Alex Budzier, CEO of Oxford Global Projects, and Fellow at the Saïd Business School, University of Oxford.
In his keynote address, Alex Budzier, CEO of Oxford Global Projects, and Fellow at the Saïd Business School, University of Oxford, discussed rare events like pandemics from a basic, statistical perspective.
Typically we assume that observations on particular events tend to cluster around the mean (forming a normal distribution). Unusually large or small observations tend to be followed by outcomes that are closer to the historical average experience. But many natural and man-made phenomena do not sit well within this ‘regression to the mean’ framework. In fact, many extreme events, including outbreaks of infectious diseases, seem to occur with more frequency and impact than expected if they followed a normal distribution. The ‘regression to the tail’ framework explicitly recognises this tendency for new events to be an even more extreme than the most extreme to date.
Fortunately, not all unusual outcomes are true Black Swans in the sense that they seemed impossible or no one had considered them before they occurred. Rather, the frequency of some outsized events such as pandemics and forest fires, as well as incidents like cyber and terrorist attacks, resemble those from standard, well-known ‘fat-tailed’ distributions which attach relatively high probabilities to extreme outliers and whose properties are well understood.
Armed with that insight, we can be better informed about the potential for extreme events to occur. However, this does not necessarily mean we have perfect foresight. It may be hard to establish which particular fat-tailed distribution best fits the pattern of past observations.
Prudent risk managers at all levels – individuals, firms and governments – would therefore be wise to recognise cognitive and other behavioural biases that may erroneously lead to a perception of mild risk, and build more contingencies into their decision-making processes so as to better absorb extreme outcomes and bounce back faster afterwards.
Applied to COVID-19, the regression to the tail framework underlines that while the timing of the infection outbreak was unclear, the probability of such an extreme event itself was massively underappreciated. Moreover, in their handling of the pandemic, many governments failed to adopt sufficiently precautionary measures to ‘cut the tail’ by reducing transmission of the disease through, for example, rapid adoption of mask wearing, effective track-and-trace mechanisms and comprehensive lockdown regimes. Sadly, as a result, many more people were affected than otherwise might have been the case.