Opinion

Felix Salmon

What was special about the Dead Presidents?

By Felix Salmon
May 12, 2010

The Dead Presidents CDOs now reportedly being investigated by the Justice Department were not your garden-variety synthetics:

One feature of the Morgan Stanley deals was a structure that could increase the magnitude of the bullish investors’ exposures to the underlying mortgage bonds. This feature, which was disclosed in some offering documents, made it more likely that such investors could lose money if the underlying bonds performed poorly.

This bears a family resemblance to what the NYT reported back in December (h/t Alphaville):

Morgan Stanley established a series of CDOs named after United States presidents (Buchanan and Jackson) with an unusual feature: short-sellers could lock in very cheap bets against mortgages, even beyond the life of the mortgage bonds. It was akin to allowing someone paying a low insurance premium for coverage on one automobile to pay the same on another one even if premiums over all had increased because of high accident rates.

I’d love to know more about this feature. I’m not looking for a quick one-sentence summary which can be dropped into a newspaper article, but a detailed explanation of exactly what it was and how it worked. Does anybody have offering documents from Citi or UBS for these deals which might include such a thing?

I think this shows the limitations of print-based journalism, and the long way we have yet to go before newspapers fully embrace the web. Both the WSJ and the NYT give the impression that they have seen and understood the structures in question, and that they’re simply summarizing them in order to make their stories easier to read for a broad audience. That’s fine — but once you’ve done that, do please give the full details online to finance geeks who want to understand the deals on a finer-grained level.

There were lots of synthetic CDOs structured and sold at the end of the subprime boom, but the ones being singled out by regulators and prosecutors seem to be the unusual ones — first the Goldman deal which was created at the behest of John Paulson, and now the Morgan Stanley deal with this mysterious embedded structure. It would be a great service if the news media, rather than just trying to report the news, also published primary documents and full details of what they’re writing about, so the rest of us can come to our own conclusions.

Update: ProPublica has found a bunch of Jackson prospectuses, if somebody fancies reading them.

Comments
13 comments so far | RSS Comments RSS

Sounds like a CPDO or CPPI to me, which were considered exotic, but not necessarily novel. Happy to be corrected if anyone thinks otherwise.

Posted by Sandrew | Report as abusive
 

Sorry Felix, I do not have the docs, but … why would you think anything other than what Sandrew said, which seems the obvious interpretation? Some variation of dynamically adjusted leverage.

I agree that the bizarre NYT analogy is anti-knowledge: reading it makes you stupider.

Posted by Greycap | Report as abusive
 

Maybe I’m wrong. As far as I know, during the short-lived CPDO craze, such deals were typically written on corporate credits (dynamically linked to OTR CDX and Itraxx indexes at first, and later to bespoke baskets). These Dead President deals are reported as mortgage exposures. Anyone aware of a CPDO backed by CDS-on-MBS?

Posted by Sandrew | Report as abusive
 

The deal feature that they are alluding to is probably Auto Reinvest. Auto Reinvest was a feature of synthetic ABS deals that worked to keep ABS CDOs outstanding as long as possible because clients didn’t want to have to continually buy more ABS CDOs as the deals amortized (and risk losing there already amazing 20-50bps of coupon). In fact, to my knowledge, a client came up with the structure in the first place!
The short explanation of how it works is as follows: Subprime mortgage bonds amortize so ABS CDOs amortize (or were supposed to). On a synthetic deal, you could structure the CDS contracts underlying the CDOs to reference a specific subprime mortgage bond but allow the notional that was referenced on the underlying bond to change over time. As the underlying bond amortizes, you simply have the amount of notional increase by the same amount. In this way, there is no amortization on the top level CDO tranches. At a certain point (usually 10-20% of original face), the notional is amortized down at 5-10 times the rate of the underlying mortgage security, thus paying off the tranche.

The rub here is that, because the underlying mezzanine subprime mortgage bonds would, in the best case, likely pay down pro rata with the rest of the structure, you are increasingly becoming exposed to the bottom 20% of the collateral pool underlying the deal. This is problematic even in a non subprime meltdown scenario because the tails of these pools, even to casual observers, were cuspy at best.

Now, one thing that I find bizarre about this whole investigation is that Morgan Stanley had no involvement in the marketing of this deal. They played a similar role to Paulson, who wasn’t investigated.

And for the record, no CPDO on ABS ever got done.

Posted by jakethesnake | Report as abusive
 

Thanks for the great explanation, Jake. Sounds like the key difference between this structure (does it have a name?) and a hypothetical CPDO-on-ABS is that only prepayment risk in this case (as opposed to overall market risk in the case of a CPDO) triggers incremental leveraging. Fairly characterized or am I missing something? e.g. Is there structuring on the liability side (i.e. tranching) in these deals?

Posted by Sandrew | Report as abusive
 

OK, Jake, the way I read you I was completely wrong (sorry Felix) and there were really no leverage adjustments at all – the idea was just to refer to the underlying bonds in % current notional terms instead of absolute. If both sides of the structure were synthetic, no portfolio adjustments were needed. The asset buyer was saved reinvestment risk on the less risky cashflows of the structure at the expense of greater capital exposure to the more risky cashflows via survivor bias. Doesn’t sound like such a great idea when you put it that way. And it seems you were still exposed to prepayment risk on the 10-20% wind-down threshold?

Posted by Greycap | Report as abusive
 

Sandrew, there really isn’t a name for the structure like a CPDO. It’s just a type of Synthetic ABS CDO. And yes, there was tranching on the liabilities side. The brilliant part about synthetic CDOs generally is that you don’t even have to place the entire capital structure (though I believe they did in these deals) because no upfront payment exist! Tranching is effectively just a rating agency output. In terms of liability “amortization” (protection notional reduction really), the structures were probably sequential, AAA then AA and so on.

Greycap, you are correct in your assessment that you are still exposed to prepayment risk in the turbo amortization phase of each security, but the real risk was actually that you write down during this phase. ABS CDS contracts, like ABS bonds, allow for writedowns over time, unlike corporate cds which only has a credit event during an insolvency, restructuring etc. For the underlying ABS, at each payment date, the amount of collateral is measured against the amount of liabilities outstanding and any difference is considered a writedown. This results in partial writedowns over time until, tranche by tranche, a securitization writes down. With Auto Reinvest, you basically end up with exposures to 5-10x the amount of original notional with the collateral dregs as backing. A 10% loss on the tranche translates into a 50-100% loss on a piece of the collateral backing your CDO bond!

Posted by jakethesnake | Report as abusive
 

We need to raise the tax on Wall Street payouts to a 60% rate

http://storyburn.com

Posted by STORYBURNcom2 | Report as abusive
 

I *think* we’re saying the same thing, but let me check.

“A 10% loss on the tranche translates into a 50-100% loss on a piece of the collateral backing your CDO bond!”

So the reference security is an ABS tranche? In the simplest case, my AAA synthetic bond refers to a single AAA ABS tranche? Then, idealized case again, imagine zero defaults in the underlying mortgage portfolio so no change in protection to the reference tranche but 80% principal reduction due to amortization + prepayment. At this point the reference tranche has only 20% of original principal but my bond still has 100%. However, the characteristics of my reference security have changed materially: in effect, I have substituted 4 copies of the worst 20% of assets for the best 80%.

My point was that I have not increased my leverage in the normal sense. My *dollar* leverage to the reference security is 5x but that is only because there is 5x less of it. If it had not changed character, that would make no difference. It would be like an equity derivative that pays 1m x 5 shares versus one that pays 5m x 1 share.

Or have I misunderstood you?

Posted by Greycap | Report as abusive
 

You’re correct Grey. In a sense, the overall amount of exposure has not increased but the relative leverage has increased. You were originally exposed 1:1 to an ABS bond in your example. At the turbo time, you were exposed 4:1 and the underlying characteristics of the instrument (which incidentally have deteriorated materially). Semantically speaking, is this a change in leverage? I would argue yes. The payout profile of the investment has changed materially since you purchased it.
Has your total notional changed? I would agree and say no.

Posted by jakethesnake | Report as abusive
 

I really really love this conversation — but does anybody know whether this is the feature that the NYT and WSJ were talking about? I’m not sure I trust myself to work that out from reading the prospectuses, I’d rather see a dealsheet…

Posted by FelixSalmon | Report as abusive
 

Auto Reinvest was one of the features that would qualify as “unusual” and definitely allowed the buyer of protection to short the bonds in an extended fashion. I’ve read today that ABSpoke and Baldwin were also now reported as being looked into. These had the same feature. To be clear though, MS was hardly the only bank or sponsor to use this feature.
I would venture that there is probably a misunderstanding related to being able to short a bond past its maturity. That doesn’t really make any sense. What they may be confusing is that the short exists past the bond’s Weighted Average Life (time weighted principal receipts, the balancing point in principal cashflows on an ABS security) but not complete maturity. That would be a very refined point to pick up on though…

Posted by jakethesnake | Report as abusive
 

As I understand it the posters above are correct, the CDS did not extinguish in line with the amortisation of the pools. So effectively it was like having 100% of your CDS on the last mortgage to repay, rather than having it exposed equally to the entire pool of mortgages. People who wanted to short the pool wanted this to make sure that they didn’t effectively pay premium for the 1st year on 100% of the pool and then watch as the pool amortised by 40-50% over the first 12 months as early repayments occurred. So they created this structure that was effectively amortisation immune. I believe that it’s described relatively well in Michael Lewis’s The Big Short.

Posted by drewiepe | Report as abusive
 

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/
  •