Tag Archives: counterfactual analysis

Reimagining History – Counterfactual Risk Analysis

Just under ten years ago, as the global financial crisis was unfolding, the book The Black Swan emerged as the most quoted critique of the financial modeling for rare events. The author, Nassim Taleb — the poster boy of sceptics — asserted that these could not be imagined, let alone predicted. Over the past decade, whenever an unmodeled catastrophe has occurred, such as the magnitude 9 earthquake and tsunami that struck Tohoku, Japan on March 11, 2011, catastrophe risk modelers have been reminded of these elusive “black swans”.

Ever since the publication of The Black Swan, I have challenged myself to develop a framework within which such events might be imagined. The solution lies in reimagining history. Since Copernicus, we no longer perceive the Earth as being specially located in the universe. Yet, the anthropocentric viewpoint has maintained that the historical past is somehow special, rather than being just one realization of what might have happened. Most events have either happened before, almost happened before, or might have happened before.

Continue reading

Learning More About Catastrophe Risk From History

In my invited presentation on October 22, 2015 at the UK Institute and Faculty of Actuaries GIRO conference in Liverpool, I discussed how modeling of extreme events can be smarter, from a counterfactual perspective.

A counterfactual perspective enables you to consider what has not yet happened, but could, would, or might have under differing circumstances. By adopting this approach, the risk community can reassess historical catastrophe events to glean insights into previously unanticipated future catastrophes, and so reduce catastrophe “surprises.”

The statistical foundation of typical disaster risk analysis is actual loss experience. The past cannot be changed and is therefore traditionally treated by insurers as fixed. The general consensus is why consider varying what happened in the past? From a scientific perspective, however, actual history is just one realization of what might have happened, given the randomness and chaotic dynamics of nature. The stochastic analysis of the past, used by catastrophe models, is an exploratory exercise in counterfactual history, considering alternative possible scenarios.

Using a stochastic approach to modeling can reveal major surprises that may be lurking in alternative realizations of historical experience. To quote Philip Roth, the eminent American writer: “History, harmless history, where everything unexpected in its own time is chronicled on the page as inevitable. The terror of the unforeseen is what the science of history hides.”  All manner of unforeseen surprising catastrophes have been close to occurring, but ultimately did not materialize, and hence are completely absent from the historical record.

Examples can be drawn from all natural and man-made hazards, covering insurance risks on land, sea, and air. A new domain of application is cyber risk: new surprise cyber attack scenarios can be envisaged with previous accidental causes of instrumentation failure being substituted by control system hacking.

The past cannot be changed—but I firmly believe that counterfactual disaster analysis can change the future and be a very useful analytical tool for underwriting management. I’d be interested to hear your thoughts on the subject.