Partial “Defense” of Rating Agencies

As I noted in a post about commercial mortgage lenders recently, two pieces of the former GMAC commercial loan group have gone to both ends of the commercial success spectrum, and done so almost simultaneously.

The portfolio gang has gone down with the Chapter 11 ship, renamed Capmark a few years ago.  The credit team (Realpoint, LLC) has been elevated to demi-god status by being designated by the SEC as an NRSRO, or Nationally Recognized Statistical Rating Organization.

For those of us who took financial registration exams or deal with insurance or bank regulations, the NRSRO’s (Rating Agencies) can change everything, because their ratings determine how much capital is required by various types of institutions to hold various assets.

In other words, they may not be the leverage cops, but they might as well be, since the real leverage cops set up their rules based on the credit ratings.  So how could I be defending the Rating Agencies after they did their part in bringing about TEOTWAWKI (The End Of The World As We Know It)?

Simple.   It’s in their name.

When you parse it, NRSRO requires their judgments to be statistically derived.  When they were rating all those subprime mortgage bonds, there were sixty  years of statistics available, comprising the history of housing prices and mortgage credit performance.

And this wasn’t some political poll with a thousand respondents and 3% confidence intervals.  This was tens of millions of loans, spread across the entire nation and a dozen boom/recession economic cycles.  Any statistician will tell you that a dataset that size is going to give you pretty precise answers, as long as you understand that all historical data are historical.

I listen to the caterwauling over how the Rating Agencies should have known that house prices were too high.  My response is:  So what if they did know?  Would you have them throw out their data and substitute an opinion?  Even if that opinion was that things weren’t going to behave the way they always did?

The guy who used to come over to chuckle at our 24/7 amateur lightning chess tournament in my sophomore year at college has just put out a best seller that tells you why that’s a really dumb idea, if you didn’t already know.

How do you decide how far past the edge of the historical data the Rating Agencies  should have gone?  If house prices had not declined nationwide year over year for a lifetime, how far down would you have the Rating Agencies expect them to go once you decided they should ignore history?

Some analysts argued that house prices had been rising above trend, so they had to come back to that long term level.  Others said they should overshoot.  Those analysts are ignoring the house price experience of the late 1970’s and late 1980’s.  In each of those periods, we also had national house prices grow above trend, but in each case the correction consisted of slower growth until the economy and/or incomes caught up.

Then what would you do with inflation?  Adjust for it?  How?  If you use CPI as your inflation index, you’re setting up a feedback loop, since CPI has a such a large housing component.

Maybe the approach would be to look at local markets that actually had negative growth, like southern California in the early 90’s, or the Oil Patch in the 80’s.  The problem with those models is that jobs were the driver, as was the case during the “bubble” years in and around Detroit.

Detroit never took part in any bubble from 2001 to 2007.  Prices declined.  Makes sense as the Big Three slowly spiraled into the ground.  Ditto for the collapse of the oil market after the strong anti-inflation medicine of the Volker Fed, and for the southern California house price adjustment that was driven by the “peace dividend” of a shrinking military-industrial complex after the Soviet bear was de-clawed.

So, what, exactly, was the lost nationwide employment driver that would force the Rating Agencies to project future nationwide declines in house prices starting in mid-2006?  To refresh the memory and point out the statistical reality, reliable studies showed over the past 50 years that local and regional housing markets rise (and decline) with about a one-year lag behind local and regional job markets.

So, unless the Rating Agencies were going to “go Austrian” (the econometric equivalent of “going galt,” they were going to have to keep rating those subprime mortgage bonds using the data they had.  Anything else would have been opinion masquerading as officially sanctioned mathematical science.

OK.

Now that you’ve decided I’m some kind of wild-eyed apologist, let me tell you where they really screwed up.

Frankly, it’s a screw-up for which the Rating Agencies had plenty of warning, but their greed led them to ignore it.  That fatal flaw was in their assumptions of correlation between debt instruments, or more precisely, lack of correlation.

It got so crazy in the past ten years that I even met people well under the age of thirty who got paid major dollars to be “correlation traders.”

The job of a correlation trader, in simplified form, was to recognize when one or another investor was putting too much or too little value on one bond or loan versus the rest of the universe.  To go the opposite way and extract the mispricing from the market, a correlation trader would then enter a private, unregulated contract (a Credit Default Swap) with a counterparty to go “long” or “short” that same credit, or a basket of credits that should perform the same over time as the mispriced single credit, or “name.”

It all comes back to the Rating Agencies’ statistical approach to the world.  Nominally, one BBB rated bond should be just as likely as any other BBB bond to default, but that wasn’t the real issue.

The real issue was that a portfolio of BBB bonds should exhibit the BBB-related propensity to default if you looked at them as a group.  Obviously you couldn’t just take BBB bonds from car parts makers, because they  be under strain at the same time and enjoy great profits at the same time.  They were obviously expected to be highly correlated credits, in other words.

On the other hand, you might be correct assuming that sneaker manufacturers, drilling equipment operators, bookstores, online ad agencies and corn liquor (gasohol) makers might be uncorrelated.

Oops!  Now that I think about it, higher oil prices might good for the corn guys, the drillers, and even the online ad agencies if driving is getting too expensive.  You could even argue that people would stay home and exercise more or read, too, thus coming up with a way that they all were correlated.

The real problem, though, is bear markets. Everything gets correlated if the bear roars loudly enough.  That’s because a margin call (or capital reserve violation) says absolutely nothing about what will be sold to raise the cash.  If there are enough capital calls in the system, pure randomness will tell you that every investable asset out there will be getting sold, by somebody.

[Before someone goes AHA! Now I’ve caught you!  What about T-bills and other Treasuries?  They rally in bear markets!  Remember that the recipient of that cash from the margin call has to park it somewhere, and that somewhere is what we call the “flight to safety” trade.]

So why should the Rating Agencies have known this before they went whacky and gave Wall Street the keys to the Ponzi kingdom with their bogus CDO (Collateralized Debt Obligation) rating criteria?

They should have known because the original correlation kings (the LTCM hedge fund) proved in no uncertain terms in 1998 that Nobel winners, real rocket scientists, and traders that can bet $10 million in personal money on dollar bill serial numbers had no way to avoid a massive auto-correlation of previously uncorrelated trades once the avalanche starts.

They should have known because the dot-com stock market implosion led to extra-high numbers of credit failures of old-line industrial companies, and commodities markets with plunging prices in spite of macro-economic reasons for inflation.  It wasn’t just routers and servers, either.

Venerable 150-year-old Corning got hurt so badly when the dot-coms stopped spending on bandwidth that it threatened their diesel truck air filter business with extinction.  I watched in amazement as glassworks (the name I call them because they trade as GLW) entered into a death spiral mandatory convertible bond financing in the summer of 2002, and hedge funds shorted 200 million shares of stock in a day.

The truly absurd coda to the Corning story?  The day after GLW ate the bitter fruit of that financing, S&P downgraded their credit.  When I got a chance to talk to them about their action, the S&P crew explained that the downgrade happened because the stock plunged, and their correlation studies had shown that companies whose stock price falls tend to get into credit trouble.  I was amazed, because the convert had been issued so that Corning would not have any debt coming due!  Their balance sheet was bolstered by an extra $200 million of equity.  How could their credit be worse after the financing? The correlation clowns were clearly in charge, and traditional credit analysis was out the window.

By 2002/2003, there were tons of Collateralized Bond Obligations built out of corporate junk bonds in 1999 and 2000 that were going overnight from AA to default!  It didn’t matter how “diversified” the portfolios were.

Did that make the Rating Agencies pause?  In a word, no.

They decided that the problem was that corporations would default on bonds since they didn’t feel a real relationship with bondholders, but the slightly more senior bank loans wouldn’t default, because companies need their bankers.

The solution?  Stop issuing CBOs, and switch to CLOs (Collateralized Loan Obligations).  Problem fixed!

Then it got totally absurd.

BBB mortgage bonds had historically performed better than any other BBB rated debt.  There were literally hundreds of billions in private label issues over decades without a loss to the investment grade bonds, other than a case or two of outright fraud.  In a sense, the only risk seemed to be sleazy management at the issuer, not a problem of unusually high defaults by the home buyers.

With fees for CDO ratings running several times the fees for rating corporate or muni bonds, the rating agencies were quite aggressive in marketing their services to rate CDO bonds backed by pools of BBB rated MBS (mortgage bonds).  Better yet, they could rate “synthetic” deals that didn’t even have to own the bonds, but simply had to have Credit Default Swaps that “referenced” a diversified pool of MBS bonds.

Needless to say, when they failed, they all failed at once, or close enough to it that even the better mortgage lenders’ bonds were soon sucked down into the maelstrom.

We’ll talk later about the Wall Street dealers controlling the servicing companies, and how they could literally affect the performance of the bonds by how they managed their servicing operations.

I’ve got to stop by my aluminum foil milliner to write that story.

hh

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: