The Weakness of Marking Subprime Bonds to Market

There's something a bit weird about the ABX prices. Let's assume, for the sake of argument, that the value of an ABX contract is related to the value of a typical mortgage-backed bond with that rating. (It isn't, but let's leave that to one side.) If you look at the current indices, the AAA contract is trading at about 66 cents on the dollar, the AA contract at about 35 cents, and the BBB- contract at about 19 cents.

Now remember that the bonds are structured on a waterfall basis. If there's any loss on a given contract, that means that all the tranches below it have been wiped out completely. If a AAA tranche is trading at distressed levels, that means there's a good chance it will see losses. In order for that to happen the AA tranche would have to get no money at all, and the same goes for all the A tranches and all the BBB tranches. But you can go all the way down the waterfall and still see significant non-zero value even in the BBB- tranches. So the weird thing is this: looking at the valuation of the AAA contracts, you'd think the BBB- contracts would be utterly worthless. But they're not.

(Update: As ABSGuy points out in the comments, this analysis is flawed. Doomed lower tranches will still get some cashflow between now and the time they default, even if the AAA-rated tranches eventually suffer losses as well. But it's not flawed in a way which really damages the rest of my argument.)

In his Fortune column today, Peter Eavis goes into some detail about the value of AAA-rated subprime bonds:

A Wall Street bank that trades AAA-rated subprime bonds is currently quoting prices for such bonds of around 88 cents on the dollar, or a 12% discount, for loans made in 2006, and 78 cents on the dollar, or a 22% discount, for loans made in 2007.

If Fannie Mae marked its AAA-rated subprime bonds to market, then, says Eavis, it would have to take a loss of well over $4 billion.

But I have a feeling that in this particular instance marking to market is a pretty crappy way of working out how much these bonds are worth. And to get an idea why, it's worth taking a look at the Paulson plan for modifying subprime mortgages. Everybody agrees that the best solution for homeowners and bondholders alike would be for smart and careful loan servicers to look at each mortgage in detail on a case-by-case basis. But everybody also agrees that the mortgage servicing industry simply doesn't have the capacity to do that.

A similar problem obtains with subprime bonds. A smart and careful buyer with a lot of expertise and a lot of time can look closely at the pool of mortgages underlying any given AAA-rated bond and determine just how likely it is to suffer impairment. But the problem is that the pool of such smart and careful buyers is small indeed, and it certainly doesn't include the mortgage-trading desks of large sell-side investment banks, who are trading in and out of thousands of different bonds and can't be expected to be up to speed on the idiosyncracies of each one.

What does mean for the market price of AAA-rated bonds? Let's say that most AAA-rated bonds really are safe and really will be paid in full, as the residual value in BBB-rated tranches would tend to imply. The problem is that no one knows which AAA-rated bonds are the bad apples which won't be paid in full. And because of that uncertainty, the market price of the bonds falls. (There's nothing the market hates more than uncertainty, especially when it surrounds a product which is meant to be risk-free.)

The mark-to-market price of AAA-rated subprime bonds, then, does not represent the market's best guess as to the present value of those bonds' future cashflows. Rather, it represents an ignorance discount. If you can do your due diligence and work out exactly which AAA-rated bonds are kosher, and if you can then find some of those bonds being offered for sale on the open market, then you can probably make out very well for yourself. But both of those ifs are very large: poring over mortgage portfolios is hard work, and then finding a specific tranche of a specific bond is harder still.

What all this means for Fannie Mae is that it might have to take a large hit to earnings in order to reflect the mark-to-market value of bonds it has zero intention of ever selling. As a result, it might bump up against the unhelpful OFHEO capital constraints, and be forced to sell high-quality assets when it should by rights be a buyer. A bit of regulatory forbearance would not go amiss here, although I doubt that Peter Eavis sees it that way.