In an oligopoly there's nothing worse than collusion -- at least that's what economists tell us. If an industry consisting of only a few firms are able to strategize together, they can do things that harm the market, like set consumer prices and drive down their supplier costs. But in the case of the credit ratings industry, that logic might not work quite as well. If they had been able to talk about the risks that were present in new mortgages being originated, might they have been more willing to act on ratings earlier?

The basis for this theory is last week's Financial Crisis Inquiry Commission testimony by D. Keith Johnson, a former president of Clayton Holdings, a firm that performed analyses on pools of mortgages. Gretchen Morgenson of the New York Times reports that after his firm found problems with many of the mortgages it looked at, he informed the agencies. But they didn't listen:

"We went to the ratings agencies and said, 'Wouldn't this information be great for you to have as you assign tranche levels of risk?' " Mr. Johnson testified last week. But none of the agencies took him up on his offer, he said, indicating that it was against their business interests to be too critical of Wall Street.

"If any one of them would have adopted it," he testified, "they would have lost market share."

The condemnation here is usually focused on the agencies. After all, they were told that these mortgages were bad, yet they continued to assign inflated ratings on the securities that they were used to produce. But perhaps this misses the point.

On some level, you can't entirely blame the agencies for being scared to challenge the investment banks and issuers. After all, as Johnson says, in their market, if one of these firms were to become more conservative, and the others refused to do so, then those others would quickly scoop up the business that the more conservative firm would lose.

This problem closely resembles the classic game theory exercise called a Prisoner's Dilemma. In the game, there are two prisoners who have committed a crime. If both stay silent, they each serve a short sentence. If one betrays the other, then the first will be released, and the second will serve a very long time. If they betray each other, then they will both serve a medium sentence. In a matrix, it looks like this:

prisoner's dilemma.png

Through some simple game theory analysis, it's pretty easy to conclude that the most rational behavior of each prisoner is to squeal, no matter the behavior of the other -- since the assumption is that they cannot collude. Think about it: if X is silent, then Y is better off betraying X (zero years versus one). And if X betrays Y, then Y is still better off betraying X (three years versus six). Y has the same analysis. So even though both would ultimately be better off staying silent, they will betray each other.

Now think about the ratings industry. Although additional participants make the game a little more complex, the analysis is similar. Let's re-invent the above matrix, simplify it to two firms, and call it the Rater's Dilemma:

rater's dilemma v2.png

These numbers are made up, but the qualitative differences seem sensible. The idea is to imagine the extreme situation in the long-term. If both revise their ratings, then they will be more credible ultimately and both flourish. If both look away, and ignore the warnings of analysts like Clayton, then they will have fewer long-term profits due to market problems and a loss of credibility. If one looks away while the other revises its ratings, then whichever looks away will satisfy Wall Street and loan issuers and get even more business, while the other will lose all of its deals.

Of course, in the Prisoner's Dilemma, the best solution would be for the two prisoners to be able to collude and trust one another that they will both stay silent. Perhaps the same thing would have worked for raters. If Moody's, S&P, and Fitch sat in a room in 2006 and talked about the grave warnings they were all hearing, perhaps they could have all agreed to heed the warnings and be more selective about mortgage collateral, without each having to worry about the other firms looking away. After all, if the raters had all agreed to take a tougher stance with banks and issuers, investment banks and issuers would have had to grudgingly accept it, since ratings are so important to investors.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.