Initially, when people witnessed exponential rise in death toll, public opposition to lockdowns was extremely limited. But as the threat appears to have passed in most western countries, the patience runs out, and economic impact becomes apparent, views questioning and even condemning lockdowns have become an integral part of the mainstream debate. Unfortunately, a lot of the debate on this issue has been either highly emotional or ideologically driven, or both. Data are often included but if data are not consistent with one’s views, then some even say data is wrong, manipulated etc.
But let’s try to look at it objectively. I will take a data driven approach and will take data as imperfect as it is.
So, when are lockdowns justified? Some civil libertarians say never: most mandatory lockdown measures in peacetime are unlawful, unconstitutional etc., and the right approach should only involve voluntary measures. But it appears that most opponents of lockdowns are closer to the view expressed by Donald Trump: the cure shouldn’t be worse than the disease. This implies that lockdowns could be justified if the disease (potential death toll) is really bad. Most critics of lockdowns are of this view; they just do not see Covid-19 as dangerous enough to justify lockdowns.
So, it comes down to the question: what is the potential death toll that justifies lockdowns? We can estimate this reference number N in a number of ways. Let’s specifically talk about the US, which in its views on lockdowns is now split right down the middle.
One reference that was mentioned in early debates was a million deaths in the US (or N1=3,000 per million of population). This number is not based on any science, it is just a big number many people do see as horrific and to be avoided at all costs. Incidentally, 3,000 per million people was the US death toll in the three and a half years of the Second World War. I am sure there will be those who will question this number as unacceptable but let’s keep it as one possible reference.
Another reference comes from influenza deaths. Over i=34,000 Americans are estimated to have died from influenza in 2018-2019 season, and this is apparently typical. Of course, they never shut down the economy for a bad influenza season, so 2 or 3 times the influenza death toll won’t justify lockdowns. But ten times? Maybe. Then let’s set another reference as 300,000, or N2=1,000 deaths per million people.
Then there is economic analysis. Economists say that a modern way to estimate how much an average life is worth is based not on the earning potential (that’s an old method) but on how much people themselves are prepared to pay to reduce the risk of their own death. Apparently, governments use this number to assess the importance of various safety regulations. US federal agencies estimate the life of an average American to be worth about L=$10 million. And it is not dependent on age even though you would think someone who has life expectancy of 2 years should be worth less than a 20 years’ old… but this is not taken into account in such calculations or it would lead to all sorts of unintended consequences (they say putting price on life is already controversial).
Based on this, we can estimate the number of projected deaths that would justify lockdowns by dividing the total economic damage E from lockdowns by cost of life L. The trouble is that it is very hard to know the total economic impact of lockdowns, or more precisely, the marginal impact of lockdowns compared to the impact of voluntary measures like in Sweden. One obvious way is to compare economic impact in Norway, Denmark or Finland compared to Sweden. But this wouldn’t be fair. Sweden is part of the EU and is affected not just by the domestic situation but also by lockdowns across Europe. One Australian economist says he estimates the cost of lockdowns in Australia to be about E=AU$100 billion. I have no idea how he got this number, looks like he took it out of thin air, but let’s think of it as a data point. Apparently, Australian governments (federal and state) use the lower cost of life than the US: L=AU$5 million (US$3.3 million; this isn’t because an Australian life is worth 3 times less than an American one; it is just they use a slightly different methodology). This will mean that Australian lockdowns would be justified if projected deaths were N=E/L=20,000, or about N3=800 deaths per million. So, we get the number again in the same (not so wide) interval between N1 and N2. An advantage of N3 as a reference (compared to N1 and N2) is that it is based on a real quantitative analysis. Its disadvantage is that cost of life L is somewhat abstract and poorly and defined while marginal cost of lockdowns E is very hard to estimate.
Now, were numbers of deaths N1, N2 or N3 in America realistic? Or more precisely, was it realistic to expect such a death toll in America without lockdowns? Early on, there were models that projected over 2 million deaths if nothing was done (and this number was repeated by Donald Trump in justification of his support of lockdowns at some point). While such projections were done with advanced models, you could get this number if you expect say 70% of people infected and mortality of one per cent. These numbers have been questioned by sceptics. But this is a moot point. It is a false dichotomy. Most opponents of lockdowns do not advocate doing nothing; they advocate voluntary measures such as in Sweden. Until now, Sweden has had over 450 deaths per million people compared to 100 per million for Denmark, 56 per million in Finland and 45 per million in Norway. Given that these countries are very similar in many ways, one can say that lockdowns reduce the number of fatalities (per million people) by a factor of 4 to 8.
The US has just reached 100,000 confirmed deaths (300 per million people) and counting. No doubt it will reach (or get close to) 400 per million people. If we multiply this number by 8, we get over N1, the emotionally unacceptable number (and above the economic-based reference N3).
If we multiply it by 4, we get 500,000 (or over 1500 per million), which is above our lower estimate N2 (but not by very much), but below N3 and N1. This is still a very big number (20 times the number of flu deaths in a season). Personally, I think this justifies lockdowns but not unequivocally.
But this isn’t the whole story. New York, New Jersey, and Massachusetts have already exceeded 1000 deaths per million. Multiply this by a modest factor of 4, and you get 4000 deaths per million people, which is higher than our upper bound N1=3000 per million people. Thus, strict lockdowns in these states appear to be fully justified.
And what about Sweden itself? Did the voluntary measured work there? Critics would point out the stark differences between the death toll between Sweden and its neighbours (most dramatically shown in the plot linked to above) to condemn the Swedish approach. But this is a wrong comparison! By this approach, Finland bungled it completely as its death toll is 10 times worse than Australia’s (4 fatalities per million people). We should compare it to the agreed reference numbers N1, N2 or N3, not against other countries. By this measure, Sweden’s eventual death toll is likely to be somewhere around 500 per million people, which is two times lower than N2, the lowest of the three reference numbers. And clearly, Sweden has flattened the curve and avoided overwhelmed health system. Many people would still say they would prefer to be in Finland than Sweden… but this opinion might not be shared by unemployed Finnish cooks, day care workers or sauna operators… Let’s stick to our reference numbers and avoid emptions.
So, if the Swedish experience is a qualified success, isn’t it the right approach for all other places? Not necessarily. We now know that even in many places that are in lockdown (New York, New Jersey, the UK etc.) the current death toll is already higher than in Sweden. This does not mean that lockdowns increase the death toll; it means that conditions are different in different countries, be it population density, age structure, lifestyle, health system capacity, climate, virus strain etc. As discussed earlier, the lockdowns are likely to reduce the death toll by at least a factor of 4, and thus bring the death toll to an acceptable level.
Several additional points need to be made:
- Another reference suggested by my daughter is the number of hospitalisations that would overwhelm the health system. Early on, there were concerns about availability of intensive care units and lung ventilators. Then the concern shifted to lack of personal protective equipment for health care workers and nursing home staff. In the end, we know that the hospitals were overwhelmed in Wuhan and in parts of Northern Italy but apparently not in many other places. But data on this is patchy and confusing so it is very hard to calculate this reference even across the western world. One thing is clear. Initial concerns about overwhelmed health systems were based on exponential growth of infection rates. It is now clear that no exponential growth occurs in countries where only voluntary Sweden-style social distancing measures are in place. Sweden’s infection rate changed from exponential to linear about four weeks into the epidemic, and its health system was never overwhelmed. That is a good thing. That this linear trend has until now continued without much of a slowdown is the downside.
- Were /are lockdowns justified in places with much smaller death toll and rate of infections, such as in some rural US states or Australia? We will never know if these death toll numbers would have eventuated in Australia. This virus is rather mysterious; it is not entirely clear why, say, Greece has completely escaped it while Italy was hit so hard. Yes, the Greek government acted swiftly but is it all there is to it? Why there are very few deaths in Lebanon and Jordan and, apparently, zero in Vietnam? Why did California come out relatively lightly even though it was one of the first states to get infection, was completely unprepared and has a large population of homeless people? It may be that Australia would have never gotten it anyway. But maybe it was reasonable for the government to expect to be hit similarly to Italy, especially with our close ties with China as well as Italy. So, from the precautionary point of view, this seems to have been a reasonable step to take.
- In the above brief analysis, I only looked at the death rate and economic damage. The virus effect is broader. Many people suffered severely but survived. Others were left with lasting injuries. Likewise, the damage from lockdowns is not just economic. It is also psychological, emotional, social and physical (lack of exercise), etc. etc. In this sense, my analysis is simplistic and incomplete, as it counts the first-order effects only. I believe the input parameters and hence the estimates have such a large uncertainty that any attempts to include all these other important factors would be pointless.
- What about poor countries where people rely on a daily wage to get food (or like in some areas of Peru, they don’t have fridges, so they need to buy food daily – and this is not even Africa!)? Well, these countries need to make a different calculation and may have to make different choices.
- And what about herd immunity? Could it be that Sweden will be immune from further waves, while Finland will be hit hard and will eventually get a similar number of deaths? Well, even if this were true, by the time the second wave hits, we may have a cure or a vaccine, or, the virus mutates into a more benign strain. In any case, hospitals and nursing homes will be better prepared.
- And finally, what about opening up? Many countries are now gradually opening up. Is this reasonable or reckless? Critics say new outbreaks are inevitable; a recent example is in Seoul, where, in response to a new and relatively large outbreak, authorities have closed all schools and shut down entertainment venues. The true picture won’t emerge for a few weeks after the beginning of opening up in many European countries but it is already becoming apparent that outbreaks may occur but will more than likely be manageable and won’t even come close to the reference numbers N1, N2 or N3. Thus, in my view, Australian states have been way overcautious in their reopening. They forgot that the lockdowns were, ostensibly, to flatten the curve and avoid overwhelming the health system, not to eliminate the virus altogether or avoid a single extra death.