When investment banks insisted that their mathematical models more accurately valued securities than market prices, they were subjected to withering scorn from regulators, investors and pundits.
But in the Department of Justice's lawsuit against Standard & Poor's the complaint is just the opposite: S&P stands accused of paying too close attention to market signals and not being faithful enough to its models.
"S&P was so concerned with the possibility of losing market share and profits that it limited, adjusted and delayed updates to the ratings criteria and analytical models it used to assess the credit risks posed by RMBS and CDOs. According to the complaint, S&P weakened those criteria and models from what S&P's own analysts believed was necessary to make them more accurate," the Justice Department says in its press release about the law suit.
It's not clear why this should be considered a description of fraud. Imagine if instead of rating bonds S&P were in the business of making breakfast cereal.
"S&P was so concerned with the possibility of losing market share and profits that it limited, adjusted and delayed updates to the recipe for corn flakes. According to the complaint, S&P changed the recipe for its cornflakes and models from what S&P's own analysts believed was necessary to make them tastier."
Doesn't sound like fraud anymore, does it? Instead it sounds like a company doing what companies often do: listening to the market instead of their internal experts.
Faced with a declining market share, S&P naturally would have asked itself the question: why do investment banks prefer to have our competitors — mainly Moody's — rate their bonds? There are three plausible answers to this question.
1.The investment banks are just making the wrong choice because either (a) they don't understand that our ratings are more accurate or (b) they are more interested in high ratings than accurate ratings.
2.The investment banks prefer our competitor because their own customers prefer our competitor. Those customers are making the wrong choice, for either reason (a) or (b) above.
3.We are offering a lower quality product.
The government claims that S&P internally concluded that the correct answer to the question of declining market share was either 1 or 2. The evidence supporting this claim is contained in emails and instant messages from people at S&P who did believe that S&P's criteria for ratings were superior.
The material offered up by Justice can look pretty damning if you don't think while you look. An analyst sings a song about mortgages to the tune of the Talking Heads' "Burning Down the House." An executive complains that the company is giving up the "search for truth." One guy complains that S&P would rate a cow.
Most of these people aren't named in the complaint. But from what scant evidence we have, it seems that the biggest internal critics of departures from S&P criteria were the people who had developed that criteria. Not the sort who would be inclined to suspect that the correct answer to the loss of market share would be a problem with S&P's ratings.
Others at the company, however, did not agree. If you do not think that the customer base you're targeting is composed of schmucks (who didn't know that S&P was more accurate) or crooks (who didn't care about ratings accuracy), then you can easily read declining market share as a sign that you are doing something wrong. Your product needs to be improved. In fact, this is usually exactly what a decline in market share means.
So even if S&P changed the "criteria and models from what S&P's own analysts believed was necessary to make them more accurate," it may well have understood itself to be adjusting to what the market believed was more accurate. There's no rule that says you have to trust your analysts over the market. That's actually a pretty bad idea. It's what got banks in so much trouble in 2007-2008.
The government case turns on the unstated claim that everyone in charge at S&P knew their customers were all schmucks or crooks. But the complaint offers no evidence that the customers actually were schmucks or crooks. Nor does it explain how it will show that the executives at S&P believed the customers were schmucks or crooks.
Changing the models to appease market appetite isn't fraudulent — even when you do it against the advice of the smart guys you hired to build the models in the first place.
Follow me on Twitter @Carney