It must be said that the Baldwin's curve already embraces an optimistic view about the impacts of recession on the epi curve. Indeed, we can expect that lockdown measures generate two effects: i) it flattens the curve; ii) it delays the peak. Both effects work in the same direction. While the curve is flatter, health systems will respond more efficiently to the peak. On the other hand, the second effect allows health systems to be better prepared. We claim that Baldwin's view is maybe optimistic because several simulations tend to show that the second effect is much more important than the first one. If theses simulations are correct, they imply that once your health system is almost ready, i.e. at the maximum of its capacity, then there is really no gain to wait more. Below, we share a figure that summarise pretty well some results obtained for Bogota by several colleagues from the engineering Faculty of the University of Los Andes.

Having said that, let us come back to our learning time and policy response issue. It is important to take into account that all the countries did not have access to the same amount of information when they had to take this decision. Indeed, some countries that were first affected (like China’s neighbours such as South-Korea, Japan, etc.) did not benefit from a lot of information dealing with the virus characteristics such as the mortality rate, the contagions vector of this new-virus, and other relevant topics. On the contrary, some countries that were among the latest affected (such as various countries in South-America) could take advantage of more informations on this epidemic, including some elements on the different policies implemented in Europe and their effectiveness.

Therefore, one could think that countries that have been affected among the latest could have implemented more quickly their containment measures.

This phenomenon could be explained by two different channels:

1. Number of observations: The first channel concerns the amount of observations. Since information on the Coronavirus can be considered as a "non rival good", countries that are lately affected have in their hands a larger amount of information. Therefore, one could expect that this would reduce the uncertainty and allow for a more efficient decision making process.

2. Political accountability: The second channel relies on political accountability. Countries that have been affected later by the epidemic have observed a lot of other countries taking drastic measure to contain the spread of the virus. Thus, it would be harder for a government to defend the fact that it is not going to implement any measures. There could be a stronger popular and political pressure on the lately affected country's leaders to implement containment policies.

Those two approaches are particularly relevant in today's world, where information is widely and freely available.

A first striking illustration of the relationship between the time of learning and the political response to the virus consists in looking at the correlation between the date of the first reported case (or death) and the lapse of time between first case (or death) and the implementation of a policy response.

The correlations revealed by those two graphs are particularly strong and striking. The latest affected countries had been engaged in quicker containment policies implementations than countries that have been affected earlier. Some of the latest affected countries even had the opportunity to implement a containment policy before that the epidemic comes to their countries. Those two correlations graphs display the implementation of internal and international travel restrictions.

In order to study this phenomenon with a broader scope, we will rely on the Stringency Index displayed in the Oxford Political response tracker (additional information on this index can be found here). As for the correlations graphs displayed above, our next two graphs that deals with the Stringency Index consider the days reported since the first case and the day reported since the first death.

We shall divide the countries in three different groups:

• The early affected countries: countries where the first death caused by Covid19 occurred between the 31st of December 2019 and the 29th of February.

• The intermediate countries: countries where the first death caused by Covid19 occurred between March 1st and March 24th.

• The lately affected countries: countries where the first death caused by Covid19 occurred between March 25th and the present day.

Those two graphs tend to corroborate our hypothesis about the learning time and the quickness of reactions implemented by governments with a broader set of measures (the stringency index contain information about 7 different public policies). One striking fact is that the lately affected countries have, on the day of their first reported case (or death), implemented more measures than the early affected countries thirty day after their first reported case (or death). This shows the importance of the learning process with regards to the rapidity of decision making. Therefore, if the implemented policies are really efficient, one could expect that the growth in cumulative number of cases (and deaths) should be lower in the lately affected countries, ceteris paribus.

Equipped with these first analysis, very soon we will be able to tackle a crucial point: Do containment policies are effective to contain the virus spread and its consequences? More to come...

References

Baldwin, R (2020d). “The supply side matters: Guns versus butter, COVID-style,” (22 March 2020).

Cordovez J-M, Santos M., Bravo C. and J. Cascante, (2020), Consideraciones para la post cuarentena. Working paper University of Los Andes.

Gourinchas, P-Or (2020). “Flattening the pandemic and recession curves,” Chapter 2 in R. Baldwin and B Weder di Mauro (eds), Mitigating the COVID economic crisis: Act fast and do whatever it takes, a VoxEU.org eBook, CEPR Press.