U.S. Department of the Treasury

 

Good afternoon and thank you for welcoming me to the National Economists Club.  It is an honor to have been invited.  In particular, I want to thank Brian Preslopsky;  I had the privilege of working with Brian at the Federal Reserve for several years, and it was good to hear from him when he invited me to join you today.

I want to speak to you today about credit rating agencies, a topic that I am assuming many of you, as economists in the nation’s capital, may not think about on a regular basis, at least not through the lens of your academic training or professional practice.  And you’d be right to wonder why someone in my role, whose responsibilities include monitoring markets here and around the world and overseeing the nation’s debt management, would choose to discuss this topic.  Most people in markets today are focusing on China, the Fed, the economy, and so forth.  All weighty and important subjects, to be sure.  But I submit that credit rating agencies, while not the headline grabbers of the day, present weighty and important issues as well.  Moreover, as an economist, there are some very interesting analytical aspects of what credit rating agencies do that let me re-connect with my inner nerdy economist.  Following markets day to day, it is often refreshing for me to go back to my roots. 

Apart from the critical job of managing the nation’s debt, a key responsibility of my office is to monitor and understand significant developments that affect market functioning and financial stability.  Credit rating agencies fit in to this area pretty neatly, and as I will outline in just a second, the financial reform following the crisis addressed these agencies.  While Treasury has no direct regulatory role with respect to credit ratings or the ratings industry, they sit squarely within our broader responsibilities with respect to U.S. and global financial markets policy.  If there is one message I want you to come away with today, it is this:  Although we are several years past the recent financial crisis, and five years since passage of the Dodd-Frank Act, credit rating agencies remain an important topic for discussion.

So what is it that credit rating agencies do?  Obviously, a fundamental function of credit rating agencies is to form judgments regarding the creditworthiness of issuers of debt securities and to communicate their assessments to investors in the form of ranked ratings.  To do this, credit rating agencies develop and employ both quantitative and qualitative models with public, and sometimes private, information about the entities they rate as key inputs.  Together, the 10 rating agencies registered with the Securities and Exchange Commission as nationally recognized statistical rating organizations, or NRSROs [N-R-S-R-Os], currently maintain more than 2.4 million credit ratings.[1]  More than 96 percent of these ratings are maintained by the three largest firms—Moody’s, S&P, and Fitch—though it should be noted that these three agencies rate many of the same debt securities.  Although credit rating agencies are private, for-profit entities, some of which are shareholder owned, their ratings have vast influence, affecting the cost of borrowing for public companies and government entities, which securities investors will hold, and the terms of many types of derivatives or other financial contracts.  In a word, credit rating agencies are an entrenched and central aspect of our capital markets.

But let’s take a step back and ask:  Why do credit rating agencies exist at all?  Private investors could form their own opinions and price financial instruments on their own.  Differences of opinions about the value of goods and services across the economy are essentially what define the price and quantity of each good and service that is produced.  At least that view is the introductory economics textbook version of the world.  One of the biggest revolutions in economic theory just as I was beginning to study economics, is the economics of information and what happens when information is imperfect—either asymmetric or costly to obtain. In that context, it is reasonable to argue that ratings agencies exist to, serve a few useful market functions.  One such function is to bridge the information gap between issuers and investors.   Issuers want to borrow at low cost and have full information about their true creditworthiness. Investors want to lend and earn a high return on their capital, but as outsiders will have incomplete information about the riskiness of a potential borrower. 

A natural model to think of is Akerlof’s well-known “lemons” problem.  In that model, only the average quality of a product is known to buyers, so they are unwilling to pay more than what is appropriate for the average.  Sellers with products that are above average in quality, withdraw from the market.  But this withdrawal further lowers the average quality, and one can easily imagine an equilibrium where the market does not function at all.  Adverse selection at its finest.  In the context of credit markets, given information asymmetry, investors demand higher rates of return because of the lower-quality, and therefore riskier, borrowers. Higher- quality borrowers might avoid the market because of the higher borrowing costs.  The result, in theory, is an adverse selection outcome in which only low quality borrowers, or “lemons,” are left in the market.  However, in reality, high quality borrowers are not categorically driven out of the credit markets.  A very related model of adverse selection for credit markets is Stiglitz and Weiss’s seminal model on credit rationing.  The same general mechanisms are at play, and the result of the imperfect information is a supply of credit that may be below the socially optimal amount. 

By producing independent judgments of issuers’ creditworthiness, rating agencies can contribute to market efficiency by helping to mitigate the information asymmetry between issuers and investors.  This effect is probably greater the more that rating agencies are able to incorporate non-public information into their assessments.  Alternately, if information is not necessarily asymmetric, just costly to obtain, efficiency gains could be greater the more costly it is to acquire information.     

Another often cited reason for the existence of credit rating agencies is due to their role as regulatory gatekeepers.  In 1975, the SEC created the NRSRO designation for select rating agencies in connection with its authority under the securities laws to set capital rules for broker-dealers.  Specifically, under the SEC’s net capital rule broker-dealers were permitted to rely on credit ratings issued by designated NRSROs to determine how much capital they had to hold against different debt securities.  Although the NRSRO designation was originally established for this narrow purpose, over time, reliance on the NRSROs’ credit ratings became widespread as references and requirements to use their ratings proliferated in federal and state law, and in rules issued by financial and other regulators.  Many other countries also adopted similar references to credit ratings in their national laws. 

This regulatory use of ratings arguably benefitted both financial institutions and regulators by reducing the costs of regulatory compliance and monitoring, respectively.  Institutionalizing NRSROs, and the resulting quasi-regulatory license they enjoyed, doubtlessly did much to entrench credit rating agencies into U.S. and global financial markets.  And indeed, three of the original NRSROs—Moody’s, S&P and Fitch—continue to dominate the global market for ratings today.  But rating agencies have existed in one form or another for more than 100 years, far pre-dating the introduction of the NRSRO designation and regulatory use of credit ratings, and strongly suggesting a market rationale for rating agencies independent of regulation.  So the regulatory license explanation for why we have rating agencies cannot be the whole story. 

In short, neither of the foregoing explanations, which I have only barely outlined here, fully accounts for the existence of credit rating agencies.  Certainly there are additional theories rationalizing their potential to add value to the capital markets, and the reality is probably some combination of proffered explanations.  But fundamentally, credit rating agencies exist to determine and issue ratings of creditworthiness with respect to bonds and other debt instruments offered by public companies, banks, governments, and other issuers.  If their ratings are objective and accurate, the theoretical explanations for their existence suggest that society is better off.  If however the ratings prove not to be objective and accurate, one would have to consider the ramifications.   

History has shown, of course, that ratings agencies can and do make mistakes.  Credit ratings are susceptible to flawed analysis and undue influences, and the confidence investors, financial institutions, and other market participants place in the rating agencies’ services can deteriorate.   The recent financial crisis demonstrated that the effects of these failures on the markets can be sudden, severe and destabilizing.  This possibility suggests that there are serious limits to what credit ratings and credit rating agencies can do in terms of improving market efficiency, and that appropriate oversight and accountability is warranted to minimize the risk that rating agencies will inadvertently harm the markets.

Although credit ratings are meant to communicate a specific borrower’s probability of default, they are often a product of backward-looking methodologies and involve more than a small amount of subjectivity.  This fact is partly a feasibility issue—rating analysts, for example, cannot observe all aspects of the entities they rate.  It is also a statistical issue—there is a limited history of past default data on which to base quantitative models.  To be fair, though, as Niels Bohr the famous physicist, the recently deceased Yogi Berra, and various others have been credited for saying, “forecasting is hard, especially about the future,” and the rating agencies have in recent years become better at making their methodologies more transparent. 

Getting back to the economics of information, an important theme to keep in mind in any discussion of ratings agencies are principal-agent problems.  These issues are fairly well understood in many contexts, but I am repeatedly surprised throughout my career at how often I am teleported back to my micro-theory classes and principal-agent problems when I think about how markets work (or don’t).  The obvious issue is, “for whom does the rating agency work?”  The issuer or the investor?  We will talk about that question in a bit.  But even if we could resolve that question, the ratings agencies themselves are commercial enterprises with commercial interest.  A potentially counterintuitive result is the fact that some recent research suggests that ratings agencies may have become overly conservative in ratings, in part to protect their reputations. 

A more serious—and perhaps more intractable—problem, however, is that credit rating agencies face certain inherent conflicts of interest.  Some conflicts of interest, such as rating an entity in which the credit rating agency has a direct ownership interest, are prohibited outright, and given effective oversight there is some assurance that credit rating agencies are avoiding such conflicts.  However, there are other conflicts of interest that the securities laws and regulations permit rating agencies to have, as long as they are properly disclosed and managed.  One such conflict of interest that presents special challenges is the business model used by most of the 10 registered NRSROs, namely the issuer-pays compensation model.  This issue is the first principal-agent problem I mentioned.

Under this model, the issuers of rated securities hire and pay one or more credit rating agencies for a rating.  The potential for conflict arises to the extent that a credit rating agency, which presumably wants to be hired for repeat business in the future, has an incentive to issue a higher rating than is otherwise warranted in order to satisfy the issuer’s economic interest in lowering its cost of borrowing.  A troubling manifestation of the issuer-pays conflict has been observed in connection with structured finance products.  In a so-called “ratings shopping” scenario, an issuer has several credit rating agencies analyze a prospective structured finance instrument and hires the one or two rating agencies that deliver a higher rating.  In a similar vein, some rating agencies have advised issuers how to structure securities that the agencies were later hired to rate. 

These and other factors came to a head in the recent financial crisis.  In its March 2008 Policy Statement on Financial Market Developments, for example, the President’s Working Group on Financial Markets detailed how during the upswing of the housing boom, faulty assumptions underlying rating methodologies and conflicts of interest led the credit rating agencies to issue top-tier AAA ratings on many residential mortgage-backed securities and other structured finance products, only to later downgrade most of them—including many recent issues—to levels often below investment grade status. In the face of these swift and severe downgrades, investors predictably lost confidence in the accuracy of credit ratings and, as a result, many structured finance markets seized up or suffered significant contractions.  There was also a breakdown in market discipline with investors relying excessively on credit rating agencies.  That fact raises for me another interesting opportunity to think about economic models of imperfect information.  Bikhchandani, Hirshleifer, and Welch’s model of information cascades provides one lens.  In that model, fully rational Bayesian agents can be led to reject their own private information and take on board public information.   I am not arguing that this mechanism fully explains what we saw, but rather, it provides another example of why investors doing their own due diligence is critical but also why it may be difficult to achieve in practice.

Of course, there were many complex and inter-related factors that led to the financial crisis.  I do not intend to talk about much of the Dodd-Frank Act.  There was, however, a part of the act that was aimed at responding to the rating agencies’ role in the financial crisis.  Congress concluded, for example, that credit rating agencies have systemic importance and therefore are matters of national public interest; that they play a “gatekeeper” role which justifies a higher level of public oversight and accountability; and that they are fundamentally commercial in character, thus providing a rationale for similar standards of liability and oversight that apply to auditors, securities analysts, and other market intermediaries.  Congress also found that credit rating agencies faced conflicts of interest and that their ratings, especially for structured finance products, proved to be inaccurate, which contributed to the mismanagement of risks by financial institutions and investors.

To address these findings, Congress included numerous provisions in Dodd-Frank to enhance the accountability and oversight of credit rating agencies, mitigate their conflicts of interest, and provide greater transparency over their ratings and operations.  Some of these reforms were self-executing through amendments to existing securities laws, but many more relied upon enhanced rulemaking and oversight authority granted to the SEC.  The SEC was also directed to establish an Office of Credit Ratings to develop and administer NRSRO rules, conduct annual examinations, and to prepare and publish annual NRSRO examination reports. Dodd-Frank also mandated several studies relating to credit rating agencies and the use of ratings. 

I should note that Congress also acknowledged the role played by government in helping cement credit rating agencies so centrally in our financial markets.  As previously noted, references to ratings proliferated over the course of many years in everything from regulatory capital requirements, to federal agency loan agreements, to rules stipulating what types of investments regulated entities may hold.  Dodd-Frank confronted this reliance directly and required that references to credit ratings be removed from federal laws and regulations, and that alternative measures of creditworthiness be used in their place. 

It is now just more than five years since President Obama signed Dodd-Frank into law.  To date, the SEC has completed all of the act’s major reports and rulemakings relating to oversight of credit rating agencies and NRSROs.  In August 2014, in one of its latest major actions, the SEC adopted a final, comprehensive set of NRSRO rules, and as of June 2015, all of these rules are in effect.  And there are indications that the rating agencies are taking these reforms seriously.  Even before it adopted final rules, the SEC acknowledged that the credit rating agencies were beginning to put certain reform provisions into place, such as those related to internal controls, disclosures on performance of credit ratings, policies and procedures for ratings methodologies, and standards of training and competency for credit rating analysts.

The SEC and other federal agencies have also made substantial progress in removing references to credit ratings from their rules and regulations.  Earlier this month [on September 16] in fact, the SEC approved the removal credit ratings references in its rules governing money market funds.  With this action, the SEC has removed references to credit ratings from 32 of its rules and forms. 

 

In addition, we are beginning to see greater competition in the credit rating agency space.  Although the three largest rating agencies continue to dominate the market, several smaller agencies are beginning to attract meaningful business in certain niche areas of the credit markets allowing them to grow and become more viable.  Kroll Bond Rating Agency [and Morningstar Credit Ratings], for example, have seen a growing market share in ratings for commercial mortgage backed securities. 

Despite these positive developments, as memories of the crisis fade, I think the thorny theoretical aspects I have laid out make it clear that we cannot become complacent.  The annual examinations of NRSROs by the SEC’s Office of Credit Ratings, as documented in their annual examination reports, continue to identify instances of poor practices and operational deficiencies among both large and small rating agencies.  Many of these problems relate to mismanagement of conflicts of interest and problems with rating agencies’ internal controls and governance procedures.  Although the SEC’s reports also note progress by the rating agencies in addressing previously identified trouble spots, clearly there remains room for improvement.  Meanwhile, it seems clear to me that investors still rely heavily on credit ratings.  Whether this fact is good or bad depends in great part on the theoretical questions I posed.  Are the agencies providing unbiased and accurate information that reduces market inefficiencies?  Are the principal-agent problems less severe than the efficiency gains?  Does the possibility of an information cascade remain, and does that pose risk?  I think all of these questions are important, and are most keenly understood through the lens of microeconomic theory. 

The steps taken by the SEC and the federal government have been constructive in pushing forward improvements in how ratings are developed and used.  I said at the top of my remarks that many of you may not have thought about credit rating agencies.  The fact is, though, there is extensive academic literature on credit ratings and the credit ratings industry.  I want to challenge you today to keep the focus on credit rating agencies, not only by expanding the theoretical literature for those of you that publish research, but also by developing and advancing additional real-world solutions to any of the remaining problems that contributed to the financial crisis. 

As I noted above, the issuer-pays business model and its attendant conflicts of interest is perhaps one of the most intractable problems in the credit ratings space.  This problem was indeed highlighted in a provision of the Dodd-Frank Act, known as the Franken Amendment, which directed the SEC to study the feasibility of establishing a system in which a public or private utility or a self-regulatory organization assigns NRSROs to determine credit ratings for structured finance products.  In a report issued in December 2012, the SEC considered several potential alternative compensation models and evaluated them in terms of how they measured up to certain desirable criteria such as independence, feasibility, accountability, transparency, and market acceptance, among others.  To be sure, it is unlikely that any alternative compensation model will be completely free of the potential for conflicts of interest of one degree or another.  But that doesn’t mean it’s impossible to identify and implement a system that might better align the interests of issuers, rating agencies, and investors.

One proposed solution that I find intriguing but have not come to conclusions about, is a combined issuer-and-investor pay model, or more precisely, a market-pay model.  Fundamentally, credit ratings present a principal-agent problem between the users of ratings and the rating agencies who determine the ratings.  The issuer-pays model, as we have seen, aligns the interests of the rating agencies and the issuers of rated debt securities; it cannot align the interests of rating agencies and investors.  The principal in the rating agency context is, broadly speaking, the market.  Therefore, there is an argument to be made that the market broadly writ should pay in order to properly align all stakeholders’ incentives.  A market-pay model could also eliminate the potential for “ratings shopping” and might lead to improved quality and accountability of ratings.

Admittedly, overhauling the compensation model of the credit ratings industry would be an enormous undertaking and it would require the buy-in and sustained commitment of stakeholders on all sides.  There are, however, other potential reforms that have been considered and are worth analyzing.  For example, the time may have come for users of ratings to call for a new or enhanced ratings framework.  Rather than a single primary rating to grade debt instruments, separate ratings for probability of default and loss given default could help investors better evaluate credit risk.  More generally, market participants need to continue to work together to search for effective alternatives to credit ratings.  Ultimately, investors and other users of credit ratings need to develop internal or alternative capabilities to assess the creditworthiness of issuers and their securities.  While credit ratings have a place in investors’ analyses, thinking of information cascades should caution us against these ratings being the sole determinant of investor decisions.

In closing, the market should not simply wait for the government to create additional practices that enhance accountability or better align the interests of investors, rating agencies, and issuers.  Or worse, to wait for the next financial crisis to consider what additional reforms are warranted.  As economists, you have unique insights and skills that can help to drive further positive outcomes in ratings reform, and I urge you direct your energies toward this end. 

Thank you.

 

 

---