I noted recently that the British Government still hasn’t published a cost-benefit analysis of lockdown, more than a year after the first one was implemented. Instead, the task has been left to academics and others working outside of government, who’ve found that thecostsalmostcertainlyoutweighed the benefits.
Now two economists based in Bolivia have attempted something similar at the global level. What they’ve done isn’t quite a cost-benefit analysis, as I’ll explain, but it puts the vast costs of lockdown into perspective.
Lykke Andersen and Alejandra Rocabado compared changes in the quantity and quality of life during the first year of the pandemic. They focus on a sample of 124 countries, which collectively account for 96% of the world’s official COVID-19 death toll.
To measure the quantity of life lost in 2020, the authors used the total number of excess life-years lost. And to measure the quality of life lost, they used the percentage reduction in the Google mobility index, averaged across different categories (retail, residential, etc.)
To simplify their analysis, the authors assume that “a 100% reduction in mobility for a year is equal to a lost year of life”. In other words, if average mobility fell by 20% in a country, then everyone in that country lost 1/5th of a quality life-year. This is a strong (and arguably unrealistic) assumption, but it’s useful for trying to get an overall sense of what happened.
The authors find that the world lost 48 million life-years due to people dying from COVID-19, but lost 1.25 billion quality life-years due to reductions in mobility. This means that the loss in quality of life was 25 times larger than the loss in quantity. The only three Western countries where the ratio was less than 2 were Denmark, Finland and Sweden.
How big is the 48 million number? As the authors note, “Every year, at least twice as many life years are lost due to children dying of diarrhea.” (Note that a child who dies of diarrhoea loses 50 or 60 life-years, whereas the average victim of COVID-19 loses only 5 or 10.)
One caveat is that data on excess deaths are not available for many developing countries, so in these cases the authors had to use official COVID-19 deaths, which are almost certainly undercounts. On the other hand, they used estimates of the average number of life-years lost per death that look to be on the high side. Overall, the 48 million number probably isn’t too far off the true amount.
Another point worth noting is that one can’t attribute the entirety of the 1.25 billion number to the impact of lockdowns. Some reduction in mobility would have happened anyway, due to voluntary social distancing. But even if it were cut in half, the total loss of quality life-years would still be 12.5 times larger than the loss in quantity.
Last spring, imposing lockdowns was sold as the only way to prevent a deadly virus from spreading unchecked in the population, and taking hundreds of thousands or even millions of lives. Since then, lockdowns have come under increasing scrutiny. Opponents claim they have upended the economy, undermined children’s education, and violated our basic civil liberties – all without having much impact on the COVID-19 death rate.
When lockdowns were first imposed in the UK and other Western countries, no attempt was made to carry out a cost-benefit analysis. It was simply taken for granted that lockdowns were the correct policy choice. This was because, so proponents argued, they would only be imposed for a limited time, in order to “flatten the curve” and prevent healthcare systems from being overwhelmed. A second justification for lockdowns, which proved influential in some jurisdictions, was that they could be used to suppress the virus completely, thereby preventing any further outbreaks until such time as a vaccine or treatment became available.
Since March 22nd 2020, the UK has spent more than five months under some form of lockdown. And in recent weeks, the country’s lockdown measures have been among the most stringent in the world. Yet its death rate – whether measured as the number of COVID-19 deaths per million people or in terms of excess mortality – is above the European average. Has it all been worth it? I will argue that no, it has not; the costs of the UK’s lockdowns have probably outweighed their benefits.
Do lockdowns work?
The case against the UK’s lockdowns begins by noting that, except in a few cases – which I shall get to – lockdowns have not been associated with substantially fewer deaths from COVID-19. This point has been made at length by the researcher Philippe Lemoine in a report for the Center for the Study of Partisanship and Ideology. As he notes, there are many places where case numbers rose in the presence of a lockdown, as well as several places where they fell in the absence of one. Although case numbers often start falling around the time a lockdown is imposed, it is frequently just before that event, rather than just after. Lemoine suggests this is because people start changing their behaviour voluntarily when they see deaths and hospitalisations rising. The Government, meanwhile, feels an increasing need to “do something”, and the subsequent imposition of a lockdown happens to coincide with the peak of the curve.
For example, the statistician Simon Wood has presented evidence that each of the three English lockdowns was only introduced after the corresponding peak of fatal infections. And in fact, the Chief Medical Officer Chris Whitty told MPs that the epidemic was probably already in retreat when the first full lockdown was imposed. Wood’s conclusions are supported by the findings of economist David Paton, who notes that seven separate indicators all appear to show infections declining before the start of January’s lockdown.
Perhaps the clearest example illustrating the argument that numbers can fall in the absence of a lockdown is South Dakota. The state’s Republican governor Kristi Noem has been stalwart in her opposition to lockdowns, arguing that “the people themselves are primarily responsible for their safety”. As a consequence, there were practically no restrictions in place when the state’s epidemic burgeoned at the end of August. Over the next three months, cases increased – slowly at first, and then rapidly – up to a peak in mid-November. However, despite no shift in policy, they then fell rapidly, and have remained low for the past three months.
What’s even more remarkable is that, according to Google mobility data, there was no major change in people’s movement around the time of the peak in South Dakota. Retail mobility decreased gradually during October and November, and residential mobility was mostly flat. Crucially, there was no sharp change that could explain the sudden decline in cases. One explanation for this anomaly (aside from the Google mobility index being a poor measure of the behaviours that drive transmission) is that the level of prior immunity has been underestimated. Another possibility is that most infections are caused by a small number of “super-spreaders”, and once these individuals have been infected the epidemic swiftly retreats.
When it comes to health outcomes, South Dakota has not fared particularly well during the pandemic – it currently has the eighth highest death rate among US states. But it has not done catastrophically either. Despite imposing almost zero restrictions on the economy, the state ended up with only a slightly higher death rate than Britain. This and other better-matched comparisons cast seriousdoubt on the epidemiological models that served as the basis for lockdowns. (It should be noted that South Dakota has probably benefited, at least to some extent, from its low population density.)
Although lockdowns have not generally been associated with fewer deaths from COVID-19, there are several Western countries where they appear to have worked: Australia, New Zealand, Finland, Norway and Cyprus. So far, these countries have kept the number of COVID-19 deaths below 300 per million; and in fact, they had negligible excess mortality in 2020. Yet, as I noted in a previous article, all five are geographically peripheral countries that imposed strict border controls at the start of the pandemic.
Since none of the five contains an international hub comparable to London, Paris or New York, each had a head start in responding to the pandemic. As a consequence, case numbers were still low at the time lockdowns were imposed, meaning that sporadic outbreaks never cohered into a full-blown epidemic. Meanwhile, the imposition of strict border controls stopped new cases being brought in from outside. It was therefore the combination of early lockdowns and early border controls, under geographically favourable conditions, that allowed countries like Australia and New Zealand to contain the virus.
While one might argue that Britain should have followed the same strategy, it is unclear whether this was ever a viable option, given the country’s size, density and connectedness. And in any case, even if it might have been possible to contain the epidemic in late January, the opportunity had almost certainly come and gone by late February. Having said this, it is somewhat concerning that the strategy was nevergiven serious consideration by the Government’s scientific advisers. For example, the minutes of a meeting on January 22nd record that “NERVTAG does not advise port of entry screening” and “NERVTAG does not advise use of screening questionnaires”.
Pre-existing differences in mortality
The second point against the UK’s lockdowns is that the increases in mortality associated with COVID-19 – even in the worst hit Western countries – have been small relative to pre-existing differences within Europe. For example, the UK’s Office for National Statistics recently calculated age-standardised mortality rates from the first week of 2015 to the last week of 2020 for most countries in Europe. The largest rise from 2019 to 2020 was seen in Bulgaria, where the mortality rate went from 28 to 31.6 – an increase of 3.6 deaths per 100,000. Yet in the year before the pandemic hit, the range of mortality rates (the difference between the highest and lowest values) was 13.8. In other words, the range of mortality rates in 2019 was larger than the largest increase seen by any European country during 2020.
One might counter that the increases in mortality associated with COVID-19 would have been much larger in the absence of lockdowns, but this seems doubtful given the availableevidence. To take one example, Sweden – the only major country in Europe that didn’t lock down – saw age-adjusted excess mortality of just 1.7% in 2020. (Incidentally, a model published last April overestimated Swedish deaths by a factor of 17.) This is not to say that lockdowns had no impact on mortality over and above that of basic restrictions (e.g. bans on large gatherings, self-isolation of symptomatic people) but any impact they did have appears to be quite limited.
The observation that COVID-19’s impact on mortality has been small relative to pre-existing differences can also be made of the UK itself. As Simon Wood noted in an article last October, “the gap in life expectancy between the richer and poorer segments of British society amounted to some 200 million life years lost for the current UK population, which is somewhere around 70 times what Covid might have caused”. He added: “Even the firmest believer in laissez-faire would have to concede that some percentage of that loss is preventable.” The fact that the Government never locked down society (or imposed costs of equivalent magnitude) to reduce much larger differences in mortality within Britain calls its coronavirus strategy into serious question.
Lockdown proponents might say this logic doesn’t apply to COVID-19, since lockdowns prevent individuals from harming others, whereas pre-existing differences in mortality are not due to such “externalities”. But I don’t find this argument very convincing. First, it’s not clear that lockdowns do have much impact on mortality over and above that of basic restrictions. Second, some of the pre-existing differences in mortality are caused by other people’s behaviour (e.g. air pollution, road accidents, flu deaths). And third, blanket lockdowns impose costs on people regardless of whether they contribute to the “externalities” of viral transmission (e.g. people who live away from major population centres, those who have already been infected).
The costs of lockdown
The third key point against the UK’s lockdowns is that their costs have been enormous: not only to the economy, but also to health, education and civil liberties. Take the economy. Britain has suffered its largest economic contraction in 300 years, with GDP falling by almost 10% in 2020. (Note that in the “Great Recession” of 2009, it only fell by 4.2%.) Of course, not all the drop in economic output can be blamed on lockdowns; some – perhaps more than half – would have happened anyway, as a result of voluntary social distancing, cancellation of large events, and reductions in international trade. But a contraction of three or four percentage points on top of that is still very significant.
However, some commentators insist that locking down the economy does not involve any trade-offs. For these “trade-off deniers” (who can count both Rishi Sunak and Chris Whitty among their number) lockdowns are a win-win – or at the very least, a win-draw. However, this argument only works if locking down allows you to completely suppress the virus, since only once complete suppression has been achieved can economic activity resume. The idea is that if you completely suppress the virus after a short, sharp lockdown, you can then re-open the economy as normal, and you end up suffering less economic damage overall than if you’d let the virus spread through the population. But as I’ve already argued, it’s unlikely that suppression was ever a realistic option for the UK. (It almost certainly wasn’t by late February 2020.) Locking down for several months as a way to “flatten the curve” might reduce death rates slightly, but it’s certainly not good for the economy.
The claim that there’s no trade-off between health and the economy appears to be based on one specific observation: virtually all Western countries and US states – regardless of their policies – saw a sharp drop in economic activity during the early weeks of the pandemic. Yet as the historian Phil Magness points out, over the following weeks and months, large differences emerged between the most and least-open US states. And recent data from OECD countries shows a clear inverse relationship between the stringency of government measures and the level of economic growth (with the UK having the most stringent measures and the lowest level of growth). Unsurprisingly, the majority of economists in a survey last November said the UK’s March lockdown did at least some damage to the economy.
I began this essay by noting that no real attempt was made to quantify the costs of lockdowns when the pandemic began. (Instead, projections from computer models weretaken as proof that, without lockdowns, healthcare systems would be inexorably overwhelmed.) Since then, several cost-benefit analyses have been attempted, andeachonehas concluded that the costs almost certainly outweighed the benefits. Of course, accurately gauging all the relevant quantities is no easy task, and these analyses are not without their limitations. But the onus is now on lockdown proponents to show that their preferred measures did pass a cost-benefit test.
Two of the cost-benefit analyses mentioned above estimated the benefits of lockdowns by multiplying the total quality-adjusted life years (QALYs) they might have saved by £30,000 – which is the amount the NHS attaches to a QALY when deciding whether to pay for new treatments that extend patients’ lives by a certain number of years. Although by no means perfect, this seems to me like a reasonable approach. I therefore attempted my own cost-benefit analysis, making what I consider generous assumptions about the public health benefits of lockdowns. I still found that the costs (which I gave as one third of last year’s decline in output) outweighed the benefits by a large margin.
Overall, the UK’s lockdowns were probably a mistake. Looking at the Western world, lockdowns have not been associated with substantially fewer deaths from COVID-19, except in geographically peripheral countries that imposed strict border controls at the start. What’s more, the increases in mortality associated with COVID-19 – even in the worst-hit Western countries – have been small relative to pre-existing differences. Finally, the societal costs of lockdowns have been substantial, and preliminary analyses suggest they almost certainly outweighed the benefits.
Of course, none of this implies that the optimal approach to COVID-19 was “do nothing”. COVID-19 is a deadly disease, and the pandemic clearly warranted government action. In a follow-up essay, I will outline what I believe (with the benefit of hindsight) could have been a more effective approach.
Noah Carl writes about COVID-19 and other topics in his Substack newsletter (where this article was originally published). You can follow him on Twitter @NoahCarl90.