In Defense of the Rating Agencies – V (summary, and hopefully final)

I write this because I was invited to be on CNBC on the topic, but I suggested that my opinion would not make for good television.  That said, I have taken my four prior posts on the topic, and assembled them into one comprehensive post.  I do not intend on posting on this again.  With that, here is my post:

The ratings agencies have come under a lot of flak recently for rating instruments that are new, where their models might not be do good, and for the conflicts of interest that they face.  Both criticisms sound good initially, and I have written about the second of them at RealMoney, but in truth both don’t hold much water, because there is no other way to do it.  Let those who criticize put forth real alternatives that show systematic thinking.  So far, I haven’t seen one.

Most of the current problems exist in exotic parts of the bond market; average retail investors don’t have much exposure to the problems there, but only less-experienced institutional investors.

Here are my five realities:

  • Somewhere in the financial system there has to be room for parties that offer opinions who don’t have to worry about being sued if their opinions are wrong.
  • Regulators need the ratings agencies, or they would need to create an internal ratings agency themselves, or something similar.  The NAIC SVO is an example of the latter, and proves why the regulators need the ratings agencies.  The NAIC SVO was never very good, and almost anyone that worked with them learned that very quickly.
  • New securities are always being created, and someone has to try to put them on a level playing field for creditworthiness purposes.
  • There is no way to get investors to pay full freight for the sum total of what the ratings agencies do.
  • Ratings can be short-term, or long-term, but not both.  The worst of all worlds is when the ratings agencies shift time horizons.

Ratings are Opinions

The fixed-income community has learned that the ratings agencies offer an opinion, and they might pay for some additional analysis through subscriptions, but if they were forced to pay the fees that issuers pay, they would balk; they have in-house analysts already.  The ratings agencies aren’t perfect, and good buy-side shops use them, but don’t rely on them.  Only rubes rely on ratings, and sophisticated investors did not trust the rating agencies on subprime.  My proof?  Look how little exposure the Life and P&C insurance industries had to subprime mortgages outside of AIG.  Teensy at best.

Please understand that institutions own most of the bonds out there.  We had a saying in a firm that I managed bonds in, “Read the write-up, but ignore the rating.”  The credit analysts at the rating agencies often knew their stuff, giving considerable insight into the bonds, but may have been hemmed in by rules inside the rating agency regarding the rating. It’s like analysts at Value Line.  They can have a strong opinion on a company, but their view can only budge the largely quantitative analysis a little.

Let’s get one thing straight here.  The rating agencies will make mistakes.  They will likely make mistakes on a correlated basis, because they compete against one another, and buyers won’t pay enough to support the ratings.

So there are systematic differences and weaknesses in bond ratings, but the investors who own most of the bonds understand those foibles.  They know that ratings are just opinions, except to the extent that they affect investment policies (”We can’t invest in junk bonds.”) or capital levels for regulated clients.

On investment policies, whether prescribed by regulators or consultants, ratings were a shorthand that allow for simplicity in monitoring (see Surowiecki’s argument).  Now, sophisticated investors knew that AAA did not always mean AAA.  How did they know this?  Because the various AAA bonds traded at decidedly different interest rates.  The more dodgy the collateral, the higher the yield, even if it had a AAA rating.  My mistake: I, for one, bought some AAA securitized franchise loan paper that went into default long before the current crisis hit.  Many who bought post-2000 AAA securitized manufactured housing loan paper are experiencing the same.  Early in the 2000s, sophisticated investors got burned, and learned.  That is why few insurers have gotten burned badly in the current crisis.  Few insurers bought any subprime residential securitizations after 2004.  But, unsophisticated investors and regulators trust the ratings and buy.

Financial institutions and regulators have to be “big boys.” If you were stupid enough to rely on the rating without further analysis, well, that was your fault.  If the ratings agencies can be sued for their opinions (out of a misguided notion of fiduciary interest), then they need to be paid a lot more so that they can fund the jury awards.  Their opinions are just that, opinions.  As I said before, smart institutional investors ignore the rating, and read the commentary.  The nuances of opinion come out there, and often tell smart investors to stay away, in spite of the rating.

How Can Regulators Model Credit Risk?

It started with this piece from FT Alphaville, which made the point that I have made, markets may not need the ratings, but regulators do.  (That said, small investors are often, but not always, better off with the summary advice that bond ratings give.)  The piece suggests that CDS spreads are better and more rapid indicators of change in credit quality than bond ratings.  Another opinion piece at the FT suggests that regulators should not use ratings from rating agencies, but does not suggest a replacement idea, aside from some weak market-based concepts.

Market based measures of creditworthiness are more rapid, no doubt.  Markets are faster than any qualitative analysis process.  But regulators need methods to control the amount of risk that regulated financial entities take.  They can do it in four ways:

  1. Let the companies tell you how much risk they think they are taking.
  2. Let market movements tell you how much risk they are taking.
  3. Let the rating agencies tell you how much risk they are taking.
  4. Create your own internal rating agency to determine how much risk they are taking.

The first option is ridiculous.  There is too much self-interest on the part of financial companies to under-report the amount of risk they are taking.  The fourth option underestimates what it costs to rate credit risk.  The NAIC SVO tried to be a rating agency, and failed because the job was too big for how it was funded.

Option two sounds plausible, but it is unstable, and subject to gaming.  Any risk-based capital system that uses short-term price, yield, or yield spread movements, will make the management of portfolios less stable.  As prices rise, capital requirements will fall, and perhaps companies will then buy more, exacerbating the rise.  As prices fall, capital requirements will rise, and perhaps companies will then sell more, exacerbating the fall.

Market-based systems are not fit for use by regulators, because ratings are supposed to be like fundamental investors, and think through the intermediate-term.  Ratings should not be like stock prices — up-down-down-up.  A market based approach to ratings is akin to having momentum investors dictating regulatory policy.

Also, their models that I am most familiar with only apply to publicly-traded corporate credit.  Could they have prevented the difficulties in structured credit that are the main problem now?

What to do with New Classes of Securities?

Financial institutions will want to buy new securities, and someone has to rate them so that proper capital levels can be held (hopefully).  But what are rating agencies to do when presented with novel financial instruments that have no significant historical loss statistics?  Many of the likely buyers are regulated, and others have investment restrictions that depend on ratings, so aside from their own profits, there is a lot of pressure to rate the novel financial instrument.  A smart rating agency would punt, saying there is no way to estimate the risk, and that their reputation is more important than profits.  Instead, they do some qualitative comparisons to similar but established financial instruments, and give a rating.

Due to competitive pressures, that rating is likely to be liberal, but during the bull phase of the credit markets, that will be hidden.  Because the error does not show up (often) so long as leverage is expanding, rating agencies are emboldened to continue the technique.  As it is, when liquidity declines and leverage follows, all manner of errors gets revealed.  Gaussian copula?  Using default rates for loans on balance sheet for those that are sold to third parties?  Ugh.

Some will say that rating agencies must say “no” to Wall Street when asked to rate exotic types of debt instruments that lack historically relevant performance data.  That’s a noble thought, but were GICs [Guaranteed Investment Contracts] exotic back in 1989-1992?  No.  Did the rating agencies get it wrong?  Yes.  History would have said that GICs almost never default.  As I have stated before, a market must fail before it matures.  After failure, a market takes account of differences previously unnoticed, and begins to prospectively price for risk.

But think of something even more pervasive.  For almost 20 years there were almost no losses on non-GSE mortgage debt.  How would you rate the situation?  Before the losses became obvious the ratings were high.  Historical statistics vetted that out.  No wonder the levels of subordination were so small, and why AAA tranches from late vintages took losses.

When prosperity has been so great for so long, it should be no surprise that if there is a shift, many parties will be embarrassed.  In this case both raters and investors have had their heads handed to them.

Now there are alternatives.  The regulators can ban asset classes until they are seasoned.  That would be smart, but there will be complaints.  I experienced in one state the unwillingness of the regulators to update their permitted asset list, which had not been touched since 1955.  In 2000, I wrote the bill that modernized the investment code for life companies; perhaps my grandson (not born yet), will write the next one.

Regulators are slow, and they genuinely don’t understand investments.  The ratings agencies aren’t regulators, and they should not be put into that role, because they are profit-seeking companies. Don’t blame the rating agencies for the failure of the regulators, because they ceded their statutory role to the rating agencies.  But if the regulators ban asset classes, expect those regulated to complain, because they can’t earn the money that they want to, while other institutions take advantage of the market inefficiencies.

Compensation and Conflicts

Some say that rating agencies must be encouraged to make their money from investor subscriptions rather than fees from issuers, to ensure more impartial ratings.  If this were realistic, it would have happened already.  The rating agencies would like nothing more than to receive fees from only buyers, but that would not provide enough to take care of the rating agencies, and provide for a profit.  They don’t want the conflicts of interest either, but if it is conflicts of interest versus death, you know what they will choose.

Short vs. Long-Term Opinions

Ratings agency opinions are long-term by nature, rating over a full credit cycle.  During panics people complain that they should be more short-term.   Hindsight is 20/20.  Given the multiple uses of credit ratings, having one time horizon is best, whether short- or long-term.  Given the whipsaw that I experienced in 2002 when the ratings agencies went from long- to short-term, I can tell you it did not add value, and that most bond manager that I knew wanted stability.


Some say rating agencies no longer should have exclusive access to nonpublic information, to even out the playing field.  That sounds good, but the regulators want the rating agencies to have the nonpublic information.  They don’t want a level playing field.  As regulators, if they are ceding their territory to the rating agencies, then they want the rating agencies to be able to demand what they could demand.  Regulators by nature have access to nonpublic information.

Some ask for greater disclosure of default rates, but that is a non-issue.  They also look to punish rating agencies that make mistakes, by pulling their registration.  Disclosing default rates is already done, and sophisticated investors know this.  Yanking the registration is killing a fly with a sledgehammer.  It would hurt the regulators more than anyone else. The rating agencies are like the market.  The market as a whole gets it wrong every now and then.  Think of tech stocks in early 2000, or housing stocks in early 2006.  To insist on perfection of rating agencies is to say that there will be no rating agencies.  It takes two to make a market, and agencies will often be wrong.


As for solutions, I would say the following are useful:

  • Competition.  I’m in favor of a free-ish market here, allowing the regulators to choose those raters that are adequate for setting capital levels, and those that are not.  For other purposes, though, the more raters, the better.  I don’t think rating fees would drop, though.  Remember, ratings are needed for regulatory purposes.  Will Basel II, NAIC, and other regulators sign off on new credit raters?  I think that process will be slow.  Diversity of opinion is tough, unless a ratings agency is willing to be paid only by buyers, and that model is untested at best.
  • Compensate with residuals and bonuses (give the raters some skin in the game)
  • Deregulation (we can live without rating agencies, but regulators will have to do a lot more work)
  • Greater disclosure (sure, let them disclose their data and formulas (perhaps with a delay).
  • Have regulators bar unseasoned asset classes.


Let those who criticize the ratings agencies bring forth a new paradigm that the market can embrace, and live with in the long term.  Until then, the current system will persist, because there is no other realistic way to get business done.  There are conflicts of interest, but those are unavoidable in multiparty arrangements.  The intelligent investor has to be aware of them, and compensate for the inherent bias.

Unless buyers would be willing to pay for a new system, it is all wishful thinking.  Watch the behavior of the users of credit ratings.  If they are unwilling to pay up, the current system will persist, regardless of what naysayers might argue.


  • AS says:

    It is big of you to NOT consort with CNBC. Thank you.

  • David,

    Really a sharp piece.

    What would you think of reform along these lines:

    1) Regulators rate the rating agencies based on competence with respect to different categories of product. That is, rather than just NRSROs, rating agencies would have to be certified for regulatory purposes as new product categories arise. Regulators can be decertified at any time, and would expect to lose certification for some period of time if they got something scandalously wrong. (Which, as you point out, would happen even to the best of ratng agencies occasionally.)

    2) No product could be rated for regulatory purposes until there are at least a certain number certified raters (ideally several more than the traditional three).

    3) We’d retain the “issuer pays” model, but eliminate the “issuer chooses” part of that. If an issuer wants regulated institutions to be able to purchase its paper, it would commit to standard fees, following which regulators would randomly assign the task to at least two raters, who would commit to independence of analysis (i.e. any collaboration or comparing of notes would be grounds for decertification, fines, etc.). Anything that would be too complicated to rate for the standard fees in a category would be too complicated to be rated for regulatory purposes.

  • @Steve. I really like the idea, but it all hinges on a competent and independent regulator. The Fed completely missed the significance of deteriorating mortgage lending standards; would they have been able to suggest or recognize a pragmatic range of scenarios for RMBS loss models? How many federal employees are there who could have understood the effect of cliff-shaped loss functions on multi-tranche securitizations, and the cascade of those effects on CDO-squareds? In May 2007 the FT made the claim that anyone who really understood structured finance was being hired away from the ratings agencies by the banks. A fortiori for government employees.

    In other words, I don’t believe that the regulators are competent to rate competence.

  • @Jim. I certainly don’t disagree with you. I wish we could find a better system that didn’t require any sort of government certification. But as David points out, as long as we will have asset-side regulation, somehow the government will have to identify categories of assets to make rules about. They could do this themselves, or outsource it to rating agencies. So we are left to choose the lesser of two incompetences: are regulators (over time and political cycles) more incompetent at rating securities directly or at judging the competence of private raters? We have to pick one, unfortunately, unless we give up on asset-side regulation entirely.

  • Brett Blake says:

    A good essay. Plenty to think about, however as we’ve heard in testimony before the Senate Permanent Subcommittee on Investigations, the chiefs of the ratings agencies admit they rate issues favorably because their businesses and/or jobs are at stake.

    Making ratings, possibly from multiple agencies, mandatory, at fixed fees, might attenuate this. I don’t know. Depends on what you think of ratings and trusting someone else to attempt your due diligence. As it is, ratings are important marketing materials, not much more.

    One could do away with the ratings agencies altogether, and this might be the most honest thing to do.

  • barry says:

    Decades ago, ratings by an agency were just a minimum criterion for making an investment decision.

    Investment guidelines had wording like “can only invest in assets rated AA or better … ”

    It didn’t mean that an asset rated AA or better automatically qualified for investment.

    There was a team whose sole job was to review financial statements, prospectuses, etc. to make an independent judgement as to whether the asset was indeed worthy. This was especially true if there was a risk of a downgrade.

    When I was at an investment bank, there was a group that tried to identify ‘mis rated’ assets (both too high and too low). Again, you needed people who actually read and understand financial statements.

  • @Steve: Of course you are right that we are stuck with regulators rating the private raters, in one way or another. And I don’t think your proposal to shift that regulation from ratings agencies as a whole to specific areas of competence would make things any worse. I’m just not very optimistic that it would make things any better.

    How about a slight variant: I used to review a lot of grant proposals for the NIH. There, the NIH employees did not aspire to being experts; rather they viewed themselves as coordinators of the process. That system works pretty well; could something like it work in this case? What if the regulators put together a panel of independent consultants, the likes of Tavakoli and Cifuentes, and used that panel to rate the various areas of expertise of the private raters?

  • Two years into one of the biggest financial crisis ever, resulting from one of the worst regulatory blunders ever, it is so sad seeing how little even “experts” understand what happened.

    You write about “proper capital levels” but, given an amount of capital, the proper mix of assets should result from managing risks and yields, and the regulator has no business whatsoever intervening with subsidizing the yields of triple-A rated assets, by means of arbitrarily assigning these the benefit of generating especially low capital requirements for the banks.

    This is why there was a stampede after AAAs and since these are by nature scarce, and the market is a market, the Potemkin ratings were supplied.

    And also, from a societal point of view, even if the ratings were perfect, why do we think that our future lies so much in never-risk land so as to be willing to make even more risk-averse normally quite risk-averse banks?

    Why for instance would we not impose lower capital requirements on banks when lending to their natural clients, small businesses and entrepreneurs who help to create jobs, and higher capital requirements when lending to what has a triple-A and has therefore reached a point where its financial needs should be satisfied in the capital markets?

  • Per, I have written elsewhere that securitized assets of equivalent ratings should be given higher capital charges because have higher default severity and are more prone to contagion effects. An approach like this would reduce the incidence of securitization.

  • John Hunt says:

    Re novel securities, you say a smart rating agency would punt b/c its reputation is more important than short-term profits. But the agency doesn’t have a reputation in rating the specific class of new instrument, and it is extremely unclear that the market exacts any reputational penalty for failure in the other parts of the agency’s business. That’s why disgorgement of profits on new products rated with poor quality makes sense. One might argue that “poor quality” can’t be discerned, even after the fact, but that implies that reputation is not an effective mechanism.