dstowell.org/write

• Are present contracts specific enough? Can the average regulator pick up a contract, read it, and know what it implies?
• How hard is it to tie a security to its underlying asset? Can I find out enough to drive to call up the guy whose mortgage backs a security?
• Can regulators ask to see any random security owned by anyone?

I don’t the know the answers to these questions, and so can’t say what kind of regulation would have prevented the current crisis. If contracts are unintelligible (or unintelligible at a large scale), I would recommend constraining the language used. A subset of a strongly-typed programming language might do the trick. You can solve new problems in a sufficiently powerful programming language without knowing what those problems are ahead of time (many economists writing about regulation fear a response that would prevent the specific problems that led to this crisis while neglecting the problems that will cause the next crisis). The main advantage of using a formal language to describe contracts would be that machines could reason about such contracts. (This checking would appeal to The Economist, which can’t resist mentioning that talented people work on Wall St. and bozos work for the SEC).

Formalizing contracts doesn’t address the human element, however. What happens when people lie? Or don’t perform due diligence? How many random contracts do you have to check before you can assess the credulity of particular actors?

I ask these questions because I refuse to believe that investment banks don’t ‘understand’ the securities they own.