A Statement by The Financial Economists Roundtable drafted by a committee, chaired by Edward Kane and Richard Herring including Edward Altman and Charles Goodhart

image002_512_24.gif

During the last few decades, securitization has become a primary channel for enlarging financial markets and transferring credit risk from lenders to investors.  Outstanding issues of privately securitized assets peaked worldwide at just under $12 trillion in 2008.[1]

When properly structured and monitored, securitization promises numerous benefits. It can generate opportunities for specialization that reduce funding costs, increase the range of financial products available, encourage financial institutions to deploy capital more efficiently, and allow borrowers, lenders, and investors to manage their risks more flexibly. However, transferring risk undermines incentives to perform due diligence at virtually every stage in the securitization process. In the last year, evident shortfalls of care and diligence in the origination, rating, and securitization of mortgages have led to a collapse in the prices of securitizations related to subprime mortgages, alt‐A mortgages and other leveraged loans. The suddenness and extent of this price decline has undermined confidence in the reliability and integrity of the ratings process for asset‐backed securities, and has reduced prices and credit flows in every market in which investors count on ratings firms to ascertain the quality of debt. Meeting in Glen Cove, New York in July 2008, the Financial Economists Roundtable (FER) discussed the need to strengthen the securitization process by changing the incentives under which Statistical Ratings Organizations (SROs) operate. SROs (profit‐making firms that prefer to call themselves credit rating “agencies”) play a central role in testing the quality of the pool of obligations being securitized and in creating and marketing “tranches” of graded claims to cash flows from the underlying mortgages or other debt. The scope and scale of ongoing ratings downgrades and defaults on securitized debt make it clear that the ways in which credit ratings are used and constructed must be reformed.

The FER sees a strong need for three types of credit-rating reform. First, FER supports strategies designed to improve SRO incentives by increasing the transparency of their modeling practices and holding their managements accountable for negligent ratings errors. Second, the FER challenges the wisdom of incorporating SRO ratings in securities and banking regulations issued by governmental entities. By outsourcing public authority to private firms, this practice intensifies the conflicts of interest that SRO personnel must resolve. Finally, to acknowledge differences in the degree of leverage that is imbedded in different issues of securitized debt, FER recommends that SROs be required to state an express margin for error in their ratings for every tranche of securitized instruments.

Some Historical Perspective

Bond markets functioned internationally for 300 years before the first rating organizations appeared in the United States.  An active corporate bond market, largely in debt issued by railroad companies, emerged in the middle of the 19th century in the United States more than half a century before the first SRO opened for business.  SROs remained largely US- focused until the 1970s, when global capital markets began to reemerge after fading in the interwar period.

In the pre-SRO era, underwriters performed some certification and monitoring for investors.  Thereafter, third-party ratings mitigated asymmetric-information problems between issuers, underwriters, and investors by credibly centralizing efforts to collect and analyze the information needed to estimate, monitor, and update the probability of default of individual bonds.

Ratings data also expanded the range of investors willing to hold corporate bonds to include parties that lacked the resources to undertake a complete and independent credit analysis.    SROs originally earned their revenue by selling ratings manuals directly to investors.

Building a reputation for accuracy is critical to the success of any SRO.  Ratings firms prospered to the extent that their predictions of the probability of default proved reliable after the fact.  Over time, the accumulation of reputational capital by successful SROs made entry difficult for new SROs.  The result is that two or three SROs have dominated the market for credit ratings, and did so long before the SEC began to designate particular SROs as Nationally Recognized Statistical Rating Organizations (NRSROs) in the 1970s.

In the early 1930s, incentives for SROs to produce reliable information for investors were complicated by introducing ratings into the regulatory process.  Regulators of banks, insurance companies and pension funds began to use ratings to limit the riskiness of the assets held by regulated entities.  Regulators now set two kinds of rules: rules that restrict the extent to which a firm can hold assets that fall below investment-grade or, as in the case of money market mutual funds, require a higher threshold than investment grade, and rules that link capital requirements to the ratings on individual securities, with lower capital charges for high-rated securities.[2]   The existence of such regulatory consequences was bound to intensify pressure on SROs to inflate the grades of lower-rated securities, because regulated clients routinely explore and develop ways of reducing their regulatory burdens.  Frank Partnoy (1999, p.684)[3] describes client pressure in this way: “[O]nce regulation … incorporates ratings, rating agencies begin to sell not only information but also valuable property rights associated with compliance with the regulation.”  As ratings became more widely used in trigger clauses in bond contracts, strong ratings conveyed additional benefits to the issuer.

Of course, a concern for protecting their reputations can act as a healthy counterincentive.  Studies of ratings accuracy during the 20th century find that SROs have done a reasonably good job of predicting the probability of default of corporate bonds relative to regulatory indicators[4] of default risk and market measures of default risk.    Still, grade inflation has occurred.  Caouette et al. (2008) observe that though the ratings do represent relative risks (on average) reasonably well, they are less reliable as indicators of absolute credit risks; default probabilities associated with specific rating levels drift over time and therefore need to be frequently updated.[5]

The spread of photocopying technology facilitated unauthorized reproduction of SRO rating manuals, which undermined the traditional user-pays revenue model.  SROs responded by shifting to a business plan in which the issuer pays for their services.  This plan intensified SRO conflicts of interest with issuers.  Issuers and underwriters actively shopped for ratings and were unwilling to pay for ratings they deemed too low.[6]  In the case of the newer securitized debt, pressure for favorable ratings has been particularly intense because the large underwriters of structured debt could direct substantial future revenue to a cooperative NRSRO, thus increasing the potential for undue influence.   SROs argued that concern for maintaining their reputational capital would nevertheless insulate ratings decisions on securitized debt from undue influence by issuers.  This argument became increasingly less persuasive as income from rating structured debt began to increase sharply and account for almost half of the revenues of the three dominant firms.

A further weakness inherent in issuer-pays arrangements is that they undercut SRO incentives to monitor and downgrade securities in the post-issuance market.  The re-rating of securities is usually paid for by a maintenance fee that is collected in advance from each issuer.  Few issuers are eager to be monitored closely, especially when monitoring is apt to result in downgrades, and so it is not surprising that ratings are seldom downgraded until long after public information has signaled an obvious deterioration in an issuer’s probability of default.[7]

Not until 1975 did the SEC confront the problem of how to determine whether a particular SRO could be relied upon to provide ratings of sufficiently high quality that they could be used in the regulatory process.  The SEC’s solution to this problem was to certify particular SROs as meeting sufficiently high standards to be designated by the SEC as an NRSRO.  Other regulatory agencies, Congress, and many private agreements made use of the SEC’s designation of qualified NRSROs.  For potential new entrants to the ratings industry, the costs and uncertainty of obtaining NRSRO status imposed an additional, legal barrier on top of their already substantial reputational disadvantage.  From 1975 to 2002, although the SEC received numerous applications from entities in the United States and abroad, only one new general-purpose NRSRO was approved.

The NRSRO designation strengthened the market power of the dominant three incumbent firms: Moody’s, Fitch, and Standard & Poors.  In turn, the oligopolistic position these firms enjoy reduces their incentives to compete in ratings methods and procedures.  For example, even though SROs inevitably lack long histories and through-the-cycle data on innovative instruments, they have all been slow to draw on the information generated by derivatives trading (especially in credit default swaps) and from secondary markets for debt and equity, both of which would help them analyze potential defaults in a forward-looking context.  Nor have SROs developed procedures for supplying information on correlations that investors need to protect against concentrations in risk exposure that might exist in a portfolio of securities.

Despite the potential benefits of strengthening competitive forces in the SRO industry, the three major NRSROs have been permitted to acquire competitors virtually without challenge.[8]  The FER believes that the regulators could enhance competition among SROs by more vigorous application of antitrust policy.    Although the SEC recently recognized a handful of additional firms as NRSROs in the last two years in response to pressure from Congress to ease barriers to entry, it will take considerable time for new entrants to wean much market share away from the three dominant firms.

FER’s Evaluation of SEC Proposals for Reform

Because some market participants are bound to base investment decisions primarily on credit ratings, efforts to improve ratings quality are important.  In June, the SEC proposed several ways to improve the work of SROs and to increase competition in the ratings industry in three ways.  The avowed and laudable purpose of these proposals is to foster increased transparency, accountability, and competition in the credit rating industry for the benefit of investors. The precise models used by SROs are proprietary and to encourage an individual SRO to invest in improving its models, the models themselves must remain proprietary.  At the same time, to hold SROs accountable for their performance requires that each SRO release enough information on data input into its models to allow outside experts to verify its conclusions or provide alternative results.

The SEC’s first proposal seeks to mitigate conflicts of interest, enhance disclosures, and improve internal policies and business practices at SROs.  The second proposal would require NRSROs to differentiate the ratings on structured products from those that they issue on traditional bonds and loans, and perhaps to provide a timely and relevant accompanying narrative.  The third proposal would nearly eliminate the role of ratings in SEC regulations.  FER supports the thrust of each proposal.  To explain why, we discuss each in turn.

In the important areas of disclosure and incentive conflicts, the SEC’s first proposal would require SROs to:

  • Publish all ratings and subsequent re-ratings in ways that facilitate comparisons of SRO performance in a timely manner.  Disclosures would include performance statistics for spans of 1, 3, and 10 years within each rating category.[9]
  • Disclose all information used to determine ratings for structured products.  In addition, this would require each SRO to explain whether and how it might rely on the due diligence of others to verify the character of the assets underlying a structured product and to include sufficient information on the changing value of underlying assets to permit outside analysts (i.e., persons who are not paid by the issuer) to evaluate the riskiness of the structured claims issued against them.
  • Explain how frequently credit ratings are reviewed, whether different models are used for ratings surveillance than for setting an initial rating, and whether, when changes are made in an SRO’s models and procedures, they are applied retroactively to existing ratings.

The FER is less enthusiastic about the SEC’s proposed prohibition against letting an SRO act as both a rater of and a paid advisor for a tranched securitization.  Although we appreciate that acting in these dual capacities intensifies SROs’ conflicts of interest, we believe that the customary industry practice of presenting alternative structures for an SRO to rate makes it impossible for the courts to distinguish ratings services from advisory services in a definitive way.  Moreover, we believe the enhanced disclosures will ease this conflict of interest.

The SEC or Congress might also impose disclosure requirements on issuers.  Every US issuer of securitized claims could be required to provide a monthly balance sheet and income statement for each and every securitization structure it creates, even if the securities are to be marketed offshore.  The revenue-generating pool of underlying assets constitutes the structure’s assets and the tranches set by the securitization structure constitute claims against these assets.  When underlying assets lose value, whether through rating downgrades or outright defaults, prospective revenues diminish and the values of affected tranches deteriorate.  These easy-to-interpret disclosures would make pending deteriorations in cash flows more visible to investors and permit the joint distribution of risk statistics for the various tranches to be studied more effectively.

The SEC’s second proposal seeks to differentiate ratings on securitizations in the future from those on ordinary bonds.  Because of their imbedded leverage, securitized instruments may have a much deeper downside loss exposure than ordinary bonds. Using the same grading scale for both kinds of instruments reduces the effectiveness of restraints on institutional risk taking built into longstanding regulatory protocols.  This renders many inherited regulatory strategies obsolete and was bound to confuse at least some investors.  As an estimate, every credit rating carries a calculable margin for error.  Introducing a differentiated scale is one way to alert investors that downside margins for error are much larger for securitized claims than for ordinary debt.  Because imbedded leverage and downside margins for error grow larger when claims on an underlying asset pool are tranched and retranched, SROs should be required to express ratings on securitized debt in a two-dimensional fashion (i.e., with an accompanying estimate of their particular margin for error).  This would be much more useful than merely developing a separate scale for securitized instruments.  SROs might either use estimates of potential downside variability to rate claims in an interval framework (e.g., a particular rating might be expressed as lying in the range from A to AAA) or prepare and publish the volatility estimates themselves.

The SEC’s third proposal addresses its practice of basing rules and reporting procedures on NRSRO ratings.  The concern is that the use of NRSRO ratings  in supervision simultaneously outsources some of the regulatory authority’s political accountability to profit-making firms and appears to confer an official seal of approval on their methods that might reduce the willingness of other parties to undertake due diligence and invest in securities analysis.  The SEC proposes to remove references to NRSRO ratings from virtually all of its rules and protocols.[10]

The FER discussion divided references to NRSRO ratings in SEC regulations into two categories:  prescriptive mandates that tell asset managers what they must do and quasi- safe-harbor provisions that provide firms, managers and directors some protection from liability for adverse outcomes.

The FER strongly endorses eliminating from SEC regulations every prescriptive mandate that is or would be based solely on credit ratings set by NRSROs.  We believe this will have three advantages.  First, the prudence of investment decisions must ultimately be evaluated in a portfolio context and cannot be assured by constraining the credit quality of individual assets an institution holds, regardless of how accurate the SRO ratings might be.  Second, depriving SRO ratings of regulatory consequences will remove a major source of pressure for ratings inflation.  Third, in the absence of SEC mandates, managers and directors can and will subject the prudence of their decisionmaking to review by a much wider array of outside monitors.  In particular, they will expand their use of directors and officers insurance and introduce letters of assurance from well-respected experts.  Whether or not these other monitors aspire to attain SRO status, they would supplement, extend, and challenge the assessments of individual securities made by SROs, thereby injecting valuable competition into the market for rating services.

The FER found it harder to assess the net benefits of quasi- safe- harbors (offered mainly to directors and officers of money market mutual funds) based on credit ratings.[11]      Some members felt that removal of quasi-safe- harbors would yield benefits from increased managerial diligence and reduced pressures for grade inflation that would more than offset the increased compliance costs and costs of defending nuisance lawsuits.  Other members believed that there are efficiencies to be achieved by use of intermediaries specialized in credit review.  They argued that the rating requirements for money market mutual funds had worked reasonably well (apart from the current credit crisis) and that increased compliance costs, especially for smaller funds, would swamp any benefits that might emerge from increased managerial effort.  Moreover, it was agreed that retaining this role for NRSROs would provide SROs with an incentive to register for NRSRO status and comply with the enhanced disclosure requirements. Even if the SEC should decide to continue to offer quasi- safe- harbors based on credit ratings, requiring a new ratings scale for securitized debt means that the content of such provisions has to be analyzed afresh to acknowledge the implications of the distinctions created.  A new scale will similarly force banking agencies and state regulatory bodies to rethink and rephrase all rules and regulations that rely on credit ratings.  In view of the importance of regulation-induced innovation in creating financial turmoil, such rethinking is long overdue.

Implications for Other Regulators

Although the SEC stressed that it had consulted with the President’s Working Group on Financial Markets, the Financial Stability Forum and the Technical Committee of the International Organization of Securities Commissions (IOSCO), the SEC’s proposed removal of references to ratings in its regulations diverges sharply from reform strategies currently being implemented by other regulators in the US and abroad.  For example, the Treasury’s temporary insurance of money market mutual funds relies on compliance with rule 2a-7 that relies on rating as a useful indicative guidance, and the Treasury’s recent plan to recapitalize banks will be contingent on ratings to some extent.   FER sees the SEC’s third proposal as providing a timely challenge to other regulators to reexamine the extent to which they plan to employ SRO ratings in their own regulatory schemes.

Although new rules and enhanced supervision might induce slightly better SRO performance, it is unlikely that increased government oversight of the production of credit ratings can improve SRO performance over time and improve the performance of investment managers as effectively as market forces can.  It is particularly important for banking regulators to reconsider their reliance on ratings decisions.  By adopting Basel II, they are linking minimum capital requirements for some banks to ratings issued by whatever SROs they recognize in each individual nation. Some banks will be free to use Basel II’s Standardized Approach, which the European Union and Japan have already begun to implement and is proposed for implementation in the United States.  In this scheme, capital charges are assigned to each bank’s assets according to their credit ratings, with unrated assets receiving a 100% risk weight.  Since loss reserves are already based on anticipated losses, capital requirements are intended to provide a buffer against unexpected risks.  Thus, it is illogical to use credit ratings to establish capital requirements, since they convey no information about the volatility of an asset’s return around expected loss experience.  In addition, ratings may be useful for establishing loss reserves for particular assets, but they say nothing about how a bank’s net worth or its portfolio of assets may vary in value.  The amount of capital that must be set aside to achieve a particular target level of safety has to be linked explicitly to measures of the volatility of its earnings, not asset ratings.

Since the subprime crisis has had a world-wide reach, regulatory authorities in other countries are also thinking about how to regulate SROs. Despite the SEC’s attempt to coordinate its actions with IOSCO, it is clear that different countries may respond to the crisis in different ways.  The use of ratings is hard-wired into many European Union regulations.  The EU’s internal market commissioner is thinking of introducing some exacting regulatory requirements to make sure ratings are not “tainted” by the conflicts of interest inherent to the ratings business. The European Commission has proposed a registration and oversight regime that would have two features.  The first charges the Committee of European Securities Regulators (CESR) with the responsibility for choosing an individual country to register, coordinate and consolidate oversight of individual SROs.  The second creates a central supervisor, financed from the EU budget, to license rating organizations.  As capital markets become more closely integrated, ratings organizations are bound to find it difficult to operate under different rules in different locations. Also differences in rules would complicate cross-country comparisons of ratings for investors and regulators.  If a single supervisory approach is to be adopted, FER strongly supports the SEC’s strategy which relies on greater transparency, increased competition and the abandonment of the practice of incorporating NRSRO ratings in regulatory mandates.   The FER hopes that other regulators will follow the SEC’s lead.

image002_512_23.gif


Footonotes:
[1] See Table 1 at the conclusion of this statement.
[2] For example, (Sylla 2002, p. 37) notes that in 1936, the US Comptroller of the Currency issued a regulation prohibiting banks from purchasing investment securities with characteristics that were “distinctly or predominantly speculative,” and then added that “the terms employed…may be found in recognized rating manuals, and where there is doubt as to the eligibility of a security for purchase, such eligibility must be supported by not less than two ratings manuals.” The latter phrasing, referring to recognized raters, was attacked as placing too much authority in the private rating agencies, and on that ground it was deleted from the regulation in 1938, although in a less formal way it remained in effect with regulators.  For additional details see Richard Sylla, 2002, “An Historical Primer on the Business of Credit Rating,” in Ratings, Rating Agencies, and the Global Financial System, edited by Richard M. Levich, Giovanni Majnoni, and Carmen Reinhart, The New York University Salomon Center Series on Financial Markets and Institutions, Kluwer Academic Publishers, pp. 19-40.
[3] Frank Partnoy, 1999, “The Siskel and Ebert of Financial Markets?  Two Thumbs Down for the Credit Rating Agencies,” Washington University Law Quarterly, 77, October.
[4] For example, Hickman (1960) used legal investment lists for savings banks adopted by regulatory authorities in the states of Maine, Massachusetts, and New York as an indicator of regulatory ratings. For additional details see W. Braddock Hickman, 1960, Statistical Measures of Corporate Bond Financing since 1900, Princeton:  Princeton University Press.
[5] See J. Caouette, E. Altman, P. Narayanan, Managing Credit Risk, 2nd edition, John Wiley & Sons, NY, 2008.  The expected dollar-denominated default rate on non-investment grade corporate bonds in 1984 was 1.6% per year, but is now 3.9% per year.  As late as 2007, Fitch reported that the default rate on structured products through 2006 was similar or lower than that on corporate bonds.  Subsequently, results for structured products deteriorated sharply.
[6] The June 2008 settlement between the New York Attorney General and the ratings agencies mandated charging separate fees for indicative ratings.  While the intent was to reduce shopping for ratings, some FER members raised concerns that it may have the opposite effect by lending tacit official approval to the practice of shopping for ratings.
[7] E. Altman, H Rijken, “How Rating Agencies Achieve Rating Stability,” Journal of Banking & Finance, 28 (2004), 2629-2714, and E. Altman & H. Rijken, “A Point in Time Perspective on Through the Cycle Ratings,” Financial Analysts Journal, 62, No. 1, (2006), 54-70.
[8] For example, Moody’s purchased the market-based credit risk and portfolio management firm, KMV, in 2001 and Duff & Phelps was purchased by Fitch in the early 1990s.  Although KMV was not formally an NRSRO, it competed directly with NRSRO firms.
[9] Although SROs provide data on default rates for bonds and loans by rating categories, data on structured products have been provided less frequently and ought to be published faster and more extensively in times of market turmoil.
[10] An exception is drawn for rules and forms that “relate to non-public reporting or recordkeeping requirements used to evaluate the financial stability of large brokers or dealers or their counterparties and are unlikely to contribute to any undue reliance on NRSRO ratings by market participants.” (Quoted from SEC 17 CFR Parts 229, 230, and 240, Release No. 33-8940; 34-458071; File No. S7-18-08, p. 5.)  These include rules which impose certain recordkeeping and reporting requirements for holding companies that own broker-dealers and of supervised investment-bank holding companies and reports regarding the risk exposures of large broker-dealers and OTC derivatives dealers.
[11] This protection is at best a quasi- safe- harbor because rule 2a-7(c) (3) states that the board must take into consideration “factors pertaining to credit quality in addition to any rating.”  It might better be viewed as indicative guidance.

image003_512_15.gif

3 Responses to "A Statement by The Financial Economists Roundtable drafted by a committee, chaired by Edward Kane and Richard Herring including Edward Altman and Charles Goodhart"

  1. Anonymous   December 2, 2008 at 2:13 pm

    Translation:Its all smoke and mirrors?

  2. Ann Rutledge   December 7, 2008 at 11:08 am

    Why aren’t there more comments? Maybe because there’s no controversy. So, let me add a little controversy:(1) strategies designed to improve SRO incentives by increasing the transparency of their modeling practices and holding their managements accountable for negligent ratings errors.The biggest contribution to opacity is the lack of formal study of structured ratings. Not a lack of transparency, but a lack of structured finance literacy. There isn’t a single university in this list that teaches the technical basis of structured finance. How can we be accountable for what we don’t understand?(In the interest of full disclosure: I am one of five people in the world who teaches structured finance. Does that mean I’m preaching for my own parish, or does it mean I’m qualified to judge? How is this any different from the argument about rating agencies?)(2) challenges the wisdom of incorporating SRO ratings in securities and banking regulations issued by governmental entities.Great: the DUI method. I agree, and why does point #1 matter if we have it.But, we must recognize that to roll back 2008-1936=72 years of history would take us to an era two decaces before the birth of modern accounting. Do we really appreciate how much rethinking that would require. How much more of an information society we would have to become not to rely on government-appointed surrogates. And – this is important – how much more responsibility we would have to shoulder individually. Are we prepared for this psychologically?(3) acknowledge differences in the degree of leverage that is imbedded in different issues of securitized debt.Most specifically, structured ratings are dependent on the original capital structure. Recycling when the associated tranche is refinanced in a different capital structure is illegitimate, statistically speaking.It follows more generally that the credit quality of all structured debt, even structured ratings of debt backed by a static pool will change, incrementally and of its own accord, merely by the deleveraging that takes place when the pool amortizes: the first principle of structured risk management.(Once again, in the interest of full disclosure: R&R is the only firm in the world that re-rates structured deals. Does that mean we’re “selling snake oil” or does it mean we’re uniquely qualified to discuss the topic? Whom would you trust: us or the rating agencies?)