The coronavirus, which is currently causing a major international crisis, is one of those unprecedented events that are taking us by surprise by making all the forecasters lie who, only three months ago for the New Year, were telling us what 2020 would be like. Should we be afraid of the virus? This is the question that each of us is asking, torn between the worry of not taking seriously what could turn out to be the epidemic of the Century and the fear of giving in to panic if its impact proves to be only modest, as Dr. Eric Caumes, head of the infectious diseases department at the French Hospital Pitie-Salpetriere, who recently declared: “If you are not afraid of the flu (up to 10.000 deaths/year in France), why are you afraid of the coronavirus?” This seems to make sense, and yet the comparison is not legitimate, because it ignores a very important distinction, that between risk and uncertainty, which a decision-maker must absolutely understand.
Risk and Uncertainty: Crucial Differences
The distinction between risk and uncertainty was introduced almost a century ago by the American economist Frank Knight in a pioneering book entitled Risk, Uncertainty and Profit. Using the vocabulary of probability, Knight defined risk as a future with a known distribution of possible states. For example, if you put three green balls and two red balls in an urn, you know the ‘risk’ of choosing a green ball (60%). Risk corresponds to known events, repeated identically, on which we are therefore able to compile statistics, which become probabilities for anticipating their future occurrence. Thus, the history of car theft enables insurers to price insurance against car theft: if 0.3% of Ford Fusions are stolen each year, yours has a 0.3% chance of being stolen.
Uncertainty, on the other hand, corresponds to a future for which the distribution of states is not only unknown, but unknowable: to take up the image of the urn, we don’t know the number of balls inside it, let alone their colors; we may not even know if there are balls and if there is an urn. This uncertainty is objective: it is not due to the difficulty of accessing information or to the incompetence of the observer, but to the unprecedented nature of the phenomenon we are facing. A new market, a war, a new technology, or an election are typically characterized by high levels of uncertainty because even if there are known aspects, each of these events is unique. Thus, saying “there is a 14% chance that the electric car will be a commercial success” is meaningless because you can’t repeat this invention a dozen times to get statistics.
The distinction between risk and uncertainty allows us to understand the difference between influenza and coronavirus. Influenza is a disease that comes back every year. It is well known, and so is its impact. It is part of the realm of risk. It is an unfortunate event, but because it happens over and over again, we know how to deal with it. Its repetition means that, despite its high morbidity (between 12,000 and 60,000 people a year in the US), it has become routine, we have become accustomed to it; it is a statistic. The coronavirus, on the other hand, is a new thing. We know several things about it, but we also ignore other things. With it, we face uncertainty, especially about its lethality. It could cause a few thousand deaths, but it could also become the epidemic of the Century, a new Spanish flu (at least 50 million deaths in 1918–1919) and anything in between. So what is frightening is not the number of deaths, however large it may be, but the uncertainty about the number of possible deaths from the virus. It is this uncertainty which is anxiety-provoking, and which shows that, yes, at this stage, fear, the emotion which accompanies the awareness of a danger, is a rational attitude.
The Challenge of Deciding in Uncertainty
When faced with a new situation, uncertainty can vary considerably. There are radically new situations, where everything is uncertain, but this is relatively rare. Even during the Black Death in the 14th century, people knew they had to avoid physical contact. More often than not, we can refer to analog situations, which will highlight similarities and differences. This is the case with the coronavirus: much is known about viruses and their modes of spread in general, about effective prophylactic measures (e.g. hand washing) and how to treat them, but its morbidity and lethality rate are not known, although figures for the last few weeks are beginning to become available. Above all, we do not know if it will mutate into a much more lethal form, as was the case with the Spanish flu.
The question, therefore, is what we are going to do about the danger, especially since we have little information and the danger is emerging. If drastic decisions are taken when the signs of an epidemic are not yet very visible, the decision-maker can be accused of scaremongering. But if they are taken when the signs are there, it may be too late. In a decision-making context with a significant lack of information, the decision-maker must therefore weigh a double risk: on the one hand, the risk of acting too drastically, which has a very high cost, both direct (mobilization of medical personnel and purchase of medical supplies, for example) and indirect (slowdown of the economy) for an epidemic that may eventually turn out to be minor; and on the other hand, the risk of not acting, or acting too slowly and thus allowing a catastrophe to occur. Work on uncertainty shows that it is often easier to assess the affordable loss than the expected gain in this situation.
Adding to this difficulty is a paradox: the very effectiveness of a response in limiting the epidemic may cast doubt on whether the threat was real in the first place. We saw this phenomenon at work in the Yom Kippur war in 1973: the Israeli generals, having learned of the Egyptians’ planned attack in April, had succeeded in convincing their government to declare a general alert, which led the Egyptians to cancel their attack. As a result, the generals were accused of crying wolf, which created the conditions for a brutal surprise a few months later when the Egyptians finally attacked.
In the end, the Coronavirus episode suggests three key points for the decision-maker faced with uncertainty: the first is to be aware of the degrees of uncertainty in a situation and to distinguish between what is known, what is assumed, what is not known and what is unknowable (i.e. uncertainty); the second is to be able to identify analog situations from which to draw inspiration, provided that one distinguishes between what is similar and what is different; the third is to weigh as much as possible the risks between acting drastically and acting too late, with the notion of affordable loss suggesting a preference for the former in the present context.