The Subprime Mortgage Ratings Scandal: How Agencies Rated Fraud as Safe

In December 2006, two analysts at Standard & Poor’s typed a short exchange over the company’s instant-messaging system that would later be read aloud to the United States Congress. One wrote that a particular deal was “ridiculous”, that they “should not be rating it”. The other agreed, then added the line that came to sum up an era: “We rate every deal. It could be structured by cows and we would rate it.” A separate S&P message, surfaced by the Financial Crisis Inquiry Commission, was blunter still — an analyst joking that a deal could collapse and the agency would be “fine” because the fees had already been collected, closing with the observation that they hoped they would all be “retired by the time this house of cards falters.” The house of cards was the American mortgage market, and within two years it fell on the world. The rating agencies — Standard & Poor’s, Moody’s, and Fitch — had stamped hundreds of billions of dollars of mortgage-backed securities with the highest possible grade, AAA, the same seal carried by the sovereign debt of the most solid governments. When the loans underneath went bad, so did the ratings, and the reason was written down in the agencies’ own files.

The seal that let strangers trust each other

To see why this was catastrophic rather than merely embarrassing, you have to understand what a credit rating is, and the quiet, load-bearing job it does. Most investors cannot examine the thousands of individual home loans bundled into a mortgage-backed security. A pension fund in Norway buying a slice of American housing debt has no way to knock on doors in Nevada and check whether the borrowers can pay. So the market outsources that judgement to a small number of rating agencies, which analyse the underlying risk and assign a letter grade. AAA means, in effect, “as safe as anything gets — this will almost certainly pay you back.” Below that runs a ladder down through AA, A, BBB, and into the speculative “junk” grades.

That letter is not decoration. Regulations, pension rules, and the internal mandates of countless institutions are written in terms of it: certain funds are permitted to hold only highly rated paper. Banks hold less capital against assets rated AAA, because the rating declares them near risk-free. The grade is a passport that lets a security travel into portfolios all over the world, and it works only because everyone agrees to treat the agencies’ judgement as objective and disinterested. The whole architecture of modern finance rests on the assumption that when Moody’s says AAA, it means AAA. This is the same kind of trusted-intermediary machinery that failed at LIBOR — a number or a grade that the entire system leans on precisely because it is supposed to be above the fray.

The flaw baked into who pays

Here is the structural rot, and it predates the crisis by decades. The agencies do not get paid by the investors who rely on their ratings. They get paid by the issuers whose securities they rate — the banks assembling and selling the mortgage bonds. This is called the “issuer-pays” model, and it contains an obvious conflict: the customer wants a high rating, and the customer holds the chequebook. If Moody’s is too tough on a Goldman Sachs deal, Goldman can take its next deal — and its fees — to S&P or Fitch. The agencies competed for business, and one thing they competed on, inevitably, was the willingness to say yes.

In the ordinary corporate-bond world this conflict had been managed, more or less, for a long time. What broke it was the explosion of structured finance in the 2000s. Banks discovered they could take pools of risky subprime mortgages — loans to borrowers with poor credit, often with little documentation of income — and slice them into layers, or “tranches”. The lowest tranches absorbed the first losses; the highest were shielded, in theory, because they would only take a hit after everything below them was wiped out. Through this financial alchemy, a pool of loans that were individually junk could be reassembled so that the top tranches qualified for AAA. Then those tranches could themselves be pooled and re-sliced into collateralised debt obligations, and re-rated AAA again, a layering that grew so intricate that even the people building it struggled to see through to the loans at the bottom.

The agencies rated all of it, and the fees were enormous. Structured-finance ratings became a huge share of their revenue in the boom years. The models the agencies used to bless these securities rested on an assumption that quietly did the whole job: that house prices across different American regions would not all fall at once, so a pool of mortgages from Florida and California and Ohio was diversified and safe. That assumption had held for as long as anyone’s data went back. It was about to fail comprehensively.

The fork: greed, blindness, or a rigged game

This is where the documented scandal and the popular telling diverge, and the fork is subtle because both versions are partly right. The proven facts are severe. The Financial Crisis Inquiry Commission, reporting in 2011, concluded that the rating agencies were “essential cogs in the wheel of financial destruction” and “key enablers of the financial meltdown”. Internal emails showed analysts privately doubting the very deals they were rubber-stamping. Congressional hearings in 2008 aired the “structured by cows” message and others like it. In 2015, Standard & Poor’s agreed to pay about $1.5 billion to settle U.S. federal and state claims that it had knowingly inflated ratings; Moody’s reached a settlement of roughly $864 million in 2017. The conflict of interest was real, the pressure to please issuers was real, and the private doubts of the analysts are on the record.

The fork comes in how deliberate people imagine it was. The satisfying story — the one that fits the “structured by cows” quote so neatly — is that the agencies knew the bonds were worthless and knowingly lied for cash: a straightforward fraud, cynics stamping garbage while laughing about it. There is real evidence for cynicism at the individual level, and the settlements acknowledged misconduct. But the fuller record, including the agencies’ spirited legal defence, points to something murkier and in some ways more troubling: a great many of the people doing the rating genuinely believed their models. They had built a system whose core assumption — that a nationwide house-price crash could not happen — was wrong, and they had every incentive not to question that assumption, because questioning it would have shut down the most profitable business they had ever had. Motivated reasoning is not the same as conscious fraud, and the line between them, inside a room full of well-paid analysts using a respectable-looking model, is genuinely hard to draw.

That distinction matters, because it is the same one that runs through Bernie Madoff’s regulators: the difference between a watchdog that is bought and a watchdog that has convinced itself, for reasons of comfort and income, not to look too hard. The popular imagination reaches for the bought version because deliberate villainy is legible. The documented version is a fog of self-deception, incentive, and models that everyone trusted because everyone else trusted them.

When the seal peeled off

The reckoning was swift and merciless once house prices turned in 2007. As subprime borrowers began to default in numbers the models had ruled out, the losses ate upward through the tranches, and securities that had been rated AAA — as safe as government debt — were downgraded, sometimes many notches at once, sometimes to junk. In 2007 and 2008 the agencies carried out mass downgrades of mortgage-linked securities on a scale never seen before. Overnight, “safe” assets held by banks, pension funds, insurers, and money-market funds around the world were revealed as impaired, and nobody knew who held how much of the poison. That uncertainty — the sudden collapse of trust in the rating itself — was a large part of what froze the financial system in 2008. When you can no longer believe the letter grade, you can no longer value anything, and a market that cannot value things stops.

The particular cruelty was borne far from Wall Street. Because the AAA passport let these securities travel anywhere, the losses landed on institutions that had done nothing but obey the rules — a German regional bank, an Australian council’s pension fund, a Norwegian municipality — bodies that had bought the paper precisely because it carried the top grade and their mandates required safety. They had trusted the seal, exactly as the system asked them to, and the seal was hollow.

What the “house of cards” email was really about

The analyst who wrote about being retired by the time the house of cards fell was not a monster. He was a person who could see, clearly and privately, that the thing he was being paid to bless was rotten — and who kept blessing it anyway, because the alternative was to be the one analyst at the one agency who said no while the fees, and the deals, and quite possibly his job, went elsewhere. That is the real horror of the ratings scandal, and it is quieter and more human than fraud. It is the ordinary tragedy of a person who knows better, trapped inside a structure that punishes knowing better.

The subprime ratings affair is the case that vindicates the diffuse, hard-to-prove suspicion that “the game is rigged for insiders” more thoroughly than almost any other, and it does so without a single grand conspirator. There was no cabal deciding to poison the world’s pension funds. There was a conflict of interest baked into who signs the cheque, a model everyone found convenient to believe, and thousands of individually rational decisions to not be the one who spoke up. The system did not need a villain. It needed only enough people willing to hope they would be retired before the reckoning came.

The reforms that followed — the Dodd-Frank Act’s provisions on rating agencies, new disclosure rules, attempts to blunt the issuer-pays conflict — chipped at the edges of the problem without resolving its core, because the core is a conflict nobody has found a clean way to remove: someone has to pay the referee, and whoever pays has a stake in the call. What the crisis proved was something more unsettling than worthless ratings. It proved that a seal of safety is only ever as honest as the incentives of the people who stamp it, and that a whole world can come to depend on a stamp without ever asking who was paying for the ink.