Rating Agencies: In Need of Some "New Math?"

One constituency that has taken a lot of heat during the subprime meltdown, bond rating agencies, have largely been called to the mat because of potential conflicts of interest. And I do believe it is an issue that warrants additional scrutiny. However, one aspect that has received less attention are the ratings models themselves; are historical analytical frameworks and paradigms used for corporate and municipal bonds applicable to derivative structures like mortgage-backed securities and their variants? An interesting article in yesterday's Wall Street Journal raised some interesting issues that got me thinking, and I thought I'd share some of my thoughts with you. From yesterday's WSJ; I have bolded phrases and sentences that are particularly noteworthy:

When it comes to bond ratings, all triple-A grades aren't created equal.

That's one reason debt-rating companies such as Standard & Poor's, Moody's Investors Service and Fitch Ratings have met such criticism over their role in the subprime-mortgage crisis.

The trio of rating companies -- owned, respectively, by McGraw-Hill Cos., Moody's Corp. and France's Fimalac SA -- has served as bond-market gatekeepers for decades, publishing letter grades designed to keep nervous investors out of risky bonds. But as the bond market has grown more complex, ratings have evolved to mean different things, depending on what's being rated.

While all ratings refer to the likelihood of default, some newer, more complex bonds don't default in the same way -- or at the same pace -- that old-fashioned corporate bonds do.

********************

The proliferation of triple-A ratings has made it easier for pension funds and other investors to pile into newfangled bonds that recently have been losing market value at an alarming rate.

The sheer volume of these ratings is noteworthy now that investors across the globe have lost confidence in the ratings and values of some bonds. Hundreds of mortgage-related bonds have been downgraded in recent months amid a decline in housing prices.

********************

Over a few days in August, S&P lowered the rating on two hard-hit structured vehicles from AAA to CC, a rating on the brink of a default. An S&P spokesman said that type of deal was rare.

********************

"The corporate rating is tried-and-true," says H. Sean Mathis, an executive at valuation-advisory firm Miller Mathis. Ratings on structured-mortgage products, by comparison, were created recently "as the product of a model."

A Credit Suisse study found that ratings on asset-backed securities, which include some mortgages, were more stable than corporate-bond ratings but were more likely to fall sharply when downgraded.

Some observations and comments:

"...ratings have evolved to mean different things, depending on what's being rated." If this really is true, this is not a good thing. Ratings, as long as I've been in and around the markets, have always had an absolute meeting with concrete standards for achieving a given rating. If there is now some "relative" ratings concept in play, where a AAA-rating on a corporate bond is different than a AAA-rating on the senior tranche of a CMBS vehicle, then I believe the entire underpinning of the ratings system is in jeopardy. S&P, Moody's, Fitch and your peers - say it ain't so. Please.

"The proliferation of triple-A ratings has made it easier for pension funds and other investors to pile into newfangled bonds..." While I don't necessarily like the use of a pejorative term like "newfangled," and think it detracts from the strength of the argument, the point itself is valid: investors rely on ratings and generally view ratings as being fungible across asset types - corporate bonds, municipal bonds, structured bonds, structured vehicles, etc. So it is hard to overstate the role rating agencies play in permitting certain instruments to be purchased by certain classes of investors. And a AAA-rating simply opens the floodgates of potential demand for a particular instrument. So once again, ratings need to be consistent, fungible and valid, because if they're not the entire system of investment guidelines being linked to ratings will crumble like a house of cards.

Now, onto the really interesting stuff.

"...more complex bonds don't default in the same way -- or at the same pace -- that old-fashioned corporate bonds do.

"Over a few days in August, S&P lowered the rating on two hard-hit structured vehicles from AAA to CC, a rating on the brink of a default."

"...ratings on asset-backed securities, which include some mortgages, were more stable than corporate-bond ratings but were more likely to fall sharply when downgraded."

I was reading this and thought to myself, this all makes sense. Given the embedded optionality and complexity of the instruments underlying these structured vehicles, there is no way the analytical models used to gauge default for more vanilla bonds can possibly used here. And I'm sure (I hope?) they're not. But why is this? If you look at the comments above, the rapid decay in credit quality of certain structured vehicles and the conclusion of the Credit Suisse study, the answer becomes clear:

The nature of the risk inherent in these vehicles is convex and, as such, subject to rapid changes in value over short periods of time.

And the more I thought about this, I began to couch risk in derivative risk management terms. It is not that a property of asset-backed securities (ABS) default risk is volatility; in fact, the Credit Suisse study highlighted the stability of asset-backed securities ratings relative to corporate bond ratings. It is that a defining property of ABS default risk is gamma. And the more I thought about it, I started to draw a parallel between the nature of ABS default risk (and, therefore, a key component of its rating) and out-of-the-money (OTM) put option risk.

In this case, the investor in the ABS vehicle is long an asset and short an OTM put option, the pricing of which appears to be wrong given what has played out in the subprime market. If an investor puts money in AAA-rated ABS security, their returns will bubble along and they will collect a premium relative to investing in Treasuries. But, one day they might wake up and find that their AAA-rated security is now trading at 92, and not only that, but that there is no depth at a 92 bid. Not exactly what they were counting on. Now, there is a correct price for this type of risk and a correct rating that takes into account the non-linear nature of this exposure. But this requires both sophisticated scenario-based modeling (because any model using normal distributions in this circumstance should be thrown out, since that is not the nature of the risk being priced) and experience in pricing and managing these high gamma risks. Basically, the major rating agencies should all have top-flight derivative risk management professionals on their staffs, because these are exactly the types of risks they are purporting to rate.

I know that I never really took the time to think about structured vehicle risks this way, but I am pretty sure I'm onto something here. And I'd be interested to see how these OTM put prices impact ratings, especially now that we have some history to serve as a guide. But I suspect what we'll see is that, indeed, not all AAA-ratings are created equal. That a bunch of stuff that has been called AAA that should have been rated AA or worse. And that the embedded losses in investors' portfolios are in the tens of billions, and they're not getting it back. And that this is the real reason for not being too happy with the rating agencies. Because unless they are pricing risks this way, they've missed the boat. Maybe I'm wrong. But just maybe I'm not.

Comments

[jck]

Roger:
The problem is that the ratings agencies themselves seem to have forgotten the difference between price fluctuations and change in credit.
In many structured products, there are "quality tests" based on the market value of the collateral and ratings changes have been based on price movements rather than change in credit. on prices.
You are onto something since the price triggers can be modelled but then we aren't talking about credit.
I think what is needed is a new rating scale for structured products, independent of what people understand as "credit".
Take the example of last year hot product: the CPDO paying 200bp above Libor with an AAA rating.There is no way you can reconcile that sort of spread with the probability of default of an AAA, it may well be an AAA in term of credit but we will never know 'cause the sample space is too small.But one thing we know for sure, the 200 bp spread isn't consistent with the default probability you would expect from an AAA.

[Clayton Davis]

Most , if not all, collateralized Mortgage Obligations are marked to the market for valuations purposes. Currently the meltdown is taking place because there is no market to mark these bonds to and thus these securities have no value. Many Institutional holders carry these positions on their balace sheet .
My questions is, what happens to these positions if large institiutions are writing them off ?

[Roger]

Jck, I know ratings have nothing to do with price. I was merely giving an example of a price movement based upon observed levels of distress and defaults that are clearly giving the market pause. My argument, if you read it, has nothing to do with price but attempting to divine the appropriate rating given the true credit risk involved. My options-based theory has everything to do with pricing default risk (and, therefore, related to credit ratings) and nothing to do with tying the market price of a security to the perceived incidence of loss. So you and I are in heated agreement. And I think I am plenty old to have an appreciation for the occasional dislocations between securities prices and default risk.

[jck]

The fact that some AAA paper is trading at 92 is meaningless, they are rating credit, not price fluctuations.
I don't know how old you are but I have been around long enough and have seen long US treasuries trading below 60.Should they have been downgraded?No and they were not because the moves were due to interest rate fluctuations not change in credit.
There is no basis and not enough data to claim that the ratings assigned to structured products are wrong...Eventually, when enough data is available, you will know if the ratings were wrong if the defaults or migrations observed don't fit the unique distribution of defaults that should be expected for a given rating level.
Repeat after me: this has nothing to do with price.What matters is :do you get your coupons and principal when due.That's what credit is all about.
By the way the coverage of this crisis by the media including the financial media has been beyond appalling.

[Tom]

It seems a bit too easy to game that investment grade AAA rating, though perhaps buyers have been all too willing not to look too closely as long as things were roaring along.

Satyajit Das has a book with many amusing (i.e., hair raising) anecdotes on the topic, "Traders, Guns and Money". In the tradition of F.I.A.S.C.O.