The response of most in the political class to the financial crisis is to call for more regulation. Markets are prone to irrational exuberance, and that exuberance can result in crashes that harm lots of people besides the traders in the markets. The problem is, regulators are prone to their own varieties of irrationality. We saw a gradual move to ease — and even require — loans that would have been unthinkable a few decades ago. Now, in response to the financial crisis, we’ll likely see a “speculative bubble” in regulation. As with post-disaster building codes, some of these regulations will be good and some won’t, and all will be gradually relaxed over time as the memory of crisis fades. That’s not rational, but it’s human, and we’re human.
My own preference would be for a more fault-tolerant system to begin with. Telecommunications switches are designed to be highly reliable; they generally use software modules that communicate only in limited and specified ways with other modules, and they often run on parallel-sync mode (or at least warm-spare mode) so that when things break down it’s not normally catastrophic. Could financial markets be structured the same way? Probably not, alas, but I like the idea. A system in which — as happened in France this year with Societe Generale — one bad trader could run up billions in losses and nearly break the bank is a system that isn’t fault-tolerant enough.
Of course, one rap on fault-tolerant systems is that problems don’t get noticed as fast, because things don’t grind to a halt when failures occur. There is an argument, instead, for crunchy systems where problems are immediately obvious, instead of “soggy” ones where they are not. Regulation, etc., tends to make systems more soggy — which is good for you when FDIC insurance protects your savings, but bad for the system when FDIC insurance makes you not care about your bank’s balance sheet or loan portfolio. Both approaches have their place, of course, but it pays to be clear about which you’re choosing, and why, and what the consequences are.