ABX Follow-up

Last Thursday's post on the ABX index was the most popular post in the history of this blog, measured by hits. I got a couple comments both on and off line, so I'll try to address a few of these here. Bear in mind that I'm not a synthetic ABS trader, so while I think I understand the ABX index pretty well, I'm sure some understand it better.

First question is how many defaults might it take to wipe out these BBB tranches. That depends on a number of factors, including how quickly the defaults happen and what the recovery is. In March of 2006, I read a scenario analysis produced by Citigroup which estimated that cumulative losses of 11% would not be enough to cause any defaults in investment-grade securities. In this case a "loss" is net of any recovery. So if you assume 50% recovery then defaults could be as high as 22%.

Just a few weeks ago Bear Stearns held a conference call on this subject. There they estimated that the "fair value" for the ABX.HE 06-2 BBB- index was around $90. That was based on assumption of defaults around 15% and recovery of 55%. They actually tailored the default assumption to each of the 20 deals underlying the index. Under their scenario, only two of the underlying deals suffered actual losses, one of 18% and the other of 0.5%.

What of the issue of loans with CLTV greater than 100%? In other words, situations where the borrower took extra cash for "improvements" but never actually made the improvements. Obviously such deals will have low recovery. We simply can't know how prevalent this is, particularly in any given deal.

Is the 70% recovery figure I threw out accurate? I tossed that out as a rule of thumb some people use, but obviously that is influenced by the long history of mostly rising home values. Banks enjoyed strong recoveries in the past because the collateral on the loan was an appreciating asset: as time passes the recovery value only improves. I think we can all agree that we are at least facing a situation where home values will rise very slowly if at all for the next couple years. So recovery is likely to be lower than 70%. How much lower? Depends on which loans go bust when.

What if we suffer a severe bear market in housing prices? This will likely cause both more defaults and lower recovery. How bad the housing market will get is an argument for another time. I reiterate here, however, that other areas of the bond market are not trading like a severe housing downturn is likely. From mortgage lenders to home builders, bond spreads are still fairly tight. The ABX index has been the only area where shorting the housing market has worked.

Finally, when I said someone looking to buy protection on the ABX.HE 06-2 BBB- would have to pay up 18 points at the outset of the deal, here is what I mean. The ABX index trades on a dollar price with a fixed spread payment. If a trade occurs at par, no cash changes hands at the outset of the trade, and the buyer of protection pays the fixed spread to the seller of protection. However, if the trade does not occur at par, one party or the other makes an up-front payment. Subsequently, the buyer of protection pays the fixed spread to the seller. So if the index is at a premium to par, the seller pays the buyer up front. If its a discount, the buyer pays the seller up front.

So yesterday the index closed at $79.04. So in addition to paying the 242bp fixed spread periodically, the buyer of protection must also pay the seller of protection $20.96 up front. Think of it like a plain vanilla corporate. If the credit weakens and the bond widens, the price will fall to compensate investors for the additional risk.

As always I welcome comments.