Introduction

On July 18th, 2007, while referring to adjustable rate mortgages (ARM) (also known as subprime mortgages) bonds, an executive of the Fitch’s residential mortgage group said “We continue to be confident that “AAA” ratings reflect the high credit quality of those bonds”. Since then, between 2008 and 2009, 140 US banks declared bankruptcy while the International Monetary Fund now estimates world banks’ global losses due to loans and credit derivatives to approximate $4.1 trillion. If the subprime crisis has been the crisis of credit, it has also been the crisis of credit rating.

Credit Rating Agencies (CRAs) (namely the tree major ones: Fitch Ratings, Moody’s Investors Service and Standard & Poor’s) have been under a lot of criticism in the recent credit crisis. Indeed, not only have CRAs been accused of making errors of judgment in rating structured debt securities, but also of operating a biased business model in an oligopolistic market.

As a matter of fact, bond issuers, government regulators and investors have now lost their blind faith in credit ratings and therefore feel the compelling need to change, reorganize and restructure the CRA current business model and industry. Even though CRAs cannot be considered the sole responsible agent for the credit crisis, they have encountered great irregularities and problems: How can they be fixed? What solutions should be implemented to prevent the next credit crisis from happening? How has the credit and CRA crisis affected the leveraged finance industry?

To tackle this question, we will first analyze what criticisms credit rating agencies have been subject to and what problems have been identified in the recent years. By evaluating different solutions and suggesting necessary changes, we will then examine how the credit rating business model and market structure could be improved. Finally, as it directly relates to the credit market and CRAs, we will study the impact of the crisis on the leveraged finance industry, with a special focus on leverage buyouts, buyout debt financing and structured finance.

Section 1: Current problems of the Credit Rating Agencies’ business model

Though many other players, such as lenders, borrowers, regulators, issuers, and macro factors, can be associated with and blamed for the current credit crunch, Credit Rating Agencies (CRAs) have been accused of being the main actors behind the malfunctioning and mispricing of the credit markets. Not only have CRAs been blamed for misrating complex structured debt products[1] and other subprime mortgage related products, but also of operating a biased business model in an oligopolistic market.

In this first section, we will summarize these three main accusations and analyze in detail the validity of each argument. Solid and pertinent recommendations can only be made if the true problems have been identified.

By analysing Moody’s financial statements, we can observe that between 2002 and 2006, Moody’s profits nearly tripled because of the growth of structured products, accounting for more than 40% of its total revenues in 2006, and the higher margins charged for these products.  Given the revenues generated, one would expect that CRAs did control the rating of these products. Now, after the default rate on adjustable rate mortgages (ARMs) reached its peak during the crisis and collateralized debt obligations (CDOs) became worthless, CRAs defended themselves by explaining how sophisticated these products were and how hard it was to rate them. This leads us to question, did CRAs rate products they did not understand?

Before the mortgage market collapse, analysts like John Paulson expressed incredulity at what appeared to be a complete mispricing of the structured debt products and began predicting that the market would crash:

“For me it was so obvious that these securities were completely mispriced and we were living in a casino. I think the other players that were involved in the business got caught up in the exuberance, […] in the competition to increase their underwriting volumes, […] to increase their fees. They were very focused on annual earnings, quarterly earnings and annual bonus pools and with the amount of the liquidity, everyone got caught up in what became a massive credit bubble.” (Distressed Volatility 2009)

Mark Zandi, an economist at Moody’s, noted in a report on U.S. Macro Outlook published in May 2006, that household debt was at a record and a fifth of such debt was classified as subprime. Unfortunately, the economic forecasting division is separate from the ratings division of the corporation. But how could CRAs not foresee the crisis and the flaws of their valuation models?

The model used to rate structured products has been criticized for two reasons. First, Moody’s rating model for assessing CDOs is a statistical model reliant on historical patterns of default. The main assumption behind this model is that past data would remain relevant, even during a period in which the mortgage industry (and its related products) was undergoing drastic change. Second, the use of this model revealed “a large failure of common sense” (Lowenstein, Triple-A failure 2008)by rating agencies as very complex securities shouldn’t have been rated as plain vanilla bonds, for which the model was designed. CRAs were checking their statistical model, but not the underlying assets.

As a consequence, Moody’s noted in April 2007 that the model “was first introduced in 2002. Since then, the mortgage market has evolved considerably with the introduction of many new products and an expansion of risks associated with them” (Mason 2007) and thus revised the model it used to evaluate subprime mortgages.

Similarly, in a response letter to Roger Lowenstein’s “Triple-A failure” article, Vickie Tillman, Executive Vice President of S&P’s Rating Services claims that her company’s rating model includes both historical data and informed assumptions to assess credit quality. This adjusted model doesn’t seem to solve the accuracy problem. Deven Sharma, president of S&P, admits “[…] historical data we used and the assumptions we made significantly underestimated the severity of what has actually occurred” (Sharma 2008)

Even though one can acknowledge the greater complexity of CDOs and the difficulty of accurately assessing the risk profile of these products, the CRAs defence doesn’t seem justifiable given the source of wealth these structure products represents to them. One would expect that CRAs would only provide a service they understood. There is still plenty of room for improvement in their models. Research led by Skreta and Veldkamp (Skreta and Veldkamp 2009) suggests that the complexity of any given asset hasn’t increased but rather that the more complex types of assets became more prevalent. Indeed, when combined with the phenomenon of rating shopping, where issuers shop from one CRA to another to pick the best rating possible, asset complexity can lead to rating inflation and biased judgment. As a consequence, failure to address potential sources of bias inherent in the business model of the ratings industry could generate future problems. This discussion leads us to the conflict of interest inherent in the issuer-pay model, the second main accusation in our analysis.

The conflict of interest between CRAs and bond issuers has been identified as the main problem because it drives the entire CRA business model. This conflict of interest between rating agencies and the bond issuers from whom they receive fees undermines the CRAs ability to give an unbiased assessment of credit risk.

There are two types of potential conflicts of interest inherent in the issuer-pay model that may arise from the activities of the CRAs. The first is that rating agencies may be enticed to give better ratings in order to continue receiving service fees. Since CRAs’ revenues come from issuers, this conflict can lead to an agency problem. The second potential conflict relates to the consulting services CRAs provide to help the issuer to better design products to meet their model’s different thresholds. In both cases, CRAs run the risk of the issuer going to a different rating agency, which leads to the phenomenon of ratings shopping.

Up until the 1970s, the investor-pay business model of credit rating agencies was straightforward: investors bought a subscription to receive ratings. It was during the 1970s that the business model evolved into an issuer-initiated ratings system where the issuers of securities began paying to be rated. Free riding by investors, leading to a reduction in profits for credit rating firms, was the main reason for this transition. As White (White 2002)observes, this shift also coincided with the rise in popularity of the photocopying machine. Although the issuer-pay business model has been around for more than forty years now, concern over ratings bias only recently emerged. Indeed, the conflict of interest, amplified by the rise of complex structured financial products, calls into question the objectivity of ratings that are critical to the efficiency of the market. (Levitt, Conflicts and the Credit Crunch 2007)

In response to these accusations, CRA executives have maintained that the issuer pay model is not contradictory to the efficiency of their business model. It seems that a firm cannot support both issuers and investors simultaneously. In fact, the Report of the Staff to the Senate Committee on Governmental Affairs during the Enron case[2] cited empirical evidence:

“The conflict appears to be particularly acute for large important issues such as […] Enron […]. In these cases investors desperately need guidance from credit rating firms, but often do not get it because of pressure from issuers, […] and in some cases, SEC officials”. (Egan and Jones 2010)

However, CRA executives have also asserted that CRAs have nothing to benefit from adjusting their ratings to their client’s needs because they have a reputation to uphold. In June 2007, S&P claimed that “reputation is more important than revenues” (Becker and Milbourn 2009) thus asserting that maintaining a good reputation had been a sufficiently strong motivating factor for CRAs to keep their high levels of efficiency and objectivity. In reference to this assertion we can ask ourselves: is reputation a sufficient motivating factor to maintain discipline among rating agencies?

As a matter of fact, research led by Mathis, McAndrews and Rochet (Mathis, McAndrews and Rochet, Rating the Raters: Are Reputation Concerns Powerful Enough to Discipline Rating Agencies? 2009)has suggested that this argument is only valid when a large fraction of the CRA revenues comes from other sources than the rating of complex products. When the reputation of a CRA is good enough, and rating complex products become a large source of revenues (more than 40% of Moody’s revenues), the CRA will become too lax and inflate its ratings. This mechanism builds on a three-step reputation cycle, ultimately resulting in crises of confidence where a single default provokes a complete loss of reputation by the CRA. First, the CRA tries to build and improve its reputation and gain investor’s trust by being very strict. Then, once a positive reputation has been gained, the CRA issues more ratings and takes advantage of its reputation. This is when CRAs become more lax and the risk of default increases.  Ultimately, when default occurs, there is a crisis of confidence: the “opportunistic” CRA is detected and its reputation is very negatively affected. This reputation cycle, which is also a confidence cycle, explains why opportunistic CRA are hard to spot and why ratings biases only recently emerged as a concern in response to inquiries from Vailiki Sketra (Sketra and Veldkamp 2009).[3] To exemplify this concept of reputation cycle, scholars find that CRAs are more likely to understate credit risk in booms than in recessions (Bolton, Freixa and Shapiro, The Credit Ratings Game 2009).

Moreover, reputation seems greatly affected by competition, as it will reduce the effectiveness of the reputational mechanism for two main reasons.  First, reputation is only valuable if there are future producer rents. As a result, the incentive for maintaining a good reputation is reduced by competition. Second, from a microeconomical approach, if the demand elasticity facing individual sellers is higher in a competitive market, the temptation to either reduce prices or otherwise attract business may be stronger which undermines the quality of output. Therefore, the conflict of interest is not solved by reputation concerns.

The second aspect of the conflict of interest relates to the collaboration between CRAs and issuers when designing a debt security. Lewis Ranieri, a pioneer in the mortgage bonds market, once said “The whole creation of mortgage securities was involved with a rating” (Norberg 2009). As a consequence, starting in the 1990’s, CRAs started to offer consulting and advisory services to issuers to improve their ratings; a process that involves extended consultations between the agency and its client. The collaborative process that ensues is as follows: issuers propose a rating structure on a pool of debt. Then, the CRA will usually request a cushion of extra capital, known as an “enhancement,” to meet the necessary conditions for a specific rating. This practice can be dangerous because it is the CRA’s responsibility to ensure that the cushion is big enough to safeguard the product, but issuers will try to minimize this extra capitalization in order to maximize their profit margin. Inside the CRAs, consultants and raters were meant to be strictly separated by a ‘Chinese wall'[4]. Regardless, CRAs (namely Moody’s) began providing unsolicited ratings and offering consultancy services to improve them. Mr. Arthur Levitt, a former chairman of the Securities and Exchange Commission, pointed out in a recent article in the Wall Street journal that the conflicts of interest arising from such activities are the central problems with CRAs:

“[Credit rating agencies] are playing both coach and referee in the debt game. They rate companies and issuers that pay them for that service. And, in the case of structured financial instruments, which make it possible to securitize all those subprime mortgages, they help issuers construct these products to obtain the highest possible rating. These conflicts are hard to spot because transparency among these agencies is murky at best, and currently it is difficult to hold these agencies accountable for any wrongdoing” (Levitt, Conflicts and the Credit Crunch 2007)

The agencies are aware of the conflicts that are inherent to their business model but they claim that they are doing their best as to avoid them. In a letter to Roger Lowenstein’s “Triple-A failure” article, Vickie Tillman, Executive Vice President of S&P’s Rating Services defends her company’s business models and practices:

“At Standard & Poor’s, we recognize the business model we use may raise potential conflicts of interest. That’s why we have always had rigorous policies in place to manage conflicts, and why we currently are implementing additional measures to further strengthen the independence and quality of our ratings opinions. […] the role ratings firms play in the market […] is to provide independent assessments of the creditworthiness of bonds.”             

In order to make up for these practices, the US Securities and Exchange Commission (SEC) issued a release in February 2007 proposing rules which would identify the issue of unsolicited credit ratings (those not issuer-initiated), as unfair, coercive, or abusive, and thus would prohibit Nationally Recognized Statistical Rating Organizations (NRSROs) from releasing unsolicited credit ratings.  Even though the SEC intervention seemed necessary, it didn’t change the industry’s business model: by 2007, the mortgage boom had already reached its peak.

Regardless of the criticism surrounding the relationship between issuers and rating agencies, the fact of the matter is that they were simply bringing bonds to market based on market demand, which clearly indicates a crisis of the issuer-based model. CRAs’ misbehaviour has played a central role in the current subprime mortgage crisis. As such, the governments and regulatory bodies should take steps forward to correct the current business model. We shall therefore investigate alternatives to this model in Section 2 of this paper.

This conflict of interest leads us to ask, who finally owns the ratings? The evidence regarding whether rating agencies bend to the issuer’s will is mixed.

A paper written by contract-theory scholars, Faure-Grimaud, Peyrache and Quesada (Faure-Grimaud, Peyrache and Quesada 2007) investigates this issue by looking at corporate governance ratings in a market with truthful CRAs and rational investors. They show that at equilibrium, in a monopoly, a CRA will fully disclose information but that issuers may prefer to suppress their ratings if they are too noisy because full disclosure is impossible even when firms have the possibility for ownership (i.e., the right to disclose the rating). Additionally, they find that competition between rating agencies can result in less information disclosure since CRAs make zero profit and fully disclose information on firms that have values higher than the CRAs’ marginal observation cost.

In fact, the current business model seems to favour the banks in their quest to receive better ratings. Dr. Joseph Mason compared default rates for corporate bonds to equally BAA-rated CDOs before the bubble burst and found that the CDOs defaulted more than ten times as often (Mason 2007). While, as we discussed earlier, it may be true that CDOs are much more complex securities than plain-vanilla bonds, another interpretation of the data is that CRAs were much more lax when dealing with a Wall Street securitizer as client. But who can blame them? While it is true that on the traditional side of the business (unsophisticated bond rating) CRAs have a large variety of clients (virtually every corporation and municipality that issues public debt), this is not the case in structured finance. On the contrary, the panel of clients is much smaller and the fees are much bigger. The only issue is that the client pays only if the CRA delivers the desired rating. If they do not, the client can either adjust the numbers or take another chance with a competitor, a process known as “ratings shopping”.

Brian Clarkson, former president and CEO of Moody’s Investor’s Service acknowledged, “There is a lot of rating shopping that goes on. What the market doesn’t know is who’s seen certain transactions but wasn’t hired to rate those deals” (Bolton, Freixa and Shapiro, The Credit Ratings Game 2009). In fact, an important feature of the credit ratings market microstructure is the capacity for a security issuer to choose which ratings to purchase. During this process, a structured debt product is issued and the issuer typically proposes a structure to a CRA. The issuer then asks for a “shadow rating,” which remains private between the CRA and the issuer, unless the issuer pays to make the rating official.

Such choices can reflect both explicit and implicit shopping for desired credit reviews and induce a selection effect in the rating process. Selection highlights the relation between the decision about whether to rely on unsolicited ratings and the potential for ratings shopping, illustrating how different types of potential conflicts of interest in the credit rating process could interact. Indeed, shopping for ratings is a practice at the heart of the different conflicts of interest we mentioned above, as it partly invalidates the reputation argument because there seems to be a trade-off between reputation concerns and the risk for ratings shopping. It also encourages CRAs to strengthen their ties and relationship with issuers, most notably by offering a wider range of services. In an interesting paper, Skreta and Veldkamp (Sketra and Veldkamp 2009) examine cherry-picking in ratings, especially for securitization, by issuers who shop for the highest ratings in order to obtain the highest price when selling to naive or little-informed investors. They highlight the influence of risk aversion in motivating the purchase of multiple ratings. Indeed, because investors are risk-averse, they will try to invest in the best-rated securities for an expected yield without having to asses the risk of every security they may be interested in, and thus rely heavily on ratings. The more ratings they have for a security, the more likely they will be to invest in it. Skreta and Veldkamp (Sketra and Veldkamp 2009) conclude that when combined with asset complexity, rating shopping can lead to rating inflation and thus biased judgment. To support that evidence, Kurt Schacht, managing director of the CFA Institute Centre explained that CRA executives

“[…] were concerned about the hype and insinuation that CRAs easily inflate their ratings in response to pressure from issuers and issuers, implicating the integrity of their process and ratings. In exploring that topic, we were very surprised by the results of our member poll where some 211 of the 1,956 respondents said they have indeed witnessed a CRA change ratings in response to external pressures” (CFA Institute 2008).

As a consequence, not only does ratings shopping enhance ratings distortion, but it also corrupts the entire rating process by giving issuers an incentive to trick their clients into buying overrated securities.

A third and final issue to investigate is the lack of competition in the credit-rating industry.  According to The Economist (The Economist 2007), Moody’s and Standard & Poor’s dominated the industry by controlling about 80% of the total market in 2007. The third-place competitor, Fitch, had only about 15% of the total share that same year. The current form of these institutions received legal status when the SEC introduced the notion-barrier of the NRSROs in 1975. The rest of the market is divided among only a few other institutions that have received legal status.  While alluding to the dominance of Moody’s and Standard & Poor’s in the credit market, the U.S. Department of Justice has referred to the credit-rating industry as a “partner duopoly” (Laing 2007). As noted by Jonathan R. Laing, a partner duopoly differs from an oligopoly because the partners in the duopoly do no face fierce competition against each other because “one’s good fortune in winning a piece of business is typically followed by the other’s receiving the same deal at the same lush fee level” (Laing 2007).This duopoly has proven quite profitable, as Moody’s operating margin is typically around 50% (if not more) better than Microsoft, Accenture, Intel, Nike or Coca-Cola. In fact, according to Congressman Henry Waxman’s statement during the Congressional hearings in October 2008, Moody’s had the highest profit margin of any company of the S&P 500 index for five years in a row. An important complaint arising from this situation is that the lack of competition permits the main players “to shirk, engaging in less effort and research that if they were true active competition” (Coffee 2006). It may therefore seem that a free market would ensure competition among its CRAs guaranteeing a higher quality and lower price of the ratings. For that reason, competition from new agencies might create a healthy diversity of opinion, leading to more accurate assessments of debt issuers’ default probabilities

Many scholars have analyzed whether this industry structure contributes to the efficiency of the global credit market. We shall investigate in further detail what seems to be the optimal market structure in the next section by examining the solutions and changes necessary to combating the various issues we have so far considered.

Other scholars recognize that the existing duopoly may present risks to the market, especially since the ‘two-rating’ norm is still in full force. Furthermore, since the CRA business model is reputational-driven business, new competitors may face very high barriers to entry. The CRA industry could therefore not allow for more participants. On the other hand, some scholars suggest that the SEC’s role in both creating and perpetuating this duopoly by which establishing the status and necessary requirements to become a NRSROs, and an official registry. Since competition can both be seen as a problem and as a solution to the CRA industry and business model, we shall now examine the different initiatives that can be undertaken to improve the overall model and functioning of the credit rating market.

Section 2: Solutions to fix the identified problems

The subprime crisis has brought to light the poor performance of CRAs in rating structured financial products and reminded investors of CRA’s past poor performance in predicting the East Asian crisis and the collapse of Enron[5]. Either directly by regulations, or by market force, there are strong signals that the credit rating business is about to change. The main accusations we previously addressed and the perception that CRAs contributed to the financial crisis led to various investigations and calls for reform. In this section, after briefly presenting CRAs’ reaction to criticism, we will first analyze the different alternatives suggested by scholars and experts to the current business model and the overall industry structure. We will then study the different reforms and regulatory recommendation that have been suggested to the current business model that would improve CRAs’ effectiveness and enhance the overall market efficiency. Finally, once these changes examined, from a regulatory standpoint, we will observe the measures recently adopted by both the European Union and the US government (and regulating agencies), determine how the approaches differ and how necessary regulation is.

CRA’s reaction to accusations

CRAs have responded to the allegations with cries of innocence.  If some rating firms claimed that they did nothing wrong and have indicated that they will cooperate openly in any investigation that comes their way, others  did acknowledge some mistakes and have announced the intention to reform their practices.  For example, spokespersons for Moody’s, Standard & Poor’s and Fitch have claimed that their organizations “will demand more data and more verification and will subject their analysts to more outside checks” (Lowenstein, Triple-A failure 2008) However, some may say that CRAs might have implemented these changes simply to avoid further criticism and regulatory intervention.  Indeed, as Lowenstein claims, “none of this […] will remove the conflict of interest in the issuer-pays model” .  We shall further analyze the case for self regulation in our analysis.

In their effort to defend themselves, the CRAs have sought to minimize their role and influence within the financial industry.  According to a spokesperson for Moody’s:

“We perform a very significant but extremely limited role in the credit markets. We issue reasoned, forward-looking opinions about credit risk. […] Our opinions are objective and not tied to any recommendations to buy and sell” (Benner and Lashinsky 2007)

The consensus of these critics is that “the agencies dropped the ball by issuing investment-grade ratings on securities backed by subprime mortgages they should have known were shaky” (Benner and Lashinsky 2007)

Rather than accept responsibility for their own lack of diligence, the major CRAs have sought to “lay the blame on the mortgage holders who turned out to be deadbeats, many of whom lied to obtain their loans” (Lowenstein 2008).  Of course, it must be noted that other groups and individuals share the responsibility for the global financial downturn.  As Laing says in regard to CRAs, “they were just one link in a subprime production line that stretched from sleazy storefront mortgage brokers, corrupt appraisers and avaricious originators to fee-crazed securitizers and, yes, mendacious borrowers” (Laing 2007).  Nonetheless, as Laing further notes, CRAs must be seen as “key enablers” in the problem’s development. 

i) New agency industry structure and business model

Proposals have been made to improve the credit-rating system and thereby reduce the problems we identified. First, it seems that CRA need more independence. As Laing suggests it, many of the changes implemented in the auditing industry with the Sarbanes-Oxley Act could be similarly carried out. (Even though one may discuss whether this Act has improved capital markets transparency or not, one must note it has enforced the implementation of internal control, due diligence and transparency procedures in firms)For instance ratings agency employees should be prohibited from accepting any favors (whether it is money of gifts) from their clients and the leading analyst should rotate from a client to another with a certain frequency and should wait at least one year before joining their clients’ firm (an issuer or investment bank in this case) Laing also suggests that the 2003 SEC proposal, which prohibits the linkage of analyst compensation with new business development, could be reenacted.

First, CRAs should be more transparent in two distinctive ways. The global credit market needs greater transparency about CRAs’ overall rating model: rating assumptions, methodologies, but also the fee structures, and past performance. To be more transparent CRAs should follow stricter disclosure requirements (as mentioned in the Rating Agency Act in 2006). Professor Charles W. Calomiris (Calomiris 2009) suggests that, more disclosure could also be required for publicly traded companies with rated debt when filling in debt-offering documents Particularly, in order to prompt CRAs to reduce or eliminate their conflicts of interest, they should disclose any structuring service or consulting-related activity (and the fees related to such practices) provided to a company in connection with the rating of fixed-income securities

Second, there is a strong need, expressed by both scholars and analysts, for a clear distinction between the rating of structured products and traditional debt products and thus different rating symbols could be used so as to avoid confusion. The issue is, not all AAA-rated securities are created equally. As demonstrated in the current credit crisis and as proven by Drexel University finance professor Joseph Mason, CDOs receiving a Baa rating from Moody’s were more than ten times as likely to default as similarly rated corporate bonds (Mason 2007). As a matter of fact, despite the identical symbols, structured products typically do not have the same risk profile as traditional corporate bonds. By nature, whereas corporate default can be estimated by very few factors (namely the level of leverage of the firm and its capacity to generate stable cash flows from operations), “default on structured debt is dependent on hundreds or thousands of individual defaults [e.g., an underlying mortgage pool] that are estimated given some distribution. They are not the same analysis so they should not be the same ratings.” (CFA Institute 2008) A different rating scale according to the risk profile of the products could be used as to not mislead investors into buying misrated securities. As an alternative, Professor Coffee at Columbia University suggests the SEC could define a maximum default rate for different class of ratings, so that if a CRA’s ratings were to exceed SEC parameters, it would loose their NRSRO status. (Coffee 2006) Building on this, the entire rating nomenclature could be changed and ratings could be expressed quantitatively as to avoid grade inflation in CRAs’ opinions. Indeed, in contrast to numerical estimates (of the probability of default (PD) and loss given default (LGD)),which do have objective and quantifiable meanings, letter grades leave more room for sub