The Need for More Realism in Thinking About Options for COVID-19 Response in Developing Countries
David Tschirley reflects on a recent McKinsey report on the COVID-19 challenge in Africa
McKinsey just published a report on tackling the unfolding health and economic crisis in Africa due to COVID-19. It is one of various rapidly emerging calls for bold action on the pandemic. Though this is encouraging, I have two issues with it:
Reported vs actual cases: The graphic on number of isolated cases, small clusters, and community transmission cases on the continent is visually pleasing, but has a major problem - these are only reported cases, and we know that actual cases are vastly higher, given lack of testing and the role of asymptomatic carriers in spreading the virus. This point is crucial and we must always emphasize it – reported cases are always and everywhere behind actual cases by some very large factor. Of course actual cases are not available as “hard data” and so people steer away from reporting them, but we then fool ourselves about how rapidly the crisis is going to evolve and containment decisions are made far too late. See here for a good treatment of the problems with reported case fatality rates (CFRs – the supposedly hard data that gets most reported) and estimates of infection fatality rates – IFRs – which is what we really need to get a handle on. Look near the bottom in the section titled “Estimating COVID-19 Infection Fatality Rates (IFR) (Update 29TH March)”
On the likelihood of “managing small clusters”: This relates directly to our previous post. The four scenarios that generate the range of GDP impacts include two in which the pandemic is “contained” in Africa, limited to small clusters that are “carefully managed.” This may be useful for scenario building and perhaps some countries will be able to it, but how, exactly? The likelihood is vanishingly small in the absence of income replacement, and without massive international assistance no country in Africa will be able to mobilize enough resources to do this at scale and contain small clusters.
Let’s hope future scenario building, and thinking about response, deals squarely with these two issues.