Robust finance

John Kay in the FT comments on a theme that is central to this blog:

Any engineer will tell you of the importance of making complex systems robust. You need inspections to prevent failure, to be sure: but since failures are inevitable it is equally important to try to ensure that the consequences of such failure are contained.

This observation is as relevant to economic and financial systems as to technological ones. Designing them with components too important to fail is a prelude to disaster, as we know. In the financial sector, the problem of disruptive linkages between components has become known as the problem of systemic risk... the main source of systemic risk is within large financial conglomerates themselves.

Kay gets the solution wrong though. He suggests that the risky component as he sees it - investment banking - should be isolated from the rest. That's foolhardy on two grounds. First, it wasn't investment banking that caused the crisis. Derivatives weren't the problem, after all: it was mortgage lending. The lesson here is that the risk often isn't where you think it is, and so isolating the risky part of the business is not straightforward.

Instead we should accept that any component might fail, and thus to keep the linkages between all components sufficiently loose that no failure can bring the whole system down. That involves increasing capital and liquidity requirements, decreasing counterparty exposure, and taking a particularly conservative view of systemically important institutions.