How does South Africa’s Covid-19 response compare globally? A preliminary analysis using the new OxCGRT dataset
A 19 May 2020 update of aspects of the working paper can be found here.
A group at the University of Oxford has recently launched a dataset, updated on a daily basis, on the stringency of the measures countries are taking in response to the Covid-19 crisis. The dataset is known as the Oxford COVID-19 Government Response Tracker, or OxCGRT. It provides countries with an opportunity to examine how typical or atypical their responses are. Interpreting these types of cross-country comparisons must of course be done carefully. Yet they can be invaluable in guiding the debates around next steps. Decisions by countries point roughly to a hierarchy of actions used when ‘the screws are tightened’ in response to the pandemic. International travel restrictions are the first to be imposed, then schools are closed and public events cancelled, then internal movement is restricted, then workplaces are closed, and lastly public transport is shut down. South Africa has more or less followed this pattern, but with an above average degree of stringency. If one examines each country’s most recent level of overall stringency, just 30 (of 139) countries had reached the maximum stringency level. One of these countries is South Africa. If one brings in additional World Bank indicators into the analysis, a multivariate analysis is possible of what characteristics of countries are associated with greater or less stringency in their Covid-19 responses. It is clear that developing countries have responded more stringently, when one takes into account where each country lies in the evolution of the pandemic. Having fewer hospital beds relative to the population is associated with a more stringent response, for instance. Thus, it appears that stricter restrictions on movement are imposed where the risk of overwhelming the health system seems greater. South Africa’s response has been stringent, even in comparison to economically similar countries. For instance, restrictions with respect to accessing the workplace have been over twice as stringent as one might expect, at South Africa’s current point in the pandemic’s trajectory. Yet South Africa is not unique. The level of workplace restrictions in the Latin America and Caribbean region, the highest in the world, is at South Africa’s level. If one examines the lag between a country’s first Covid-19 case and workplace restrictions of maximum stringency, South Africa was about average. Absolute numbers of deaths, or Covid-19 deaths relative to how many deaths a country could have expected anyway in 2020, provide what is probably the best basis for comparing, across multiple countries, the speed with which Covid-19 multiplies. There is clearly a large variety of trajectories for Covid-19 deaths across countries. South Africa’s trajectory is not that unusual. Predicting how sensitive these pathways are to restrictions imposed by governments is hugely important, and will preoccupy analysts in the coming months and years. Datasets such as OxCGRT will be important for this work. Some very preliminary analysis done for the current paper points to the difficulty of detecting meaningful correlations, let alone cause and effect.