Keynesian uncertainty and the shaky foundations of statistical risk assessment models

With the financialization of the economy, increasing reliance has been put on statistical models for derivatives pricing and for risk assessment in the day-to-day business of financial operators as well as in financial regulation. This practice had already been criticized from many quarters and on different accounts before the crisis, but these criticisms were simply ignored by the prevailing consensus. This work reconsiders such criticisms from a different standpoint: though they are justified, they could have been put forward in even stronger terms had they relied on Keynes’s work on probability and his notion of uncertainty. I shall thus focus on the conceptual views underlying statistical risk assessment techniques rather than on the techniques in themselves. Finally, I set out some policy implications for regulation.