"Felix Salmon at Portfolio’s market movers blog has a very thorough post on super-senior tranches of CDOs, following on from his exploration of synthetic CDOs a couple of days before. Here is his spot-on conclusion:
Ultimately, then, the error was one of management, not of financial technology. The banks’ balance sheets — and those of their off-balance-sheet vehicles — were expanding faster than the banks’ executives and risk managers could really keep a handle on. And rather than call a halt to that which they didn’t fully understand, they handed down edicts instructing the CDO desks to keep on dancing for as long as the music was playing. Most of the executives probably never even heard the term “super-senior” until those tranches started getting written down. It was their own incuriousness, rather than any CDS technology, which was really their undoing.
Felix’s contention is that CDS technology - which is an intrinsic part of the super-senior gambit, has so far actually worked. It’s the misuse of it which has failed.
First, though, a little explanation."
I agree with Mr. Salmon. The Super Senior Tranches aren't nanobots independently fiddling around with our tranches for the pleasure of seeing us lose our shirts. The not understanding stuff I don't credit.
"As Felix notes, the super-senior tranche, a curiosity (originally) peculiar to synthetic CDOs, is something of a conjurer’s trick. The super senior tranche is essentially a large slice of the CDO pie which is senior even to a triple-A tranche, hence its moniker. This all cuts back to the magic of structuring: the single most important invention in finance, if not economics, in the past few decades.We’ve knocked up the below, loosely based on the kind of thing Janet Tavakoli - a structured finance expert par excellence - has been saying for years:
On the left is a run-of-the-mill vanilla structured finance vehicle. Based on assumptions from the rating agencies (and a lot of to-ing and fro-ing, skydiving, lunching and clay pigeon-shooting, sometimes even portfolio analysis) a subordination level - known as an attachment point - will be set which will determine the percentage of the portfolio against which AAA rated bonds can be issued.
The attachment point is based on stress tests which model millions of scenarios to see how the portfolio will perform under those scenarios. On our above diagram, the attachment point is at 80 per cent. Notwithstanding the erroneous assumptions by the rating agencies that led to the mis-setting of those attachment points, the structure is a thing of genius and beauty.
The right hand diagram shows the same structure, but with a super-senior tranche incorporated. This is where things start to get rather messy. And indeed, this is where the technological snafu occurred that caused a lot of the current crisis."
To me, this graph is pretty clear, so let's go on:
"The first thing to understand is why super-senior technology started to be used: low yields.
Yields on AAA ABS were getting low by 2004/2005, low enough that there wasn’t really much interest in buying them. Banks were increasingly finding that with their securitisations and CDOs they couldn’t pass on the AAA tranches or the equity tranches - only the middle stuff. Of course, to complete a deal, you’d need to sell the lot. Or else take some of the risk yourself."
They wanted to make more money. Fine.
"Super-senior was in effect, a proposition that killed two of those birds with one stone. The super-senior position was basically a way of further dividing the risk within the AAA tier of the structure. And by doing that, enabling an effectively “subordinate” AAA tranche to yield more. Banks kept an interest in the CDOs, but a very low risk one, and investors got higher yielding AAA notes."Not the original risk though on the tranches. They basically gave one guy the first losses. That's it.
"To boot: from the rating agencies point of view - and indeed that of most investors - nothing has changed between the two structures. The AAA notes in both are still above the 20 per cent attachment point specified in the rating agency models."
However, a commenter says that within a tranche, this narrowing, Moody's demands some extra capital for the small, more risky, tranche.
However, strictly speaking, the original risk hasn't changed. But the risk within the tranche has, so doing what the commenter says makes perfect sense.
"But there is a fundamental difference. Say we invested $10m in both structures. We’ve shown this as the grey shading. Our investment in the left hand vanilla CDO yields +20bps. Our investment in the right hand CDO (with super senior) yields +40bps. But say we have a crisis, and the losses on the underlying portfolio exceed the 20 per cent attachment point modelled by the rating agencies. We’ve shown this on the diagrams above as a horizontal red line.
Whereas in the left hand vanilla structure, this only eats into a small portion of our “vertical” stake in the AAA notes, in the super-senior structure, our investment is devastated.
Clearly then, an AAA-rated tranche subordinated to a super-senior tranche is riskier than a normal AAA stake. So why didn’t the rating agencies capture that? Because, believe it or not, they pretend the super-senior stake doesn’t exist. The rating agencies only go to AAA. There is nothing above that. How very un-spinal tap."
We've already explained this. The whole point is that someone takes the losses first, or in a higher percentage. I've made it as simple as I can.
"Here’s where we’d add a wrinkle to Felix’s post. Salmon writes:
So the bank’s remaining risk, after selling off that triple-A-rated synthetic tranche, has been brought down to safer-than-triple-A levels. Some of the banks referred to it as a “quadruple-A” risk, although that’s not a real-world rating. But the banks were so comfortable that defaults at that level could never happen that they didn’t feel any need to hedge themselves against it happening.
But in fact, the banks did want to hedge that risk out. The banks weren’t just happy to hold on to all those low yielding super-senior positions. Although the Basel risk-weighting requirements were relatively minuscule, they did still tie up reserve capital. Besides, banks had a way of doing better: by buying cheap protection on those super-senior positions using CDS.
This is where we might also part with Salmon slightly in his conclusion, and say there is a risk inherent in the financial technology itself, albeit one exploited to absurdity by gnomic structurers at banks and willfully ignored by banks’ management teams."
This is correct. It's riskier. That's the point. But the point about the nanobots still applies.
Not necessarily. The main point was for the remaining trancher to make more money.
"Monolines and insurance companies like AIG thus wrote billions in super-senior CDO protection while not treating it as a highly levered position. And in turn, they also ignored the leverage risk because they were spared the need to meet collateral calls by virtue of their own AAA ratings. A delusion which didn’t last long."
That's true, but within the tranche. You have to keep the two things separate. The original tranches, and the splitting up of the original tranches.
"Citi found another way of getting cheap protection on its super-seniors. It set up special LSS - leveraged super senior - conduits which issued commercial paper and used the funds to write back CDS positions. It was a method that allowed the bank to increase its CDO structuring business massively though 2005 to become the world’s largest CDO issuer in 2006.
And then what happened? Many ABS-backed CDOs have imploded. The banks and rating agencies’ attachment point assumptions were so wrong that even super-senior positions have been affected in CDOs which used them.
For the monolines and insurers, this wouldn’t have been such a dreadful problem had they not also invested in the underlying notes of many CDO structures: a move that led to the rating agencies downgrading them, and thus exposing them to collateral calls on their billions of leveraged super-senior swaps.
Citi’s dalliance with LSS conduits and the commercial paper markets was equally catastrophic. In the summer of 2007, it caused the collapse of several large conduits in Canada. That in turn precipitated a global buyers strike in asset-backed CP. Which spread, in turn, to a buyers strike of all financial CP. Thus ratcheting up the threat of banking collapses, and indeed, leading directly to them, in Germany, and in the UK (Northern Rock).
Alas, the leveraged super senior catastrophe is not over. Most of the severe unwinds have been taken against the value destruction in ABS CDOs. The synthetic CDO market - particularly the corporate synthetic CDO market - ironically the place where super-senior technology originated, has not yet collapsed.
One article which has been doing the rounds is a piece in Australia’s Business Spectator by Alan Kohler. He writes:
As the world slips into recession, it is also on the brink of a synthetic CDO cataclysm that could actually save the global banking system.
It is a truly great irony that the world’s banks could end up being saved not by governments, but by the synthetic CDO time bomb that they set ticking with their own questionable practices during the credit boom.
Kohler’s assumption is that most synthetic CDOs were written as naked positions: banks simply bought CDS protection from the vehicles they created in a one-sided bet. The banks, in other words, would only win in the event of a default, when the vehicles they created would pay them out.
Unfortunately that’s not the case. The very invention of the synthetic CDO was born out of the desire for banks to hedge existing loan positions on their books. JPM started it all with a deal called BISTRO. In fact, there are already examples of synthetic deals coming a cropper and forcing banks to unload corporate loan exposures in order to try and avert onerous capital charges. Barclays and the unwind of the appropriately named Black Diamond being a case in point.
And where banks didn’t own the underlying assets the portfolio CDS referenced, they would almost always write protection and sell it into the market to offset their “naked” position with the synthetic CDO they sponsored.
The actual CDS underlying synthetic CDOs are probably then best thought of as being part of a zero-sum game played by the financial sector.
The risk is with the noteholders of the synthetic CDOs. And just as with ABS CDOs, those noteholders are likely to see some very severe losses. Synthetic CDOs are only now about to experience the same kind of dramatic collapse that plagued ABS CDOs way back in late 2007 and early 2008.
The trigger will be the growing number of corporate defaults, which just like assumptions on subprime mortgage defaults, was, in many synthetic structures, underestimated. Barclays analysts see a “rising tide” of synthetic CDO downgrades on the horizon. Downgrades which could well have huge regulatory capital requirements on the super-senior positions banks have on their books."
This is a lot of speculation, which might be correct, but I don't know.
As to our original problem, here's my quote, which might or might not make things clearer:
Posted by Don the libertarian Democrat [report]I’m puzzled by the problem. The whole point is to divide up risk, going from least risky to more risky. Clearly, on the ladder of risk, to make this simple, what’s below is riskier than what’s on top. If you’re an investor, that would seem to be the thing you’d really like to know, and it’s simply drawn and simply understood.
In order to create the various levels, a math model was used that allowed you to see bumps or groupings of defaults that you could assign risk to and sell it. The original division has, say, 3 levels, and now you’ve created a fourth. The previous poster’s point is how did this happen. Your model gave you three levels of risk that you could package, so where did the fourth come from? Is it a new model? Has a fourth bump or group been added?
The answer looks like “no”. What’s happened is that within that particular level of risk, one person has agreed to take the losses first. In a sense, that’s what the whole thing looks like. Who takes the hit first. But now that’s all it is, while the original model included default rates. Now, within that grouping of rates, you’ve created a parceling of losses.
In essence, what this Super Senior did was distill the original model down to its basic concept: Who takes the losses first. Within the original risk model which applied to the whole pie, you’ve simply taken one slice, still under the original risk model, and had two people agree that when the risk comes calling, the first person will lose money before the second. It’s a risk within a risk, but , again, when you look at the graph or ladder, it’s pretty obvious where the agreed upon risk stands.
Having said it’s all obvious, I’m not sure that anyone will understand what I just said. I’m not sure I do. But I gave it a shot.
Posts
No comments:
Post a Comment