Credit rating agencies are back in the public eye as the Senate Permanent Subcommittee on Investigations releases another trove of embarrassing agency e-mails and the financial reform bill nears enactment. In this context, a proposal by two professors at NYU's Stern School of Management, Lawrence White and Matthew Richardson, to improve credit rating agency performance by having the SEC decide which agencies will rate each instrument has attracted favorable attention from commentators such as Paul Krugman. Although the proposal is refreshing in its boldness and offers a potentially useful way to address one of the many problems besetting the agencies, it should not be understood as a complete solution. The complementary issue of rating-agency accountability for poor quality should also be addressed.
By now, the background story is familiar: Rating agencies - Moody's, Standard & Poor's, and similar firms whose business it is to assess the likelihood that debt obligations will be paid as agreed - gave their stamp of approval to innovative financial products that were in fact incomprehensible and/or based on the premise of an unending real-estate boom. Panic set in when everyone realized that the ratings were wrong or unsupported. The details and even the basic correctness of this narrative are disputed, but the major rating agencies have all more or less conceded that there was a problem of some kind: their ratings on novel financial products didn't do as well as they could have.
What exactly are the problems with the rating agency market? There are several candidates:
(1) Lack of competition. The SEC says that the three largest agencies have over 97% of the market, as measured by number of ratings outstanding.
(2) Absence of transparency. Market participants complain that it is hard for users to know what the agencies do and how well their ratings perform.
(3) Financial regulators' reliance on credit ratings in their rules. If the rules say that regulated firms like banks and insurance companies have to own financial instruments with credit ratings, then there will be a demand for credit ratings, even if the agencies have no idea what they are doing.
(4) The "issuer pays" business model. The firms selling the financial products usually are the ones who pay for the ratings. "Issuer pays" poses an obvious conflict of interest, as rating agencies have an incentive to please their customers, who are the people selling the products, not those buying them.
(5) Absence of accountability. There is no clear way to hold rating agencies liable for poor performance unless it rises to the level of fraud.
Congress' and the SEC's actions to date have focused mainly on the first two issues, competition and transparency: Legislation passed in 2006 and rules adopted since then have focused on trying to get more rating agencies into the market and on increasing disclosure about what the agencies are doing, what data they're using, and how they're performing. There has been fitful action on the third issue. Starting in 2008, the SEC and other regulators have considered reducing their use of credit ratings in their rules, but they have not eliminated their reliance on credit ratings and do not seem to be on track to do so, although deliberations are ongoing. The problem here is that financial regulators need a measure of credit risk, and it is not clear what would take the place of credit rating agencies, an issue I took up in this article last year.
The White and Richardson proposal focuses on the fourth issue. The idea is to remove issuers' ability to shop for high ratings from agreeable rating agencies by using the SEC to assign the agency that will rate each debt instrument. Under the proposal, the issuers still pay for the rating, but they pay the agency that the SEC selects to do the rating. The SEC will make its selection based on its assessment of which agency is likely to do the best job, thus eliminating the conflict of interest that arises from the issuer-pays business model. This is an innovative idea, and its boldness is refreshing given the limited scope of the reforms that have even been considered to date. White and Richardson would fundamentally restructure the rating market - indeed, they apparently would eliminate the rating "market" and substitute SEC assignment.
Of course, those who think that government can't do anything right will oppose this idea, arguing that the regulators are corruptible, capturable, and/or unskilled at evaluating the performance of rating agencies. Those who think the SEC in particular is the wrong choice - pointing perhaps to the Madoff affair, to the SEC's oversight of the Wall Street investment banks leading up to the financial crisis, or to allegations that the SEC has been biased toward the large incumbent rating agencies - will oppose the choice of this particular regulator. Those who think that government approval of rating agencies led to excessive reliance on them will observe that the proposal could exacerbate that problem. I'll note all three sets of objections and set them to the side for the moment.
I have two different concerns about this proposal. First, it seems overbroad to prohibit agencies from expressing their opinions unless authorized to do so by the SEC. If the SEC selects Moody's to rate a bond, should S&P really be barred from opening its mouth about that bond? Even setting to one side the agencies' more extravagant First Amendment claims, this seems problematic. The overbreadth concern could be addressed without changing the essence of the proposal by saying either that the SEC-selected agency is the only one whose ratings "count" for regulatory purposes or that only the SEC-selected can be paid by an issuer, leaving other agencies free to express their opinions in other contexts.
Second, the White and Richardson proposal doesn't address what I see as a central problem: When confronted with a large and growing market for a set of novel products, agencies that are paid by the rating (or otherwise based on the volume of their business) have a financial incentive to issue ratings on those products even if they don't know what they are doing. After all, the more ratings, the more revenue - even if the technical complexity or novelty of the product means that the agency can't do a good job. Indeed, White & Richardson acknowledge this issue, stating that "it's surprising that rating agencies would even attempt to rate" certain types of novel products because of the technical difficulty of doing so. But under their proposal, agencies apparently still are paid by the rating even though they are selected by the SEC: The more ratings, the more revenue. That enticement to poor quality still exists even though the issuer-pays problem may be eliminated.
This problem can be addressed by taking on the fifth issue, rating agency accountability. Eventually, poor-quality ratings will be discovered and if the agencies know they will have to give up their profits from the poor-quality ratings in that event, that reduces their incentive to issue ratings when they don't know what they are doing. An article I wrote two years ago addressing this point can be found here.
The Dodd bill takes steps in the direction of accountability by clarifying that a rating agency can commit fraud by failing to conduct a reasonable investigation of facts upon which it relies and by empowering the SEC to decertify rating agencies that consistently produce poor-quality ratings. Neither of these provisions really addresses agencies' temptation to issue ratings when they don't know what they're doing. Decertification is a weak remedy, as credit rating agencies can operate without being certified, and the requirement to conduct a factual investigation doesn't go to the fundamental question, which is whether the agency knows what to do with the facts that it has.
The Dodd bill does provide for further study of agencies' incentives to produce high-quality ratings, so even if neither the White and Richardson proposal nor a stronger rating-agency accountability provision makes it into the financial reform bill that seems likely to be passed soon, there is some chance that the complementary issues of issuer-pays and accountability will be addressed.