true and analytically useless. Roughly a trillion dollars of private credit has migrated onto insurance balance sheets totaling about ten trillion dollars, an asset base he sizes at roughly 150 percent of the Federal Reserve's. The comparison he reaches for is not 2008 but 1929, on the grounds that subprime was a 1.2 trillion dollar problem and this one is larger, less visible, and sitting behind entities with no FDIC equivalent. State guaranty funds exist, but he describes them as largely unfunded promises tied to tax credits rather than pools of cash, against 22 trillion dollars of in force life insurance. The trigger he identifies is not a default wave in the abstract. It is a modest uptick in annuity surrenders meeting balance sheets he says are levered at multiples that would embarrass a 2008 investment bank.

Why It Matters

• This is an observability problem, not a forecasting problem: The thesis rests on loan level work across BDC filings, footnote reading on lien structures, and cross referencing insurer statutory disclosures. All of it public, none of it assembled.

• The constraint is data, not talent: If he is directionally right, neither regulators nor allocators lack analysts. They lack a consolidated view across lenders, borrowers, and the insurance entities warehousing the exposure.

• Cost per insight is the real barrier: Seven months of manual work to reach a conclusion that should be a query is a platform gap, and platform gaps get filled.

• Reputation is currently doing the work of underwriting: Flows follow brand rather than portfolio quality, which is only sustainable while nobody can compare portfolios directly.

Big Picture Drivers

• Adjusted EBITDA as the load bearing wall: Underwriting runs at roughly seven times EBITDA, but that EBITDA is adjusted for synergies that S&P vintage data shows miss by 25 percent about half the time. Some sponsors add back rent, which converts seven times into nine or ten and a half times once PIK is layered on.

• Layered leverage that nobody aggregates: Leverage sits at the operating company, the fund, the BDC, the LP, and the GP simultaneously. Sovereign allocators can repo treasuries into ten to twenty times leverage before contributing to funds, while GPs borrow against their own stakes at roughly seven percent cost of capital.

• Ratings arbitrage two levels deep: Insurance debt is rated off CLO exposures, and CLO underlying loans are rated by smaller agencies rather than the major three. Nemeth's view is that the large agencies are structurally paid not to ask questions rather than incapable of answering them.

• Triple B mezzanine as the regulatory optimum: He describes triple B CLO tranches as the minmax point between yield and required capital reserves, which explains why insurance capital concentrates there rather than in higher quality paper that would not cover an eight percent cost of capital.

• Software as the concentrated exposure: Software is the largest deal category, and he argues many portfolio companies are thin wrappers around workflows that frontier models can now handle. One to two year contracts mean the repricing arrives on a schedule rather than all at once.

• Surrender economics that were never stress tested: Actuaries model rate driven surrenders well and reputational surrenders not at all. A three percent penalty is not a deterrent against a policyholder who believes the carrier is impaired.

By The Numbers

• Ten trillion dollars of insurance balance sheet, roughly a trillion of it in private credit, against a 1.2 trillion dollar subprime market that broke the system in 2008.

• Under ten percent of Athene's assets are level one, with roughly 40 percent level two and 50 percent level three, meaning half the book is marked by model rather than by market.

• Six point three percent direct lending default rate, above 2008 levels, in an economy he describes as not particularly bad. Software sits at 2.3 to 2.4 percent, flattered by PIK.

• Six quarters of sustained elevated defaults is his threshold for forced CLO downgrades, the mechanism that converts a credit problem into an insurance capital problem.

• Over 100 of 680 insurers would breach risk based capital constraints on a low single digit rise in surrenders, on his math.

• Four to six billion dollars of what he calculates as real capital at Athene against roughly 300 billion in assets, versus the 20 to 30 billion he says is presented.

• Ten percent first year sales commission on some annuity products, against surrender penalties that fall to three percent by year three, a mismatch created at the point of sale.

Key Trends to Watch

• The refinancing wall meets the AI repricing: Loans underwritten on 10 to 15 percent growth assumptions come due against portfolio companies whose growth case has weakened. He expects recoveries well below marked expectations, particularly in software where collateral is thin.

• Rational arbitrage into closed end vehicles: Allocators can exit funds marked at par and buy substantially similar exposure in listed BDCs at discounts. ARCC trades inside five percent while FSK sits near 50 percent, both with roughly 60 percent software. Once that trade starts, it is outflow from the asset class by definition.

• Reflexivity in fundraising: Inflows are the load bearing assumption. Spreads and multiples are priced off continued growth, so a deceleration from ten percent to five percent is itself the shock, not merely a warning.

• Contrarian dispersion among the managers: He rates Ares as the widest gap between brand and reality, Blackstone as substantially a marketing organization, and Blue Owl as the most underrated underwriter despite what he calls disastrous public relations. Apollo he credits with genuine underwriting skill and legal aggression, while questioning the capital cushion.

• Regulatory capacity as the binding constraint: State insurance departments are not equipped to value a dental practice rollup in a windown. His view is that Dodd Frank relocated the risk to the balance sheets least able to resolve it.

Memorable Quotes

• "These insurers balance sheets are levered up in many cases more than Lehman Brothers" This is the load bearing claim of the entire thesis, and the one that separates it from routine private credit bearishness.

• "They are not dumb. They are paid to be dumb." On the ratings agencies. He frames the failure as a business model outcome rather than an analytical one, which is precisely why he expects it to repeat.

• "Top of book liquidity is not liquidity" The distinction between a quotable price and a liquidation price, and the reason he does not accept broadly syndicated loans as meaningfully safer than direct lending.

The Wrap

Strip away the crisis framing and what remains is a claim about measurement. Every element of the argument, the adjusted EBITDA, the stacked leverage, the level three marks, the ratings chain, the surrender sensitivity, is knowable from filings that already exist and are already public. The reason nobody has priced it is that assembling the picture requires stitching together BDC disclosures, insurer statutory filings, and loan level footnotes across hundreds of entities, which is exactly the work no single participant is incentivized to fund. For technology providers serving this market, the opportunity is not a better risk model. It is the cross lender, cross entity data layer that makes the aggregate position visible before the surrender window rather than during it.