| Unfortunately, back in January 2007 when this column was disparaging Constant Proportion Debt Obligations (CPDOs) few paid attention. In the previous 12 months more than $600 billion in synthetic CDOs had been issued. The structured credit party was in full swing, the punchbowl overflowing and Chuck Prince (remember him?) was out on the dance floor.
Perhaps the biggest surprise of the credit crunch has been that CPDOs actually lasted longer than many of the institutions that created them. However, last week Bloomberg reported that three of the very first CPDOs, launched by ABN Amro in 2006, had finally succumbed. Once upon a time supposedly clever people at the credit rating agencies had thought these structures were worthy of triple-A status. Last month they were downgraded to D. Investors will be lucky to get back 10 cents on the dollar.
CPDOs are the apogee of the excesses of the credit bubble. Financial engineers created an instrument that combined opacity, complexity and leverage in spades. If a banker sat down with an investor today and offered them a structured product with a 200bp spread over a CDS index there would be a robust debate about its merits. If it was then explained that this yield was only attainable because the structure was 15 times leveraged, incredulity would probably turn rapidly to laugh-out loud mockery.
In reality such a meeting would not take place. Both the seller and the potential buyer (most likely a structured credit hedge fund) have probably lost their jobs. There are some parts of the structured credit demimonde that will not be coming back. Ever. CPDOs have ceased to be. However, what about the broader market for credit risk transfer via CDOs, CLOs and the like, the three-lettered products that were popular long before anyone had ever conceived of a CPDO? Will they suffer a similar fate?
Before we can answer that question it might help to delve further into what went wrong with CPDOs. Unsurprisingly, if you lever something 15 times you create an instrument that is inherently fragile. However, it was the leverage that was supposed to protect against normal spread widening in the underlying CDS indices (iTraxx and CDX). The idea was the leverage would enable the CPDO to capture higher premiums that would outweigh the mark-to-impact of spread widening. This is a post-modern Ponzi scheme.
But spreads widened remorselessly. This had not been in the script. All of the models had been built on the assumption that spreads are mean reverting. Strike one. Secondly, liquidity disappeared. Strike two. Finally, defaults rose and leverage really did turn out to be an Achilles heel. Strike three.
Moodys and Standard & Poors have admitted to basic modelling errors. But the biggest mistake was their largesse with triple-A status. A CPDO getting a better rating than Belgium or oil-rich Qatar is the most egregious example. Triple-A paper ought to be so safe that your granny can buy it, see it trade no lower than 99.5 and get paid out at par at maturity. That is what triple-A is supposed to mean. It precludes analysis and modelling. That is the standard that the rating agencies should now apply.
What went wrong with CDOs? A CDO is relatively easy to understand. Take a bunch of BB-rated corporate loans, say, and then tranche the cash flows by applying historic default assumptions. If 85% of the cash flows get paid to investors in the senior tranches, then it is certainly the case that this tranche of the structure deserves a higher credit rating than BB. However, some believe that the very act of credit risk transfer creates moral hazard, as the lender no longer cares how the borrower performs. It is surely an issue regulators will be thinking about.
The problem specific to CDOs of ABS has been a failure to understand how correlation works. No one can say that we were not warned. A correlation storm hit credit markets in May 2005. The downgrading of General Motors and Ford to junk, arguably a relatively minor credit event as they did not default, caused knock-on effects far removed from the direct holders of the auto-makers debt. In particular, the pricing models for certain tranches of CDOs and synthetic CDOs were shown to be faulty.
Delta, the sensitivity of CDOs to underlying spreads, was underestimated. Wind the clock forward to a significant credit event, real defaults across the US mortgage markets, and neat models of correlation go out of the window. To get to grips with how these structures will behave what needs to be teased apart are all of the pair-wise correlations of all of MBS bundles of all of the underlying mortgages. To say this is difficult is an understatement. But if CDOs are going to survive it will have to be done.
The credit crash has claimed another victim, so farewell CPDOs (20062008). CDO-squared and -cubed are dead men walking. Will more plain vanilla CDO structures survive? CDOs backed by bank loans will. Credit risk transfer is too important an innovation and the clock cannot be turned back. Those backed by ABS might also slowly return to favour. However, investors will need to be persuaded that the new models rating agencies introduce work, or be prepared to do some pretty arduous analysis themselves.
Andrew Capon is editor-in-chief at State Street Global Markets, the research and trading business of State Street Corp. He was formerly senior editor at Institutional Investor and has won numerous awards for journalism on fund management and investment issues. The views expressed are the authors own
Euromoney's CPDO coverage
From the archive:
CPDOs: When will the tail wag the dog?
How much CPDO volume can be produced before it starts to affect the market of which it is a derivative?