Has anyone read the famous by Arora et al. on computational complexity and financial derivatives. I'd like other people's thoughts.If you do a search for 'computational complexity and financial derivatives', it should come up on google.

Hi UVAstudent,That's a very nice article. Original, and thought provoking. However, I don't think it's really useful in practice. In their example it's pretty essential that the structurer knows what the lemons are. If somebody knows this type of things, it's probably easier for them to take advantage of this information directly, than by hiding them in a complicated CDO. Even if for whatever reasons, the person with this information can only exploit it via CDO, the incentive structure on Wall Street is in such does not reward this. That's because a bank doesn't keep two separate sets of books, one with the inside information reflected in the price, and one with the generally available information. If you as a structurer know that the realized loss in a CDO will be higher than everybody else thinks, how will you convince your boss, your market risk department, etc, etc, that you deserve a better bonus, short of making public your information? It's true that one may argue that structurers do this thing subconsciously. Further research may shed more light into this. Anyway, thanks for pointing us to this nice article. Best,V.

if the structurers do this thing subconsciously then this would not be a problem because the lemons will be distributed among the derivatives.Being a devils advocate here ....But in a world where volume is king and such products are sold off (i.e., not on the structure's books), surely there is incentive to sell as much as possible to make as much money, hence one would not need to convince anyone. I don't know how the market or internet company structures work but I would think it would be easy to hide lemons given that you are selling it off.But thinking of it differently. Every set of assets will have lemons and the structurer will still know which ones they are. They can distribute it among all CDOs in a fair fashion (i.e., random) or hide it in a few so they know for certain that there would not be any payoff for those bad derivatives. If they do it in a random fashion then they would have to pay out for every derivative (in expectation of course).Mortgages is one place where one might know where the lemons are.