By Nicolas Véron Credit Rating Agencies (CRAs) are prominent participants in the assessment of credit risk by financial markets. They determine and publish credit ratings， which represent the CRA’s opinions on issuers’ relative probability of default. The market for credit ratings is currently dominated in most western countries by three players:Standard Poor’s (SP) is a division of the McGraw- Hill Companies， a US-based media group whose ownership is dispersed(the largest shareholder is Capital Group， with 12 percent of shares);Moody’s Corporation is an autonomous US-based listed company with dispersed ownership (the largest shareholder is Berkshire Hathaway， with 12.5 percent of shares);Fitch Ratings is a division of the Fitch Group which is jointly owned by Fimalac， a Paris-based listed investment vehicle (60 percent of shares)， and the US-based Hearst Corporation (40 percent of shares).CRAs rate different types of issuers or issuances. For example， Fitch reports that in 2009-10 it rated around 6，000 financial institutions， 2，000 non-financial corporate， 100 sovereign states and 200 territorial communities， 300 infrastructure bond issuances， 46，000 US municipal bond issuances， and 8，500 structured product issuances.3 A simplistic but common way of segmenting the market is between sovereign ratings， corporate ratings， and structured credit ratings.CRAs are UnreliableMeasuring the accuracy of credit ratings is intrinsically difficult， even with hindsight， because they correspond to probabilities. A poorly-rated issuer may avoid a default even though the probability of default was high. Conversely， a highly-rated issuer may default even though the probability of it was low. Thus， strictly speaking ratings quality can only be measured on average over many rating opinions， based on the law of large numbers， and not on individual ratings. Moreover， according to the CRAs， their ratings measure relative probabilities of default， not absolute ones. An AA rating signals a lower probability of default than a BBB， but CRAs do not provide a numerical estimate of the respective probabilities (even though they do publish historical data on the frequency of default associated with different past ratings).From this standpoint there was a clear failure of CRAs when it came to US mortgage-based structured products in the mid-2000s. Many mortgage-based securities were highly rated but had to be downgraded in large numbers following the housing market downturn in 2006-07， especially in the subprime segment. Subsequent enquiries， in particular the Securities and Exchange Commission (SEC 2008) and Financial Crisis Inquiry Commission (FCIC 2011)， have convincingly linked the CRAs’ failure to a quest for market share in a rapidly growing and highly profitable market segment. Under com- mercial pressure， CRAs failed to devote sufficient time and resources to the analysis of individual transactions， and also neglected to back single transaction assessments with top-down macroeconomic analysis that could have alerted them to the possibility of a US nationwide property market downturn.While the CRAs fully merit blame for this failure， it is to be noted that credit ratings for residential mortgage-based securities in the 2000s was a relatively recent activity compared with corporate and sovereign ratings， and that the subprime segment was new within the larger US mortgage market. In other segments， including corporate and sovereign ratings， CRAs could rely on much longer and deeper experience of risk factors and past failure patterns. In these more “traditional” segments， statistical tables published by the CRAs and others (e.g.， IMF (2010)， figure 3.7) suggest a generally strong correlation between past ratings and relative average probabilities of default as observed ex post over large numbers.That said， there have been several past cases in which rating agencies clearly failed to spot deteriorations of sovereign or corporate creditworthiness in due time. This was particularly true of Lehman， AIG， and Washington Mutual， which kept investment-grade credit ratings until September 15， 2008. CRAs were similarly criticized for their failure to anticipate the Asian crisis of 1997–98 or the Enron bankruptcy in late 2001.It appears fair to conclude that all three main CRAs have a decent though far from spotless record in sovereign and corporate ratings， but that their hardly excusable failure on rating US residential mortgage-based securities in the mid-2000s has lastingly damaged their brands and reputations.CRA Downgrades Can Trigger Sudden Shifts in Risk PerceptionsSince the beginning of the crisis， CRAs have frequently been accused of timing their downgrades badly and of precipitating sudden negative shifts in investor consensus. However， it is infrequent that rating downgrades surprise markets， generally they follow degradations of market sentiment rather than precede it. When CRAs do anticipate， they are often not given much attention by investors， such as when SP started downgrading Greece in 2004.Specifically， the evolution of euro area sovereign yields since 2008 suggests that the biggest and most sudden shifts in investor sentiment have been triggered by new information from the policy sphere， such as， among others， the announcement by Greece of worse deficits than previously disclosed， the French-German Deauville declaration of October 18， 2010.May 16， 2011. These policy signals have had demonstrable impacts on risk perceptions， as the Bank for International Settlements has noted in the case of the Deauville declaration. By comparison， CRA downgrades of euro area countries so far have had limited market impact， if any.The sovereign downgrade of the United States by SP on August 5， 2011 was a special case， to the extent that the US sovereign debt market has a specific anchoring role for the global financial system and there had never been a downgrade of US sovereign debt in living memory. Ironically， it coincided with a sharp increase in risk aversion which resulted in a short-term decrease of yields on US debt. The downgrade may have contributed to market jitters about France’s creditworthiness and French banks’ prospects in the days that followed. However， at the time of writing there does not appear to be an analytical consensus on its role in triggering these market developments compared to other simultaneous factors.The upshot is that instances in which CRA downgrades materially affect general market sentiment seem to be possible but rare， and that none has been compellingly observed recently in the context of the euro area crisis.References to credit ratings are embedded in a number of contractual and regulatory provisions throughout the financial system. Thus， even though CRAs argue that their ratings are mere opinions intended for the judgment of market participants， they can have a mechanical， pro-cyclical effect if such provisions result in， for example， forced selling of a security as a consequence of its downgrade. The collateral policy of the European Central Bank (ECB) is one example.However， the actual extent of such mechanical pro-cyclical effects is limited by several factors. Most investment mandates now have significant built-in flexibility to reduce dependency on individual rating changes. The ECB has displayed considerable flexibility in adapting its collateral policy to new developments， including rating downgrades， throughout the crisis. Strikingly， as observed above， US debt markets has been observed following the downgrade of the United States by SP in August 2011.While credit ratings are affected by economic cycles， they tend to be much more stable than market-based indicators of creditworthiness(Moody’s 2009). Thus， replacing credit ratings with market-based measures would reinforce pro-cyclicality. In short， mechanical pro-cyclical effects of rating downgrades are a legitimate concern， even though CRAs are not to blame for them. However， these effects seem to be already mitigated to a significant extent.CRAs Escape Local RegulationsCRAs started life outside of the scope of public regulation， often in connection with media and/or advisory businesses. The United States introduced the Nationally Recognized Statistical Rating Organization (NRSRO)6 process of administrative recognition of CRAs in 1975， and a more hands-on registration regime was introduced by the US Credit RatingReform Act of 2006. In the European Union， the regulatory framework was not intrusive until the crisis， but has evolved rapidly with the adoption of successive regulations in September 2009 (known as CRA 1) and May 2011(CRA 2); a third regulation is at an early stage of elaboration. Other jurisdictions， including Japan， Australia， and Hong Kong， have also adopted a new CRA regulatory framework since the crisis.As with other financial information intermediaries， territoriality is a difficult issue in the context of such regulations. In principle， creditworthiness analysis of any issuer can be done from any location. Moreover， the global consistency of credit ratings is viewed by most market participants as a significant benefit. The combination of these two factors potentially reduces the scope and effectiveness of territorial regulation.Within the European Union this issue has been addressed with the devolution of most regulatory and supervisory tasks regarding CRAs to the recently created European Securities and Markets Authority (ESMA)， which in principle guarantees regulatory consistency across all EU member states. However， the risk remains of inconsistency or interference with regulatory regimes in non-EU jurisdictions.Public Standardization of Ratings MethodologiesThe methodologies and criteria used by CRAs to prepare ratings have a significant impact on ratings outcomes， and are inevitably open to debate. For example， SP has been criticized for having included an analysis of political dynamics in its recent downgrade of US creditworthiness. However， the temptation to publicly regulate ratings methodology should be resisted， as it would collide with the justification of ratings as independent opinions. In the absence of a global level of public standardization， such an approach would also threaten the international comparability of ratings.Thus， the United States and European Union have been right to commit themselves to refraining from the direct regulation of CRA methodologies so far. A US Treasury official declared in the Dodd-Frank legislative debate that“the government should not be in the business of regulating or evaluating the[CRAs’] methodologies themselves.”The European Union’s second regulation on rating agencies (May 11， 2011) specifies: “In carrying out their duties under this Regulation， ESMA， the[European] Commission or any public authorities of a Member State shall not interfere with the content of credit ratings or methodologies”.Changes in the CRAs’ Business ModelDuring their first decades of activities， CRAs mostly relied on investors as their main customers， but shifted to their current “issuer-pays” business model around the 1970s as their activity expanded significantly. This raises the possibility of a conflict of interest as an issuer may leverage the commercial relationship to obtain a higher rating. A different business model could be imposed as a condition for public registration.Whether this measure would be beneficial， however， is questionable. The most likely outcome would be a significant decrease in the overall resources of regulated CRAs， as investors have until now seemed unwilling to pay significant amounts for credit ratings. One US-based CRA， Egan-Jones， is financed by investors but its size remains limited: It has five credit analysts and a total staff of 22， or 200 times fewer than Moody’s.12 Moreover， an “investor-pays” model would by no means eliminate conflicts of interest， as (for example) investors who hold a security may desire it to be highly rated. Finally， it is to be noted that while conflicts of interest have been a significant issue in structured ratings， and to an arguably lesser extent in corporate ratings， they are essentially absent from the sovereign ratings market from which CRAs drive a very small fraction of their revenue.Tighter Regulation and SupervisionThe 2006 US Credit Rating Reform Act and the Dodd-Frank Act， and the two successive EU regulations adopted in 2009 and 2011， result in significant regulatory and supervisory powers for the Securities and Exchange Commission (SEC) and ESMA respectively. Unfortunately， there is no reason to believe such regulation can be sufficient to eliminate imperfections in the credit ratings market. Obviously， regulation of CRAs by the SEC (in place since 1975 and reinforced by the Credit Rating Reform Act) did not prevent the sub-prime debacle. Europe’s regulatory screws on CRAs are already quite tight， and full implementation of the existing regulations would be warranted before envisaging their further tightening.Moreover， both theory and experience suggest that regulation generally reinforces barriers to entry in concentrated markets， and there are no reasons to believe the market for credit ratings is an exception. Thus， tighter regulation and/ or supervision are unlikely to contribute to addressing the problem of market concentration. If anything， they could make it more intractable still. Moreover， CRA regulatory regimes in the European Union and elsewhere are highly prescriptive as to how CRAs should organize themselves and conduct their business， which could discourage innovation and the pursuit of new organizational or operational practices that may eventually lead to better ratings.A Global Regulatory / Supervisory Regime for CRAsTo the extent that credit ratings are useful， their comparability across borders is a global public good in the context of international financial markets integration. The risk of fragmentation due to regulatory differences was minimal before the crisis， as only one major jurisdiction (the United States) materially constrained the behavior of CRAs through regulation. Now that the European Union， Japan， India， Australia， Hong Kong， and other jurisdictions have started to regulate CRAs， however， there is an increasingly material risk of different regulators imposing different standards resulting in a reduction of cross-border ratings comparability.A constructive step to address this risk would be the adoption of global standards that would determine the content of jurisdictional regulations applicable to CRAs with the aim of maximum harmonization. The International Organization of Securities Commissions (IOSCO) would be the logical forum for the discussion and preparation of such standards， as it has done in the past for less hands-on regulatory approaches such as its 2004 “code of conduct fundamentals for CRAs”(IOSCO 2004). A more radical initiative would be the establishment of a global public (treaty-based) authority to which individual jurisdictions would delegate the supervision of CRAs， thus ensuring global supervisory consistency.This proposal may sound overly ambitious on grounds of national sovereignty， but it is to be noted that similar arguments were long made against the establishment of European Supervisory Authorities in the financial sector， which were eventually superseded with the adoption of the EU financial supervisory legislative package in 2010. The continued global integration of financial markets may require unprecedented steps of international supervisory cooperation， and CRAs are arguably one of the categories of regulated market participants for which such efforts could be considered most necessary.(Author: Visiting fellow at the Peterson Institute for International Economics)