The economic and social damage of lock downs in Australia is starting to get noticed so much that even academic economists are paying attention. After months of resisting actual data, some Australian economists who previously refused to even contemplate the idea that an economic collapse would also cost lives are finally trying their hands at data and have produced cost-benefit analyses for the corona crisis. Unfortunately, it is clearly novel territory for them and they have made basic, yet grave mistakes. Let me dissect their writings.
First off, Richard Holden and Bruce Preston, previously active in that “infamous letter by economists” which Sinclair Davidson rightfully has termed bizarre, tried their hands at a cost-benefit calculation in the Conversation. Their calculation is of great simplicity: they say the economic collapse will cost Australia at least 180 billion AUS and then they look for how much the lives saved by the lock downs would be worth. To do that they use the “statistical value of life” estimate used in some government calculations, ie 5 million dollars. I have used the same number in some of my writings here on Troppo, so though I would argue it is not the most appropriate number, it at least is defensible to say that one thinks it will eventually cost one whole life if GDP is reduced by 5 million.
Then they claim 1% of Australia would have died without the lockdowns, which is around 220,000 Australians. They multiply that 220,000 by 5 million and say the lock downs saved Australia 1.1 trillion dollars. That “estimated benefit” is bigger than their claimed “cost” so they conclude the lock downs are worth it.
They make three big mistakes. One is a rookie mistake, two are less serious but still bad mistakes.
The rookie mistake is that they use the statistical value of life for someone who dies of corona. Yet, the statistical value of life holds for a whole life, ie 80 years of life. By contrast, the corona victims would have expected to live only 3-5 years more. So one should only count 5% of the 5 million as the appropriate value of those years, ie 4 years out of the 80 in a full life. That, after all, is how the statistical value of life estimates are used in government allocation decisions. Holden and Preston seem just not to know this, thus confusing their molehill for a mountain. Using the statistical value of life properly would get them a “saving” of only 45 billion Australian dollars of the lock downs, which is 4 times less than they themselves claim is the economic loss.
So if Richard Holden and Bruce Preston are scientifically honest they should immediately update their own figures and own up to the fact that using their own methodology they themselves now think the economic collapse costs at least 4 times more than the lock downs saved.
They make 2 more mistakes, less serious, but still grave. Another mistake they make is that they essentially assume that the recession is over within a year: they quote the IMF numbers on the 8 trillion US dollars the world economy is expected to lose from now till then end of 2021 and essentially apportion Australia its share of the world economy in that loss (about 1.5%). Yet, that is a mighty strange argument because it would be the only deep recession ever that would not only be over within 18 months, but where the economy would have recovered all of its lost ground as well within 18 months. So not merely is everyone who lost a job then re-employed, but we’ve even caught up with the growth we would have had. The IMF is certainly not predicting that.
A much more realistic economic forecast is to think it will take years to recover lost ground and that we’ll thus have a lower GDP than otherwise long after 2021. Stock markets thus look far further than merely 2021 and have tanked far more than 8 trillion. So a more reasonable market-based estimate of the discounted losses to the world economy over the next 10 years is then 50 trillion. That was already clear 2 months ago. This is just basic macro-economics.
Thus a more reasonable assumption on what happens in this depression mean the economic damage is 6 times higher than these authors say. So not only is the benefit of the lives saved 20 times lower than the figure they use, but the economic damage is probably 6 times higher. These two mistakes together conveniently change the cost-benefit ratio of these authors by a factor of 120.
Their third mistake is to presume a whole 1% of Australians would have died without lock downs. Let me merely suffice to say that 1% deaths due to the corona virus is 40 times higher than the death toll in the US. It is 30 times higher than the death toll in Sweden. It is over a 1,000 times higher than the death toll in India, Indonesia, the whole of Africa, etc. It is hence an outlandish assumption. It might have been reasonable in February 2020 to still buy into these apocalyptic numbers, but no longer. No country, whatever policy they have pursued, has even 1/10th of the number of casualties these authors claim would have happened without the lock downs. And that in a country which only had a few hundred cases anyway before it locked down!
So if you add a factor 10 exaggeration on the death toll avoided by the lock downs, the 3 mistakes by these two authors conveniently change the cost-benefit ratio by a factor of 1200 in the direction of making the lock downs look sensible. I hope they have the honesty to admit this and rectify their position accordingly. It would be even better if they would then also have a look at the many other costs of the lock downs: abused women, the suffering of the lonely, the cancelled hospital services, etc. Yet, to do that properly needs a WELLBY approach, which might be a bit much to ask of them.
Holden and Preston could have done even worse and write a Conversation piece like Neil Bailey, who argued the huge reduction in economic resources will not have any effect on lives at all. It is a somewhat strange argument for an economist to make because it basically says government expenses, including all health services, have no health benefits. One wonders why we even bother to have hospitals then? That truly bizarre assumption allows Neil Bailey to say the economic collapse will not cost any lives via reduced government services. As I have calculated several times now, the economic collapse will cost the equivalent of hundreds of thousands of corona virus victims in Australia alone, and many millions worldwide. At least Preston and Holden are aware of that fact!
In conclusion, I am happy to see more economists finally starting to look at numbers in order to come to statements on costs and benefits of lock downs and other policies. Its where the Conversation should be. Correcting for their rookie mistake in not knowing what the term “statistical value of life” actually stands for, Richard Holden and Bruce Preston have basically concluded the economic damage is at least 6 times higher than a generous estimate of the benefits of the lock downs. Correcting for their further mistakes in applying standard macro-economics and the fatality rates pertaining to this virus, they would have come to the conclusion that the costs of lock downs outweigh the benefits by a factor of 240 in Australia. I hope they have the honesty to own that position from now on.
The 3-5 year estimate is probably excessive because if I’m correct you’re using average life expectancy. However, there are interactions between the virus and people’s pre-existing conditions (as you note for younger people), so it’s probably less as it would be killing people more likely to die sooner, so you could really only attribute part of the reduction to it.
That being said, I doubt people would behave normally with hospitals over-flowing, bodies in the streets etc. for a long time, so using a big bang approach is probably economically bad in other ways. There is also data which is hard to evaluate suggesting that people may be getting neurological and other damage, and less difficult to evaluate data showing people get permanent damage to their lungs (which they got in SARs). This would reduce people’s life expectancy and productivity as well as increase their medical expenses across their lifespan, which would presumably be expensive.
Hi Conrad,
I dealt with the conditional life expectancy issue in the previous post. If one takes the residual life expectancy of someone who dies being the same as the average person of that age, hen you get about 10 more “quality of life years”, at least in the UK and most of Europe (also Australia). Yet, when you indeed count the existing conditions, both the conditional expected risidual length of life and the qulaity of those years reduces such that we’re in the 3-5 year range.
Luigi Zingales made the same mistake.
It seems very strange to assume the value of life as a perpetuity when valuing the cost of early death.
Thank you, very interesting blog post. You ought to contact Holden et al, and offer to write an article for The Conversation, which they can reply to.
I would be very disappointed if The Conversation did not publish your article. After all, its own title suggests that we ought to be having conversations with one another.
Holden et al can reply on the Conversation – or here. Paul doesn’t need to restate his case on the Conversation.
Isn’t the audience different? That said, up to Paul. It was just a suggestion. I agree Holden et al can reply in the manner you suggest.
Paul
Rome and Milan were as far as I know under the same national rules viz lockdown and they are both fairly crowded ,frequently visited Italian cities, yet the difference in per capita numbers of deaths between the two cities is enormous. Sweden did a lot less ‘lockdown’ than the UK ( mind Sweden was far from, let it rip either) , yet the Swedes per capita rate is better than the UK and Sweden overall is about middle of the peloton viz Europe.
The Imperial College model predicted that the UK , under lockdown, would have 20,000 deaths. And that Sweden’s policy would result in something like 40,000 deaths. Both predictions have proved to be so out that they probably would have done better using a Chinese astrologer.
Australia’s R figure dropped to below 1 by mid March; much too soon for the level three restrictions announced late in March to be the cause.
The observable results are chaotic , seem more dependant on small variations in starting conditions or plain luck than anything else. For example we were lucky that the pandemic got going when most of our OS student population had gone home for ‘Christmas’ and the globally headlined fire emergency also really crueled OS visitor numbers to SE Australia .
I don’t believe that any policies ,apart from restricting OS arrivals, washing hands and Swedish style distancing,( which kicked in in Australia by about March 10) have really made any difference.
indeed, the whole period has been a salutatory lesson in how wrong supposed experts can be, and how easily their conclusions fall in line with the hysteria of the day. The unscientific nature of most scientific editors and academies has also come into view. The cover-up will be enormous.
Paul only modelling the cost benefit of , do nothing vs mandated lockdown, seems akin to only modelling the risks of car travel at either , 200kph or at 5kph.
We will see plenty of ‘ we saved 20,000 lives’ in Australia. But in the UK where deaths under lockdown are more than twice what Imperial predicted making such claims would be hard work.
oh dear,
a letter is termed infamous and left at that,
Poor writing. explain yourself.
Davidson as usual does not show anything is bizarre, He links it to an article in the Australian. Silly you for linking anything from Davidson.
I do hope you realise which neither Foster nor Davidson did, that consumption would have dropped as people would have been loathe to go out from their homes
with coronanvirus running rampart. Also Employers would have needed to take action to protect employees. this again would have has dampening effects.
Again this seems to have been missed.
The economic damage of the COVID-19 virus isn’t mainly down to imposed distancing and isolation measures.
We’ve already good information showing that in the USA states without lockdown suffered nearly all the economic damage that states with lockdown did. See eg NBER working paper 27127 of May 2020. We have pretty solid information from historical studies that economic recovery was stronger and earlier, after the Kansas flu (the ‘Spanish’ flu), in USA jurisdictions which used stronger and earlier lockdown than in those which didn’t.
The cost benefit work you attack certainly has problems.
It treats avoided deaths as the only benefit. It ignores costs of serious illness. It measures deaths in a highly restricted way, completely inconsistent with the way other epidemic disease rates are worked out (from overall, excess deaths methodologies).
It treats all economic contraction from the virus as cost of lockdown.
But those are the problems you ignore in your own work, and they overstate cost and understate benefit.
The Spanish ‘flu was totally different because it killed so many young people. Corona attacks the elderly and people with existing conditions. That makes all the difference.
“USA states without lockdown suffered nearly all the economic damage that states with lockdown did”
Supply chains. Pigs slaughtered and buried , beer flushed down the drain etc because their end market was locked down….
There has been a huge amount of uncertainty about Covid-19, and government policy had to be made “on the go” without the benefit of information we now have. Even today, there is still a lot we don’t know about Covid-19, including whether surviving an infection confers only partial immunity.
Given the uncertainty, the decision to implement severe containment measures could be defended on “precautionary” grounds. As I understand it, there is a “real option” component to the containment measures taken. For example, it was thought (and perhaps still is thought) that delaying implementing containment measures would have irreversible effects on the rate of infection.
Anyway, the point is that there was (and still is) a lot of uncertainty with Covid-19.
I think we know enough today to end the shutdowns. Specifically the low fatality rate and that the virus attacks the elderly or people with existing conditions. Both provide a clear path forward. The current containment measures make zero sense any more. The politicians have painted themselves into a corner, though.
I do not think an honest assessment is being made on the cost side of the shutdown and many countries are touting the “test/trace/isolate” mantra without wanting to admit it’s futile at this point. The virus is here to stay and should be dealt with in the practical fashion that Sweden is doing.
You can forever say there is uncertainty and delay any decision making, but that would be a huge mistake now.
Yes, it is a fair point that the coronavirus was, based on Wuhan and northern Italy (the early outbreaks), expected to be a lot more lethal to the population than it has turned out to be. The case for drastic action was strong in early March.
So the experts were wrong, but it is not fair to imply incompetence for this – they worked off what could be known at the time. Paul is right that they should have listened to the economists more and the epidemiologists less, but I think he is wrong to think that it would have changed the decisions at that time much because the potential direct costs of covid then appeared to be larger than they have turned out to be.
It is clear, though, that we should now be worrying about covid less and recovery a lot more.
In an ideal world the experts and politicians would say “Sorry we jumped the gun a bit. Couldn’t be helped”. But I expect Paul is also right that the real response is going to be denials that they ever got it wrong, and the more people point they were wrong the more hysterical those denials will be.
My impression is that most economists were (and remain) supportive of stringent containment measures. It’s only a minority of economists that oppose. (I think the same heterogeneity applies to epidemiologists, but less sure.)
Of the majority of economists that support stringent containment measures, my impression is that they simply took the containment measures as some kind of “moral constraint” or “lexicographic preference”, and then sought to optimise within those constraints/preferences. Whereas the minority were prepared to question the constraints/preferences.
Anyway, it will be interesting to see if any economist from either side of the debate changes her/his mind. Holden et al should at least acknowledge any “technical errors” even if they remain committed to their larger world view that stringent containment measures were the right policy.
you are right, it is a lexicographic preference, but a mighty strange one. Its not a preference for life, which has been sacrificed by the lock downs. It is not a preference for worthy deaths, which have also been sacrificed during the lock downs. It is not a preference for health either, which has been heavily adversely affected by the lock downs.
Rather, it seems a lexicographic preference for avoiding visible deaths involving this particular virus. Other deaths are not counted at all.
Very, very odd.
I do think many economists will change their minds. The ring leaders of the “pro-lock down” group of course will simply put their heads in the sand. But the policy economists in the central banks will not want a second wave of lock downs. And the public sentiment will likely change a lot when they start feeling the pain of unemployment and loneliness, which will sway many economists into an appreciation of the bigger picture.
Hope you are right about economists changing their minds. There were a bunch of economists that warned about quantitative easing and called for austerity (and put their names to an open letter), and so far I don’t know of any that have changed their minds on that matter!
Probably most of the “majority economist” will just say circumstances have changed, and so containment measures can be relaxed. Will they actually say they were wrong in their original judgement? If you know of any economist willing to admit error, let me know! Must commend them for their intellectual honesty.
changing one’s mind and openly admitting error are totally different things!
The only example I know of is Paul Krugman admitting he was wrong about expecting the economy to tank following Trump’s election; and changing his mind about minimum wage.
I personally don’t think any lesser of Krugman because he admitted to error. (In fact, I have less respect for people who refuse to admit error.)
So, if there are any economist who wants to admit error about containment policies, be assured that my esteem of you will rise!
Paul it may be “Very, very odd.”
But Pandemics are odd times. Its not much point proposing policy that just dismisses community anxiety as “hysteria”.
Pandemics would be easy to manage without the hysteria. Govt could simply adopt one of your back of the envelope cost benefit analyses and the problem would be solved. I am sure they are pretty accurate, but not that useful in the current circumstances.
You talk about economists changing their mind, like its a bad thing. Well of course they will. No one is proposing a permanent lock down. The aim is to develop a dynamic set of economic policies that manage the pandemic as community attitudes to the pandemic evolve over time.
David,
“Its not much point proposing policy that just dismisses community anxiety as “hysteria”.
The same was once said about villagers afraid of vampires, communities convinced they saw UFOs, or the widespread belief in the imminent arrive of the 4 horses of the apocalypse. The question of to counter hysteria and counter-productive beliefs was a strong part of the enlightenment, so whilst I agree government cannot easily dismiss hysteria, it should nevertheless think of systems that prevent it getting that bad, and how to counter it if it does.
So I venture its the duty of rational government to ascertain how real a threat is before it goes along with draconian actions that it could reasonably know will lead to mass poverty and immense suffering. Our governments should also think of how to prevent and counter hysteria. Western societies have failed that test and rushed into unproven policies which have damaged the community as a whole. To great applause of many economists, it must be admitted.
I once read a nice description of an economist I found very apt: a good economist is someone who doesn’t tell you want to you to hear, but who tells you what you need to hear.
When there is radical uncertainty, standard economic analysis is just speculation. There is uncertainty about the dynamics of the disease and uncertainty about the negative economic impacts because we do not have a previous example in history that we can build our modelling upon. Therefore, lots of the discussion I am reading is much ado about nothing, except that Paul is right about Holden and Preston’s first mistake. That is what happens when you engage in speculation!
We also do not have a previous example in history of locking up all of humanity and expecting to eradicate a virus that way. Yet we engaged in and it has failed miserably. Time to try something different and to learn from the mistakes.
How can you possibly know if it is a ‘miserable’ failure until afterwards? Then we can, indeed, learn from our mistakes. This is not a scientific option right now.
Hmm
A second wave ,within the next year is likely ,so questions about what to do next time can’t wait till “afterwards “.
It’s also likely there will never be an “afterwards “ vaccines ‘strong ’ enough to achieve what we did viz smallpox are unusual.
Your are still in the zone of speculation (like many others). How do you know about the strength (or timing) of a vaccine? What is the point of such discussion? It seems to me that a lot of this is just admonition of the decisions of governments that locked down and inconvenienced everyone. This is, in fact, political position, not a scientific one.
You don’t know the time or cost on anything perfectly. That’s why you estimate it (like everything else, such as the speed the virus spreads). Give there are billions of dollars on offer for a decent vaccine, and given there are very large numbers in development, even though the chance of any single one of them working is small, the chance that any single one will work is large.
It is thus reasonable to speculate that human greed, altruism, and ingenuity, are good motivators for coming up with a solution for things that can be obviously solved within a reasonable time frame (c.f., complex problems like global warming), especially because many of the vaccines have some efficacy in primates similar to humans (and apparently the Chinese government worked a lot on vaccines for the first SARs until it was shelved because it didn’t exist any more), people and governments will poor money into, and they will bend the rules that otherwise make drug testing so slow.
John
If there is a second wave in say August September , what will be the policy?
Will we ( somehow) try to do it all over again , or what ?
Conrad
hope you are right however my understanding is that even among the optimists few expect a working vaccine, in quantity , before this time next year.
I agree with the year minimum — at present, I believe possibly incorrectly that the vaccines in human trials (phase I) are basically old-style technology which means that even if they were ok and made it through phase III trials (however long that takes…), they will still take ages to manufacture in the millions upon millions.
Of the newer style ones which don’t have this problem, I think there are a few that have passed animals trials (i.e., they cause a decent immune response in monkeys) but have not advanced as far as this.
Thanks
Can only pray for a miracle.
Evidence is obviously patchy ,however evidence that severe lockdowns made much difference re health outcomes is poor.
For example the UK not only did worse than Sweden, the majority of UK deaths happened well after their level 4 lockdown commenced.
In Italy Rome’s excess deaths was tiny but Milan’s excess was enormous, both are large cities under the same rules.
Local factors seem to have been more significant ,
for example Vietnam, which is yet to see even one death , had a public health system that was well prepared for a SARS epidemic ( a different kettle to a ‘flu’ epidemic). Vietnam restricted OS arrivals on the first of Feb and mobilised a ‘street by street and village by village ‘ response.
We were lucky that most of our OS student population were out of the country and not due back till the Uni year commenced. (And the fires also really reduced OS visitor numbers for December January.)
The UK in contrast opted to protect their hospitals by moving a large number of patients into aged care homes, but did not adequately test those patients before transferring them. And those aged care homes could not have had that many RNs and ENs or training in barrier nursing either. So about half of the UKs deaths have happened in aged cared homes.
In the US it seems that the large underclass living in poor conditions with a high rate of diabetes etc has been a large local factor . (And bizarrely because so many have lost their jobs and consequently their health cover. about 15 or more US hospitals have actually closed , no paying customers no hospital.)
Obviously Sweden has voluntarily shutdown a fair bit of their economy but as one of them said “losing 60 percent is a lot better than losing 100 percent.”
Speaking as a sole trader :what’s the governments likely response to the next wave ? Is a question that Cannot be left unanswered if there is going to be any investment in producing anything much ( apart from vaccines, treatments and PPE)
A few comments.
Firstly we may have a depression but not a recession. Now that is something to try and get your head through.
now,
Why is anybody examining anything now. does not the second stage count? It did for the spanish flu. Will a lockdown mean a second stage is less deadly?
How are the deaths measured. Is it the excess mortality rate? If not why not.
Why do people like Foster think this recession/ depression is like the rest. CBA card data shows this is a silly assumption. household income has held up well. Put this together with mortgage and rent ‘relief’ then things are different this time much different.
What about other benefits like a lot less people getting flu and thus not dying?
although the social costs of the lockdown were written about little was written about the benefits. Is it a net positive or negative?
One thing that we do not know is how many lives were saved because ICUs were not locked up with coronavirus patients.
Lastly people who live in glass houses should not use terms like rookie errors unless they can show they have learnt from their own
Sweden has about 10 million people and about 3,500 Covid19 deaths ( and the Swedes have a very accurate accounting system)
If the UK had the same per capita death rate as Sweden it would be about 22,000 to 25,000 deaths The UK under mandated lockdown has had more than 40,000 Covid19 deaths (and the UKs system has probably missed quite a few Covid19 deaths.)
The benefits of lockdown are questionable.
It could well be that most of the benefit is in voluntary distancing, hand washing and restrictions on international arrivals etc but too much of the cost was in mandatory lockdowns.
John the vital point to a lockdown is when it occurs, do it too late and it is less effective.
Higher case fatality rates are associated with late lockdowns.
Other measures contribute too of course but an initial period of movement control should allow contemplation of how to prioritise the next steps.
It is less effective at stopping the virus. Unfortunately it is every bit as effective as an early lockdown at stopping the economy.
So logically the argument should be that while an early lockdown may be worth doing, a late one is not – you get all the pain for little gain.
Might the outcome have been even worse if the UK had not locked down at all, and followed the Swedish path?
Just raising the issue. I don’t know the answer, and the uncertainty makes it so difficult to work out the appropriate policy response to Covid-19.
But with the benefit of hindsight, it seems fair to say the earlier the response, the less stringent the lockdown has to be. I’m thinking of S Korea, Taiwan, Vietnam which responded early, and I understand the containment measures there have not been as tight as in the West.
On the other hand, Singapore & HK responded early, but have had to tighten their containment measures over time. So even an early response is no guarantee of success.
No one would disagree with the general idea that a late late lock down is less effective but just as damaging.
Political posturing seems to have got in the way of early aggressive action in some countries.
In reply to Kien Choong it is reasonable to wonder if the death rate could have been much worse in the UK , a counterpoint is to watch Brazil and how that fares with its uneven approach.
A few problems. Google stats show little difference between Sweden abd Germany yet only one had a lockdown. The UK only recently had people from overseas go into a 14 day self isolation.
Murph
The UKs death toll is more than twice what Imperial predicted ,if the lockdown started on about the date it did start. And Sweden’s death toll with no mandated lockdown is about one tenth of what Imperial predicted.
Claims about ‘how many lives’ were saved by mandatory lockdown seem to have little or no basis in observable evidence. And the claim that lockdowns have made any net difference are equally unsupported.
I might as well claim that it’s my daily prayers that made the difference….
Elephant, meet room. Room elephant.
Paul, you have landed some good blows.
Now when are you going to protect your own reputation for intellectual honesty which must surely require you to reign in the rude health of your confidence that the vast costs you document are attributable to the lockdown?
You don’t seem to have engaged with this when Quiggin put the heat on you on Twitter. I’m not following the debate as closely as you, but no one I respect thinks the case fatality rate is 0.2% as you’ve been claiming and haven’t backed off from when challenged (perhaps you have lately or even here – I’ve not read every comment.)
More to the point, there’s the singular fact that Sweden with the mildest coercion of its community doesn’t seem to have an economic dividend to show for it – another stylised fact I’ve picked up in the ether rather than scrutinise myself so you may wish to correct me on it. But the point is that a no-lockdown scenario contains many of the economic costs that you attribute to the lockdown. If all or most of the costs of the lockdown would exist without the lockdown, then your whole case needs to be reconsidered/rebuilt.
In fairness, the counter-factual is very difficult to work out. Do we assume that the entire world adopts a policy of preserving the economy over containment? If only one country prioritises the economy while the rest of the world goes for lockdown, that one country would not avoid a down turn.
So it is hard for Sweden to avoid a downturn when the global economy is going into recession.
(first answered your economic points below. Will answer here your curious request that I take JQ’s twitter nonsense seriously)
On the 0.2% John Quiggin simply pulls one of his usual swifties. “Case fatality rates” are something totally different to the main statistic policy makers and the population rightfully worry about, which is the question of how many people would die of the virus if we did not have lock downs. That alternative statistic is something like the “minimum percentage of the population that will die if we stop repressing this virus and let it spread until the country reaches herd immunity and hence cases get close to zero”. JQ keeps droning on about the first statistic, but its the second one that matters for public policy.
The difference is three-fold: in an optimal spread scenario, of the type I sketched in the previous post and that is somewhat followed by Sweden, a fair proportion of the population would never get the virus and thus never become a case. Obviously the optimal scenario is where that is the most vulnerable part of the population.
Secondly, of course, many of the estimates now doing the round on case-fatalities are based on the deaths per recorded case, which is hence heavily skewed to the fatality within the groups exposed. Observed cases probably strongly over-represent the most vulnerable groups (because they are actually the hardest to protect in lock-downs: the vulnerable have to mingle since they need help, whilst the healthy do not, so the average case is not the same as the average person in the population, though this varies greatly by country). So the reported estimates on case fatalities are probably over-estimates of how dangerous the virus is to the average member of the population. If you like, the “case fatality rate” of healthy people getting a normal low-level infection with this virus is probably well below 0.2% as I have argued in previous posts, but that you will also see in the medical literature, particularly the studies on immunity levels.
Lastly, there is of course the question of just what a “case” is. There are various degrees of infection possible with this virus (and the optimal one is a fairly minor nasal infection which helps build immunity without the large kick one gets if the entire lungs are infected). Superficial infections often dont even count as cases (which is particularly relevant for children), but you’d want them in your calculations for the whole population.
Now, I am sure JQ is actually aware of most of the above, but he doesnt care and simply wants to appear to “win an argument” and so tries to shift the debate to a “yes/no” on a different statistic than the one I have been talking about, which is the percentage deaths in an optimal “no lock down” scenario. I for a long time have said this was probably 0.2%. I would now actually revise that downwards to 0.05% given how low the death rates around the world are, but also because they are actually quite low in Sweden whilst their virologist Geisecke claims that they are likely to get to herd immunity in June already. The extremely low death rates in over 4/5 of the world (whatever their policies) also make clear this virus is just not that contagious or deadly enough to make high numbers of victims in a “do little” scenario in many countries realistic.
So, to put it bluntly, you have been fooled by JQ. He is just value signalling by playing a “shift the debate” game and not engaging in a proper discussion on likely scenarios, cost, and benefits.
This is just typical of JQ: he in my experience almost never argues in good faith but tries to pick a statistic that seems like the relevant one, but that isnt truly the relevant one, and then goes on about that one as if that is all that matters, meanwhile presenting what he wants you to believe as the default implication of the discussion on the small thing he points to. Its his “distraction” technique that works well with the gullible and ill informed, but it makes for poor discussions and little enlightenment. It makes debating JQ a distraction in itself.
Having said this, he is welcome to debate the issues here on Troppo. I am also willing to take him on in a public forum on the broad issue of whether lock downs were smart to have done and to continue with in the future. It is not difficult to debate JQ once you see through his tricks, because it is always the same trick really, which is one of distraction: “surely you agree trees are actually green and therefore we should impose lock downs. How can you possibly disagree trees are not green? Don’t you know how many studies show trees are green?”.
Hmm – you’re jumping all round the place and telling us how bad others’ motives are. That’s your right I guess – I’m not making a moral point. I’m making one about the need to take what others say seriously in order to make progress in a debate. Here’s Hans Georg Gadamer on the point.
In your latest comment, you keep pronouncing on what really matters rather than addressing the issues I’m raising. Case fatality rate seems like a pretty important stat to me. It’s conceptually simple – though raises various challenges to do with sampling bias. It’s something that people are focusing on. I certainly want to know it for my own sake (though of course, I understand that I’d have my own CFR).
It’s certainly much simpler than “the main statistic policymakers and the population rightfully worry about, which is the question of how many people would die of the virus if we did not have lockdown”. I’m uninterested in people’s guesses as to what the number of deaths would be without lockdown because there is no such thing as no lockdown so it’s a hopelessly confused debate. As people see people dying they lock themselves down. In that sense, we can agree that the estimates of early maximum possible death rates from the virus were always overstated. They were still legitimate it seems to me in drawing people’s attention to estimates of what would happen if there was no change of behaviour. But you can disagree that they were legitimate. It makes no difference to my goal which is to try to arrive at reasonable conclusions about the practical options available to us – and also some conclusions about who to trust in the debate. As we both know, there are a lot of dodgies out there.
Here’s the epidemiologist interviewed by 80,000 hours. “The true infection fatality rate in the US [is] I think … clearly under, say, one and a half percent, and it’s very likely under 1%. It’s clearly above 0.2% and probably above 0.4%. So I think if I really had to cover all my bases, I would say between 0.2 and 1.5, but I would put most of my money in the intermediate range”.
I think one of the challenges is that it’s so variable by your risk factors, that a little bit of unrepresentativeness in the sampling of cases can translate into a really wrong inference about the death rate.”
NYC currently has 16,153 confirmed COVID deaths and 4,781 probable deaths according to this website which, summed, is 0.25% of the NYC population given on this website. I hope you can see why, starting from a position of ignorance – I just had to look these things up now – I’m not feeling reassured that you’re helping me understand where the truth probably lies.
You haven’t responded to most of my previous economic points, and nor are you trying to prosecute your case with some grace towards those who disagree with you. If you don’t like the way JQ does it, fine. I think you should pick someone else who is knowledgeable, who disagrees with you in ways that are relevant to the issues I’ve raised, isolate where you disagree and try to enlighten us as to why you’re right and they’re wrong.
Hi Nick,
you sound a bit grumpy here. Maybe we both need a good night’s sleep before we recommence a long debate?
We could have a long argument about infection fatality rate.
We could talk about how you seem to be saying the same I was saying when you said “They were still legitimate it seems to me in drawing people’s attention to estimates of what would happen if there was no change of behaviour. “. My reply above makes clear that the case fatality rate is not an “estimate of what would happen”, but I am also interested in the question of “what would happen”.
Indeed: “what would happen if there was no change of behaviour”? That’s not the case fatality rate, or the infection fatality rate. And normal behaviour always involves some precautions and adaptations. We didnt shut down the economy over AIDS, but neither did we pretend it wasn’t there.
Anyhow, it seems to me we are unlikely to truly disagree much if we’d work through the issues.
As to good graces, it is true I am finding it hard to remain courteous given the vitriol I now get sent on a daily basis. Ponder that, my friend. Ditto for not responding to everything you raise. Three lengthy replies seemed quite a lot to me. Do you want me to write a paper on every question you have?
Oh, and on New York, I am quoting the figures from here: https://www.worldometers.info/coronavirus/country/us/ which footnotes the NYC government. Maybe a different New York to the figures you quote?
Regrettably a lot of comments on blog posts seem more about putting down others vs advancing an argument.
There may be a biological explanation – our amygdala and other parts of our limbic system are overly sensitive towards challenges to our status, and so innocent arguments can come across as a challenge to our status. It takes a great deal of cognitive effort to keep our amygdala in check. I learnt this from reading Robert Sapolsky’s excellent book “Behave – The Biology of Humans at our Best and Worst”.
The only way to make progress is by engaging in reasoned discussion. Reacting to perceived challenges to our status doesn’t help our cause; it only divides people into Us vs Them.
I think it also helps to recognise that reasoned discussion can only narrow down our differences, it rarely leads to a single “right” answer. But I would argue that narrowing down our differences is a substantial achievement that we have reason to celebrate.
you are right. Reasoned discussion is the basic philosophy of cluptroppo.
Kien,
Yes, there is the related question of how a group of individuals who are of good will but each limited in their own way, would endeavour to get to agreement on what to do and how best to view developments.
Maybe something like Nick’s citizen assembly? How would one set one up for this issue though, given that the challenges are ongoing? In many countries they have set up small groups of selected experts to do this, but it doesn’t seem to me to have worked out so well in many places. Ordinary citizens come with the advantage of having no reputation at stake in the outcome of their deliberations.
Once I would have said that public discussion takes place through the mainstream, online and social media (including blogs). But the mainstream media has been disappointingly prejudicial to good reasoning. Our mainstream journalists seem to be “captured” by vested interests in matters like climate change, the refugee crisis and inclusive growth.
Still, we must be patient, persistent and continue to engage in the online and social media. I like to think that public discussion and good reasoning have long-term effects and will eventually make the world better, even if nothing much happens during my own lifetime.
In any event, I personally learn from reasoned discussion; for example, these blog posts have taught me about Welby vs QALY. In fact, I think I learn more from good blogs than from the mainstream media (with some honorable exceptions).
Nick – there are numerous studies converging on an infection fatality rate of 0.2%. I think that’s what you mean when you say “case fatality rate”.
Examples:
https://www.medrxiv.org/content/10.1101/2020.05.13.20101253v1
https://jamanetwork.com/journals/jama/fullarticle/2766367
Thanks Kien,
But you don’t seem to be trying very hard.
The counterfactual of Australia locking down is that it didn’t lock down or did much less.
And vice versa with Sweden. I’ve been expecting Sweden to be doing a lot better economically than its neighbours in which case we get a good discussion about trade-offs – but people are saying that this hasn’t happened.
In which case there aren’t trade-offs. One just has more COVID deaths than the other. And if it’s the case that Sweden has at least some economic advantages from its alternative policy, the numbers for the trade-offs are very different.
Hi Nick,
I must have not been clear enough in my arguments, though lord knows I have tried.
There are several social and economic costs that are directly tied to domestic lock downs, independent of what any other country does. The costs of loneliness, the disrupted IVF services, the disrupted health services, the long-run costs of inactivity, etc. These alone are already a large multiple of the potential benefits in terms of lives saved, so your basic contention that the only major effect of “no lock downs” would have been more deaths is simply untrue.
There are also direct economic costs to lock downs, though these are harder to calculate, but most models and analyses now acknowledge these. The additional costs to the government of keeping jobs afloat (furlough), which are incurred to a lesser degree by Sweden which is keeping places open. So one place to look for their wins is in government borrowing, which is not updated daily as a statistic. Also, many businesses will become unviable if not still operational, even at a lower rate, so lock downs cause more bankruptcies and unemployment, so most economic models now link local lock downs to more local economic misery. You see that very clearly reflected now in government choices in Europe, where several governments are trying to save their tourist industries by opening up quicker. The governments of France, Greece, and Italy thus now have become convinced local lock downs involve local economic costs and are scrambling to save their tourism sectors. Australia is not scrambling, but tourism businesses here are certainly really worried. Ditto for various other industries, like the universities.
There is of course a large public good element to the economic costs in that other countries are hurt if we lock down. Many dont care about this, but if all populations feel the same, they collectively all lose. So one should count part of the costs of the economic collapse in the world as due to local choices.
Finally, the question of whether Sweden is doing better economically. For one, it will be able to operate with open borders much quicker than its neighbours, who will be faced with second waves and other problems when they open up more. So the economic benefit is in the future. Second, its stock exchange, which is one of the few indicators we have of future economic activity, has tanked a lot less than that of Germany, France, the UK, and many other European countries (though not Denmark, where the stock market for reasons I do not understand has had only a very modest fall. Maybe its the type of firms listed on their exchange). I do expect the Swedes to look much stronger economically than its neighbours though in the next 12 months, essentially because it can operate normally and the others cannot until they pursue a similar strategy (or the miracle cure comes).
In terms of the statistics that people look at, Nick, there are two that are hardly in the public debate but that are far more important than the daily case numbers that keep people fixated and that seem to have you occupied as well. One is the percentage of the population that has immunity per corona death and/or per economic disruption. On that statistic, Sweden is probably the world leader by a mile (but the data is a few dozen studies and predictions, not thousands of updates as with the case rates). The second statistic is total misery, both current and expected into the future. We do now have quite a lot of statistics on actual misery (people have been running a lot of surveys), but it gets little attention. Expected misery by necessity comes from a view of how things will progress.
I think you misunderstood my question.
It had nothing to do with the international context – which I clarified in my comment to Kien.
The fact is, as I think you mentioned pretty early on in your commentary a few months ago, even if the government doesn’t lock people down, the ‘hysteria’ would see people locking themselves down. That’s what’s going on in Sweden.
Your first substantial paragraph seems to assume this away – viz:
It’s definitely not simply untrue. It may be complicatedly untrue. People are locking themselves down in droves in Sweden – I’m guessing they’re lonely. So there’s no difference between the ‘factual’ and the ‘counterfactual’ unless you want to unpack it.
To speak personally (and not saying that this is a justification for policy), I’m glad there’s a lockdown because I’d be locked down come what may and the community lockdown both eases anxiety for me and also makes me more relaxed – I lockdown a little less obsessively as I watch the numbers fall and the risk to me lightens. Eradication is also closer at hand. There’s therefore a substantial public good to compelling people to lockdown (as well no doubt as various private and public bads that you point to).
So both the benefits and the costs of the lockdown are hugely overstated, and need recalibrating if the argument is to be focused on practical options. It seems plausible to argue that some of the costs of lockdown are avoided by avoiding compulsion. You’ve pointed to some plausible ones, but they’re not nearly as large as you’re making out and the data isn’t in yet. I’d have expected that even on the data available so far we should be able to see Sweden doing substantially better.
For instance to quote a factoid I’ve heard – if it’s not true it can serve simply for illustration – Swedish road traffic is down 70% and in Australia it’s down 85%. You’d think that would drive differences that were readily observable in economic impact now between Sweden and its neighbours. But that doesn’t seem to be the case.
yes, fear and hysteria are big factors in themselves, which is why I have consistently argued it is the job of governments to argue strongly against the fear. One of the things the lock downs do is to fan that fear. Even in Sweden, there are far too many restrictions and advises in place that keep fear high there as well, so they could certainly do much better.
You miss a crucial issue in this though, which is that a month delay in the transition to normality should be valued in terms of the benefits of normality versus the outcome during the delay period.
So whilst you are quite right to say that the fear and hysteria has infested “no lock down” in many countries with social distancing, that does not make it the right counterfactual because the countries that openly resist the fear narrative will sooner return to normal. In that sense, the Australian choices are even worse than I have sketched, because its low infection rates have dug it a hole that will make it much harder to return to normal (something Hugh Sibly rightly pointed out). The UK has dug a similar hole in its return to normality, because the population is now super-afraid, which has for instance lead to nearly 2,000 primary schools refusing to re-open.
So when I compare the pain in the lock downs with normality, I have the whole trajectory in mind between getting closer to normality a month earlier by not doing a lock down and openly resisting the fear. I am pretty sure I said this somewhere before in the comments to previous posts. Its being generous to lock downs: I should probably count the resisting as speeding up normality faster than by a mere month as lock downs accentuate and normalise the fear.
Hence one of the crucial things to know about Sweden is that the majority of its population is now accepting the fact they will probably get the virus and are not particularly worried about it (I forgot the link, but its somewhere online). Their chief scientists on this thus openly states they will get back to normal much sooner than other countries (https://www.ft.com/content/a2b4c18c-a5e8-4edc-8047-ade4a82a548d).
Dont forget also that I have linked in the previous posts to evidence on the finding that its the people who cant go to work that have the lost wellbeing, not those still allowed to go to work, which is hence purely regulation related.
I tend to agree that government policy influences behaviour. Admittedly this is subjective, but I was quite happy to continue watching movies in theatres, going to out to eat in food courts. But having now spent so many weeks in lockdown, I feel much more conscious about maintaining social distancing. On the other hand, I suspect my I will be quite happy to go out and watch movies again over time. I think I would even be OK to board trains and planes, but I would follow whatever guidance the government gives.
I also agree that my behaviour would likely change even if the government did not implement strict containment measures. I just do not know how much my behaviour would change.
I agree in principle we could take the global economic downturn as a given in comparing the factual with the counter-factual. I just think it is all very speculative trying to unentangle the global effects from the local effects. However, Nicholas may be right that I am not trying hard enough.
yes, its a tough issue to get your head around. So many scientific disciplines are involved in thinking through some of the main counterfactuals, and one really needs to have thought of many of them before it even becomes clear what the most important questions are. And there is so much desinformation, posturing, deliberate shifts in statistical definitions, jargon, wild claims, competing agendas, and misunderstandings involved. Some people try harder than others to see the woods for the trees, but its tough for everyone.
The international context can be taken as given with the counterfactual analysis being country by country. The international context does have an influence but in Australia, and most other countries, most of the downturn will be generated domestically.
I tend to take a “social choice” perspective to policy-making. In particular, government policy should be made democratically. While it is right that we pay close attention to what experts have to say, people may reasonably care about matters that haven’t been taken into account within a (say) Wellby framework. For example, we may have deontological commitments, and so a consequentialist framework doesn’t predetermine government policy.
Our deontological commitment (a “lexicographer preference”) might mean that we ought to ensure no one dies because of the lack of medical care even if that involves temporarily sacrificing personal wellbeing.
However, it is right that social scientists draw attention to the consequential implications of our social choice. Due to “availability bias”, we give excessive weight to the easily observable adversity and neglect less observable adversity.
I agree one should never let “the experts” be in charge, even if that expert is me. Not merely for the reason you state, which is that people can care about different things than experts, but also because it is so easy for specialists to convince themselves of something really stupid. A group more naturally at least thinks of areas the specialist doesnt even imagine exist. Also, power corrupts experts even more than it does others, I think. Democratic deliberations and decisions are thus far preferable to being slaves to experts, even in times of war.
I do personally take a consequentialist approach very far, much further than merely doing my best to present the likely consequences of various options to some public. I also take a step back and wonder why the public is even wondering about the choices on the table and whether it should think of other choices. I think of how the public has come to be susceptible to this or that influence and belief, this or that habit. On such matters rational argument are often but part of what a consequentialist should do. Giving the right example, exuding the right emotions, and partaking in life are also important.
If you like, there are armchair consequentialist and involved consequentialists. Which one are you? Both, neither?
I like to think of myself as a consequentialist that also takes account of deontological commitments, if that makes sense. It’s a comprehensive perspective that attends to both consequences and the processes that lead to those consequences.
As to whether “arm chair” vs “involved”, I think Hamlet didn’t quite get it right when he asked, “Whether tis nobler in the mind to suffer the slings and arrows of outrageous fortuner, or to take up arms against a sea of troubles, and by opposing, end them.”
Sometimes, it is right to take up arms. Sometimes, it is wiser to be patient and bear the slings and arrows of outrageous fortune. But as far as possible, we should engage with people who have different views than us over matters of social choice, and try our best to narrow down our differences and identify the areas that we agree on. But always recognising that even good reasoning cannot guarantee unanimity in everything.
Have a broad church and give everyone a place in the tent, even the dastardly Chinese Communist Party!
https://www.youtube.com/watch?v=DKh6kJ-RSMI&utm_medium=email&utm_source=CampaignMonitor_Editorial&utm_campaign=LNCH++20200521++Facebook++SM+CID_68cde2c35f19a74b1f241cabac1cb2f7
“Remaining in a state of lockdown is extremely dangerous from the point of view of the vulnerability of the entire population to new pathogens. Effectively we used to live in a state approximating lockdown 100 years ago, and that was what created the conditions for the Spanish Flu to come in and kill 50 million people.”
– Professor Sunetra Gupta, in her interview with Unherd.
yes, I read that interview too. Certainly interesting observations regarding the possibility that the anti-body tests are a poor measure of immunity levels because there are other forms of immunity, such as antibodies against the common cold.
But I do doubt she is right that many countries have over 50% immunity levels already. It seems an unlikely story on the propagation of the virus side, ie that so many got infected despite the lock downs. It flies against what we know about the fast spread in some situations (crowded rooms) and low in others (outdoors). Also, the regional pattern of immunity shown by these anti-body tests are sort of what you expect (10% in the big cities, 2-3% in the countryside). If these anti-bodies would only pick up a small part of the actual immunity, then you’d still have to conclude outside of the major cities immunity levels are quite low: there’s be a multiplier on the immunity so if the cities had 50% immunity, the rest would need to have 10%. Cities with 20 or 30% anti-bodies would then be impossible. We should soon know that and that would kill off her story.
I do agree with her that the death statistics, despite all their problems, are the main solid data we have on the fatality one should expect at the level of the whole country. One simply cannot maintain that 1-2% of the population would die without controls if there is no country in the world where more than 0.1% has died, whatever the background stories on immunology and testing. It really does anchor the range one should think about.
And of course she is right on the huge costs of the lock downs and our hysteria. Brave of her to say it so openly when many think it but few dare whisper it.
Paul the whole thing is similar to wildfire.
If you can get in early and get close when it’s just a few hectares and hit it hard ,you can eradicate it. However once it’s spread ,really got going , which doesn’t take long, and it is spotting kilometres away, it’s too late , it can’t be controlled . It becomes evacuate the vulnerable defend what you can accept that there will be losses wait for it to burn itself out and pray for rain
not a bad analogy!
Btw, did you noticed in that article you linked to, but also in other european and american newspaper articles talking about immunity levels, how the conversation has subtly, but completely flipped?
Low immunity levels are now seen as disappointing by the virologists and journalists alike. Whereas a month ago, low immunity levels were the thing they cheered about because it meant few people had contracted the virus. Now, implicitly and totally on the sly, the opposite narrative has become the dominant one: that we actually want far more people to be contracted. Not the ones who would die of the virus, obviously, but the others.
Yes on both counts.
In hindsight we were lucky that the big fire to the west of us got going early when conditions were still cool and assets were not stretched. Mind it was pretty scary really would not judge others in a time of the fog of war
[…] Once you take into account Quality-adjusted life years (QALYs) or Wellbeing-adjusted life-years (WELLBYs) then the cost to life from the shutdowns is staggering, and something like 24 times higher in lives lost compared to lives saved. […]
Firstly, without lockdown, about 10% of Australians would have died. 9 out of 10 severely ill patients can be saved – but only if there are ICU beds available for them. Too many would have fallen ill all at once for the health system to be able to cope. Having had Covid does not mean people will be immune. The first German Covid patient who fell sick in January had no antibodies by April. Some who recover never have antibodies in their blood. This means that every few months every Australian can get Covid again. The severity of the disease is determined by the inoculum – not the age. This is a virus like none before it. 10% of those who have a severe or moderate case of Covid have permanent lung and/or kidney damage, putting further pressure onto the health system. Pity Australia didn’t get it right first time.
There is no spike in Stockholm.
If people with COVID aren’t having an immune reaction you can kiss any hope of a vaccine goodbye.
Can anyone here clarify whether the original post was correct? I see the numbers, including reference to the $4.9m/life and $1.1trillion saved, were quoted in a SMH article today (8th August). However, as soon as I read it, I could see the problem with using $4.9m. Surely that is such a basic error that it would have never been made, let alone not picked up, and, consequently, it is not an error?
it was indeed a basic error, picked up by several people, such as in the post above. However, instead of admit their mistake those authors doubled down, called the whole methodology to arrive at statistical value of life bogus and nevertheless falsely used its figures in an inappropriate manner ever since. Scientific deception, pure and simple, though I did see a weasel phrase in the article to the effect that they take the value of life numbers as if they are not based on value per year of life.
What they do is a bit like saying you presume the weight of a drop of water to be a kilo because a liter weighs a kilo and you find it unfair to drops of water to give them less weight than a whole kilo. But you preface it by saying that you can use the kilo measure of a liter for a drop as if that kilo ts not based on a whole liter of water.
Their continued and now deliberate deception on this matter is an embarrassment.
Thanks for your reply. I know enough about population modelling and economics to be able to identify basic errors, but I still had this niggling doubt that I was missing something.
You’d think, though, that a senior economics writer at the SMH would also know enough to see the problem, but apparently not.
[…] The corona cost-benefit analyses of Richard Holden, Bruce Preston and Neil Bailey: ooops! The economic and social damage of lock downs in Australia is starting to get noticed so much that even academic economists are paying attention. After months of resisting actual data, some Australian economists who previously refused to even contemplate the idea that an economic collapse would also cost lives are finally trying their hands at data and have produced cost-benefit analyses for the corona crisis. Unfortunately, it is clearly novel territory for them and they have made basic, yet grave mistakes. Let me dissect their writings. CLICK LINK TO READ FULL STORY […]
[…] https://clubtroppo.lateraleconomics.com.au/2020/05/18/the–corona–cost–benefit–analyses–of–richard–holden–brucepreston–and–neil–bailey–ooops/ […]
[…] https://clubtroppo.lateraleconomics.com.au/2020/05/18/the–corona–cost–benefit–analyses–of–richard–holden–brucepreston–and–neil–bailey–ooops/ […]
FYI: the above three “pingbacks” are due to economists in Australia using this post and the holden-Presten-Bailey article it refers to as exam questions for their students. Those students are asked to figure out what is going on.
[…] mistake would likely spell the end of their careers. The same is true for many economists who jumped on the initial bandwagon. They are now doubling down too. In for a penny, in for a […]