Guest post by Lawrence J. White, Robert Kavesh Professor in Economics, NYU Stern School of Business
The major credit rating agencies (CRAs)—Moody’s, Standard & Poor’s (S&P), and Fitch—contributed significantly to the financial crisis of 2007-09. Their excessively high initial ratings of residential mortgage-backed securities (RMBS) helped fuel the bubble of mortgage finance that ultimately burst, with near catastrophic consequences for the U.S. financial sector.
These disastrous failings motivated the post-crisis urge to tighten regulation of the CRAs. It’s not hard to share the (metaphorical) desire—reflected in the Dodd-Frank Act of 2010—to grab them by the lapels and shout “Do a better job!”
There is, however, a better way, albeit one that is less intuitive and possibly less gratifying: namely, eliminate—or at least greatly reduce—the regulation of the CRAs. This would encourage entry into the credit rating business, stimulate innovation and, eventually, improve the efficiency of capital markets. Surprisingly, the Financial CHOICE Act, the bill to reform Dodd-Frank that was passed by the U.S. House of Representatives last month, did not seize the opportunity to roll back the largely ineffective regulation that Dodd-Frank applied to the CRAs.
To understand the argument for less regulation of the CRAs, let’s start with what they do: CRAs provide opinions about the creditworthiness of bonds that have been issued by corporations, municipalities, sovereign governments, and the like. Such information provides the foundation for the operation of any credit market, where lenders naturally want to know about the creditworthiness of borrowers (but the lenders may be too small or lack the expertise to collect or analyze data themselves), and the more creditworthy borrowers want to have a credible means of conveying their creditworthiness. Put differently, the fixed costs of information gathering and analysis lead many investors to outsource this professional service.
The three large CRAs are the most prominent providers of this information—but they are not the only sources. There are also: smaller CRAs; smaller firms that may not call themselves CRAs but that provide similar information to clients; “fixed income analysts” employed by securities firms who provide such information to investor-clients (as well as to the securities traders for these firms); and similar personnel who are employed for in-house analyses by large banks, insurance companies, pension funds, and other significant investors in bonds.
A few dates are important in the history of CRAs. The first is 1909, when John Moody created the first publicly available ratings for bonds. Since the ratings were published in “rating manuals,” Moody’s business model was that of “investor pays.” Moody was followed by other rating firms over the next two decades, all of which had the publishing business model.
The next major event came in 1936, when federal prudential bank regulators mandated that U.S. banks—if they were going to invest in bonds—could hold only “investment grade” bonds. The determination of what was an investment-grade bond (a rating of BBB- or better, in the modern S&P system) was left entirely to the major CRAs.
This regulatory change had two major consequences: First, it meant that prudential regulators had “outsourced” a part of their regulatory process to third parties; and, second, it guaranteed an audience for the major CRAs’ ratings, since banks would need to know the status of any bonds that they held or might want to hold (and the other participants in the bond market would also want to have this information when transacting with the banks).
Over the next few decades, state regulators of insurance companies followed suit, and started to rely on the major CRAs’ bond ratings. In the 1970s, the Department of Labor (DOL) adopted similar rules for corporate defined-benefit pension plans, and the Securities and Exchange Commission (SEC) began to make use of the CRAs’ ratings for capital requirements for broker-dealers. Then, in the early 1990s, the SEC extended similar requirements for the bond holdings of money market mutual funds. All of these actions, of course, enhanced the importance and stature of the major CRAs.
In developing its rules for the broker-dealers in 1975, the SEC realized that the rules were potentially vague as to which CRAs would be eligible to create the ratings of the bonds held by broker-dealers. To eliminate any ambiguity, the SEC created a category—“nationally recognized statistical rating organizations” (NRSROs)—for specifying which CRAs’ ratings were relevant; and the SEC immediately designated Moody’s, S&P, and Fitch as NRSROs. Other regulators that used ratings soon adopted the NRSRO system.
Over the next 25 years the SEC became a substantial barrier to entry: The agency designated only four additional CRAs as NRSROs; but mergers among the new four and with Fitch meant that as of year-end 2000 the number and identity of NRSROs was back down to the original three. Also, the SEC was quite opaque regarding the criteria it used to designate new NRSROs.
In the aftermath of the prominent bankruptcy of the Enron Corporation in November 2001, financial journalists discovered that the three major CRAs/NRSROs had maintained investment-grade ratings for Enron’s bonds until five days before its bankruptcy. In subsequent Congressional hearings, SEC officials were asked about the NRSRO system, how the SEC had managed it, and why there were only three NRSROs.
In response to this pressure, in the next two years the SEC certified two additional NRSROs. But the Congress, still unhappy, enacted legislation in 2006 that established criteria that the SEC should use in designating new NRSROs and gave the SEC information-gathering powers and limited oversight powers vis-à-vis the NRSROs. The SEC “got the hint” and subsequently designated more NRSROs, so that today there are 10.
Tables 1 and 2 provide end-2015 data for the NRSROs’ absolute and relative numbers of outstanding ratings and credit analysts. The three large CRAs remain dominant in all categories. No other NRSRO attains even a double-digit percentage of the ratings or employment in any category.
Table 1: Number of NRSRO Credit Ratings Outstanding and Credit Analysts Employed (Ranked by Total Ratings), 2015
Source: U.S. Securities and Exchange Commission, Annual Report on Nationally Recognized Securities Rating Organizations (Dec. 2016).
Table 2: Percentage of NRSRO Credit Ratings Outstanding and Credit Analysts Employed (Ranked by Total Ratings), 2015
Source: U.S. Securities and Exchange Commission, Annual Report on Nationally Recognized Securities Rating Organizations (Dec. 2016).
There is one final pre-financial-crisis “event” that is important to mention: in the late 1960s and early 1970s, the major CRAs shifted their business models from the original investor-pays to an “issuer-pays” regime. That is, instead of John Moody’s arrangement where the investor purchased the ratings book, the CRAs adopted a model in which the issuer of the bonds pays a fee for its bond to be rated. The reasons for this switch were twofold: first, the CRAs appear to have been concerned that high-speed photocopying would lead to free-riding by substantial numbers of subscribers (much as digital copying damaged the recorded music business in the late 1990s). Second, the CRAs may have realized that the expanding regulatory use of their ratings meant that the issuers really did need to have their bonds rated and were willing to pay for the “privilege.”
The newer issuer-pays model had (and still has) an obvious potential conflict of interest: in order to secure business, a CRA may offer the issuer special “favors”―like a higher rating than the issuer’s financial posture would otherwise justify. Nevertheless, for the next 30 years, this possibility did not become a reality―and it still has not arisen in the areas of U.S. corporate bonds, municipal bonds, and sovereign government bonds. The CRAs’ long-run reputational concerns as accurate raters generally prevailed over any temptations that may have arisen.
This all changed in the early 2000s when the large tide of residential mortgage-backed securities (RMBS) issues led the major CRAs to “cater” to the RMBS issuers. It is important to understand why the CRAs’ reputational discipline broke down for RMBS but didn’t break down (and still hasn’t broken down) for traditional “plain vanilla” ratings.
In the traditional arena, there are thousands of corporate (and muni and government) issuers; there is a great deal of publicly available information about the issuers; and the profit margins for the CRAs have traditionally been thin. Together, these elements have meant that there is relatively little gain for a CRA to do a favor for any individual issuer, while the risk of critics’ discovering that a rating is anomalously high (and asking why) is substantial.
By contrast, for RMBS, there were far fewer issuers—about 12 issuers accounted for 80 to 90 percent of RMBS issuances; the information available to the public was far more opaque; the volumes were large; and the profit margins were relatively wide. As a result, the temptation to do a favor was greater (and the likelihoods that critics would find out were smaller).
The evidence is clear that the CRAs’ reputation discipline did weaken for RMBS ratings.
In response, two of the Dodd-Frank Act’s many provisions applied to the NRSROs: One instructed federal financial regulatory agencies to eliminate their reliance on the NRSROs wherever possible and to find alternative means to achieve those agencies’ prudential regulatory goals. By 2015 most federal regulators had complied. The Department of Labor, however, still includes references to NRSROs in its regulation of pension funds. And state insurance regulators, which are not covered by Dodd-Frank, continue to rely on NRSROs in their prudential regulation of insurance companies.
The second set of Dodd-Frank provisions strengthened the SEC’s regulatory powers over the NRSROs―specifically with respect to conflicts of interest and to achieving greater transparency with respect to the CRAs’ methodologies. However, the SEC is not supposed to influence either the specific methodologies of the CRAs or their specific ratings.
As mentioned, the Financial CHOICE Act leaves these two Dodd-Frank provisions largely unchanged. This is a missed opportunity―especially with regard to the second set of regulations.
Despite the social urge to use regulation to get the CRAs “to do a better job,” it is far from clear that the SEC can do much that is beneficial. Importantly, the bond market is largely a wholesale market: except for muni bonds (where households hold about 50% of outstanding bonds), the overwhelming percentages of holdings are by financial institutions. Thus, the primary “investor” in the bond market is a bond portfolio manager for a financial institution. Such managers—even if they don’t have the resources and expertise to do the research on bonds themselves—should know enough to be able to decide which third-party providers of creditworthiness are reliable, and why. They shouldn’t need guidance from the SEC.
Further, the CRAs’ concerns for their long-run reputations—especially in the context of a wholesale market—should be powerful. And in potentially challenging areas such as RMBS, the SEC could strengthen CRA discipline by requiring that issuers provide more information directly to the public. And, surely, after the experience of the 2000s, bond portfolio managers should be wary of blindly accepting RMBS ratings.
At the same time, let’s remember the substantial drawbacks to the SEC’s regulation: it tends to have high fixed costs of compliance arising from the need for lawyers, accountants and the like. This favors the larger CRAs, buttressing their market shares (see Tables 1 and 2!). But entrants are often the source of innovation in an industry. Thus, the SEC regulation may well be discouraging new ideas, new methodologies, new technologies, perhaps even new business models for credit rating firms. And, of course, the regulation raises the direct costs of compliance for the regulated NRSROs.
With regulatory reliance on ratings substantially reduced, the need for specifying which CRAs’ ratings should be heeded is reduced. This weakens the argument for maintaining the NRSRO category sufficiently that the category ought simply to be eliminated. With less of a government implicit imprimatur for the incumbents, entry would again be encouraged.
Change in the credit rating industry has traditionally been slow. Rating users’ reliance on the experience and reputation of the major CRAs is an important contributor to this sluggishness. But government barriers to entry are significant. The elimination—or at least the substantial reduction of—the regulation of CRAs and the elimination of the NRSRO category offer a better path to achieving change than maintaining the status quo.
Note: Professor White's previous guest post was Walmart and Banking: It’s Time To Reconsider.