howard esaki is the former global head of securitization research at Standard & Poor's. larry white is a professor of economics at New York University's Stern School of Business.
Illustrations by The Heads of State
Published August 4, 2017
Financial markets are certainly safer today after the tighter regulation imposed in the wake of the 2008 crisis. But one of the glaring sources of instability in 2008, the collapse of ratings-agency standards that fed the bubble in residential mortgage-backed securities, has not been dealt with in a conclusive way. Here, we suggest an approach to changing ratings-agency incentives — one that would require minimal government intervention yet prevent the race to the bottom that gave investors false confidence in the quality of these complex financial assets while the bubble inflated.
Stated simply, our proposal is parallel to the process used to limit the impact of biased judges in Olympic events ranging from gymnastics to diving. In these sports, the score proposed by the most lenient judge on a multi-judge panel is automatically dropped. In our proposal, the ratings agency that proposes to rate a securities issue most leniently would be dropped from consideration or, alternatively, its rating fee would be withheld.
The proposal is, of course, aimed at residential mortgage-backed securities rating. But it would also serve to reduce the incentives of these for-profit ratings agencies to compete for market share by inflating ratings of commercial mortgage-backed securities, collateralized loan obligations and other asset-backed securities, such as automobile loans and credit-card receivables.
We can already detect some grumbling here: why not just leave it to investors to decide which ratings are worth trusting and which aren't? The long answer, which turns on why markets sometimes fail, is, well … long. The short answer, which should satisfy even free-market zealots, is that the realistic choice is not between our proposal and caveat emptor but between lightly applied indirect regulation and increased direct regulation by the Securities and Exchange Commission that would cost more and be less effective.
Origins of the Fall
Residential mortgage-backed securities are anchored by financial claims against pools containing thousands of mortgages, which are divided into classes with different maturities and credit ratings. Mortgages with guarantees by federal agencies (Ginnie Mae, Fannie Mae and Freddie Mac) backed the initial MBS offerings in the 1970s. The credit ratings of these securities were thus almost always the top-of-the-line AAA/Aaa.
But for better and worse, the financial industry is constantly in search of newly engineered products that appeal to investors and generate profits. In the 1990s, a market developed for private-label residential mortgage-backed securities that were composed of mortgages that lacked government guarantees. The credit ratings agencies initially rated various classes of these securities from AA/Aa2 to B, based on various criteria. But the main difference between the highest- and lowest-rated deals was the amount of credit support or "subordination" — that portion of the collateral that could be lost without leaving the owners vulnerable to losses.
If one of the many mortgages backing one of these securities defaults, goes through foreclosure and takes a loss, the pool's administrator applies the loss to the most junior class (or tranche) of securities that was created when the issue was assembled. Additional losses from other mortgage defaults are similarly applied until that most junior tranche is no longer backed by collateral. Then, if more losses accumulate, the next most junior class becomes vulnerable, and so forth. The structures of securities-backed commercial mortgages, automobile receivables and other collateralized loan obligations are similar.
In the 1990s, the senior-most class of most residential mortgage-backed securities made up 70 percent of the deals, with 30 percent of the collateral subordinate to it. In other words, the collateral represented by the pool of mortgages could withstand a loss of up to 30 percent of its value without jeopardizing the claims of the owners of the most senior class.
A moment for digression. Some readers may wonder why investors were willing to put up with all this intricate slicing and dicing that could not have been done without fast computers and modern software. For one thing, securitization (which diversifies the bet made by any individual investor) is an incredibly efficient way to funnel credit into a heterogeneous market like housing (or cars or strip malls or warehouses). For another, dividing securities into senior and junior tranches opens the door to investors with widely varying willingness to bear risk in exchange for a higher expected return.
But, of course, the inherent complexity of these securities puts a premium on the availability of accurate information about the risks that owners of the securities must bear. Which brings us back to the ratings agencies.
The Race to the Bottom
Moody's and Standard & Poor's initially dominated the market for rating private-label residential mortgage-backed securities. Subordination levels were fairly steady, and the rating of the most senior tranches was typically a healthy AA/Aa2 rather than the gold-standard AAA/Aaa. The issuers of most of these securities solicited (and paid for) at least two ratings, so the two major agencies could each maintain nearly a 100 percent market share. But as the private-label market grew and profits from rating such securities increased, the market drew new entrants, including Fitch Ratings and Duff & Phelps. Market share began to fall for Moody's and S&P as the new kids on the block chose to compete by offering slightly looser rating standards (i.e., less subordination).
But, not surprisingly, this provoked a competitive response from everyone else in the business. All ratings agencies began to refine their mathematical models for determining which classes of security drew which rating, and this almost always resulted in declines in the collateral coverage required to earn a given rating.
In part, this was a sensible response. Many of the early ratings were too conservative, as underwriting standards in the 1990s were fairly strict and senior classes could withstand default rates of up to 75 percent (under the assumption of a 40 percent loss on foreclosed loans), which nobody expected short of the launch of nuclear war.
But subordination continued to drop even after underwriting standards started to weaken with the growth of subprime mortgage lending that largely disregarded (a) the creditworthiness of the individual borrowers, and (b) that house prices were rising to record levels suggesting the potential for a big fall in the value of the collateral in the event of a systemic crisis. Many of the ratings-agency changes could be justified case by case. But cumulatively, they resulted in massive drops in the credit quality of residential mortgage-backed securities that were not reflected in the ratings.
In the years before the financial crisis, issuers mostly sought two or three ratings by asking for estimates of how much collateral would be needed to obtain a given rating from four to six ratings agencies. The issuers then typically chose the lowest level that was "bid" by either S&P or Moody's since many institutional investors required a rating from one of these two agencies to buy a security. Another (perhaps the only other) rating chosen was usually the most lenient rating from among the other bids by less-established agencies.
Now, purchasing ratings from two or more agencies has the advantage of providing some protection against errors in risk-modeling or computation that would otherwise cause the use of an outlier as a final rating level. But it also reduces the risk for any given ratings agency in cutting corners on the level of protection required since the agency knows that the issuer is likely to not use the new low estimate, but the next-to-lowest level. It thus reduces the metaphoric speed bumps in the race to the bottom.
By the time of the 2008 financial crisis, subordination levels for the most senior AAA/Aaa class had fallen to as low as 2 percent (no misprint). This meant that 98 percent of a collateral pool of unrated (and likely well below-investment grade) mortgages could receive the highest rating (AAA/Aaa), which previously had been reserved for only a handful of corporations and countries. To restate that in starkest fashion: the rating on securities protected by the thinnest of collateral in housing rose to higher than S&P's current (since 2011) rating of the full faith and credit of the U.S. government. Support had fallen from nearly 30 percent for lower-rated AA/Aa2 securities 20 years earlier.
In for a penny, in for a pound: bankers even repackaged small slices of BBB and lower-rated tranches of existing residential mortgage-backed securities into hybrid securities called collateralized debt obligations. And as if by magic, the most senior tranches of these hybrids created from the cats and dogs lying around in the bankers' basements were awarded AAA ratings.
Back to the Sports Analogy
To beat this incentive problem without removing the profit motive from the equation, we propose the following approach.
1. Securities issuers may solicit any number of agencies to submit credit-support-level estimates of the collateral they would require to rate a security (as they do now).
2. The issuer must pay a modest "breakage fee" — i.e., a bid-preparation fee — for each estimate.
3. The issuer may select any group of agencies to rate a deal and may choose to base the selection on the credit support levels required by the agencies — again, the same as the current system. The breakage fees are a credit against the total rating fees.
4. If chosen to rate, the agency accepting the lowest quality backing for the security issue at the end of the evaluation process receives only the breakage fee instead of its full fee. By the same token, this agency would not receive any ongoing surveillance fees to maintain oversight of the mortgage pool. Possible alternatives to the restriction of payment to the agency are:
- Simply barring the agency with the lowest credit support from rating the issue.
- Taxing the fee of the most lenient agency at a punitive rate.
- Preventing the use of the most lenient agency's rating for any regulatory purposes, such as the determinants of capital requirements for banks and other prudentially regulated financial institutions.
5. If one agency is a major outlier it would be barred from rating the deal.
6. All final bids are kept confidential until the rating is made public, at which time all bids (winning and losing) are made public.
The reduction in (or elimination of) payment to the bidder requiring the least credit backing creates an incentive not to undercut competitors. The issuer retains the choice over which agencies rate a deal. However, a ratings agency that demanded a lot of collateral simply to avoid being the low bidder would run the risk of elimination for being too stringent to satisfy the issuer.
Our proposal would not entirely eliminate the possibility of what game theorists call "strategic" bidding behavior in which each ratings agency factors in what it expects other agencies to bid. But our approach would balance the incentive to make the dash for the bottom to get the deal.
Remember, too, that the rules would require all bids to be made public after the fact. This would reduce the incentive of the issuer to eliminate bidders demanding a lot of collateral, since potential investors would have access to the cautionary bid information, anyway.
Yes, But …
Our proposal would not satisfy the impulse to radically alter the current system, as many analysts wanted in the wake of the ratings agencies' culpability in the financial crisis. But we think it would be a plus to achieve the goal of making ratings more reliable without starting over.
Our approach would be somewhat vulnerable to cheating, either by collusion among ratings agencies or by issuers making secret side payments to influence the process. But there's nothing new here. One would hope that a combination of regulatory oversight and standard financial auditing for public corporations — not to mention the threat of criminal and civil penalties — would suffice to minimize rule-breaking.
Another potential objection to our focus on changing incentives on credit-support bids is that collateral is only one dimension of a securities deal. That focus would give short shrift to other factors such as structure and legal provisions. This, however, seems to be a problem that is solving itself, since these other dimensions of securities are becoming standardized in the marketplace.
The application of our proposal could be problematic in cases in which one ratings agency made the most lenient bid for one tranche of a security offering and a higher level than competitors for another tranche. We would manage this concern by applying the rules on a class-by-class basis. In other words, a ratings agency that is not paid for a senior AAA rating on a senior tranche could still be compensated for a BBB rating on a junior tranche if met the criteria outlined above.
Another potential complication is that pools of mortgages can change characteristics during the rating process. To prevent endless rounds of bidding, we propose that only the firms that qualify in an initial round be allowed to submit subsequent bids in the event of a change in the quality of the pool.
While we've focused here on residential-mortgage securities, we believe our proposal would work well for other sorts of collateralized securities. These other sectors have experienced some decline in rating standards that has been driven by the same sorts of competitive pressure affecting the standards of residential mortgage-backed securities.
Note, however, that we don't think that our reformed approach is needed for corporate bond ratings, as that sector has been subject to much less competition among ratings agencies and there is little evidence of a trend toward more lenient standards. Moreover, the opportunities for manipulating a corporate financial structure so as to achieve a better rating target are far fewer.
Regulation with a Light Touch
The markets for fixed-income securities are largely institutional ones, with professional portfolio managers representing the bulk of the buyers. And as we suggested earlier, these professionals should have memories and thus be both inclined and able to give more standing to ratings agencies that have good track records. In turn, this should motivate the ratings agencies to pay more attention to their reputations in the long term and to forswear the temptations to make easy money by catering to issuers' desires for unduly favorable ratings.
But, to paraphrase a great economist writing in a very different context, in the long run, many of the principals will be dead — or at least no longer in the line of fire — when hard times arrive. And that applies to the aforementioned portfolio managers as well as to the analysts responsible for deciding ratings criteria. Something more than the discipline of the free market is needed.
The "something more" mandated by the Dodd-Frank Act of 2010 was a substantial increase in the Securities and Exchange Commission's regulation of the 10 largest credit-ratings agencies. By contrast, our proposal would not involve any increase in direct regulation, but only a set of rules about how issuers select their ratings agencies.
The financial crisis undermined the credibility of the best of the residential mortgage-backed securities sector along with the worst. By no coincidence, the volume of such securities is less than 5 percent of its peak issuance rates, and other securitization sectors are also much smaller than before the financial crisis.
The benefits (in terms of the efficiency of capital markets) of having a much larger volume would almost certainly be substantial. But that can only happen if the private-label residential mortgage-backed securities sector regains the trust of big institutional investors like insurance companies and mutual funds. This could probably be achieved by imposing sufficiently tough direct regulation.
However, in light of the 2016 election results, tougher regulation seems unlikely. In any event, we believe that it would be far better to build self-regulating mechanisms into the market to do the job than to rely on fallible regulators subject to myriad outside pressures.