PrefBlog

As reported by the WSJ, the BIS Quarterly Review deprecated the widespread use of the ABX indices when estimating credit losses on sub-prime.

They make three major points regarding pitfalls in using the ABX:

• Accounting Treatment – many subprime RMBS are held by investors who do not mark-to-market, resulting in a wide gap between reported writedowns and estimated fair value of losses.
• Market Coverage – the indices only sample the universe … but this is probably not a big deal, the sample is reasonable.
• Deal-Level coverage – “Similarly, ABX prices may not be representative because each index series covers only part of the capital structure of the 20 deals included in the index … In particular, tranches referenced by the AAA indices are not the most senior pieces in the capital structure, but those with the longest duration (expected average life) – the so-called “last cash flow bonds”.”

The last point is very important and forms the core of their argument.

This information has been available to non-specialists for some time … I could have sworn I mentioned it specifically on PrefBlog at one point, but can’t find the reference … and at any rate I should have emphasized it myself when discussing the fair value estimates. The best tracing to this information I can give is … in my discussion of the Greenlaw paper, I referenced the comments to Econbrowser’s Mortgage Securitization post, in which I referenced Felix Salmon’s How to test the accuracy of the ABX post, which referenced his prior ABX RIP post, which referenced Alea’s ABX Extra piece, which … highlighted the information.

The guts of the BIS argument are given only in the notes:

Incomplete coverage at the deal level further reduces effective market coverage: typical subprime MBS structures have some 15 tranches per deal, of which only five were originally included in the ABX indices. As a result, each series references less than 15% of the underlying deal volume at issuance.

Duration effects at the AAA level are bound to be significant for overall loss estimates as the AAA classes account for the lion’s share of MBS capital structures. Using prices for the newly instituted PENAAA indices, which reference “second to last” AAA bonds, to calculate AAA mark to market losses generates an estimate of $73 billion. This, in turn, translates into an overall valuation loss of $205 billion (ie some 18% below the unadjusted estimate of $250 billion).

I will suggest that even the PENAAA indices will be not very well corellated with actual credit analysis, but these data certainly provide an indication of the value of subordination.

The last review of loss estimates was the discussion of the OECD paper; there is not really enough data in the BIS note to put it on the board as an estimate … but it certainly seems to support the “lowball Bank of England estimate” rather than the “terrifying IMF estimate”!

The main articles – apart from the “Overview” and “Highlights” – in the BIS Review are:

• International Banking Activity Amidst the Turmoil
• Managing International Reserves: How Does Diversification Affect Financial Costs?
• Credit Derivatives and Structured Credit: the Nascent Markets of Asia and the Pacific
• Asian Banks and the International Interbank Market

Update, 2008-6-11: This post was picked up by iStockAnalyst and attracted a puzzled comment on the Housing Doom blog:

Here’s a technical criticism of the ABX index that was posted yesterday. If you can understand what this guy is complaining about you’re doing better than me.

“BIS Quarterly Review Deprecates ABX Benchmark for SubPrime”, James Hymas, iStockAnalysis, June 10, 2008.

Well, I guess for new readers who have not been assiduously reading my remarks, this post will be a little cryptic!

The gist is: in order to make a sub-prime RMBS with a large AAA component, it must be tranched; for example, the Bear Stearns ABS I use as a model had a total value of USD 395-million, which was divided into seven publicly marketed and three private tranches … payments went first to the USD 314-million senior tranche, then on down the line until the final (public) tranche of USD 4.5-million, initially rated at BBB- and downgraded to B on August 24, 2007 gets paid … if it ever does! I looked at the economics of tranching very early on.

This particular issue is relatively simple, but there are issues with more tranches … as the BIS piece above notes, the average is 15 tranches per deal of which … maybe five? I’m guessing … would be rated AAA.

So you have five AAA tranches that get paid one after the other. Obviously, the first one to be paid is the safest and most likely to meet its committments; the ratings agencies, in their infinite wisdom, determined that tranche #5 was also good enough to warrant an AAA rating. The ten that came after that would be sold to the public with worse ratings and higher yields.

So … the market goes blahooey and all of sudden banks and brokerages, with a need to mark their inventories to market to meet the accounting rules, are stuck with the problem: how to assign a market price to inventory comprised of relatively small issues representing a class of security that simply isn’t trading at all. After discussion with their accountants, they determine that the methodology least likely to get them into trouble is to use the Markit ABX indices as a benchmark. This methodology is also used by third parties (e.g., the OECD, referenced above) to estimate what the total losses for the entire universe of about USD 1.4-trillion might be.

There are a number of problems with this approach. Firstly, the Markit ABX index only rates the worst tranche for each credit rating … the value for the AAA-rated index is based entirely on tranche #5 of our example, even though there are four other tranches rated AAA in this deal, each of which (this is the important bit) are safer than the chosen tranche by definition.

Secondly, the ABX index is based on Credit Default Swaps, a market that is now basically dysfunctional.

Thirdly, we are interesting times; getting out of sub-prime paper is currently a “crowded trade” and the cash market itself is dysfunctional (although it is starting to show signs of life). The market price of the securities does not have a lot to do with the present value of its expected cash flows.

All these factors mean that estimates of sub-prime losses that mark-to-market off the Markit ABX index are (a) highly imprecise, and (b) greatly overstated.

The new PENAAA index referred to above is based on the penultimate tranche in the AAA tranche – tranche #4 in our 15-tranche mini example. A quality spread is quite evident; using the value of this index to estimate market values over the universe results in BIS computing an estimate for losses that is much, much lower than the initial estimate.

This entry was posted on Tuesday, June 10th, 2008 at 12:59 pm and is filed under Sub-Prime!. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.