CDOs are inherently vulnerable to fraud

Great Debates
#1

…and here’s the proof. A team of computer scientists and economists have shown that a malevolent seller can bury lemon mortgages in a CDO and its impossible (i.e. computationally intractable) to discover that this has occurred, even in very simple models, and even if the buyer knows almost all of the relevant information.

I admit, I know next to nothing about CDOs, what they’re used for, why they’re beneficial, so starting this thread was mostly for my own educational purposes. However, a few people (see here for instance) aren’t fans, and are calling for them to be banned. What would be the result of banning CDOs? Are they a fundamental to the well being of the economy? Does the above cited paper change anybody’s mind?

#2

Well, but how is that more insightful than saying any financial transaction involving any two parties is inherently vulnerable to fraud?

I prefer collateralized to non-collateralized debt, but the fact that it is collateralized (and here, typically though not technically with mortgages) does not mean that the whole pipeline from broker to mortgagee to consolidator to reseller to…etc etc…is somehow not “inherently vulnerable to fraud.”

Somehow it doesn’t seem an enormous surprise that derivatives amplify the effect of asymmetric information…in any case, with mortgage-backed CDOs in particular, I’d argue there was significant asymmetric information even on the collateralized part of the whole structure. I’d argue the borrowers were horrible risks, and the presumed information was that the collateral was good when in fact it was not. The collateralized part was very information-sensitive, and the information was lousy–dead wrong.
How would a CDO derivative mitigate that risk? It’s not separating risk from rock-solid collateral; it’s separating derivative-related risk from risky collateral.

I do agree the various computerized processes overlying the fundamental mortgages amplified the disaster.

#3

ETA: I hope it’s clear my reply was referring to CDOs and the question referenced in the title. You seem to be taking up the CDS discussion in the second part of your main post–at least, that is what is being singled out in the second link.

#4

Here’s the press release from Princeton.