There is no clear economic trade-off for saving lives during the pandemic

Contrary to popular belief, it seems that countries that have protected their residents better during COVID-19 have not experienced a worse economic outcome.

This article was inspired by my sense that some commentators consider that a tougher stance against COVID-19 may have saved lives but will have come at a large economic cost. This appears to be the subtext in New Zealand, for example, where early pride in the strong response to COVID-19 has given way to a sense that the economic cost was too high.

A study[1] published in The Lancet in 2021 was one of several analyses to show that countries pursuing elimination strategies (including Australia and New Zealand) had better-performing economies during 2020 than those choosing mitigation. The study found that during 2020, COVID-19 deaths per 1 million population in OECD countries that opted for elimination (Australia, Iceland, Japan, New Zealand, and South Korea) had been about 25 times lower than in other OECD countries that favoured mitigation.

At the same time:

Elimination is superior to mitigation for GDP growth[2] on average and at almost all time periods. GDP growth returned to pre-pandemic levels in early 2021 in the five countries that opted for elimination, whereas growth is still negative for the other 32 OECD countries.

However, perhaps this was a short-term phenomenon, with other countries now performing better as the elimination strategy has ended, borders have opened, and most mandatory defence measures have gone?

I could not find any up-to-date analysis of the economic cost (or otherwise) of saving lives, so I decided to see what I could deduce from readily-available data.

The non-existent trade-off

Figure 1 – No strong relationship can be seen between lives lost (excess mortality) and economic impact (increase in GDP) for 39 countries, shown with standard 2-character codes

Figure 1 compares cumulative excess mortality and excess real GDP growth for 39 countries, over the course of the pandemic. I will discuss the selection of these measures below. For now, the point is that there has been no clear economic cost for countries that have done a better job of saving lives[3]. Indeed, as Figure 1 shows, there is no significant relationship between excess mortality and excess GPD across the pandemic.

I have highlighted five “Anglo[4]” countries in Figure 1, on the basis that Australia might best be compared with the other four. This smaller sample (in which the UK – ironically – appears to be a clear outlier) also shows no clear economic cost for countries that have done a better job of saving lives.

Methodology and data

I suggest that we can get a good indication of the relative economic cost of saving lives by comparing the increase in real quarterly GDP with cumulative excess mortality, for a range of countries. I have selected 39 countries, being those[5] for which relevant GDP data is available from OECD.Stat and excess mortality data is available from Our World in Data (OWID).

My proposition is that excess mortality is a sufficient proxy for the effort made to save lives during the pandemic. In this context, I note that:

  • most defence measures reduce the spread of respiratory disease, as well as COVID-19;
  • vaccination reduces the direct mortality impact of COVID-19;
  • excessive stress on health services, including hospitals, leads to higher mortality rates from other causes; and
  • COVID-19 damages the body in ways that we are only beginning to understand, leading to higher post-COVID-19 mortality rates from various causes including heart disease, diabetes and dementia. It seems reasonable to expect that vaccination would reduce this risk.

 

This approach has the benefit of consistency, since only one source is used for each of GDP and excess mortality, but it does exclude some countries that may be of interest, such as Singapore.

Lockdowns and other measures have had an adverse effect on the wellbeing of many people, whether through mental stress or because of lower use of appropriate medical services. However, the Actuaries Institute’s COVID-19 Mortality Working Group did not find any evidence that this had a significant impact on mortality in Australia, at least in 2020 and 2021[6]. This is not to say that there was no impact[7], nor does it downplay the effect on individuals. Of necessity, however, this analysis can only be general, addressing overall totals and averages. In addition, governments and their advisers have been aware of these potential outcomes and have had the opportunity to take action to prevent or address them – e.g. by funding additional mental health support or by encouraging people to continue to attend cancer screening tests.

Therefore, if we measure the cumulative excess mortality relative to projections based on pre-pandemic mortality, since the start of 2020, we will get a reasonable picture of the cumulative effort made to save lives.

Similarly, while no measure can fully capture the different lived experiences of individuals, I submit that the change in real GDP since before the start of the pandemic is a good measure of the economic experience of a country.  However, because countries seem to have different natural growth rates, I think that the actual change in real GDP needs to be compared with the change that would have happened if real GDP had continued to grow at the pre-pandemic trend growth rate[8]. I call the difference “excess real GDP”.

OECD.Stat has GDP data for these 39 countries up to 2022 Q3. I have used the LNBQRSA measure, with subject B1_GE.  This gives seasonally adjusted quarterly GDP under the expenditure approach, in national currency, with chained volume estimates – i.e., quarterly real GDP. I have calculated the percentage increase in real GDP since 2019 Q4 and have compared this with the expected percentage increase using the trend growth rate since 2011 Q1 (the earliest date in the OECD dataset).

OWID has data (periodic actual deaths and cumulative projected deaths) to enable calculation of cumulative excess mortality since 1 January 2020, as a percentage of projected deaths. Data is available at 30 September 2022 (or 25 September for countries with weekly data) except for Australia (28 August), Canada (10 July) and South Korea (31 July).  For these three countries, I have used the latest available cumulative excess mortality rate without adjustment.  I do not expect this to have a significant impact on my analysis.

Referring back to Figure 1, for example, the United Kingdom (GB) saw 9.6% (160,000) more deaths than expected over the 33 months from January 2020 to September 2022 and its real GDP in 2022 Q3 was 6.6% less than expected (growth of -0.8% since 2019 Q4, compared with expected growth of 6.2%, giving 99.2 / 106.2 = 93.4% of expected).

Five Anglo countries

Let’s focus on my five Anglo countries. Note that these countries were my starting point when I first thought about this analysis, on the basis that I wanted to look at Australia and New Zealand and the other three should be reasonably comparable. When I later realised that arguably comparable data was available for more countries, I recast my analysis to cover the 39 in this article.

I will start with excess mortality. As discussed above, I use this as a (negative) proxy for the protection of lives during the pandemic.

First, let’s look at weekly excess mortality. To simplify the picture, I have produced a 13-week centred[9] average.

Figure 2 – There have been waves of excess mortality in the Anglo countries during the pandemic

Despite the simplification, Figure 2 shows the early UK peak and the high mortality in the US and UK in 2020 and, with Delta and Omicron, from the second half of 2021. Consistent with their low levels of COVID-19 until reopening, Australia and NZ had negative excess mortality in 2020 and did not experience much positive excess mortality until the Omicron wave. Canada has oscillated between 0% and 10% excess mortality.

This can be seen in the way in which cumulative excess mortality has developed during the pandemic.

Figure 3 – Cumulative excess mortality is broadly flat in Canada, UK and US, and climbing in Australia and NZ

Figure 3 shows that, of these five Anglo countries, only New Zealand had (slightly) negative cumulative mortality at 30 September 2022. Australia and Canada sat just below 5%, UK at 10% and US at 15%. The differences between these levels appear to be significant, so we would expect clear differences in GDP outcomes.

Figure 4 – The five Anglo countries experienced generally similar GDP growth before the pandemic but had quite different trajectories after 2020 Q1

Figure 4 shows that pre-pandemic growth was reasonably similar for all five countries, with trend CAGR ranging from 2.0% (Canada) to 3.6% (NZ). During the pandemic, growth trajectories have been very different, albeit with the familiar V-shaped recession. This is no surprise, given the differences in mortality outcomes. Interestingly, however, the relationship appears to be a broadly inverse one.

New Zealand has had the highest increase in real GDP during the pandemic, despite its most effective efforts to save lives. The United Kingdom still has not achieved pre-pandemic real GDP, despite having been much more open during the pandemic (as evidenced by the high loss of lives). It is not clear how much impact Brexit has had on the UK economy, noting that its formal withdrawal from the EU took place in February 2020. 

Of course, Figure 4 doesn’t compare GDP with pre-pandemic trend growth, which is the GDP measure used in my analysis. Adjusting for this and combining with excess mortality gives us the points used in Figure 1. Let’s look at these five points in isolation.

Figure 5 – For the Anglo countries, saving lives may have produced a better economic outcome, even when the UK (GB) is treated as an outlier

Figure 5 shows a weak[10] negative correlation between excess mortality and excess GDP, with the United Kingdom (GB) treated as an outlier[11] and excluded[12] from the trend.

Of course, it is clear from Figure 1 that it would be possible to pick five other countries that showed a positive correlation between excess mortality and excess GDP. However, my contention remains that there is no real evidence that saving lives in the pandemic came at an economic cost.

Calibration to the Lancet study

I have carried out my analysis using data up to 2020 Q4, which is essentially the period covered by the Lancet study referred to at the start of this article. Remember that that study had effectively shown that a lower mortality outcome (due to pursuit of an elimination strategy) was correlated to a better economic outcome (higher GDP).

If my data and methodology are appropriate, I should see this relationship.

Figure 6 – At the end of 2020, countries that had better mortality outcomes also tended to have had better economic outcomes

Figure 6 does indeed show a relatively strong[13] (and steep) negative correlation between excess mortality and excess GDP.

In fact, looking at the trend line in each quarter between 2020 Q4 and 2022 Q3, I get Table 1, which charts the change from a clear economic benefit from minimising excess mortality to an apparent economic cost with no statistical significance whatsoever.

Table 1 – For the 39 countries in my analysis, the relationship between economic outcomes and excess mortality has become positive and weaker since 2020 Q4

Conclusion

Figure 1 and Figure 5 do not support the popular proposition that countries that have been better at saving lives during the pandemic have paid an economic price. The positive relationship seen in Figure 1 is far too weak to be statistically significant, while Figure 5 demonstrates a (still very weak) negative relationship.

Figure 6, meanwhile, demonstrates that the opposite was probably true at the end of 2020, as asserted in the Lancet article that I have referred to.

Caveat

The analysis supporting this article is not especially rigorous. I have used publicly available data without adjustment, other than to manipulate it into consistent format.  There is some variation between countries in the calculation of GDP used by OECD.Stat. Similarly, OWID relies on mortality data and projections that are not necessarily consistent between countries.

I suggest that this does not invalidate my conclusions but would be happy to be shown the error of my ways!

Of course, I would also be delighted to be pointed to an academic study of equivalent currency, whether it supports or demolishes my analysis.

References 

[1] SARS-CoV-2 elimination, not mitigation, creates best outcomes for health, the economy, and civil liberties – ScienceDirect

[2] This appears to mean the growth in weekly GDP in comparison with the equivalent week in 2019, but the Lancet article does not give a clear definition

[3] For the technically minded, I note that there is a slight positive slope in the chart (as shown), with an R2 of 0.01

[4] I use this name as shorthand for “English-speaking, with broadly similar British-influenced legal structures”

[5] This approach has the benefit of consistency, since only one source is used for each of GDP and excess mortality, but it does exclude some countries that may be of interest, such as Singapore

[6] See, for example, Research Note 3 and Research Note 4 of the Actuaries Institute

[7] It seems likely that excess mortality in Australia in 2022 includes more than a trivial number of deaths due to lower use of medical services.  Also, I do not know the extent to which other countries have been affected.

[8] This is analogous to using pre-pandemic mortality trends when calculating excess mortality

[9] This means that peaks and troughs occur at the right time, rather than being delayed

[10] R2 = 0.02

[11] I leave it to the reader to speculate on why the UK is such an outlier!

[12] Including the UK would give a stronger and steeper negative correlation between excess mortality and excess GDP, but this would likely be misleading – unless the US is actually the outlier!

[13] R2 = 0.2

CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.