I attended an interesting discussion of risk management in the City this week, bringing together insurers with bankers. The two sectors manage risk quite differently, which is why there are rarely insurance crises and frequently banking crises. Insurance crises tend to occur when insurers act like banks (AIG Financial Products, MBIA and other monolines). Bank crises tend to occur when banks act like investment banks.
Insurers must not underwrite risks that they will not be able to cover in the event, and must therefore have reserves sufficient to perform at all times. This makes the insurers much more cautious about taking on risk, about pricing risk accurately at the time of contracting, and about managing reserves to be liquid when claims require payment. Regulation is fundamentally about solvency and selling.
Banks undertake risks on their books that they can only cover so long as they continue to have access to liquidity (funding, deposits, repos or central bank support). Bank capital is never enough to ensure performance without market liquidity for reserve assets. Banks are generally much less cautious about taking on risk, rely overmuch on incomplete models to price risk, and manage capital to optimise returns rather than ensure survival. Regulation focuses on capital (never enough on its own) rather than conduct, common sense and functional suitability.
One risk manager observed that in insurance the risks are exogenous, generally independent in occurrence, and finite. In banking the risks are too often endogenous, correlated in unpredictable ways, and of unknowable magnitude. As a result, a single bank failing has systemic consequences for the banking system, where a single insurance company failure has no systemic consequences for the insurance sector.
An interesting observation both insurers and bankers agreed on was that international harmonisation of regulation had driven formerly diverse business models and management preferences toward uniformity by enforcing preferred models for capital and solvency. As a result, the risks of total systemic failure are much, much higher than before Basel II and Solvency II, because when the models are wrong, the whole financial system is compromised.
Models are always wrong because they are partial, approximate, and use historic data and correlations. In internationally harmonised regulation, the failure of models is even more assured as many domestic factors which have great implications for financial risk are ignored or discounted. Quite simply, models are illuminating, not correct.
What this means is that the 25 year drive to harmonise regulation using financial models is almost certainly counterproductive if the aim was to ensure wider financial integrity and stability. Instead of a global financial system constructed as a spider's web, such that the breaking of one strand does not compromise the whole web, we have a system that has bound all the threads into a single cable. And if that cable frays under stress . . .
There was controversy around the idea of functionally segregating the pedestrian but systemically important functions like payments and mortgage intermediation from the riskier eccentricities of modern of investment banking. About half the room thought it perfectly sensible, and half thought it couldn't be done. I'm of the view that "narrow banking" for some functions might be a very reasonable way to secure the taxpayer from future losses by reducing the scope for contagion in the banking system. And if we could do that, we could allow bad banks to fail, restoring some morals and some hazard to the management of banking.