The Perils of Bogus Quantification
(Bloomberg Opinion) -- The British government recently sent a letter to every household saying it would do whatever was necessary to beat the coronavirus “if that is what the scientific and medical advice tells us we must do.” At the same time, numerous organizations — commercial, national and international — produced forecasts of the path of the economy over the next two years. But science, and especially models, cannot tell us what to do. In particular, using black-box models, in which the key assumptions are not laid out explicitly and debated, too often becomes an exercise in “bogus quantification,” a common policy failure John Kay and I describe in our new book, “Radical Uncertainty.”
In the context of Covid-19, there are two reasons for caution before rushing to judgment about the way forward. First, the science is uncertain — and how to take those uncertainties into account is not a question of fact. Second, the cost of the economic shutdown is not just a matter of forgone wages and gross domestic product, but also includes the harm the restrictions cause to the health and well-being of the tens of millions of people affected.
To note that there are many uncertainties about Covid-19 is not to criticize the science. Although the genetic code of the virus has been known since mid-January, the fatality rate of the disease is uncertain because researchers don’t know how far the infection has spread in the population. The effect of relaxing the shutdown on the reproduction rate of the virus is also unclear.
Mathematical models of epidemics date back to Daniel Bernoulli’s 1766 paper on smallpox in England. The modern structure of such models was laid out by two Scottish scientists, William Kermack and Anderson McKendrick, in a series of papers between 1927 and 1933, and the dynamic modeling of infectious diseases culminated in the 1991 book by Roy Anderson of Imperial College and Bob May of Oxford University, both of whom have been scientific advisers to the U.K. government. British scientists have played an extremely important role in understanding epidemics, and the government has access to the best minds in the field. That, however, does not eliminate the inevitable uncertainty about the future path of the epidemic or what might have happened without the current shutdown. Where models are helpful is in guiding researchers to collect the right information — such as testing a large random sample of the population to determine the likely spread of the disease and hence its mortality rate.
The timing and design of an exit strategy must be based on an assessment of the economic costs — including the broader costs in terms of health — of maintaining the current shutdown as well as the direct impact of the epidemic. Economic forecasts are most successful when nothing much happens and the future looks rather like the past. No one could believe that is true today.
Official statisticians will be hard put to measure economic activity and the prices paid for goods when shops are shut. It is almost impossible to make accurate forecasts about the likely size of the contraction of the economy during 2020 and its possible recovery in 2021 because much will depend on the response of people to an unprecedented situation. Even if the restrictions are relaxed, how many people will choose to observe social distancing in order to protect their own health before a vaccine is available? And what will be the effect of the inevitably prolonged restrictions on international travel? Useful models lay out the issues and help to identify the crucial parameters. Black-box models conceal these features.
Policy should certainly be informed by the science. But the judgments that have to be made are, and should remain, the responsibility of elected governments.
It is striking that the biggest challenges facing the U.K. and U.S. governments today are logistical. How can we ensure that loan schemes for small businesses are providing the intended volume of lending? How can we secure sufficient personal protection equipment for health workers and distribute it to where it is needed? And, crucially, how can we use our industrial might to permit testing for Covid-19 on a scale that offers a viable exit strategy from the shutdown? So far, the U.K. has managed to test, at most, around 20,000 people a day. For testing to offer an exit route, that number would have to rise not to the government’s target of 100,000 a day but to several million a day. Certainly, more investment is needed to develop a vaccine at scale — if indeed such a vaccine can be found — and improve the speed and reduce the cost of testing. But solving the logistical problems is essential.
In assessing when and how to lift the current shutdown, the government shouldn’t delegate responsibility to scientific and medical advisers, no matter how expert. Coping with the Covid-19 crisis is a classic example of decision-making under radical uncertainty. Governments have to steer between the Scylla of the shutdown’s costs and the Charybdis of more infections as restrictions are eased. Good decision-takers listen to advice, but probe and challenge it. Above all, they understand that scientific advice does not “tell us what we must do.”
For an excellent example of the qualitative, not quantitative, path of an epidemic with and without social distancing, see Roy M. Anderson, Hans Heesterbeek, Don Klinkenberg and T. Déirdre Hollingsworth: “How will country-based mitigation measures inﬂuence the course of the Covid-19 epidemic?”
Daniel Bernoulli, “Essai d’une nouvelle analyse de la mortalité causeé par la petite vérole.”
R.M. Anderson and R.M. May, “Infectious Diseases of Humans: Dynamics and Control.”
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
Mervyn King is a Bloomberg Opinion columnist. He is a member of the U.K. House of Lords, and a professor of economics and law at New York University. He was governor of the Bank of England from 2003 to 2013.
©2020 Bloomberg L.P.