Risk Management: CDO Valuation and Ethics

In late 2007, I was unemployed, but had a line on a job with a minority broker-dealer who would allow me to work from home, something that I needed for family reasons at that point.  The fellow who would eventually be my boss called me and said he had a client  that needed valuation help with some trust preferred CDOs that they owned.

Wait, let’s unpack that:

  • CDO — Collateralized Debt Obligation.  Take a bunch of debts, throw them into a trust, and then sell participations which vary with respect to credit risk.  Risky classes get high returns if there are few losses, and lose it all if there are many losses.
  • Trust preferred securities are a type of junior debt.  For more information look here.

I got to work, and within four days, I had a working model, which I mentioned here.  It was:

  • A knockoff of the KMV model, using equity market-oriented variables to price credit.
  • Uncorrelated reduced discrepancy point sets for the random number generator.
  • A regime-switching boom-bust cycle for credit
  • Differing default intensities for trust preferred securities vs. CMBS vs. senior unsecured notes.

It was a total scrounge job, begging, borrowing, and grabbing resources to create a significant model.  I was really proud of it.

But will the client like the answer?  My job was to tell the truth.  The client had bought tranches originally rated single-A from three deals originated by one originator.  There had been losses in the collateral, and the rating agencies had downgraded the formerly BBB tranches, but had not touched the single-A tranches yet.  The junk classes were wiped out.

Thus they were shocked when I told them their securities were worth $20 per $100 of par.  They had them marked in the $80s.

Bank: “$20?! how can they be worth $20.  Moody’s tells us they are worth $85!”

Me: “Then sell them to Moody’s.  By the way, you do know what the last trade on these bonds was?”

B: “$5, but that was a tax-related sale.”

Me: “Yes, but it shows the desperation, and from what I have heard, Bear Stearns is having a hard time unloading it above $5.  Look, you have to get the idea that you are holding the equity in these deals now, and equity has to offer at least a 20% yield in order attract capital now.”

B: “20%?! Can’t you give us a schedule for bond is worth at varying discount rates, and let us decide what the right rate should be?”

Me: “I can do that, so long as you don’t say that I backed a return rate under 20% to the regulators.”

B: “Fine.  Produce the report.”

I wrote the report, and they chose an 11% discount rate, which corresponded to a $60 price.  As an aside, the report from Moody’s was garbage, taking prices from single-A securitizations generally, and not focusing on the long-duration junky collateral relevant to these deals.

In late 2008, amid the crisis, they came back to me and asked what I thought the bonds were worth.  Looking at the additional defaults, and that the bonds no longer paid interest to the single-A tranches, I told them $5.  There was a chance if the credit markets rallied that the bonds might be worth something, but the odds were remote — it would mean no more defaults, and in late 2008 with a lot of junior debt financial exposure, that wasn’t likely.

They never talked to me again.  The bonds never paid a dime again.  I didn’t get paid for running my models a second time.

The bank wrote down the losses one more time, and another time, etc.  How do you eat an elephant?  One bite at a time.  It did not comply well with GAAP, and eventually the bank sold itself to another bank in its area, for a considerably lower price than when they first talked to me.

So what are the lessons here?

  • Ethics matter.  Don’t sign off on an analysis to make a buck if the assumptions are wrong.
  • Run your bank in such a way that you can take the hit, rather than spreading the losses over time.  (Like P&C reinsurers did during the 1980s.)  But that’s not how GAAP works, and the CEO & CFO had to sign off on Sarbox.
  • A model is only as good as the client’s willingness to use it.  There are lots of charlatans willing to provide bogus analyses — but if you use them, you know that you are committing fraud.
  • Beware of firms that won’t accept bad news.

I don’t know.  Wait, yes, I do know — I just don’t like it.  This is a reason to be skeptical of companies that are flexible in their accounting, and that means most financials.  So be wary, particularly when financials are near or in the “bust” phase — when the credit markets sour.

By David Merkel, CFA of Aleph Blog