In my mortgage math post, I claimed that:
... The main incentive for making financial products complicated is to make them hard to value and thus easier to sell for more than they are worth. ...
A commenter responded in part:
Maybe I've just bought into the hype, but I was under the impression that, at least in some cases, the complex, derivative whole can be more than the sum of its simple, nonderivative parts, because different investors, due to their different circumstances, differ in their valuation of the derivative pieces.
I am not claiming that complexity never adds value (or that it is theoretically impossible or anything like that) just that in practice the apparent value added often turns out to have been illusory. And that this was the case for the mortgage backed securities that recently proved so problematic.
Now obviously these products were being created because they could be sold for more than their cost. Cost being the amount required to purchase the underlying mortgages plus the fees and overhead required to create and sell the derivative securities. So investors in these securities did think value was being created. But they were mistaken. The buyers were relying on the credit ratings of the securities and these ratings were optimistic. This is not too surprising. The people creating these securities could test a million different ways of reassigning cash flows to create derivative securities, examine the resulting ratings (they had access to the rating company software) and pick the assignment that gave the highest combined value to the derivative securities created. Now if the ratings had been perfect this would not have been a problem but of course the ratings were far from perfect and this procedure found the cases where the ratings were most inflated (whether from model flaws or actual bugs in the software). The security buyers didn't adequately discount for this effect and thus overpaid. I believe any actual value created was generally less than the extra overhead costs so that these securities had little real reason to exist and will largely disappear now that buyers are more wary.
Another related problem with these products is that they turned out to have no effective defense against fraud in and/or misrepresentation of the underlying mortgages. The mortgage brokers who arranged the mortgages had no incentive to see that the buyer's income and credit rating and the appraised value and sales price of the property were accurately reported since the more inflated these values were the more they could resell the mortgage for. And the resale price directly impacted the broker's income since they could give a mortgage for $200000, resell it for $220000 (if the buyer's monthly payment would actually have supported the larger mortgage in the current mortgage market) and pocket the $20000 difference. Again the rating agencies and buyers relying on the ratings didn't adequately allow for this effect which got steadily worse as underwriting standards completely collapsed during the housing bubble. Of course this could have been a problem with simple mortgage pools as well but I believe the greater distance between the ultimate buyer and the underlying mortgages made things worse.