The role of credit rating agencies

The subprime crisis has been called an "accident waiting to happen" by many market observers during the past few months. More than a year before the turmoil broke out, experts had warned of likely negative developments on the US real estate market, and as early as the fourth quarter of 2006 loan losses in the subprime segment were seen to be rising rapidly.

Despite this, CRAs were very slow to react to the flaws in structured finance products market that were beginning to appear. Only between 10th and 12th July 2007 did the big CRAs downgrade a few hundred securities, doing so by several notches at once. While this was not the only reason for the turmoil on the financial markets, it was one of the triggers and a clear sign of a crisis in the marketplace - with all the resulting consequences.

At the same time, this illustrates the enormous importance of ratings - and of the conduct of CRAs - for the securitisation market. It makes the errors of judgment that all the major CRAs were guilty of when rating structured finance products doubly serious. As a result, CRAs have rightly come under fire.

No flourishing securitisation market without ratings

In the corporate bond sector investors have relatively detailed information on the issuers at their disposal, so they do not have to rely solely on the reports provided by CRAs. This is what makes bond ratings so different from structured finance product ratings, where investors do not usually have access to the required information themselves. Apart from that, the cost of a full analysis of securitised asset portfolios is so high that business would not be worthwhile for either the issuer or the investor without a division of labour cutting transaction costs. An assessment of the quality of such portfolios and the individual tranches thereof by impartial third parties recognised as highly reputable is seen as the solution. CRAs have undertaken this job.

These special conditions on the market for structured finance product ratings call for special care and transparency from CRAs. While it is true that the ratings of structured finance products are normally relatively stable because of the diversified pools of assets, this finding needs to be put into proper perspective: If there are changes in ratings, e.g. following changes in the assumed default probability for asset classes, these are usually relatively large.

Another point is that - also unlike bonds - credit rating actions on structured financial products can scarcely be foreseen by market participants. This explains in part why the large-scale downgrades last year produced a general distrust of structured finance and market players with large exposures in such products. At the same time, it possibly also illuminates an important reason why CRAs did not react to the changed situation by taking rating actions until relatively late (1).

Rating process needs to be optimised

Looking back at market developments and rating activities during the past few months, it becomes clear that a number of factors have adversely affected the quality of the work done by CRAs:

The ratings of structured finance products are based, because of product structuring, on mathematical models whose results depend heavily on assumptions regarding default probabilities for portfolio components and their correlations. The assumptions regarding the risk of default are based on historical performance. In the case of subprime loans and other complex products the historical data available were very limited, however, and not sufficient to reliably calculate correlations. In addition, real estate prices in the United States rose continuously until 2005. So no experience or information had been gathered on how ABS portfolios behave in difficult times. Moreover, it was often wrongly assumed that a pool of similar loans from different parts of the US could already be regarded as sufficiently diversified.

Caution would also have been wise as the market indicators were sending contradictory signals: the growth of the subprime sector continued to accelerate even after interest rates started to climb and the price surge in the real estate market was already flagging. This shows that macroeconomic determinants or market-related events ought to have been taken into account much more than was actually the case.

On top of this, ratings were not monitored properly. The time span between the end of the real estate boom in the US at the end of 2005 and the rating actions in the summer of 2007 was very long, even after allowing for the length of CRAs' decision-making processes.

CRAs also rate issues by conduits and SIVs. They should have seen the risks arising from intensive maturity transformation in the case of vehicles whose assets were made up of securities of steadily deteriorating quality.

In the face of the strong expansion of the securitisation market, the high staff turnover at CRAs may well have led to personnel shortages. At the same time, however, there was a strong incentive to deploy personnel mainly in the "new business" segment, as the fees for first-time ratings far exceed those for monitoring existing ratings.

Finally, CRAs disclosure practice was inadequate. So that investors can take decisions on their own, they need sufficient information about the main characteristics of ratings and the models and model assumptions underpinning these. While CRAs did disclose information, some important aspects remained unclear (2).


The picture of the ratings dilemma would be inaccurate if it were only to spotlight shortcomings on the part of CRAs. It is undoubtedly just as true that investors often accepted ratings uncritically and overestimated their meaningfulness. Not enough attention was paid to the fact that ratings are only estimates of the relative probability of default or expected loss on a debt instrument. They are not a detailed assessment of risk and they say nothing about the price quality or liquidity of an instrument. Ratings are no substitute for investment risk management, particularly as the information provided by CRAs is limited.

Even if the need for corrective action in the CRA sector is obvious, it would not be fair to put the blame for the financial market turmoil solely or mainly on this - or to jettison the existing system as a whole. Yet it is also true that CRAs' prominent position in the structured finance products market imposes on them an obligation to help ensure the long-term viability of this market. This involves a kind of "early warning" function - a function that CRAs have failed to perform properly to date.

Regaining confidence is a must

To make the securitisation market fully functional again, CRAs need to restore their dented reputation. This calls, in particular, for modifications to rating procedures, improved transparency and a review of internal processes to ensure a high rating quality. A whole raft of conceivable measures is currently under discussion. To what extent they are each likely to improve the situation is a matter for debate, however.

What appears to be indisputable is the need for rating procedures to make more allowance for macroeconomic determinants and developments on the markets affecting rated portfolios, e.g. real estate, credit card or automobile markets. That applies equally to first-time ratings and monitoring of existing ratings. In this way, pure - and inadequate - monitoring of default performance would be accompanied by a kind of economic plausibility check.

In connection with improved transparency, an alternative rating scale for structured finance products or adding scales to the existing rating scales are widely seen as a possible solution. This idea is supported particularly by central banks and supervisors. The recently revised IOSCO Code of Conduct therefore calls for a distinction to be made in future between ratings of structured finance products on the one hand and traditional bonds on the other hand.

In contrast, market participants are taking more of a cautious approach. Such a change needs to be carefully weighed up. Naturally the product categories mentioned display different risk characteristics. At the same time, ratings are particularly important for financial market participants precisely because they allow a comparison of different types of securities in terms of their probability of default. This advantage should not be surrendered by introducing complex new rating information.

Summarily identifying certain risk aspects by means of additional rating tags would also be a solution, though this raises a number of questions at the same time: That goes firstly for the scope of this new "rating dimension". If structured finance products are treated separately, they need to be defined precisely and uniformly, since this term has been used differently in the marketplace so far. In particular, a clear-cut distinction between structured finance products and covered bonds, e.g. Pfandbriefe, is necessary.

What would not be a solution - but, on the contrary, actually a step towards more complexity and considerable confusion - is if every CRA were to apply its own definitions. What is ultimately at stake here is more transparency with regard to the risk structure and stability of ratings. No matter whether CRAs use additional identifiers or their own rating scales, they should provide information on the assumptions, default probabilities and correlations between individual components underpinning asset portfolio ratings, on stress test scenarios and results, on anticipated rating volatility and on rating history.

Finally, operations and processes within CRAs need to be improved. Personnel resources, employee know-how and staff organisation at all levels of the rating process must meet market requirements. Moreover, the objectivity of rating procedures should be beyond doubt. Conflicts of interest must therefore be avoided in future.

Regulation of CRAs

Present system: The subprime crisis more or less inevitably raises another question: the content, type and rigidity of possible CRA regulation. Given the oligopolistic structure of the ratings market, the regulatory set-up currently in place is based mainly on self-regulation within certain supervisory "crash barriers" in the form of the Fundamentals of a Code of Conduct for Credit Rating Agencies issued by the International Organization of Securities Commissions (IOSCO).

These fundamentals provide the foundation for individual CRA codes which are ultimately designed to prevent conflicts of interest, ensure the integrity and independence of ratings and avert problems between CRAs as well as issuers and investors. CRAs have to disclose how far they have adopted the IOSCO fundamentals and give reasons for any deviations from these ("comply or explain" principle). Their conduct is monitored by supervisors.

All CRAs operating in the structured finance products market are registered in Europe as External Credit Assessment Institutions (ECAIs) and, as such, additionally vetted by a banking supervisory authority. In the United States, CRAs are officially registered as Nationally Recognized Statistical Rating Organizations (NRSROs) if they satisfy the requirements set by the Securities and Exchange Commission (SEC) and are subject to comprehensive vetting. The SEC requirements incorporate many elements of the IOSCO Code of Conduct.

Reform plans: There were - and are - good reasons for this regulatory approach: CRAs analyse companies, issues and finance products and estimate the risk of default on debt instruments. This business eludes tight regulation to some extent, as it is mainly about subjectively assessing the information available.

Practice has, however, shown that CRAs should be set more rules than hitherto. The rating process indisputably contains incentive structures that can - theoretically at any rate - provoke misconduct by CRAs. Because of the competitive situation on the ratings market, the pressure that clients can potentially exert is, on its own, not enough to force CRAs to behave properly. In addition, public authorities are increasingly using ratings for regulatory purposes. Even if CRAs have not pushed for such a central role in the regulatory process, this places special demands on the integrity of rating procedures and business models.

At the same time, it should be borne in mind that national responses to these challenges will not suffice. Even the European Union would be sub-optimal as a stage for action, since ratings are usually addressed to all investors worldwide. This means that internationally uniform rules are called for. If these are very strict, there is the danger that they will be interpreted and applied very differently from region to region. Compared to such a situation, the IOSCO Code of Conduct takes much more account of the internationality of ratings and national or regional specifities. Generally speaking, despite the fact that some CRA codes of conduct deviate somewhat from the IOSCO fundamentals, the regulatory system has worked.

In this connection, it should also be remembered that CRAs themselves have made extensive changes in all areas where weaknesses became apparent. This is in their own interest. They have to make clear to everyone that they are seriously trying to regain their reputation, as this is what their business model is built on.

Supervisors have drawn the right conclusion: A number of changes were made to the IOSCO Code of Conduct, and the revised version was published at the end of May. The new rules contain numerous additional requirements that CRAs have to comply with. Most of these are undoubtedly sensible and, once they have been implemented by CRAs, should help to ensure the quality of ratings and restore confidence in their reliability (3).

Creation of an enforcement authority?

The main problem continues to be how to duly monitor compliance with the tighter IOSCO rules. Particularly problematic is the lack of sanctions that can be imposed in the event that CRAs fail to adhere to the IOSCO Code. This is a question that has been dominating the policy debate in Europe and the rest of the world lately.

Despite this, governments have refrained from making any quick fixes and instead asked supervisors to draft recommendations on how to regulate CRAs in the structured finance products market. At the same time, policymakers see CRAs as having a special responsibility to demonstrably remove weaknesses themselves by taking concrete measures of their own (4).

The European Commission decided in 2006 not to regulate CRAs any further as long as there are no new findings that make it necessary to change this position. Against this backdrop, the Committee of European Securities Regulators (CESR), the European Securities Market Expert (ESME) Group - both on behalf of the Commission - and the US Securities and Exchange Commission (SEC) presented proposals that appear to show a way to keep up the required pressure, even if these need to be examined more closely. That goes particularly for the division of responsibilities between the more globally-oriented IOSCO and national supervisors and for the cardinal problem of dovetailing and consistency of CRA requirements on both sides of the Atlantic.

At European level, there is also the question of what powers national supervisors are to have with an optimal division of responsibilities and adequate cooperation between the European authorities in mind. To avoid the danger of any competency overlaps, but also of regulatory loopholes, solutions to these questions have to be found before any far-reaching policy decisions are taken. Against this backdrop, too, the regulatory plans announced recently by Internal Market Commissioner McCreevy are premature and seen by experts as generally going in the wrong direction (5).

Conclusions

High-quality, widely recognised ratings are a basic condition for a functioning securitisation market. This is why it is to be welcomed that CRAs are undertaking substantial efforts to restore confidence in their ratings.

Users of ratings also need to change their behaviour. The past few months have shown that uncritically adopting CRA ratings is not enough if investors are to make proper investment decisions. This also involves an obligation for CRAs to educate investors better. An important task for them in the future will be making the difference between ratings and risk assessments even clearer in order to prevent wrong conclusions that ultimately pose a potentially serious threat to market stability.

The subprime crisis has also shown how problematic it is that ratings are being used increasingly for regulatory purposes. At the end of the day, this has made supervisors and CRAs dependent on one another - a development that gives cause for concern. The president of BaFin (the German financial supervisory authority) put it this way: "Ratings which turn out in retrospect to be fictive are also a disaster for us supervisors." This is an unsatisfactory state of affairs, and it is essential to consider what consequences need to be drawn from it.

The financial market turmoil has inevitably revived the question of the best possible regulation of CRAs, particularly as the oligopolistic structure of the ratings market remains unsatisfactory. A clear understanding of the role and competencies of supervisors when implementing and overseeing the necessary optimisation measures is needed.

It is doubtful whether tighter regulation or registration of CRAs would have prevented the subprime crisis. In this light, the discussion of how to optimise ratings of structured finance products is highly welcome. At the same time, it is merely a necessary but not yet adequate response to the many questions thrown up by the difficult situation on the financial markets in recent months.