The government's lawsuit against S&P has triggered speculation about why DOJ singled out just one agency and whether cases against the other two (Moody's and Fitch) will be forthcoming. One theory is that S&P was chosen because it downgraded US Treasury bonds in 2011. Two other options seem more likely: (1) the other two agencies may still be in settlement talks with DOJ, or (2) DOJ has a better case against S&P.
I was in Structured Finance at another rating firm in 2006 and 2007, and recall the headiness of the time. Revenues were exploding and half the money fell to the bottom line. Analysts were under pressure to keep up with the rapid flow of new securitization deals pouring in from Wall Street. Management had trouble hiring good people in the highly competitive environment. Meanwhile, everyone was aware that business could quickly be lost to competing rating agencies if investment banking clients were dissatisfied with the speed or nature of our conclusions.
In short, it was an environment that encouraged the cutting of corners - in terms of research quality, and, if the government allegations hold true, in terms of ethics - in fact, if the allegations are true, it seems S&P transcended the realm of ethical lapses and entered the land of outright fraud.
According to the complaint, S&P management instructed employees not to publish software and data updates that would have resulted in lower ratings. For example, pages 42-48 of the complaint detail how S&P management suppressed an update to the agency's LEVELS tool that relied on a much larger and more representative set of mortgages. (Recall the US Senate testimony of former S&P analyst Frank Raiter: “…S&P had developed better methods for determining default which did capture some of the variations among products that were to become evident at the advent of the crisis. It is my opinion that had these models been implemented we would have had an earlier warning about the performance of many of the new products that subsequently lead to such substantial losses. That, in turn, should have caused the loss estimates mentioned above to increase and could have thus caused some of these products to be withdrawn from the market as they would have been too expensive to put into bonds.”).
By cancelling a previously announced upgrade to LEVELS at the end of 2004, S&P was allegedly able to perpetuate the use of a flawed methodology which allowed investment banks to create deals with insufficient collateral subordinated to the senior AAA tranche. While cancelling this upgrade allowed S&P to remain competitive with Moody's and Fitch, it (allegedly) did a huge disservice to AAA investors such as the Western Federal Credit Union, on whose behalf the government filed its complaint.
According to the complaint, S&P management instructed employees not to publish software and data updates that would have resulted in lower ratings. For example, pages 42-48 of the complaint detail how S&P management suppressed an update to the agency's LEVELS tool that relied on a much larger and more representative set of mortgages. (Recall the US Senate testimony of former S&P analyst Frank Raiter: “…S&P had developed better methods for determining default which did capture some of the variations among products that were to become evident at the advent of the crisis. It is my opinion that had these models been implemented we would have had an earlier warning about the performance of many of the new products that subsequently lead to such substantial losses. That, in turn, should have caused the loss estimates mentioned above to increase and could have thus caused some of these products to be withdrawn from the market as they would have been too expensive to put into bonds.”).
By cancelling a previously announced upgrade to LEVELS at the end of 2004, S&P was allegedly able to perpetuate the use of a flawed methodology which allowed investment banks to create deals with insufficient collateral subordinated to the senior AAA tranche. While cancelling this upgrade allowed S&P to remain competitive with Moody's and Fitch, it (allegedly) did a huge disservice to AAA investors such as the Western Federal Credit Union, on whose behalf the government filed its complaint.
Naturally, S&P denies this allegation and it remains to be seen whether the government can prove its case. While the gory details of who knew what will undoubtedly fascinate, I hope that the debate around this lawsuit has room for a discussion about how to solve the fundamental rating agency problem. Rather than merely consider who is to blame, we should focus on how to change the institutional structure of the industry to incent more positive behavior.
First, why should we even care about the rating agency business enough to bother reforming it? After all, as depicted by Michael Lewis in The Big Short and in other financial crisis chronicles, rating agency employees are just a bunch of bottom feeders wearing J.C. Penney suits and sucking up to the investment bankers who might one day hire them.
Whatever we think of rating agency employees (I, for one, never shopped at Penney's), the inescapable fact is that their output shapes much of our financial conversation. Discussion around the US budget deficit and the Eurozone sovereign debt crisis often focuses on how rating agencies will respond to political measures. S&P's upgrade of California - raising it above Illinois in state bond rating purgatory - was a major local news story last week. Enron filed for bankruptcy because it lost its investment grade rating. And, of course, toxic assets poisoned the financial system in the years leading up to 2007 because of the high ratings they received.
Ratings are essential to the financial system because they help direct the flow of capital. By bucketing debt instruments into different risk categories, rating agencies help determine their interest rates. This function - if executed well - optimizes the use of society's savings and thus contributes to economic growth.
Given the importance of ratings, we need alternatives to the way they are now produced, i.e. by for profit companies with known conflicts of interest using proprietary data and analytics together with closed door rating committee meetings.
A much better alternative would be a system based on open source rating software, with fully transparent inputs and outputs, and no rating committee discretion. This fully open, fully deterministic approach controls biases regardless of whether the analysis is funded by investors, issuers, foundations or governments. It also allows a distributed peer review process to occur over the internet. An excellent case for open source ratings appeared recently on Naked Capitalism. PF2 has advanced this idea by supporting my Public Sector Credit Framework - a simulation tool for rating government bonds.
Given the importance of ratings, we need alternatives to the way they are now produced, i.e. by for profit companies with known conflicts of interest using proprietary data and analytics together with closed door rating committee meetings.
A much better alternative would be a system based on open source rating software, with fully transparent inputs and outputs, and no rating committee discretion. This fully open, fully deterministic approach controls biases regardless of whether the analysis is funded by investors, issuers, foundations or governments. It also allows a distributed peer review process to occur over the internet. An excellent case for open source ratings appeared recently on Naked Capitalism. PF2 has advanced this idea by supporting my Public Sector Credit Framework - a simulation tool for rating government bonds.
The first question I get when I propose such an approach is how are you going to make money? I have thoughts about that, but let me respond here with another question: why aren't more academics, pundits, politicians and regulators thinking of ways to make this operational model work?
Universities and foundations could fund rating transparency projects. The only such example right now is the National University of Singapore's Risk Management Institute. I have yet to find any foundation or academic willing to create such an institution in North America or Europe.
Easy to blame a bunch of greedy people at rating agencies for the financial crisis. Much harder to put the proper incentives in place, to do the heavy intellectual lifting needed to really fix the rating system.
3 comments:
Great article and couldn't agree more, Marc. The rating system/industry is badly flawed and the availability of good open-source analytics to provide an objective calculation/rating is a huge step in the right direction. Thanks
Frank Raiter is a good, honest person as I recall. Michael Lewis was wrong about this:
"rating agency employees are just a bunch of bottom feeders wearing J.C. Penney suits and sucking up to the investment bankers who might one day hire them."
I worked on the buy-side as a proprietary trader, for a non-US bank and for a hedge fund. It made me nervous, worried. So I did the next best thing, which was to work for S&P. I did not "suck up" to investment bankers. Ratings agency employees earn a fraction of what I-bank employees earn. It is, or was, prior to 2000, much better, intellectually challenging work.
Open source ratings software... you mention that no interest is expressed in funding it. I can think of many reasons, am not certain of any of them though. Personally, I don't believe that ratings with zero rating committee discretion would have much value. But there are far less worthwhile projects that have received all sorts of enthusiastic financial backing! If it were up to me, I'd much rather some of those funds go toward what you suggested instead. Even if it weren't successful in the long run, it would increase understanding, and could lead to something that might be an improvement over what we have now. Credit ratings are important. Without them, or without public access to them in some form, capital markets become more opaque.
Their need to be a different group of companies doing the ratings.
Post a Comment