Insurers are in the business of making promises and keeping them and, historically, the insurance industry has been one of the most reliable in fulfilling its commitments.
The heartbreaking loss of life and stunning economic aftermath brought about by the coronavirus outbreak has heightened the focus on the low-probability, high-severity events that underlie those promises.
The Black Swan, written by Nassim Nicholas Taleb, suggests that it is difficult to know both the probability and size of such extreme tail events, and explains how an event can become a black swan:
“First, it is an outlier, as it lies outside the realm of regular expectations, because nothing in the past can convincingly point to its possibility. Second, it carries an extreme “impact.” Third, in spite of its outlier status, human nature makes us concoct explanations for its occurrence after the fact, making it explainable and predictable.”
So is the COVID-19 crisis a black swan? Based on Taleb’s criteria, it doesn’t lie outside the realm of regular expectations because, historically, pandemics have been one of the top mass killers of people, exceeding natural disasters and wars. Yet while it’s clear that the current pandemic will have a massive impact on people and the economy and it is too early to predict how it may ultimately unfold, it does not appear that COVID-19 will produce global mortality outcomes close to, for example, the 1918 flu outbreak. Although the nature of the virus and its manifestations could not have been predicted exactly, experts had been warning for decades that a global pandemic involving a highly infectious respiratory disease virus was a plausible scenario – it wasn’t a question of if, but of when. The number of recent, smaller outbreaks, such as SARS, MERS, and H1N1, also point to this reality.
Another way to provide some perspective is to think about what constitutes a 1-in-200 event – the scenario often used for capital stress testing. Thinking historically, this is equivalent to considering the 10 most significant events in the past 2,000 years, a time period that includes the fall of the Roman Empire, the discovery of the New World, the invention of the printing press, and the rise of democracy as a major form of government, to name a few. Currently, numerous factors indicate that the current pandemic will likely not be considered a 1-in-200 event, let alone a black swan; however, it has brought renewed awareness of the potential magnitude that infectious disease outbreaks can have despite the impressive medical, technological, and financial resources of the modern world.
So, what does this mean for insurers taking on low-probability, high-impact risk? Real adversity preparedness requires anticipating what might come, not just learning from historical records of what caught the industry by surprise in the past. When attempting to create a useful adverse scenario, it’s important to note the attached limitations:
- The output is only as good as the input, and the assumptions are frequently rounded to a few significant digits. The results are only accurate to the same number of digits.
- Assumptions about the probability distribution of risks are difficult to validate in the tail. Given an economic record of just 100 years or so in the modern era, the amount of experience on many economic risks is insufficient to effectively extrapolate the tail.
- Models only contain the risks that are understood and included – many tail risks are completely removed from the process. For example, data from countries where a regime change or other major economic adjustments have occurred are frequently excluded. However, these unusual experiences being excluded may be the tail risk data that is needed.
- Risks that are usually independent can change in a tail event. Integrated scenario analysis that considers how things can go wrong together may be more revealing than computer models for these effects.
While multiple warnings and relative historic importance may not make the COVID-19 pandemic a black swan, its impacts will still be significant and widespread and could help us better anticipate and prevent a severe pandemic in the future. And if we’re unable to prevent it, at least we’ll know more about how such an event might unfold, what kind of impact it could have on life insurers, and how to react to it.