White Paper on Rating Competition and Structured Finance (Part 2)

Implications: Continued from part 1.

Analysis: failure that permeates the rating industry is a network effect. An example of a network effect is a language. A specific language gains currency and, hence utility, through wide adaptation. The larger the number of speakers, the greater is the language’s usefulness. This imparts a monopoly status to an established language. Competition – that might arise from a parallel language – if anything, wastes resources through the need to employ interpreters and duplicate documents. Ratings are a type of language. They too gain currency when widely “spoken” and understood. When discussing the attributes of a bond, traders and investors prefer to speak in one rating language. They want to be sure, for example, when told that a bond is rated BBB, it is of a known credit quality. Clarifying that the BBB is from XYZ rating agency often simply confuses the matter.9 Consequently, any emerging, competing risk language will face much resistance until it reaches a critical mass of users. Like a language, a rating system gains currency when both coverage and distribution are broad. Wide coverage – across obligations, issuers, sectors and regions – facilitates investment comparisons. Wide distribution also increases the chances that users will prefer one rating system to another. Unfortunately, the high costs of achieving broad coverage and wide distribution form a barrier to entry. Most financial news and data providers allocate space for just one or two rating systems. Large investors and others buy feeds from the major rating agencies and must configure their databases and display systems to handle each rating system. Issuers generally do not enjoy meeting with rating agencies. Beyond enduring uncomfortable questions, they must prepare presentations and allocate scarce time and personnel for meetings. They do not want to meet with 10 rating agencies. Nor do they want to buy the services of 10 rating agencies. In other words, there is a network effect at the rating industry level and smaller network effect with respect to an individual rating agency. In order for a competing system to displace the established rating paradigm, it must entice a critical mass of users. Such early adopters must be willing to bear costs without yet benefiting from the network effect. And in order for a new rating agency to become successful, it must achieve broad coverage and distribution at a substantial financial risk. As illustrated below, under certain conditions, competition between rating agencies leads to rating shopping and thus to sub-par rating opinions.

Shopping for Structured Finance
The recent failure of rating agencies to signal in a timely and accurate fashion the condition of many securities backed by subprime housing loans can be traced to weaknesses (or outright failures) in the protections against conflicts of interest cited above. It is instructive to describe first the rating process for structured transactions. The structured finance industry arose as a partnership between Wall Street and the rating industry. In principle, modern structured transactions can trace their lineage to practices used to package the obligations of the US government sponsored enterprises (GSEs), such as those of Fannie Mae and Freddie Mac. These organizations issue securities backed by so-called “conforming” loans used to finance home purchases. Even though the underlying loans are subject to credit risk, the GSEs guarantee the securities backed by risky loans. The principal risks, therefore, stem from changes in interest rates and the somewhat related risk of prepayment. What Wall Street brought to the table was the repackaging of GSE securities into bonds that represented various bets on the direction of these risks. Because the GSEs only buy so-called conforming loans, there appeared an opportunity to issue securities backed by non-conforming loans: those with balances greater than the GSE limits as well as those to borrowers of less than stellar credit standing. To issue these, a banker or arranger creates a bankruptcy-remote Special Purpose Vehicle (SPV) whose function is to buy a pool of loans and issue securities to finance the purchase of the pool.10 The obligations of the SPV are tranched in such away as to insure that any losses (from delinquent and foreclosed loans) accrue first to the lowest tranche (or security class), and then to the next higher tranche, and so forth. In order to assign a rating to the highest (or any other) tranche, typical practice is to model the loss distribution of the entire loan pool. The objective is to calculate the expected loss for the tranche and match that loss to historical loss rates observed in the corporate bond market. Consequently, a structured finance rating committee for a new issuance does not vote on a rating, per se, but rather on the amount of support (subordinated to the rated tranche) needed to achieve the desired expected loss rate.11 There are a number of ways to model expected pool losses for residential mortgage-backed securities. Most are designed to build a certain amount of leeway into the assigned rating. Common practice is to replicate a severe down in the housing market through statistical methods. Unfortunately for the modelers, outside of the Great Depression, there have been few instances of a widespread decline in home prices and little historical experience with subprime borrowers. Other asset types have been used as underlying collateral for structured transactions. The most common are based on pools of automobile loans, credit card receivables, commercial mortgage loans and corporate bonds. Securitization techniques have also been applied to the tranches of existing securitizations and to pools of derivative securities.

Rating Shopping
Somewhat unique to the structured finance market is the opacity of rated securities. In certain situations, the details of the underlying asset pool and often the structure of the transaction are not publicly available for external scrutiny.12 And unless the banker/originator brings a transaction to a rating agency for evaluation, the agency will generally not have enough information to assign it a rating. The role of rating agencies is particularly important to the structured finance process. Investors rely on agency ratings when making purchase decisions because of the opacity described above. Moreover, the tools to analyze credit risk, even with transparent assets, are beyond the grasp of many investors. Rating methods are quite technical, often relying on advanced statistical techniques. Documentation supporting a transaction can be equally daunting, reading more like a legal brief than helpful financial guidance. In turn, a solid understanding of how to value structured securities remains elusive. No one “model” dominates pricing practices in structured finance. Instead, there are literally dozens to choose from. The business of rating structured finance securities is highly competitive. For one thing, the fees (and corresponding margins) tend to be high relative to other product lines.13 Structured finance is perhaps the largest single product line for the major rating agencies, representing 40% or more of total revenues. Moreover, growth in structured finance helped fuel high price/earnings multiples for rating agency shares. So there is intense pressure for each agency to see its structured finance practice thrive. In addition to being profitable for rating agencies, structured finance is very profitable for the arranging banks. Consequently, the incentive to see a transaction close is strong. This has led rating agencies to compete on standards of credit support. The rating agency most willing to assign a low level of support to a given transaction is most likely to receive the mandate to rate it. The result is a situation in which rating shopping dominates the structured finance business.14 Reputation risk is in effect traded for short-term financial gain. Bankers can wield tremendous power and play the rating agencies off of one another.15 They are able to do this because many investors see the ratings of Moody’s, S&P and Fitch as interchangeable. Consequently, many investors will purchase a security with ratings from two (or sometimes just one) of the three major agencies. Support levels migrate to the lowest possible values as agencies maneuver to maintain market share. The agency with the highest support level on a given transaction will lose the deal and, over time, its structured finance business. This situation was sustainable in the subprime area so long as pool losses remained subdued. Low support levels (which offer minimal protection to senior securities) seemed sufficient in an atmosphere of easy money and rising home prices. But when losses began to materialize, ratings began to fall, and investor losses surfaced.16

Role of Regulation
Certain market observers point to the use of ratings in regulation as a contributing factor to rating shopping. Because many regulations stipulate minimum rating levels without referencing a specific rating agency, such use contributes to the “commoditization” of ratings. When the SEC, the NAIC or a banking regulator officially recognizes a rating agency, its ratings assume a quasi-official status. One reason this occurs is because many investment professionals simply do not place importance on a rating opinion as long as it meets a regulatory (or institutionally approved) minimum. Most investment managers try to maximize return subject to a specified risk level. In the fixed income universe, ratings are the language used to establish maximum allowable risk levels and are often seen as a constraint. There is in principle a link between ratings and expected return. But investment managers add value through independent research and finding opportunities regardless of the published rating. Official status is thought to be a major contributor to demand for an agency’s ratings. Yet it is extremely difficult to estimate the benefits from official recognition. What is clear is that such recognition is highly sought after by new entrants. Finally, regulators have indicated a bias against unsolicited ratings. In fact, unsolicited ratings provide an opening for new competitors – sometimes the only opening – and form an important defense against rating shopping. As discussed below, when an issuer or its banker cannot suppress an unwanted rating opinion, the incentive to “shop” rating agencies is reduced.

Alternative Business Models
Given the conflicts inherent in the issuer-pays arrangement, it is worth considering alternative business models. We describe two of these here.

Investor-Pays
As previously noted, over its first sixty years, the bond rating industry did not charge issuers for ratings. Instead, ratings were financed mainly by manual sales and certain other investor-oriented services.17 The primary purchasers of manuals were bankers and libraries. In order to have access to ratings, one needed access to a manual from at least one rating agency. The manuals themselves were typically published annually. If ratings changed in the interim, investors would not have access to the updated ratings until publication of the next edition. This did not pose much of a problem during the relatively slow investing world of the mid-20th Century. But it certainly would not work today. With the increasing breadth of the bond markets, the number of rating manuals expanded dramatically. By 1928, Moody’s alone published separate manuals for Railroads (subsequently re-named Transportation), Industrials, Public Utilities, Governments and Municipals and Bank & Finance (subsequently re-named Bank, Insurance, Real Estate and Investment Trusts). In turn, the costs of access to a comprehensive set of ratings rose. The rating manuals were not designed to provide ready information about the actual ratings of various bonds. Rather than simple rating lists, they focused on financial information and issuer history. Many were not organized along obvious lines, such as the alphabet. They catered to a time when bond investing was a serious, deliberate affair and when the typical buyer was a buy-and-hold investor. In time, however, bond rating “surveys” and “guides” offered timely and concise rating coverage. Ready access to the rating on any obligation became the norm, reflecting the trend towards increased trading of bonds. These quick reference guides opened the rating industry to a new threat, the photocopier. With the advent of photocopying, rating agencies suffered the “free rider” problem, whereby non-paying investors could benefit from relatively easy (even if illegal) access to rating lists. Moreover, in the US public finance market, the investor base per bond was too small to support adequate analysis. Sensing this, S&P began charging issuers for public finance ratings and was quickly followed by the other rating agencies. Soon thereafter, all issuers were charged for ratings. One could argue that ratings are a “public good,” and should be available to all market participants. For one thing, rating analysts in the US are exempt from fair disclosure laws. These securities laws prevent the disclosure of non-public information to market participants. Rating agency personnel are viewed as special because of the role of ratings in the securities markets. And because ratings are used in many regulations, open access to ratings would seem to fulfill a broader public policy purpose. Restricting ratings to a select group of investors willing (and able) to pay for them may stoke populist fears. An investor-pays model therefore faces economic hurdles relating to the free rider problem and from ratings’ public good status. Newspapers are subject both to the former, and to a lesser extent, the latter hurdles. They typically rely on advertising revenues to meet the costs of supporting a large editorial staff, printing and distribution. This poses a conflict, in that a newspaper might refrain from negative reporting about a large advertiser. Many newspapers carry official notices, lending them minor public good status.

A Mutual Rating Organization
Today’s major rating agencies are large, complex organizations with staff numbering in the thousands. They have offices in countries around the world. In order to attract qualified personnel, they must compete with insurance companies, mutual funds, commercial banks and investment banks. Mutual ownership, once common in the insurance industry and still widely used by credit unions, can offer an alternative means to garner the resources necessary to compete on a global scale. Unlike public ownership, a mutual organization operates for the benefit of its members. Members supply capital and receive shares in the mutual. As owners, they also share in any profits accruing to the enterprise. Such co-ops offer myriad benefits to their owner-customers, particularly in a non-profit setting. A mutually owned rating agency would have as shareholders commercial banks, investment banks, mutual funds and other institutional investors. One need only look at the client list of any large rating agency to identify potential candidates. Unlike the shareholders of many mutual organizations, these are large, sophisticated entities. A mutual rating agency would not need to charge issuers for ratings. Its shareholder/customers would pay for access to ratings and any affiliated commentary. Access to ratings themselves would be free to all, but the costs would be borne chiefly by the shareholders. Any profits would be distributed back to the shareholders as dividends or reinvested in the agency. Losses would be covered by shareholder assessments. Because each shareholder would have a stake in the success of the rating agency, its ratings would likely crowd out those of any competitor. The more widely owned, the quicker this might happen. For one thing, issuers would generally prefer the “free” ratings of the mutual agency, doubly so if they knew that leading investors followed such ratings. The shareholders would in theory operate the rating agency to provide the best possible ratings within prudent cost guidelines. Managers would answer to the owner/users, rather than to a public board of directors. Resources would be channeled to meet the mutual needs of shareholders. Unfortunately, organizing such a diverse group of natural competitors itself provides a sufficient barrier to success. There are institutional, regulatory and legal constraints to deal with. And unless the world is convinced that the current system is broken, generating enthusiasm for such an endeavor may be difficult.

A Prescription for Structured Finance
If we accept that the issuer-pays model is the only viable alternative, an effort must be made to minimize the risk of rating shopping. To summarize, relatively tranquil markets, little reputation risk, high fees and opaque structures facilitated rating shopping within structured finance. Some of these features will self correct, others must be addressed head-on. Recent market turmoil has removed the appetite for highly complex, opaque structured products. The market for CDOs and CDOs of CDOs (“CDOs squared”) has seen its heyday and may never return in the same shape or form. The same might be said for Structured Investment Vehicles (SIVs) and many synthetic-based products, which rely on derivatives. Demand for securitizations of subprime mortgages is certainly likely to remain low for many years to come. In order to prevent rating shopping in any future structured finance business, fundamental changes must be made in the way transactions are created and marketed. As mentioned previously, banker/originators decide which rating agency can or cannot view the details of a new securitization. This prevents certain agencies from offering an unsolicited rating opinion. The power to suppress an unwanted rating opinion is at the heart of the rating shopping problem in structured finance. There must be a shift in the balance of power if rating shopping is to be contained. Specifically, any stigma associated with unsolicited ratings must be banished and biases in regulation eliminated. For example, section 3.9 of the International Organization of Securities Commissions (IOSCO) Code of Conduct Fundamentals for Credit Rating Agencies (CRAs) states: For each rating, the CRA should disclose whether the issuer participated in the rating process. Each rating not initiated at the request of the issuer should be identified as such. The CRA should also disclose its policies and procedures regarding unsolicited ratings. The implication is that unsolicited ratings are necessarily inferior to those solicited and paid for by the issuer. Paragraph 108 of the Basel Committee on Bank Supervision’s International Convergence of Capital Measurement and Capital Standards (Basel II) raises further concerns when discussing External Credit Assessment Institutions (ECAIs): As a general rule, banks should use solicited ratings from eligible ECAIs. National supervisory authorities may, however, allow banks to use unsolicited ratings in the same way as solicited ratings. However, there may be the potential for ECAIs to use unsolicited ratings to put pressure on entities to obtain solicited ratings. Such behaviour, when identified, should cause supervisors to consider whether to continue recognising such ECAIs as eligible for capital adequacy purposes. Although ostensibly addressing potentially anticompetitive practices, intense pressure from European corporate issuers was likely the motivation for this guidance.18 Competitive forces will not improve the quality of ratings without the ability to offer an unsolicited opinion. For their part, rating agencies must not use unsolicited ratings in anticompetitive ways. Anticompetitive behavior might occur, for example, if a major rating agency were to enter a new market and “dump” free ratings in order to drive out potential new entrants. Regulatory authorities must insist that future structured finance transactions be sufficiently transparent in their structure and in the details of the underlying collateral that any rating agency may offer a credible opinion, whether it was selected by the originator or not. Rating agencies must publish and abide by transparent methodologies for rating structured securities. They must provide these for every asset class and each must meet certain minimum criteria. The overriding principle is that an outside party following the methodology should be able to conclude to the rating (or, more specifically, support level) reached by a rating committee. At most, any deviation would be minor. If the rating methodology is expressed in terms of a model, that model should be available to all (for free) so that any sufficiently competent user can replicate the rating/support outcome. In those instances where a rating agency deviates from its methodology, it must explain why. For existing transactions, there should be adequate public disclosure of the underlying asset pool performance and sufficient disclosure of the parameters and thresholds that might lead to a rating change. This assumes that transactions are indeed monitored once completed. It does not mean, however, that rating agencies should operate in a totally mechanical fashion. There must be room for human judgment and effort should be made to encourage forward-looking criteria. But the emphasis should, first and foremost, be on transparency. Even if market forces do not render them extinct going forward, the rating of complex structures should be avoided or prohibited. Complexity is attractive to rating agencies because fees can be much higher than those for simpler, generic securitizations. Complexity, however, is anathema to transparency, and thus opens the gate to rating shopping. These suggestions are meant to assure the survival of the rating industry, not to drive it into poverty. Unless the incentive to shop ratings is removed, the industry risks obsolescence. Alternative risk measures will emerge and gain currency. While this may happen on its own accord, better ratings may forestall such an outcome.

Conclusion
Rating agencies perform a valuable social service. Their opinions can help improve the efficiency of capital markets. Conflicts inherent in the issuer-pays business model have instead contributed to faulty ratings for many structured finance securities. Increased competition alone will not fix the problem. Rather, fundamental changes must be made in the way structured finance securities are created and marketed. Increased transparency in the structure and performance of individual transactions, along with increased transparency in rating agency methodologies, will allow investors and rating agency competitors to assure standards are being met. In order for competition to succeed, biases against unsolicited ratings must fall.