Ranking countries economic and health performance during Covid19
Tim Vlandas, University of Oxford. This post first appeared on LSE Europe blog
What has been the economic and health performance of different countries since the covid19 crisis began? I propose to rank countries on the basis of how they have fared since the ongoing pandemic began by combining data on two dimensions: a health dimension capturing mortality data; and an economic dimension capturing increases in unemployment.
While the two indicators I select cannot provide an exhaustive picture, they are nevertheless useful in giving us some sense of how countries have fared across two of the dimensions that the covid19 crisis has affected most.
A measure of health costs
The health dimension is based on the so-called p-score. This data is available from the ourworldindata website (data extracted on 14th November 2020) and captures the weekly deviation of current mortality from the 5 years average for that week.
Excess mortality have two advantages over more direct measures of covid19 deaths. First, they do not depend on the testing capacity of the country under consideration, nor are they dependent on definitions of what it means to have died from covid19.
Second, they include the total ‘health cost’ of the pandemic in terms of mortality, i.e. the excess deaths that are the product of both the pandemic and the policy responses to the pandemic.
A measure of economic costs
However, as is well recognised and widely discussed, the pandemic and our policy responses entail significant economic costs. Many economic indicators could be relevant to capture the economic costs. For my purpose, I focus on monthly unemployment rate data, which is extracted from the OECD website.
Partly, this choice is based on data availability constraints, since the alternative to use instead GDP growth data would be hindered by more limited, less recent and less frequent data at the time of writing.
But partly this choice finds inspiration in the so-called misery index which was created following the stagflation in the 1970s. At the time, governments were attempting – and more or less able – to jointly minimise inflation and unemployment.
Adverse unemployment performance can be captured in two distinct ways. The first is simply to look at the average monthly unemployment rate. However, this does not account for the fact that when the pandemic hit, countries started from different relative position. Since the p-score is calculated as a percentage increase from a previous average, I calculate an ‘unemployment score’ as the percentage increase in unemployment from one month to the next.
Country coverage and time period
The following countries are included in my analysis: Austria, Belgium, Canada, Chile, Czech Republic, Denmark, England & Wales, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Israel, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Norway, Poland, Portugal, Slovakia, Slovenia, South Korea, Spain, Sweden, and the United States.
Because ourworldindata.org does not report excess deaths by age groups for the UK as a whole but instead for England&Wales, this is the data I use throughout. For most countries. I have data from January to September 2020 (inclusive). More recent data was not available for all countries at the time of writing so what follows does not capture what has happened since the ‘second wave’.
Excess mortality by age group
I start by showing the weekly evolution of excess mortality by four age groups (Figure 1): 15-64; 65-74; 75-84 and 85+.
The figure reveals the (by now) familiar worse performers, most notably Belgium and England. Note further that English experience is not atypical in the UK given the similar picture for Scotland and to a lesser extent Northern Ireland.
Other countries that did not fare well include Chile, Spain, Netherlands, Italy; although we can observe differences in the timing of the peak. In all cases, excess mortality is strongly a function of age, as has been well documented elsewhere.
Figure 1: Weekly excess mortality by age group
Economic and health costs over time
If we abstract from the cross-national variation, we can see February was actually below the excess mortality average of the last five years for that time of the year and in March the average for our countries only experienced a mild increase. By contrast, in April most countries experienced very significant rises in their excess mortality (Figure 2).
Figure 2: Monthly excess mortality and unemployment score
Bringing in the labour market deterioration into the picture, the increase in unemployment associated with the lockdown measures many countries introduced is also apparent. From May onwards and well into September, both unemployment and mortality stopped to increase in any substantial way across this sample of countries.
The correlation between the unemployment score and different measures of excess mortality is statistically significant and positive but modest (between 0.13 and 0.15), which suggests that countries’ performance in one dimension does not relate strongly for its performance on the other dimension.
This contrasts with claims about the automatic inevitable adverse effects of addressing pandemic for the economy, but also about the presumed positive effects of addressing the pandemic for the economy.
Finally, the heterogeneity is apparent when it comes to excess mortality by age group (figure 3): while some countries experienced the highest increases in mortality for the very old (85+), in others the figures were worst for the 65 to 84, and we can observe both below and above average excess mortality rates for the 16-64 age group.
Figure 3: Excess mortality by country and age group over whole period (from Febuary 2020 onwards)
Pandemic Misery Index (PMI)
Policy makers are therefore faced with a joint minimisation problem whereby they are trying to minimise both health and economic costs. It is in this respect worth keeping in mind that both are – at least in the medium to long term – intrinsically linked to each one another.
On the one hand, mass health issues end up undermining economic productivity and growth. On the other hand, economic decline, insecurity and deprivation generate health problems while also limiting our ability to fund health interventions.
If we combine our economic and health performance indicators into a single pandemic misery index (PMI), we can see that the peak in April hides significant cross-national variation as captured by the standard deviation, and that as the mean of the PMI falls, so does its standard deviation (Figure 4).
In Figure 5, we plot the cross-national variation in the PMI. In the top worst performers we find three liberal market economies (the UK and the US) and two southern European countries (Spain, Italy and Portugal). Although not in top five worst performers, two small open economies in continental Europe – Belgium and Netherlands – also fared poorly.
Figure 4: Pandemic misery index over time
Note: the standard deviation statistic with weights returns the bias-corrected standard deviation, which is based on the factor sqrt(N_i/(N_i-1)), where N_i is the number of observations.
Figure 5: Pandemic misery index across countries
To assess what’s driving poor performance we can disaggregate the PMI along its two dimensions (Figure 6): the PMI in Spain, Italy and the UK is driven by mortality rates with relatively mild increases in unemployment, compared to the US and to Canada, where the increases in unemployment was much more acute.
Among good performers, we find several central and eastern European countries, including Latvia, Hungary, and Slovakia; and also Nordic countries such as Norway, Denmark and Iceland.
Of course, different countries started with different labour market positions, so plotting average levels (instead of changes) in unemployment rates, reveals a slightly different picture (Figure 7). Greece and Spain now look worsts in terms of unemployment rate over the period, followed by Chile, Canada, US and Italy. Czech Republic, Poland, Netherlands, England and Wales, among others kept a low level of unemployment rate.
Finally, excess mortality among the very old (85+) reveals especially dire numbers for parts of Southern Europe (Figure 8), but contrast Spain and Italy, and to a lesser extent Portugal on one hand, and Greece on the other hand. The US, England and Wales, and Canada score high but are not in the top 6 worst performers, while Iceland, Norway, Hungary, Denmark and Slovakia do especially well.
Figure 6: Disaggregating the pandemic misery index across countries
Figure 7: Disaggregated PMI using levels in average monthly unemployment
Figure 8: Disaggregated PMI using mortality rates for over 85 rather than all ages