The White House Coronavirus task force briefing on 31 March followed the cardinal rule of successful market strategists: provide a number or a date, but never both. The task force predicted 100,000-240,000 deaths related to COVID-19 and explained that the high number reflected an absence of mitigation, that it did not “accept” the 100,000 number, and that it would do its best to keep deaths below that. It also withheld specifics on timeframe other than indicating that infections would peak only in mid-April.
These predictions seem extraordinarily high against the current 5,000 (approx.) deaths attributed to COVID-19. Based on the current US mortality rate of 2.4%, even the “low” death forecast would imply 4.2m infections, against 200,000 currently. Were there such an increase over, say, the next few months, it would likely trigger a deep panic. But if deaths stay well below these numbers, as seems likely, the task force could lose credibility.
This article is only available to Macro Hive subscribers. Sign-up to receive world-class macro analysis with a daily curated newsletter, podcast, original content from award-winning researchers, cross market strategy, equity insights, trade ideas, crypto flow frameworks, academic paper summaries, explanation and analysis of market-moving events, community investor chat room, and more.
The White House Coronavirus task force briefing on 31 March followed the cardinal rule of successful market strategists: provide a number or a date, but never both. The task force predicted 100,000-240,000 deaths related to COVID-19 and explained that the high number reflected an absence of mitigation, that it did not “accept” the 100,000 number, and that it would do its best to keep deaths below that. It also withheld specifics on timeframe other than indicating that infections would peak only in mid-April.
These predictions seem extraordinarily high against the current 5,000 (approx.) deaths attributed to COVID-19. Based on the current US mortality rate of 2.4%, even the “low” death forecast would imply 4.2m infections, against 200,000 currently. Were there such an increase over, say, the next few months, it would likely trigger a deep panic. But if deaths stay well below these numbers, as seems likely, the task force could lose credibility.
Radical Uncertainty
In all fairness, experts have not made the task force’s job easy: their predictions have been all over the place. The task force’s approach to the abundance of conflicting forecasts has been to average model predictions and add a few tweaks of their own. Yet perhaps the task force would be better served by relying on less precise but more robust models of the epidemic. In their recent book, Radical Uncertainty (2020), John Kay and Mervyn King argue that in most real-life situations we lack enough information to make decisions based on complex quantitative models.
The coronavirus epidemic is a case in point. We have limited knowledge of the actual number of infections, complications and mortality rates, as well as rates of transmission. In such a situation, Kay and King believe:
Models should be treated not as forecasting tools but as ways of organising our thinking. Their construction and interpretation require judgment. Their value depends on our understanding of the processes that give rise to the data we observe, and the quality of that data.
So, inspired by Kay and King, radical uncertainty, here is a highly judgmental and imperfect attempt to assess the likelihood of the task force’s dire predictions. The key parameters are mortality rates and quality of containment.
Mortality Rate
Because of the difficulty disentangling the role played by COVID-19 from co-morbidities, the more reliable way of assessing mortality would be the number of excess deaths. However, in the US over the past 10 years, deaths in March have averaged 237,000 with a standard deviation of 13,000. The 5,000 COVID-caused deaths are therefore statistically insignificant.
Table 1: Testing and Mortality, as of March 31st, 2020
Source: Countries’ Websites, Macro Hive
An alternative approach is to examine the death rates in countries that have engaged in extensive testing. Table 1 shows mortality and testing rates for the top 10 testing countries with more than a thousand confirmed cases. Of course, mortality reporting can be as biased as case reporting – for instance, mortality rates for Spain and Italy seem implausibly high.
Therefore, we’ll use, for our US base case scenario, the average mortality rate of the other eight countries, weighted by their testing intensity, which is about 1%. Against the current mortality rate of about 2%, this implies that the actual number of US infections is twice as high as reported. Assuming that it takes two weeks to go from identification to death, the number of infections in mid June should predict the number of deaths at end-June.
Quality of Containment
Two key factors drive this. First, the completeness of the lockdown. In theory, a full lockdown should contain the epidemic in two weeks since 99.9% of infected individuals will show symptoms with that period (and those infected but asymptomatic will have run the course of their illness also within two weeks). In reality, of course, a full lockdown is unfeasible since essential (including medical) workers need to be out and about.
And in the US, a decentralised public governance system makes it difficult to enforce a strict, countrywide lockdown. While the CDC issued its current guidelines on 16 March, Florida, Georgia, Mississippi, Nevada and Pennsylvania only implemented state-wide stay-home orders on 1 April. Because there are no borders between states, weak containment measures in some states have negative consequences for the whole country.
Chart 1: New Cases in Italy Stabilise After Tightened Lockdown
Source: John Hopkins University, Macro Hive
Italy’s trajectory illustrates the importance of completeness and compliance. Italy first introduced a limited lockdown on 22 February in 11 northern cities. But its new cases stabilised only after the Italian government repeatedly tightened the lockdown (Chart 1).
Chart 2: Korea’s Number of New Cases Decline
Source: John Hopkins University, Macro Hive
The second factor affecting quality of containment is the quality of contact tracing and testing: testing must shift to tracking the epidemic in the general population rather than only the population deemed at risk. Contact tracing must allow stay-home orders to be proactive and targeted, the holy grail of containment with economic normality. Korea’s S curve demonstrates how to achieve this (Chart 2).
The Predictions
This suggests two scenarios for the US. In a good scenario, the CDC stay-home order gets implemented across all US states and, as predicted by the task force, new cases peak in mid-April. The US takes advantage of the lockdown to develop strong testing and contact tracing. In this scenario, new infections peak at 200 per million in mid-April, against the current 150/m. The total number of infections would reach about 11,600/m by mid-June, resulting in a total of 116 deaths per million members of the population (or 38,200 in total), well below the task force’s “good” number (Chart 3 and Table 2).
Chart 3: US Infection Scenarios
Source: Macro Hive
Table 2: Task Force’s Infection Scenarios
Source: Macro Hive
Note: Assumes Mortality Rate of 1%
In the bad scenario, implementation of the lockdown and of testing and tracing is uneven. The number of new infections peaks at 370/m in early May. The total number of infections and deaths ends up twice as high as in the good scenario but still less than one third of the task force’s 240,000.
The Bottom Line
These numbers are not meant to be predictions but rather to identify key variables to monitor and indicate the magnitudes involved in each scenario so as to benchmark upcoming information. The CDC extension of its guidelines to end-April should provide enough time to slow the epidemic, provided that compliance is strong across the US. Early signs are that lagging states are starting to follow the CDC guidelines, which adds to the likelihood of a favorable scenario.
Dominique Dwor-Frecaut is a macro strategist based in Southern California. She has worked on EM and DMs at hedge funds, on the sell side, the NY Fed , the IMF and the World Bank. She publishes the blog Macro Sis that discusses the drivers of macro returns.
(The commentary contained in the above article does not constitute an offer or a solicitation, or a recommendation to implement or liquidate an investment or to carry out any other transaction. It should not be used as a basis for any investment decision or other decision. Any investment decision should be based on appropriate professional advice specific to your needs.)