Credit Rating Agencies and 2008

April, two thousand seven. Two analysts at Standard and Poor's are messaging each other about a deal sitting on their screens. One types... that deal is ridiculous, we should not be rating it. The other writes back... I know, the model does not capture half of the risk.

Wait. Those are the people GRADING it?

Those are the people grading it. And then comes the line that ends up in a federal lawsuit. She writes... we rate every deal. It could be structured by cows and we would rate it.

Cows. Okay. Then here's the thing I've genuinely never understood. How do you take a pile of loans to people who might not pay... and turn it into the safest grade that exists?

That's the magic trick. And the uncomfortable part is, it isn't fraud. It's maths. You don't rate the loans. You rate a pool of them, then you slice the cash coming out of that pool into layers. Bottom layer eats the first losses. Next one up only gets hurt once the bottom is wiped out. By the time you reach the top slice, a frightening number of borrowers have to fail before a single dollar touches it.

So the top layer is safe because it's got human shields underneath it.

Crudely, yes. And the whole thing rests on one assumption. That the borrowers fail independently. A guy loses his job in Ohio, that tells you nothing about a family in Nevada.

And if that's wrong?

Then the human shields all fall over at the same moment, and your triple-A is a subprime loan in a very good suit.

Hang on though. Three serious companies, decades of reputation. Why would they lowball that? What's in it for them?

Follow the invoice. You'd assume investors pay for ratings, because investors want the truth. They don't. The issuer pays. The bank building the deal writes the cheque to the agency that grades it.

Oh, that's fine. I'll just pay the examiner directly and we'll see how my test goes.

And you can shop. If Moody's wants more cushion under your top tranche, you walk down the street to S and P. Jerome Fons, a former chief economist in Moody's mortgage-backed group, told Congress in two thousand eight that a large part of the blame sat with the conflicts in issuer-pays, and with rating shopping. The drive to keep market share, he said, made the agencies willing participants.

That's not a critic. That's one of their own people.

Saying it at a public hearing. And another internal document read out that day... rating agencies continue to create an ever bigger monster, the CDO market. Let's hope we are all wealthy and retired by the time this house of cards falters.

They knew.

Some of them clearly suspected. Then summer two thousand seven, the downgrades come in waves, everyone dumps the paper at once. The Financial Crisis Inquiry Commission's verdict, four years later, is worth hearing exactly. The failures of the credit rating agencies were... essential cogs in the wheel of financial destruction. Those mortgage securities, they said, could not have been marketed and sold without that seal of approval.

So what's the punishment for being an essential cog?

Money, eventually. Twenty fifteen, S and P's parent settles with the Justice Department and the states for one point three seven five billion dollars, plus a hundred and twenty-five million to the California pension fund. January twenty seventeen, Moody's pays nearly eight hundred and sixty-four million.

Did anyone actually admit anything?

Moody's acknowledged it hadn't followed its own standards, and agreed to keep analysts out of commercial conversations, with the chief executive certifying compliance for at least five years. The S and P settlement contained no findings of violations of law. Both firms are still here. Still enormous.

Fine. But the payment model. Please tell me they fixed the actual mechanism.

They did not. New rules, new disclosure, a dedicated ratings office at the SEC. Issuer-pays survived almost untouched, because nobody could agree on what would replace it.

So the cows are still in the building.

And the slicing machine is running at full speed on something new. CLOs. Pools of corporate loans instead of mortgages. Increasingly, loans to mid-sized private companies that no public market ever prices.

Wait, wait. If nobody prices them, where does the default assumption even COME from?

From the agency itself. It's called a credit estimate. And last year Moody's reported that among smaller borrowers in middle-market CLOs, the share with weak liquidity had jumped to about thirty-one percent, from around twenty. Leverage climbed to six and a half times earnings, from five and a half.

More debt, less cash. That's the bad quadrant.

And here's the bit that made me sit up. Researchers looking at the CLO market found that after the crisis, the most complex deals were MORE likely to carry just one rating instead of two.

...which is the fingerprint of shopping.

You collect the opinions. You publish the one you like.

Same move. Different asset, exactly the same move.

The agencies would say the models are tougher now, the data's better, triple-A CLO tranches came through the last stress tests intact. That's a fair defence, and I want to be straight about it. But the disagreement between agencies on these private-credit pools is real, and it's live right now.

So what do I do with that? I don't buy CLOs.

Your pension fund might. Your insurer almost certainly does, because regulators let them hold less capital against highly rated paper. So a letter, produced by a company paid by the people being graded, is quietly setting how much risk sits behind your retirement.

And triple-A doesn't mean safe. It means somebody's model said so.

It means an opinion with very good typography. In two thousand seven that opinion was wrong at a scale that took the world economy with it. The question now isn't whether the agencies are honest. It's whether anyone outside the building is checking the maths... or whether we're all just hoping to be wealthy and retired before the cows come home.