Marginal Revolution linked to an interesting paper today.
These two lines in the abstract was particularly interesting:
Using a cost/benefit method proposed by Professor Bryan Caplan, and using two extreme assumptions of lockdown effectiveness, the cost/benefit ratio of lockdowns in Canada, in terms of life-years saved, is between 3.6–282. That is, it is possible that lockdown will go down as one of the greatest peacetime policy failures in Canada’s history.
So the Caplan ‘test’ is described in this blog post.
Taking quality of life into account, how many life-years has the reaction to COVID destroyed? To see what I’m getting at, ask yourself: “Suppose you could either live a year of life in the COVID era, or X months under normal conditions. What’s the value of X?” Given the enormous social disruption and dire social isolation that most people have endured, X=10 months seems like a conservative estimate.
With a little multiplication Caplan estimates that the cost of Covid prevention was about 15 times great than the benefits.
Douglas Allen replicates that analysis for Canada.
As of March 2021 the pandemic has lasted one year. That means that the average Canadian has lost two months of normal life. The population of Canada is about 37.7 million people, which means that 6.3 million years of life have been lost due to lockdown.
The average age of reported Covid-19 deaths in Canada is about 80.47 In Canada an average 80 year old has a life expectancy of 9.79 years. This means that the 6.3 million years of lost life is equivalent to the deaths of 643,513 80 year olds. As of March 22, 2021 Canada has had a total of 22,716 deaths due to Covid-19. That amounts to 222,389 lost years of life.
The question is, however, how many lost years of life would have resulted from Covid-19 deaths if there had been no lockdown? Consider two extremes:
a. Assume that the number of Covid-19 deaths would have been 10% higher had there been no lockdown. Then Canada would have experienced an additional 2,271 deaths, which means there would have been additional 22,333 years of lost life due to Covid-19 deaths. The benefit of lockdown, therefore, was the avoidance of this extra 22,333 years of lost life. However, the cost of lockdown, as noted, was 6,300,000 years of lost life. The cost/benefit ratio of lockdown is 282 = 6,300,000/22,333.
b. Assume that the initial ICL model forecasts were correct and without a lockdown Canada would have experienced 200,000 deaths. This would mean that Canada’s lockdown policies prevented 177,281 (200, 000 – 22,716) deaths. Under the same age and life expectancy assumptions lockdown prevented the loss of 1,735,580 life years. The cost/benefit ratio of lockdown is 3.6 = 6,300,000/1,735,580.
So let’s plug some Australian numbers into that calculation.
Let’s assume X = 10. The population of Australia = 25,693,059. The median age of Covid deaths in Australia is 86. Life expectancy for Australians in 2017 – 2019 was 81.31 for males, 85.27 for females, and 83.28 on average. Life expectancy for Australians aged 80 in 2020 was 9.87 years. Finally 910 Australians have died from Covid. (Of those, 820 deaths occurred in Victoria). Unfortunately, I’m going to have to use the life expectancy of an 80 year old instead of an 86 year old in the calculations. This is will bias the costs upwards. The age distribution of death is here – to the extent that people have died under the median age this under-estimates the costs.
So quickly calculating:
4,282,177 years of life have been lost – equivalent to the loss of 433,858 80-year old people. As of today, Australia has lost 910 people to Covid – that is 8,981.7 of actual years lost.
So we can do the plus 10% analysis – assume 1,000 people would have died from Covid. The cost-benefit ratio is in the thousands (4,821). Similarly if we assume a number based on the Imperial College London analysis (I don’t recall them producing an Australian number but I guestimate their equivalent figure for Australia would be about 130,000 people dying from Covid). That would produce a ratio cost-benefit ratio of 3.26.
But we can do a bit more than that. We can turn the analysis around.
I estimate the cost-benefit analysis breaks even at 43,957 people dying. That figure is much smaller than the Imperial College London equivalent number. But that figure was a worst case scenario – and to be fair to them, it was advertised as such in their paper. It is, however, within the range of estimated deaths in a paper produced by Warwick McKibbin and Roshen Fernando. In their scenario analysis they estimated deaths between 21,000 and 96,000 for Australia. So, if Australian policy makers were following McKibbin and Fernando’s figures in March of 2020 then their policy ex ante is not unreasonable given the Caplan definition of costs (and the caveats to that definition that I haven’t yet spelled out).
The break-even point is 434,760 deaths in total and 433,858 deaths avoided.
But … and this is a huge ‘but’ …
One could argue that the Covid-19 lockdown policy was only wrong ex post. Hindsight is 20/20, and looking back is unfair. In March of 2020, faced with an unknown virus and expert advice that millions of people would die without lockdown and isolation, politicians and public health officials made the correct decision at the time.
Such an argument is reasonable for March of 2020, and even possibly for April 2020. However, as noted in the literature review, by late April it was already known that i) the empirical predictions of the SIRS based models were wrong, ii) that the models made a number of questionable assumptions, iii) that the deaths were highly skewed to the elderly, and iv) that the costs were large.
There was a lot of radical uncertainty in March and April of 2020. By May, however, that radical uncertainty had reduced substantially. I have been making this argument for a while – most recently here.
Now for the caveats.
This analysis only considers the number of years of lost life. A proper cost/benefit analysis would consider the value of these lost years. As noted above, the value of life is not constant across age. Since the life years lost to Covid-19 deaths were mostly among those older than 60, and since the years of lost life because of lockdown have mostly been among the young, adjusting the the above cost/benefit ratios for the value of life will make lockdown an even worse policy.
After all that – what is the bottom line?
The review of the literature suggests that Case (a) is closer to reality. If lock-down only had a marginal effect on deaths, then by cost/benefit standards, lockdown has been a public policy disaster.
I suspect the policy responses to Covid will be debated for decades. I have little doubt that the lockdown policy will come to be seen as having been an over-reaction. What happened in my view was that policy makers did not update their policy information and/or preferences after March and April of 2020. Why that is the case is an open question. Perhaps they are just stupid (unlikely), too scared (perhaps), locked into some path dependency (perhaps), [insert hypotheses here].
Update: It turns out that the Imperial College London team did produce numbers for Australia (and most other countries too).
Depending upon the assumptions in the scenarios, they estimated a range from 6,882 through to 194,953 people dying from Covid. Let’s look at the worst case scenario – 194,953 would have died with an unmitigated response (i.e. not mandatory social distancing and no voluntary social distancing either). The cost-benefit ratio is then 2.24. Even in the very worst case the Australian lockdown cost twice as much as the benefits.