The credit grade you're not allowed to read

My mother has an annuity.

She bought it with the money from selling her house. She asked me twice whether it was safe. I said yes, and I meant it.

An annuity is a simple promise. You hand over a lump sum. A life insurance company pays you every month until you die.

The promise is only as good as what the company owns.

So I went and looked at what they own. Mostly bonds. And a rising share of those bonds carry a credit grade you are not permitted to read.

Not because you lack a terminal. The grade arrives as a confidential letter to whoever paid for it. It goes to the insurer and to the regulator. It goes nowhere else, ever.

In 2018, US life insurers held $46 billion of these privately rated bonds. Last year they held $481 billion.

That count comes from a Columbia Business School paper dated May 31 this year. Xuelin Li, Sangmin Oh and Giacomo Ricciardi went through the regulatory filings of 650 life insurers and more than 2.8 million bond records.

Then they did the interesting part. They lined up bonds carrying identical letter grades. Some graded in public. Some graded in a letter.

The privately graded ones ran into trouble about twice as often. And they got downgraded less, not more.

Worse outcomes, fewer warnings, same letter on the page.

The authors think they found the reason. They isolated the small set of bonds that one insurer holds with a private grade and another insurer holds with a public one. Same security, two doors.

When a public grade exists on the same bond, the accuracy gap disappears.

So it is not the method that slips. It is the missing audience. A published grade gets argued with by people who own the bond and by people who want to short it. A letter gets read by three parties who all want the same answer.

I have no idea when this matters. Credit is calm. Spreads are tight. It may sit here for years.

But my mother's monthly cheque sits on top of it. So does every pension a company has handed to an insurer to look after.

And I told her it was safe.

This connects directly to how the money actually moves. Not the argument about opacity. The purchase orders.

S&P Global Market Intelligence went through what US life insurers bought in the third quarter of 2025. Bonds carrying a private letter rating made up 9.7% of everything they bought that quarter. Close to $30 billion at cost.

Of those, 94.7% came with a regulatory designation equal to BBB minus or better.

Holdings went from roughly $366 billion at the end of 2024 to more than $408 billion nine months later, on the same measure.

Apollo put its side of it to investors last November. Nine percent of Athene's net invested assets carry a private letter rating, and 98% of those are investment grade. Apollo's argument is that confidentiality is the only thing separating a private grade from a public one. Same cost, same work.

MetLife's chief investment officer took a different line on a call the same week. He said the firm's primary source of credit underwriting is the work MetLife does itself, not a rating letter.

That is a sentence worth reading twice. One of the largest holders of these assets is telling you not to lean on the grade.

And here's where it spreads. The work has to get done by somebody, and the somebody is small.

Egan-Jones Ratings operated for years out of a four-bedroom house on Haverford Station Road, outside Philadelphia. In 2024 it graded more than 3,000 private credit investments. Bloomberg counted the analysts who did it. About twenty.

The firm calls itself the most prolific grader in the market and says it has issued roughly 16,000 ratings. It is an accepted rating provider with the insurance regulators.

Senator Elizabeth Warren wrote to eight rating firms in July 2025 asking how they manage conflicts of interest in this market. Egan-Jones was one of them.

Then in November, Bloomberg reported that SEC enforcement lawyers were examining whether the firm and some senior executives had pushed commercial considerations into the rating process. No accusation of wrongdoing was made. The firm says it takes compliance seriously and remains in good standing.

In April this year the SEC pushed back on the firm's application to rate government debt and asset-backed paper again, saying the application raised questions about whether it had the resources to consistently produce credit ratings with integrity.

And in January, Bermuda's regulator stopped recognising the firm's ratings altogether.

One jurisdiction has already made its decision. The rest are still reading the letters.

Meanwhile, the exposure is not spread evenly. It never is.

Moody's put the US life industry's private illiquid bonds at $807 billion. The ten largest insurers hold $352 billion of that, or 44%. Those same ten hold 24% of the industry's fixed income overall.

So a small group is carrying roughly double its share.

The Bank for International Settlements said the same thing from Basel last October. Private credit grades used by insurance companies cluster among the smaller ratings firms, and that raises the risk of inflated assessments of creditworthiness.

Somebody has read all of this and acted on it. Hedge funds have built more than $5 billion of short positions against US life insurance stocks.

The regulators can now argue with the letter. On January 1 they gave themselves the power.

The insurance regulators' own analysts can now challenge a grade that sits three or more notches away from their own view. If the challenge sticks, the insurer holds more capital against the bond.

Fitch has already warned that carriers with thin buffers and heavy exposure could lose capital headroom over the next twelve to eighteen months. Nobody knows yet how often the challenge will be used.

Three more things worth keeping track of.

The shipyards. Average container freight rates fell 13% in 2025. Owners responded by ordering a record 4.8 million TEU of new capacity, and they kept going into this year. BIMCO put the containership orderbook at 37% of the existing fleet in the first quarter, and LNG carriers at 40%. Those ships arrive in 2027 and 2028 whether the cargo does or not. Scrapping has nearly stopped, so the old tonnage stays too.

The repossession lots. At buy-here-pay-here car dealers, the ones who sell you the car and finance it themselves, about 5% of loan balances were in active repossession in the third quarter of last year. At ordinary auto lenders the figure was under half a percent. Those borrowers are roughly two and a half times more likely to be late, and sixteen times more likely to have somebody coming for the car. Senator Warren opened a look at the industry in February.

The bid that switches off. Citadel Securities counts more than $1.1 trillion of announced buyback authorisations that moved back into open windows through August, and notes that two thirds of the biggest ones this year sit outside technology. Those windows start closing again around September 12 as companies enter pre-earnings blackout. September is also the weakest month of the year for retail buying on their platform, and retail buys down days at about half its usual pace. So the two most reliable buyers both step back at once. We'll see.

Nobody voted on this.

On February 15, 1936, the Office of the Comptroller of the Currency issued a rule about what America's national banks were allowed to buy. They could no longer purchase securities that were, in the words of the rule, distinctly and predominantly speculative.

Which raises an obvious question. Who decides what is speculative?

A footnote answered it. The terms could be found in recognised rating manuals. Where there was doubt, two manuals had to agree.

There were four manuals. Moody's, Poor's, Standard and Fitch. Four private publishing businesses that sold books of opinions to subscribers.

That morning their opinions became federal law.

Washington never defined what safe meant. It named four publishers and let them define it.

The phrase investment grade was born that day. It had no meaning before it.

And it spread. State insurance regulators copied the approach. Pension regulators copied it. In 1975 the SEC gave certain firms an official seal and wrote them into the capital rule for broker-dealers. In 1989 the ban reached savings institutions, which had to sell their low-rated bonds, and the junk bond market fell hard that year.

Now here is the part that should stop you. Somebody checked the record before 1936, after the rule was written.

An economist named Melchior Palyi went back to the railroad bonds that defaulted in 1924. He found that 70% of them had carried an investment grade from Moody's. He published it in January 1938.

Two years too late. The footnote was already law, and the country was already building on top of it.

Ninety years on, that rule still applies to banks.

We never did learn to define safe. We just kept asking somebody else. We'll see.