Op risk and Black Swans

Scarce data is a well-recognised problem for the assessment of operational risk. In such circumstances, David Rowe argues, it is necessary to blend professional judgement with the available data. In doing so, however, it is crucial to counter some well-documented psychological biases in our subjective estimates of probability – and a healthy dose of humility is also advisable

Nassim Nicholas Taleb (2004) makes an interesting distinction between what he calls type 1 and type 2 environments.¹ He defines a type 1 environment as one where most of the contribution to randomness² comes from the body of a distribution (let’s say the middle 99.9%). This broadly characterises randomness in the physical world, where the normal distribution is common. A type 2 environment is one where most of the contribution to randomness comes from a small number of extreme events in the