The recent financial market turbulence has brought credit ratings agencies under fire. Finance ministers from Britain, France, Germany and Italy met last Thursday to discuss the financial turmoil and strengthening government regulation. There are indeed problems with the agencies, but many suggested policy remedies are equally problematic.
Ratings agencies exist to deal with principal-agent problems and asymmetric information.1 Company managers or sovereign finance ministers may seek to mislead investors. Requiring a minimum rating can limit the risk for asset owners and guarantors if an asset manager would otherwise invest principals’ funds in high-risk assets. The agencies help investors overcome their lack of information about the variables that will determine whether a borrower will service debt. The agencies are ‘gatekeepers’, like auditors, investment analysts, and journalists. But they are more profitable and have higher price to earnings ratios and more acute conflicts of interest.
There are so few ratings agencies2 partly because of network effects, insofar as investors want consistency of ratings across issuers. But the natural monopoly characteristics are enhanced by the dependence of regulators on ratings, as (for example) formalised in the National Recognized Statistical Rating Organizations status created in the US in the mid-1970s and in the Basle 2 regulations. A wide range of investors are required not to hold securities whose ratings are below ‘investment grade’, and ratings affect the risk weightings of banks’ assets in calculating capital adequacy ratios. The regulators determine whether an agency holds this ‘regulatory license’, and this is a barrier to entry.3
There are several problems associated with the agencies, in terms of both the incentives they face and the performance of their roles. There is an apparent conflict of interest most acute with structured finance instruments, insofar as an agency may first advise on how the construction of a security would affect its rating and then issue a rating that confirms its advice – earning two separate fees in the process. This problem has grown hugely: 44% of Moody’s revenues in 2006 came from its structured finance activities. Moreover, there is an apparent incentive (in)compatibility issue: the issuer pays for the rating and may shop around for the best deal (a favourable rating), while the agency may be inclined to reward an issuer that chooses it over the other agencies.
The agencies’ performance is also problematic. They are blamed for reacting ex post rather than anticipating – the ratings are lagging indicators. Ratings changes may be procyclical (an effect that might be accentuated by Basle 2) and may create herd effects, magnifying instability. Both were strong criticisms during and after the Asian crisis, and Fitch, for example, accepted their validity in their mea culpa of February 1998.4
The agencies’ data and their models are suspect. In rating residential mortgage-backed securities involving subprime mortgages, for example, the agencies used data from an extended period of rapidly rising house prices, during which doubtful mortgages had been validated as householders’ equity grew. And rating complex structured finance instruments on the basis of model simulations may not be helpful when markets become disorderly, tail risk materializes, actual correlation risk far exceeds the models’ parameters, and the models blow up. Moreover, it may be inappropriate to use the same metric to evaluate sovereign risk, corporate bond risk, and complex instruments like collateralized debt obligations. In each context, the rating reflects the agency’s estimate of the probability of default over a given period – nothing more. It ignores, for example, the possibility that the market for the security may become illiquid; and it ignores the likely recovery rate if the security defaults.
Most importantly, a significant literature finds that the agencies simply do not add value – the quality of information they provide is often no better than that which a good analyst could extract from publicly available data. Detailed studies cast doubt on their ability to assess credit quality better than measures based on market spreads or to predict major changes.5
There are a number of identifiable sources of these problems. They suggest some directions for policy and some constraints on policy. There is a clear public good aspect of the information that the agencies provide. Hence there is a free rider problem, and payment by the user of the information will be either suboptimal or unenforceable. (Nevertheless, until the early 1970s, it was in fact the users of the ratings who paid, by subscription.) This aspect gives an efficiency argument for market concentration, which eliminates duplication of effort in generating information that will be available to all.
The agencies do not take full responsibility for their ratings. In fact, they have successfully (so far) maintained legal immunity from malfeasance claims on the ground that they are only financial journalists publishing their opinions, which are protected free speech. That Moody’s is much more profitable than the Financial Times or the Wall Street Journal may suggest, however, that they are in fact earning some rents. In addition to a return on their reputational capital, which is what they claim to sell,6 they are also selling the ‘regulatory license’ conferred by their roles in the regulatory regime. Doesn’t this status make their ‘speech’ rather different from that of a securities analyst or an FT columnist? One might also infer rents attributable to the regulatory license from the profitability of collateralized debt obligations. After all, these just repackage existing securities – the apparent source of ‘value’ is the rating gain.
Potential policy solutions
Academics and policy-makers have considered numerous proposals, from nationalising the agencies to abolishing official recognition of their ratings. The underlying incentive difficulties create a classic ‘mechanism design’ problem, but there is so far no formal analysis that could inform policy. And there are no easy answers.
Officials often support a voluntary code of conduct, since the market participants will lobby heavily against anything stronger. But the International Organization of Securities Commissions already promulgated a code of conduct at the end of 2004. According to the French Autorité des Marchés Financiers, by early 2007 it had been implemented in a ‘globally satisfactory’ manner. But the AMF still expressed concerns at that time about the roles and performance of the agencies in the structured finance markets – and rightly so, because the code has no teeth. Voluntary codes cannot solve the incentive problems.
Some have argued that public goods should have public funding. But there are obvious dangers in effectively nationalizing the agencies. A feasible alternative may be reviving subscription: a levy on users (investors).7 Some observers suggest that standardization of ratings across agencies would be helpful. If that just means using the same notation for a given probability of default, it is trivial. Anyone can convert a Moody’s rating into an S&P rating. If it means standardizing valuation models, it would eliminate competition – not entirely, because the agencies might feed different data into the same model, but one would then like to know why the data differ.
Regulators could require the agencies to provide more information than just a specific rating: an assessment of the liquidity characteristics of the instrument, of the likely volatility of its market price, etc. But the agencies do not seem well equipped for this – ‘As a result of unprecedented price volatility, Moody’s has adapted its methodology [for rating structured investment vehicles].’8 That does not say much for Moody’s data analysis: in fact, by all measures, volatilities during August 2007 were not significantly higher than in May 2006 (for example) and much lower than in autumn 1998. But the analytical problems here are formidable. The extensive academic literature on liquidity risk and market risk gives little guidance on how to estimate them quantitatively. And the underlying conditions change more rapidly than the fundamentals governing default risk, so the corresponding ratings would have to adjust frequently. That might confuse investors and add to market volatility.
The agencies should at least, however, provide a range for the risk of each instrument rather than a point estimate; or they should develop a distinct rating scale for structured finance products.9
Some propose introducing explicit legal liability for negligence or malfeasance. But this is likely to lead to the demise of the agencies – they would get sued out of business.
Separating rating from consultancy and advisory functions seems obviously desirable, and Chinese walls will not do. But forcing the agencies to give up the highly remunerative advisory work will be extremely difficult politically. Resistance might weaken if the structured finance business disappears, as some suggest it will – but then the problem disappears too.
There should be more competition among agencies: new entrants. Of course we all believe in competition, or at least market contestability, but as noted, there are aspects of the industry that suggest natural monopoly. And with more agencies, we might see a race to the bottom as issuers seek the agency that will rate them most favourably. Some observers report that investment banks shopped around for higher ratings in securitising subprime mortgages.
Could not the regulators substitute market valuations (spreads, say) for ratings? The agencies maintain that these are too volatile, but one could use a smoothed moving average. More important is that many securities effectively have no market – they are bought by ‘buy-and-hold’ investors. And many others are fairly illiquid – the average number of trades per day for a UK corporate bond is two (three for a euro-denominated bond).10 Credit default swap prices might deal with that problem, but that will not help in the primary market, which is where the impact of the ratings has been so pronounced in the recent period.
Some suggest eliminating the ‘regulatory license’ by abolishing recognition – that is, removing the NRSRO designation and merely requiring agencies to register with the regulators. This would confer no official status on the ratings. It would also vastly increase the burden on the regulators – but with increased budgets, they could hire people from the agencies (Moody’s has just announced layoffs). This proposal would also suppress the role of the ratings in Basle 2. After all the effort put into Basle 2, however, the regulators as well as the agencies have strong vested interests in it.
Perhaps recent experience will give enough support to the critics to override the political and lobbying obstacles to some of the more promising proposals. Without policy changes, the structural problems will surely persist.
1 See R. Levich, G. Majnoni and C. Reinhart (eds.), 2002, Ratings, Rating Agencies and the Global Financial System, Kluwer.
2 The big three (S&P, Moody’s, and Fitch) take an estimated 95% of the market (Variances 32, ENSAE, December 2007).
3 See F. Partnoy, 2006, ‘How and why credit rating agencies are not like other gatekeepers’, Legal Studies Research Paper No. 07-46, University of San Diego School of Law. Some argue that the agencies are now just a ‘whipping boy’ or scapegoat. Sophisticated investors, it is said, should have recognized the dangers in the new, complex financial instruments, and others should not have invested in them. But the regulatory license exists precisely because the regulators do not accept these arguments.
4 Financial Times, 13 February 1998.
5 There are several such papers in Levich et al. (2002).
6 It is argued that the recent fall in Moody’s share price indeed reflects a downgrading of its reputation. More likely, however, it is due to the collapse of the structured finance business.
7 A newly certified NRSRO, Egan-Jones, operates with a subscription model, giving subscribers immediate access to new ratings information and releasing it publicly with a lag (John Dizard, Financial Times, 14 January 2008).
8 Moody’s ‘Update’, 5 September 2007
10 B. Biais, et al., 2006, European Corporate Bond Markets: Transparency, Liquidity, Efficiency, CEPR.