6 March 2010

Information asymmetry

Filed under: Commentary — András Salamon @ 16:14

Rahul Santhanam suggested I re-read the recent paper by Arora, Barak, Brunnermeier, and Ge on information asymmetry in financial products, presented at ICS2010. I agree with Richard Lipton (and Rahul) that this will spawn many papers. It seems most likely that further work will start with the general framework of informational asymmetry but investigate other problems in economics, using the tools of computational complexity.

The main result of the paper is that CDO instruments suffer from informational asymmetry. It is possible for the seller to benefit from choosing how to distribute low quality assets into CDOs, but so that the buyer cannot detect this, and nor can a regulator, unless they can solve NP-hard problems efficiently. This is different to previous work which has related various problems in economics to the hardness of finding Nash equilibria, which seems somewhat easier.

Back in October, I thought the paper had missed an important part of the story. It seems worthwhile to explain in more detail what I mean by this glib comment.

Few people pre-2008 seemed to care about CDOs being hard to value, or even if they were built out of NINJA loans on tin shacks in hurricane territory, as long as CDS instruments were available to hedge them at reasonable cost. As far as I can tell, the cost of the matching CDS was based mostly on the ratings assigned by agencies to the CDO, which it now seems was mostly done by running a few Monte Carlo simulations with samples from a normal distribution, and with an underlying assumption that any CDO was largely independent of every other CDO being rated.

When Lehman Brothers was allowed to fail in 2008, the CDS side of the equation became doubtful, and the CDO side suddenly went from irrelevant to rather crucial. This is why AIG appears to have became too important to fail — with a value of the CDS market that exceeded the annual output of the global economy, if AIG as one of the biggest issuers of CDS instruments had failed, then the entire global economy could well have collapsed in the ensuing chaos.

The real problem here was the vast disconnect that built up between the size of the underlying asset pool and the value of securities built on top of it. (George Soros has identified such a disconnect as the key component of any economic bubble.) In a 2007 report, leverage of 100 to 200 times was reported as quite common in CDOs. With more realistic pricing of CDS instruments on the part of sellers, based on more realistic measurements of risk by ratings agencies, there would have been little market for high-risk CDOs.

Monte Carlo simulations based on normal distributions have little relation to actual risk, especially when the underlying assets are highly correlated but the modelling assumes they are independent or nearly so. It is a pity that the ratings agencies escaped as lightly as they did from the chaos of the last three years, and if they are still using risk methodologies like this, then they add close to zero value to the financial services industry. I hope the state of the art has advanced since the heady days of 2004.

Oh yes, don’t forget: most of these CDO and CDS instruments are still out there. They are typically legal entities with complex contractual arrangements regulating their existence, and many will take years, if not decades, to fully unwind. In the meantime they are polluting the balance sheets of insurance companies, pension funds, banks, and probably your national government.


Leave a Comment »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at

%d bloggers like this: