Lockdown cost-benefit analysis for Australia by Martin Lally

Our most popular cartoons of the yearMartin Lally is a kiwi economist who late in 2020 decided to calculate for himself what his own country was losing by locking itself away from the world, coming to the conclusion that New Zealand was sacrificing something like 26 life-years in the future to ‘save’ 1 life-year. The way he arrived at that number was to essentially calculate how much less the government would have available to spend in the future as a direct result of the effects and costs of lockdowns, and then compare that reduction with how much governments historically had to pay to produce healthy years of life. The logic is that if you reduce government expenditures to that of Russia, then you will get the life expectancy and health of Russians.

Helpfully, he has now done the same type of calculation for Australia (THE COSTS AND BENEFITS OF A COVID LOCKDOWN-6).

You will not be surprised that his headline conclusion is that the lockdowns in Australia will cost around 20 times more than they save, though even that number is achieved by being extremely optimistic about the benefits of lockdowns and by ignoring many of its costs.

Let me point out some interesting aspects of Martin’s effort:

  1. Looking at the data around the world till the end of 2020, Martin estimates that on average, the more stringent the lockdowns, the higher the claimed subsequent covid death rates. In these estimations, Martin takes account of population density, date of first death, population size, GDP, the proportion over 65, the severity of the previous flu waves, the proportion of nursing home beds per head, and average household size. His estimates are repeated in many other publications, but it has really not yet hit the public consciousness: lockdowns do not merely destroy happiness and are extremely costly, but they actually are associated with more covid deaths as well. One likely mechanism is that lockdowns make people unhealthy and thus more susceptible to lots of health problems, including covid. Image
  2. In order to arrive at some ‘possible benefit’ of lockdowns to compare with costs, Martin essentially has to assume they help avoid covid deaths, against his own estimates. It is somewhat humorous to see that Martin thus essentially has to disregard his own best estimates so as not to have to dismiss the case for lockdowns at the outset. Many others doing cost-benefit analyses of lockdowns have struggled with this as well: if there is no benefit in ANY dimension, there is no cost versus benefits, only costs. Like Martin, in my own calculations I also just often presumed lockdowns to reduce covid-deaths by a huge amount, contrary to all the evidence (ie, the totality of the evidence), just so as to be able to say “even if….”.
  3. In some of his calculations, Martin appeals to estimates of the WELLBY literature in order to value the misery of unemployment.
  4. Martin uses Swedish data to estimate that the residual life-expectancy of the average covid-death is 5 more years, which he then uses for Australia to infer the number of saved years of life from lockdowns. What he does not do in that calculation is use the information about the high proportion of covid-deaths that are in nursing homes and care homes, which would reduce that number of years lost per covid-death further because we know that it is the relatively unhealthy going into those homes. Also, Martin does not adjust those 5 years by the (health) quality of those years. Had he done those things, he would have arrived at around 3 healthy years of life (QALY) lost per covid-deaths. So he is, once again, loading his calculations in favour of lockdowns. It’s a common refrain.
  5. To get at costs of lockdowns, Martin estimates the GDP effects and unemployment effects of lockdowns, using middle-of-the-road estimates from around the world. He leaves out of these calculations the huge effects of debts incurred during lockdowns and that provide an easier method of calculating the effects of lockdowns as governments themselves publish the changes and expected further changes in debt. Using those official estimates bypasses the need to do own calculations and hence to disagree with governments: by their own estimates one already gets huge effects of lockdowns. So once more, Martin loads the case for lockdowns by essentially ignoring most of the costs already incurred.
  6. Martin does not calculate many known negative direct effects of lockdowns, such as disruption of IVF services, the mental health costs, the human capital loss of disrupted schooling, the costs of disrupted health services, etc. Martin hence in a way does his very best to underestimate the costs of lockdowns and to overestimate the benefits, still arriving at a cost-benefit ratio of about 20 to 1.
This entry was posted in Coronavirus crisis, Death and taxes, Economics and public policy, Employment, Health, Politics - national, Science, Social Policy. Bookmark the permalink.

140 Responses to Lockdown cost-benefit analysis for Australia by Martin Lally

  1. Nicholas Gruen says:

    Except that it’s not ‘contrary to all the evidence’ that lockdowns reduce deaths in places where they’ve been used sensibly and decisively and not eased up until they’ve achieved plausible local eradication – China, New Zealand, Australia. It’s obviously the case.

    If that can’t be conceded, then we’re in wing-nut territory, something I was confident we didn’t do here on Troppo until these posts started (as interesting and worthwhile as some of their content undoubtedly is).

    Also, what’s his counterfactual of the extent of self-lockdowns in the absence of coordinated – i.e. mandated lockdowns?

    • Nick,

      if you look at as much variation across the world and over time as you can find, which is what Martin tries to do, then you do not find lockdowns (as measured by that Blavatnik stringency measure) achieving local eradication. Rather, you find places without lockdowns where waves dies down (NSW being a recent case in point). What you are arguing is that you observe places in which there were both lockdowns and local eradication, and then you ascribe causality. The problem is that there are lots of other places with lockdowns and no local eradication, and places without lockdowns and yet eradication. Also, lots of governments have insisted that they would achieve eradication by following the best-practise lockdowns their advisers would come up with. To insist that the places you know about ‘prove’ lockdowns achieve local eradication whilst the similar claims of all those other governments have been proven false, is then unwarranted. One must open up to be possibility that what seemed like causality hid something else.

      You should note that over time I changed my position on this as the evidence came in. Round about April 2020 I without much hesitation bought the argument that lockdows must have some benefit in terms of reducing covid cases and deaths. Round about September 2020 the evidence from Europe, Asia, and the Americas was making me doubt they were all that useful in containing covid, for instance because lengthy and quite draconian lockdowns in Peru or Rhode Island were associated with worse covid outcomes, not better. Round about February 2021, I really started to seriously look at what could possibly make lockdowns cause more covid-deaths as all the top covid-death countries in Europe were lockdown-countries (Sweden just dropped out of the top 20!). I wrote a lengthy post on something I called covid-congestion, which was still optimistic about possible benefits of lockdowns in China and Oceania. With the added information since then, I am even less positive about covid-containment effects of lockdowns. Now, you simply do get more and more papers written by scholars around the world in that same vein. They are also looking at other explanations for the experiences in Oceania and Asia. That is how science goes: with new information, one re-assesses the earlier hypotheses. One starts to ask questions like “why else did covid-waves die down in Oceania”?

      Of course one can always say “oh, but there is something about the ones I am interested in that makes them work and no-one else did that. I dont know well what that is, but it must be something because we achieved eradication so it must be our type of lockdowns” That is close to a solipsism: irrefutable, but also useless. I am hence in the post simply stating the state of play as it now is in empirical studies looking to find the effects of lockdowns by looking at what happens across countries over time.

      We have had this discussion before btw, such as on https://clubtroppo.com.au/2021/03/10/what-experiments-on-cult-behaviour-tell-us-about-lockdown-beliefs/ . You essentially state the same belief, asking me the same question to which I then gave a lengthy reply. Did you totally forget that discussion?

      With the information that now exists around the world, the plausibility now goes the other way, so why dont you make an effort to explain why lockdowns don’t “work” in so many other places?

      • No – the earlier discussion left me totally unsatisfied.

        We introduced lockdowns in New Zealand and six different states in Australia in umpteen population centres and in numerous places in China. The basic trajectory of the virus was very similar.

        This doesn’t refute the proposition that we might have achieved it in other ways, but it does refute the idea that, as a proposition ‘lockdowns don’t work’. Nor does it refute the proposition that you can use lockdowns in ways that don’t work. Indeed, one of my main preoccupations in my own writing on COVID has been how many obvious things we’ve not done. As Conrad says below, if you use lockdowns late, they work less well, but more to the point, if you use lockdowns to ‘flatten the curve’ and then subsidise people to ‘eat out to help out’ before you’ve eradicated the virus from circulation, then lockdowns won’t work. Nothing very difficult to understand there.

        Most good empirical work involves a crabwalk between useful techniques and commonsense. Sticking Australia’s lockdowns in with Boris Johnson’s, might get your paper published, might increase ‘n’ by 1, but in the context of a question like ‘can lockdowns work’ to eliminate the virus’, including lockdowns in the sample that were reluctantly imposed and then lifted before they’d worked is pretty silly.

        Anyway Paul, if you can’t accept what I’m getting at above, then I’m afraid I grant you honorary wing-nut status on this topic and I’ll go on respecting your views and your knowledge in numerous other areas (with newfound wariness). But you shouldn’t take it to heart. After all, despite my flattering myself that I’m groping, however inadequately towards the truth, these thoughts you see above are really the fevered rationalisations of (yet another) panic merchant ;)

        • paul frijters says:

          what took you so long? I decided around April 2020 you had joined the wing-nuts on this topic, ie when you started advocating masks as the wonder solution. I was hoping you’d gradually see sense. I still do.

          Perfectly fine with me if you reciprocate in the meantime though. In July 2020 I would have agreed with your conclusions on NZ and Aus (and included the resulting type of tradeoff thinking in my estimates). Now I am indeed no longer convinced that lockdowns were either necessary or even instrumental in the waves dying down. Its still possible, but no longer beyond doubt because of the failure in other places to replicate that supposed effect.
          I do see you make no real effort to engage with all the evidence of what happened around the world so also see no point in engaging on this issue.

          • Jerry Roberts says:

            You economists are doing what economists do — analysing data and failing to convince one another. Do you think it is wise to use a word such as “eradication” for a corona virus? I wish you would turn your minds to a more urgent question, namely the safety of the vaccines. I expect Australia to cop it much the same as everywhere else. There was a nasty incident in Perth in recent days when police harassed a woman who was not wearing a mask — outside in the open air. I remain where I was a year ago, much more worried about a police state than I am about the virus.

            • conrad says:

              I’m not sure why people need to worry about the safety of the vaccines — even the ‘unsafe’ AZ and J&J ones are hardly worth worrying about compared to any number of risks that people don’t care about much and can reduce the risks of getting (heart attack, cancer, and so on) — even more so if you over 50.

              To me the interesting things are

              1) People clearly have no ability to trade off risks. The chance coronavirus will break out in Australia, you will catch it, and you will get nasty side effects (including deaths) seems a lot more 1 in 250,000 (and that’s not even the death rate), especially if you are old or vulnerable somehow.
              2) Since this is fairly transparent compared to monetary trade-offs between what we do now and how that relates to debts socialized across the community in the future, it is pretty clear why Paul (even if he his right) is going to have a very difficult time convincing people of his position.
              3) That people don’t understand base rates and trade-offs is of course well known.
              4) It will be curious to see what the Oz government does if a reasonable proportion of people won’t get the vaccine.

              Scientifically, I think the vaccine stuff is interesting.

              At least to the arguments and discussion we’ve here over some months, to me the most interesting part is that you can clearly be contagious after having some of the vaccines, and so the big benefit is reducing symptoms. I imagine this suggests that a lot of people are getting a T-cell benefit (and other stuff), rather than immunity to the virus via other systems.

              This means things like vaccine passports don’t make much sense if you want to stop virus transmission. It also means the standard way vaccines are evaluated doesn’t make sense (where T-cells are down the list) — if there are tablets or nasal sprays that are safe to use that just give you T-cell resistance, that’s probably really worth while.

            • paul frijters says:

              Hi Jerry,

              “You economists are doing what economists do — analysing data and failing to convince one another.”

              what a fantastic phrase! I hear you and totally agree with your police state argument.
              My own reading on the data about side-effects of vaccines is that they are so rare as not to need bother with them in a cost-benefit analysis. Let’s put it like this: the risks of the vaccines relative to covid are similar to the risk of covid relative to the risks of lockdowns.

              Conrad’s point that populations are spectacularly bad at tradeoffs rings very true. But does that mean economists should not talk about them or embed them in institutions? There is a nice saying on economists that I subscribe to which captures this perfectly: a real economist does not tell you what you want to hear, but what you need to hear.

    • ianl says:

      >” … China, New Zealand, Australia. It’s obviously the case. ”

      Taiwan, South Korea, Japan, Vietnam, Cambodia … it’s obviously NOT the case.

      Well done, Nick.

      • I am and will always be Not Trampis says:

        Alas Ian you have to remember what Nick was referring to.
        Our newly crowned wing nut states all lockdowns fail.

        nick was merely showing this clearly is not the is not the case.
        Perhaps you have missed that not all lockdowns are the same and there are qualitative differences between them as well.
        Nor was he asserting lockdowns are the only solution

    • This discussion is wrong way round
      IF you can do containment suppression of small outbreaks really well then lockdowns are a useful component of a balanced strategy . Lockdowns without effective containment and suppression of small outbreaks are simply shifting deck chairs .

      • conrad says:

        Clearly NZ must have had effective containment then given their lockdown worked.

        • They blocked all arrivals ,when they had very few comunity cases .

          • conrad says:

            They cleaned up a small number of cases, so there is no evidence that they didn’t have decent tracing (c.f., Victoria) — basically, their strategy worked in terms of what they wanted.

            • Don’t understand what it is your trying to say.
              Can you name a place nation where containment suppression was crap, a nation place that over the past 14 months relied almost solely on lockdowns, where the results were, even remotely OK?

              • conrad says:

                I was agreeing with you. To make things work with lockdown, you also need good tracing. This is why the lockdown worked in NZ.

                • Thanks
                  you need good containment suppression it not lockdowns is central. If containment suppression can’t be done then lockdowns are cost but no even medium term benefit.

    • Abe Froman says:

      Australia and New Zealand are geographically separated from much of the rest of the world. Both countries also have very small populations in relation to their sizes and have societies which are not cosmopolitan; I.e. both countries are in a societally imposed lockdown anyway. Neither country is a passthrough port either so neither country sees a high proportion of travellers compared to other destinations.

      That said, Australia and New Zealand also suffer from being somewhat behind the rest of the world. It may be the case that Australia and New Zealand will have their corona moments while the rest of the world has moved on.

      Either way, the price of the lockdowns will have economic costs which will far outweigh the costs of coronavirus, costs which could have been reduced through a better funded health system. WA’s health system, for example, is already greatly under pressure and it has dealt with a negligent number of cases.

      Now more than ever we need to be scrutinizing government behaviour but your “wing-nut” comment speaks to the dismissive nature of the true believer.

      And China…well they do like a bit of state propaganda.

  2. Conrad says:

    I’ll offer the proper counter-condition based on what we saw elsewhere in white-guy countries where you had reasonable proportions of vulnerable people (e.g., 30%+), as you do in NZ (overweight, old etc.):

    1) Instead of an initial lockdown, you wait for ages
    2) This causes cases to explode, as they did everywhere that waited too long
    3) At some point, your medical system is overwhelmed, so no-one can turn up apart from covid patients.
    4) People modify their behaviour anyway, so even if they can turn up to hospital for other stuff, they don’t. They also don’t go out much, don’t get restaurants etc.
    5) The cases keep on increasing. Your medical workforce gets sick a lot (many with long covid), and they threaten to go onstrike, many don’t turn up even if they are not sick because they don’t want to get sick, don’t want to work 12 shifts etc..
    6) The government starts putting in more and more rules on things like opening hours, social gatherings etc. to try and reduce cases.
    7) The cases keep going on. Things are now very critical.
    8) The government capitulates and locks down for a few weeks to try and keep the medical system going and get voted in next election.
    9) Cases go down.
    10) The government re-opens, except since there are still lots of cases everywhere, you get invasive permanent restrictions (e.g., no cinema, no restaurants after 6:00 etc.).
    11) Cases go up again and this just cycles over and over until you get vaccines.
    12) You never work out a way to convince much of the population to act as normal, so you have basically worked a method to get permanent behavioural change that hurts the economy, unlike NZ and similar where things pretty much went back to normal.

    • Nicholas Gruen says:

      13) You remain blissfully untouched by self-doubt throughout.

    • paul frijters says:

      I see you are essentially taking the occurrence of a panic as a given. The idea that overflowing hospitals are a terrible event that simply necessitates a government action is a bit weird considering how they are normal for many countries in most winters.
      Crucially though your story essentially supposes that lockdowns have containment effects. That presumption is basically wishful thinking by now.

      But I do note the essential similarity between your stories and those of Nick. Both of you think you can see a causality on the basis of only very small numbers of cases, and then want any alternative to be proven to your satisfaction following your rules of what counts as evidence. And both of you manage on that basis to take no responsibility for the damage of the policies you have supported.

      That is not reason or humility as we know it!

      I can just see you and Nick trying to argue with a Melbourne woman who was abused by her husband during the lockdowns because she wasnt allowed to leave the home, whose kid is now a basket case because of the months at home, and whose cancer was undiagnosed and thus now untreatable, that she shouldnt complain about her lot because she cannot convince them to your 100% satisfaction that lockdowns haven’t saved her from covid. Besides, Conrad then adds, she cant prove either that all these bad things wouldnt be worse without the lockdowns.

      • conrad says:

        They are a political given in most places. All I did was describe the case of France (not dissimilar to many other places), which was a country that tried to avoid lockdowns as much as possible for as long as possible.

        On this note, you never came up with any strategy that stops this and stops people modifying their behavior to avoid catching covid (and pressuring the government politically for that matter).

        Maybe your strategy really would have been first-best if you were able to solve the above problem. However, the only country with lots of high risk people where this happened as far as I know was Brazil (although they apparently had some more localised rule setting). Sweden might be a better second best for you, but it is not clear how well they have done, and they also have very high levels of trust in the government, which most countries don’t have (they also had to implement rules too).

        So if this is the case, all you are describing is an unachievable ideal in most places, which is of course the case for any number of things the government wishes it could change. Apart from this, it generally takes decades for public health and other campaigns to change behavior to work (smoking, HIV, littering, sun-smart et.), suggesting behavior modification across the population is hard even for things that might seem comparatively easy (some never work, like obesity).

  3. I am and will always be Not Trampis says:

    Conrad is onto something here.
    First as George Orwell would say no all lockdowns are equal. Those here and NZ were vastly different to Europe and the USA for reasons I have explained previously. If you cannot acknowledge that then…
    however there is anther substantial qualitative reason. The former had proactive lockdowns. The latter reactive lockdowns. conrad describes them well. The former appears to have substantial popular support and were educative in social distancing etc and there had long term benefits.

    I would have thought India shows what occurs when you do nothing.
    Talking of India is our favourite cultist ever going to admit he was wrong about the stats there.

  4. Luke Thorburn says:

    Readers of this post might be interested in some work by Tony Blakely and his team at the University of Melbourne which tries to get at a similar question of which strategy of restrictions (lockdowns) for managing COVID-19 minimises public health losses, in the context of the vaccine rollout. In a nutshell, it found that a ‘moderate elimination’ strategy that relaxes over time tends to minimise health losses under most scenarios.

    There is an interactive web tool that allows you to play with the model parameters, link below.

    COVID-19 Pandemic Trade-offs

    I was peripherally involved in this work so happy to answer in any questions.

    • paul frijters says:

      I saw that when it came out. I see the economic impacts are yet ‘to come’ telling me even after 14 months you are in no hurry.
      I also missed the IVF disruptions of high-level lockdowns.
      I see many of the health disruption effects are still work in progress.

      It wasnt entirely quite clear to me but in your R0 and IFR you seem to take the parameter values that cannot rationalise what happened around the world (ie, the absence of huge surges in cases when lockdowns are abandoned, as they were in many US states), so the health simulations are essentially totally skewed towards the ICL-model type predictions.

      • Luke Thorburn says:

        Hi Paul, a few comments:

        – The ‘coming soon’ label on the economic impacts refers only to their inclusion in the interactive web tool, and to the latest re-run of the models to incorporate the effects of the vaccine rollout. The research group was incorporating economic modelling throughout last year, though not all of the research outputs were public. See this preprint for a recent example that includes economic effects. Key graphs on page 17.

        – The two health disruption effects that we are still working to include are those related to intimate partner violence and reductions in physical activity. There are of course many other potential effects, but for inclusion in the model we needed quality studies that estimated the size of the effect of lockdowns on a particular health effect as distinct from the effect of the pandemic in general. Most studies do not attempt to disentangle these two causes. The effects we include (and intend to include) are all those for which we could find good estimates of the lockdown-caused effect sizes.

        – If you have good data on IVF disruptions we’d be open to building it in.

        – The assumptions about the values of R0 and IFR were made to fit an Australian (particularly Victorian) context.

        • paul frijters says:

          the IVF restrictions are specific to states and hospitals. What I did to get a sense of the maginitude of the disruptions was to randomly sample a few hospitals and check their websites for what they were doing at particular times. There is no central registry that I now off from which to easily pull data, though IVF births are registered I think, so you will know about the disruption afterwards. I suggest the issue is important enough for your team to do what I did, namely to sample some hospitals yourself and talk to the IVF groups about when they stopped and how much disruption they had at particular times.

          What is the IFR assumed in this model and how did you estimate it? I am also curious how on earth you got to a R0 of 3 with Victorian data (there are a lot of issues as to how one estimates such a thing since you want to use it as a population R0 and not one particular to initial groups, which is what the earliest data would be on).

          It is symptomatic of how all this has gone that in the website you refer me to the economic effects are not displayed after 14 months, but that you refer me to page 17 in some other report for their existence.

          • Luke Thorburn says:

            Do you have reason to think that the IVF disruptions were due to lockdowns and not due to self-imposed precautions in light of the circulating virus?

            I should clarify that my role in this work was on the data visualisation side of things (ie. building the web interface), not in the modelling, so I will need to clarify with the epidemiologists who did the modelling regarding the calculation of R0 and which IFR was used.

            I’m not sure where you get the R0 of 3 figure from. The baseline scenario available in the tool is R0 = 2.5, which is intended to correspond to most of the 2020 virus strains. Two other scenarios (R0 = 3.125 and R0 = 3.75) are also available, to represent various mixes of ‘vanilla’ SARS-CoV-2 with more virulant strains that are emerging and may come to dominate. What R0 and IFR do you think is accurate? We can probably include them as additional scenarios in the next update of the tool.

            I share your frustration that this kind of analysis integrating net public health effects with economic impacts has not been available earlier in the pandemic. However it takes time to do this work, and we are working as quickly as we can whilst balancing other commitments.

            • paul frijters says:

              Hi Luke,

              thanks for clarifying your role. I wont hold you responsible for the parameter choices then!

              Yes, I mentioned a R0 of 3 as a kind of average of the different ones mentioned. If I would have to hazard a guess, I’d say a R0 of about 1.5 would be the one normally applying in Victoria, dropping to 0.9 with closure of large indoor gatherings (which is the one measure for which the case is still very strong that they must have a big effect on the R). You basically need a model with lots of sub-communities though to meaningfully simulate what happens if the average population R0 is 0.9 (because you will then still get flare-ups locally, but only in a non-representative population model). Yes, do suggest running simulations with that. Should prove humorous just to see their reactions to the suggestion.

              On the IVF stuff, yes. In some countries like the UK it was basically an executive decision to halt them as non-essential. Decentralised in Oz the clinics and hospitals were following general advice on restricting services to essentials and to avoid seeing any ‘patients’ not in acute danger themselves (which thus excludes women needing to go to hospital to get fertilised eggs inserted). So I definitely chalk them up to lockdowns. You see them start up again after lockdowns.

  5. KT2 says:

    Can ANYONE tell me what a lockdown is please. Martin Lally or Paul, I am sure you have a specification for your most important word. Is lockdown in the literature? Where is your database of dates, times, covid dynamics, culture, early late speed end etc etc etc???

    Nick Gruen, wing nut territory is close to Paul – “soviet economy”, “we’re all murderes” and “maximum deaths”.

    As Paul says:
    “I want maximum cases, minimum deaths as a means of getting to herd immunity with minimum loss.” & “I essentially see all of us as murderers”. 

    So your statememt Nick- “then we’re in wing-nut territory, something I was confident we didn’t do here on Troppo until these posts started” is true, and shows you have magnanimously let Paul’s many many wingnut territory friends and claims slide by. You must be a good friend. Eternal vigilance. No more winging it in a nutty way.

    +1 & Thanks NG “After all, these thoughts you see above aren’t really trying to get at the truth. They’re the fevered rationalizations of panic ;)”.

  6. KT2 says:

    Paul, we understand you will, to support your position, as NG says go “in wing-nut territory,” [ouch].

    Paul, Nick was trying to get you to accept ” that it’s not ‘contrary to all the evidence’ that lockdowns reduce deaths in places where they’ve been used sensibly and decisively and not eased up until they’ve achieved plausible local eradication”… simple. But not for covidiots. Or panicking hysterical people.

    As stated below in “Probabilistic Model for Control of an Epidemic by Isolation and Quarantine”, with full paper & equations: “effectiveness [isolation & quarantine ] highly sensitive to small changes of intervention strength” and speed and if introduced early or late in response.

    So to assist your in unwinging your nuts Paul & Martin Lally, here is a set of equations to show us all you are correct. I am sure, Paul, Gigi & Martin will have no problem using such to finally have success at, as Paul states “maximum cases, minimum deaths as a means of getting to herd immunity with minimum loss.”.

    So Martin, Paul, Gigi and tada! The IPA, when will you use this paper to provide a report on Isolation & Quarantine. Please.

    Bulletin of Mathematical Biology
    “Probabilistic Model for Control of an Epidemic by Isolation and Quarantine
    David V. Kalbaugh

    …”we discover an unexpected pattern in maximum fraction of population infected. We develop from first principles of probability an eighth-order system of ordinary differential equations to model effects of isolation and quarantine. We derive analytical expressions for reproduction numbers modeling isolation and quarantine when applied separately and together and verify them numerically. We quantify strength and speed required of these interventions to contain epidemics of varying severity and examine how their effectiveness depends on when they begin. We find that effectiveness is highly sensitive to small changes of intervention strength in a critical region. Finally, adding two more differential equations to capture natural population dynamics, we calculate endemic disease equilibria when affected by isolation and examine dynamics of coming to an equilibrium state.”

    “Figures 6 and ​and77 quantify strength and speed of the two processes needed to contain an epidemic, on condition that the processes start early enough. 

    …” In sum, assuming a homogeneous population we developed a model based on first principles of probability, with eight differential equations, six parameters and five reproduction numbers, and quantified important aspects of isolation and quarantine effectiveness in controlling an epidemic.
    “If quarantine starts early enough, the process can essentially be completed by the time the fraction of infected people becomes appreciable and, using Eq. (35), we can write

    “If quarantine starts late, and we will quantify “early” and “late” in the next section, dynamics of quarantine are simultaneous with those of infection and removal. In this case, we turn once more to the definition of reproduction number based on one infected individual in a population of susceptible but well people. Using Eq. (27), we find. ..” …
    “When isolation and quarantine are combined, we again have different formulas for early and late start of quarantine. In an early start, A and B processes are uncoupled in time and it is easy to see that the reproduction number for isolation and quarantine combined is(Equation 62)

    “For the late start, given the forms of and  above we might expect the combined reproduction number to be in the form (Equation 63)…

    …” In sum, assuming a homogeneous population we developed a model based on first principles of probability, with eight differential equations, six parameters and five reproduction numbers, and quantified important aspects of isolation and quarantine effectiveness in controlling an epidemic.”

    Referenced in above;
    “Mathematical Models of Isolation and Quarantine
    ” The model predicted that these policies [isolation & quarantine] would help and showed how dramatically they could reduce the size of a SARS outbreak (by a factor of 1000). These results agreed with actual observations.1,2 ”
    …”  but the model was able to illustrate the power of I&Q as control measures.
    “Models can provide rapid estimates of the impact of control strategies even before data from other areas are available (before epidemic spread occurs) and when experimental data may be incomplete or inaccurate.”

    “One of the most important contributions in mathematical epidemiology has been to show that the most important factor in any I&Q or treatment campaign is the speed of response.”

    Carlos Castillo-Chavez, PhD; Carlos W. Castillo-Garsow;Abdul-Aziz Yakubu, PhD

    Paul Fritjers quo vardis and personal decent into darkness from “The descent into Darkness of the UK and Victoria. Quo Vadis?”

    Paul F says…” In both the UK and Victoria, governments are trying to run a Soviet-style economy…”

    To which John Wormald says: at 5:25 am
    “I can’t think of a sillier statement than this.”

    Which,  after jocularity – insensitivity – saw Saupreiss say at 7:28 am
    “You just want people to die, admit it!

    Paul Frijters says:
    September 17, 2020 at 3:11 pm

    “hej, I am not the Prussian! I come from the pacifist tribes to the West, remember? :-)”


    I am and will always be Not Trampis says:
    September 10, 2020 at 9:21 am

    “The only thing sillier would be to do nothing and allow the virus to permeate society and see what happens.

    “A soviet economy eh? Good to see hyperbole is blowing hard.


    Paul Frijters says:
    September 10, 2020 at 6:23 pm
    “what would you call an economy wherein the vast majority of businesses and workers get their incomes directly from the government in return for (not) doing their job? Sure ain’t capitalism.”

    Paul then goes on to confirm in his own words that yes, he’d like to see maximum deaths:-

    Paul Frijters says: in reply to Conrad: September 12, 2020 at 5:42 am

    [Conrad said] ” if all you care about is reducing cases”

    PF reply: “that’s not something I care about. Indeed, instrumentally I want maximum cases, minimum deaths as a means of getting to herd immunity with minimum loss.
    what I’ve set as the goal is the wellbeing of society. All the rest is instrumental. What other goal would you advocate?

    I am and will always be Not Trampis says:
    September 12, 2020 at 7:32 am

    “herd immunity would mean at a minimum 175,000 deaths so your theory of minimum deaths is bunkum”

    Paul Frijters says:
    September 12, 2020 at 5:49 am

    “…. Also, I essentially see all of us as murderers in terms of the outcomes we support as voters and tax payers, which you might see as cynical but once you accept it as a fact is also liberating. None of us can possibly be the ideal our society asks for but we can try to improve.
    On covid, I hope you will join my point of view, as I presume you hope I join yours. Make the case persuasively and who knows? :-)

    The persuaders;
    “AN ALTERNATIVE TO LOCKDOWNS – Institute of Public Affaits
    “1 Sept 2020 — Daniel Wild, Director of Research … 15 Department of Social Services, “JobSeeker payment and youth … on the model prepared by Dr Martin Lally of Capital Financial…” 

  7. Nicholas Gruen says:

    Hi Paul,

    Can I just have a quick reality check here?

    I’m claiming that the evidence is overwhelming that in the cases I know of – in Australia, China and New Zealand – that lockdowns were a central part of eradicating the virus.

    I think it follows from that that if such methods were properly reproduced in similar circumstances – e.g. similar levels of infection in the community and closure of additional potential vectors of infection via closing borders and so on – that it’s likely they’d work.

    I’m not claiming (though nor am I conceding) that
    * All lockdowns work
    * Other methods would not have worked
    * That the effect of lockdowns in getting on top of the virus couldn’t have been done at lower cost

    Is your claim that that might not be right in my core claim – viz that the lockdowns might not have played a central part of eradicating the virus in the cases I’ve cited. If so what else do you think caused the eradication? To clarify, I’m not after any kind of counterfactual analysis about what we could have done without the lockdown, I’m after a causal story based on what did happen in which we eradicated the virus with a lockdown but the lockdown didn’t in fact make a major contribution to helping us do that.

    • No the central part is what you do, after you have reduced ( not eliminated) the opponent . (Eliminating something that was by the first of February was already endemic is not possible lest within the medium term )

      I Don’t blame you , you are after all an economist, however you don’t understand tactics and strategy ( or the art of “how”) at all.

    • paul frijters says:

      Hi Nick,

      yes, I understood your reasoning. Your phrase “that the lockdowns might not have played a central part of eradicating the virus in the cases I’ve cited. ” is indeed something that I now see as a possibility.

      Note that I am not claiming with any certainty that lockdowns were not central to the dying down of the covid waves in Australia and NZ. What I am now saying is that i) on the totality of the evidence around the world, the average lockdown in the average region doing them has no covid-benefits, and ii) that claims that lockdowns were central to covid-outcomes in Aus/NZ is not the certainty it seemed around July 2020. The doubt has come in because of the failure of lockdowns to have clear dampening effects elsewhere to such an extent that on average covid-outcomes in the Americas and Europe are worse in the lockdown places.

      That still leaves a lot of possibilities. It might be the type of lockdown, the type of restrictions, the combination of populations, weather, prior health, etc. My various posts have talked a lot about this the last few months, whereby more and more the spotlight has come on i) what is so unhealthy about lockdowns in many places, and ii) why are these covid waves dying down in some regions of the world but not others.

      It basically is now a serious proposition that the dying down of covid waves in much of E-Asia and Aus/NZ is due mainly to attributes the population already had in February 2020m (such as prior immunity). This is not proven at all, but neither do I think it is now beyond doubt that lockdowns helped in these regions.

      I told you my alternative possibility for Aus/NZ the last time you asked the same question (where you seem to forget everything you then hear as an answer. Its like you get a memory wipe every month or so on this topic). I think its well possible that in Aus/NZ the natural R0 is close to 1 anyway and that with just the banning of large indoor events where vulnerable people go to, any covid waves brought in from outside die down anyway without large numbers of deaths. All the other measures are then marginal or even counter-productive in terms of short-term covid-outcomes (a few months). Conrad and I discussed various aspects of that (whilst you tuned out).

      For China, as conrad has also made clear btw, things are more difficult as its totally unclear whether waves died down totally there or that small waves were simply ignored after a while, whereas at least for Aus and NZ it cannot be denied that the covid waves really died down.

      Your parting wish to play this “convince me to my satisfaction about what happened in my backyard” is not a game I am going to play. We have been there too many times by now and its not an open-minded premise. It is humility theater.

      • Chris Lloyd says:

        “I think its well possible that in Aus/NZ the natural R0 is close to 1 ” No Paul it is not possible at all. I reduced my interactions with people by at least a factor of at least 10, probably 50. So did most people. Yet it took many months for the daily infections to get down to less than 5.

        Partly this was due to sub-populations ignoring the rules (in SE Melbourne mainly) but there is no way that R0 was less than 1. You have seen how the numbers were increasing prior to the lockdown haven’t you??? Why would you suggest that R0 was only marginally above 1?

  8. Chris Lloyd says:

    “Martin estimates that on average, the more stringent the lockdowns, the higher the claimed subsequent covid death rates.” Sorry Paul. I call bullshit. Apparently human kind have just been silly to stay indoors during the hundreds of plagues over our history. These poor pre-historic humanoids had never heard of Wellbys. Lockdowns not reducing deaths is about as believable as their being a rape crisis on campus. Politically motivated ideological bullshit. Send me a link to a paper published in a reputable journal and then I will go to the trouble of debunking it.

    • paul frijters says:

      why dont you read his articles. Or simply look at the covid-deaths per million across Europe and note that the top 20 are all heavy lockdown places.

      Your answer displays a disinterest what has really gone on: you seem to imagine that populations all stayed indoors with everybody separated from others during lockdowns. Because you think that that would surely help, you thus deduce that lockdowns must be useful. The essential issue is that that image is totally false as it is not what happens with lockdowns. Many people are forced closer together. And exactly the ‘wrong’ type of mixing often gets higher because the vulnerable need constant help (which is why they are vulnerable). Its basically the healthy who run little risk and who are little risk that are kept indoors in lockdowns. Etc. But dont let reason bother you. Or data.

  9. There were two critical assumptions underlying the standard response to covid.
    One was that net reductions of movement greater than about 60 to 70 percent were actually achievable ( and that they would also not in them selves create their own transmission problem ).
    The second assumption was of the two assumptions the bigger gamble , it was assumed that it would be largely be finished by mid 2021.

    Nicholas I remember quoting the following to you back in about May of last year:

    “ In the first six to twelve months of a war with the United States and Great Britain I will run wild and win victory upon victory. But then, if the war continues after that, I have no expectation of success.”

  10. paul frijters says:

    An interview yesterday between BBC presenter McVey and Oxford Professor Sunetra Gupta, who has been pushing a very similar line to me from the beginning, almost perfectly mirrors some of the discussions above.

    [Gupta] said that a model of “where there was no effect of lockdown” on infections would “completely mirror what we have seen in many countries throughout the world”.
    At one point, Ms McVey says: “Sorry to interrupt you professor but multiple scientists say lockdowns have worked, the Government says so too.”
    This prompted Dr Gupta to hit back: “That doesn’t mean it’s true!”

    • Jerry Roberts says:

      I agree with you and Sunetra Gupta and have done from the outset and I am still hoping you will have a closer look at the vaccines. I have just watched Bret and Heather 78th Podcast where Heather discusses the latest animal research from the Salk Institute showing that the spike protein on its own causes cardio-vascular damage. When added to the work of Geert Vanden Bossche and Mike Yeadon, this points to Mike Whitney’s call to withdraw the vaccines and complete Phase 3 trials. The Israeli People Committee’s most recent report concludes “there has never been a vaccine that has harmed so many people.”

      • Hi Jerry,

        yes, I have seen some of that evidence, as well the blood clotting evidence. There certainly is evidence of harmful, sometimes deadly, side-effects. The problem is that there are side-effects with almost any vaccine or treatment, so the question is not whether side-effects exists but just how prevalent they are. On the prevalence, the main data I look at is the spikes in excess deaths for age groups across countries. The advantage of that data is that it does not rely on reported side-effects (which will be a small percentage of the actual side-effects). The over 80s were vaccinnated first in many countries and if the side-effects were really bad, we should have seen a very clear spike upwards following the vaccination programs (such as in the UK in February). But we didnt see a spike up. Rather, we saw a clear reduction in all European/American places. Now, the fact that we saw that reduction everywhere means vaccination is only one factor (another will be weather and natural herd immunity), but it does mean that the deadly side-effects cannot be very prevalent.

        I do remain open to new information on the prevalence of the side-effects. It is clear deadly side-effects exists, but their prevalence matters. There are always lots of claims around, but on this kind of question a reasonable view of the data is the only way forward.

        • Jerry Roberts says:

          Keep watching the data. The story is only just beginning. I doubt if we will ever find out what this is all about. Laboratory leakage seems the most likely but the fellas wot done it are unlikely to “fess up. I look for opinion that has no skin in the game and that includes you — people who are not making money out of the virus and are not part of the mainstream cheer squad. Ramesh Thakur has an excellent piece in today’s Pearls and Irritations.

      • Chris Lloyd says:

        “I agree with you and Sunetra Gupta and have done from the outset.” Thanks for saving me from having to read the rest of your post. You knew lockdowns wouldn’t work from the outset.

        • Jerry Roberts says:

          Hi Chris. Recommend you read the Ramesh Thakur piece in today’s Pearls and Irritations blog. In Australia we are postponing the onset of Covid-19. I think we will regret not spending the money on staffing our hospitals instead of hotel accommodation and building up the police state.

    • Paul
      suggest that it’s clear that if lockdowns are truly seen as delaying tactics ,aimed at buying the time needed to get containment suppression really truly up to speed then they are an effective ,if relatively small and easy to do, part of a integrated strategy that does’ work’- obviously it leaves the ‘ what next ?’ question- covid is clearly endemic , unlikely to be eliminated in our lifetimes.

      • in the debates in Europe it has often been suggested that lockdowns are delaying something. That is certainly how they started out being marketed (“flatten the curve” = delay). The problem is that empirically it just does not seem to work that way.

  11. I am and will always be Not Trampis says:

    time to retire our wingnut and tell him to write for Catallaxy where his talents will be appreciated.
    gosh hospitals are only going through what occurs in winter. well no. ABC interviewed a surgeon who is assisted nurses in the UK ie he is nursing. Why. Because covid patients are crowding every other patient. until the covid surge is falling to a significant degree he won’t be doing any surgery. covid patients rise at an exponential rate until policies are implemented to resist this. this is why they crowd out every other disease in an hospital.
    Only an embicile would believe a lockdown has no effect on covid infections. There is zero evidence lockdowns of any variety do not curb infections and reduce pressure on hospitals.
    It is covid infections that determine if people who wish to have IVF can do so.
    A person saying lockdowns prevent these people getting IVF treatment is either completely stupid or an inveterate liar.
    Burt let us go to India where our wingnut embarrassed himself yet again.
    firstly the data is not robust. the Federal government has now accepted more peeople have died from covid. Of course they are not to blame. A State government has been caught redhanded lying about how many died. Thay have been Fritjered so to speak.
    Next Fritjers simply does not understand people dying from covid lags people being infected from covid. Lagging is simply something beyond his comprehension.
    Lastly and I could go on but won’t if a hospital system is gone to gowings and patients being treated on the street is confirmation of that and workers are over worked to the nth degree then it follows there will be more deaths not less from covid.
    never let it be said logic enters the head of the anti-lockdown cultists .

    Fritjers would not know data if is was put in front of his face

    • Nicholas Gruen says:

      Homer, can you go easy on the abuse.

      Debate is what we’re after here. Besides, we all have our wing-nut moments whether we share them with the world or not.

      I’d like Paul to reply to my own comment in a way that he thought was more likely to persuade me, I’d like him to respond to Chris’s point as if Chris wasn’t just an ignorance peddling fellow.

      Not much point in him responding to abuse.

      • “I’d like him to respond to Chris’s point as if Chris wasn’t just an ignorance peddling fellow. ”

        then it would help if Chris didnt use the word “bullsh*t” and made a clearer effort to see where Martin was coming from.
        As to responding to your comments, we seem to be having the same conversation every month. By now, they feel like a waste of time. But we could do something more constructive, perhaps an open online podcast or something? One with symmetry in the questioning?

        I do agree with you though that it is important to have no abusive language on Troppo no matter how passionately we feel about the issues. Even if we are certain the other is acting in bad faith, there is still no point to abusive language.

        • Chris Lloyd says:

          There is no need for you to respond. I am done on this particular topic until some real econometric evidence is provided (which does not mean some random NZer).

          Saying that “I call bullshit” on a theory is surely within the bounds. OK, maybe I overdid it with the F-bomb. But I am an Aussie after all and my dear old Dad enjoyed the odd expletive in an argument. However, I see no reason to argue with you.

          I spent a few weeks arguing with a flat earther FB friend last year before I unfriended him. I simply am not required to check empirical claims which have so little chance of being correct. Ufologists point out “evidence” all the time, as well as crop circle nuts. It makes no sense for me to spend time in assessing their evidence. The same for ESP experiments. Not only is a mental force ruled out by quantum mechanics (source Sean Carroll) but decades of research has failed to find it. I put your peddling the notion that locking people up in their homes will not reduce the spread of disease in the same boat. Once a study appears in an A* econometrics journal I might take it more seriously.

          It is a pity you have ruined an initially useful set of posts with stuff I consider wing-but (Nick’s phrase. I would add the f-bomb myself). Government listening too much to health experts instead of doing a CB analysis? Check. Government using populist rhetoric to make people more compliant? Check. Dan Andrews claiming he was using the science when he wasn’t? Check. Government overreach banning protests? Check. Refusal of anyone in the MSM to take a contrary line? Check. Betrayal of youngest in society to possibly save a small number of people in their 80s? Check. But saying the lockdowns have no effect on infection rates? I do not feel the need to be polite on that topic

          • Nicholas Gruen says:

            Got to say, apart from the politeness stuff – being rude is basically a loss of one’s own self-control (what’s the point?) – I agree with pretty much everything you’ve said Chris. I don’t mind Paul arguing that in some general sense lockdowns may not work qua lockdowns. In other words that there may be plenty of cases where they’ve not worked.

            However I can’t come at the proposition that lockdowns haven’t made contributions to reducing and effectively eliminating COVID in numerous cases – which I’ve cited – China, Australia and New Zealand.

            Fwiw, I also agree with you that your use of the term ‘calling bullshit’ is use of the idiom that need not be taken in our culture – and shouldn’t be taken – as abuse.

            • Chris Lloyd says:

              I apologise for swearing. The upper case was not supposed to be angry shouting. It was more like Woody Allen raising his hands and shoulders in a WTF gesture?Text is very deceptive, especially without emojis.

              Actually, I swear a lot in my “real life”. I think our generation swear much more than millenials, though I do not think you ever have. Fred was very polite in a the Zweig tradition, Alan not so much. Lovely guy but had a foul mouth which I lived around for 40 years. My swearing is partly my upbringing but possible a conceit that I use to present as not part of the academic elite!

              Pretty sure I was not abusing Paul. I was just challenging him to provide convincing evidence for a prima-facie absurd proposition.

  12. I am and will always be Not Trampis says:

    What do you term abuse nick?

  13. BTW
    hospital emergency bays are overloaded much of the time – in Victoria the other day a woman feeling sick rang for an ambulance by the time the ambulance arrived about six hours later she was dead and in nsw trainers and office staff have recently had to join the frontline to make up for serious staff shortages.

    As hospital treatments have got evermore hi tec and very capital intensive, more and more of our smaller suburban and regional hospitals have been closed and not all of the beds they once housed were transferred to the big central hi tech hospitals that have replaced them.
    From memory over past twenty years the total number of hospital beds in the UK shrank by about 20 percent.

    Most of those capital intensive hospital treatments are principally focused on extending the lives of people who are already post three score and ten, so we have planned well for an epidemic of venerable old age but not so well for a widespread, infectious disease epidemic

  14. KT2 says:

    Paul, another nail. Please feel free to verbally verbal this study. And do try to answer Nic’s questions please.

    “SARS-CoV-2 elimination, not mitigation, creates best outcomes for health, the economy, and civil liberties

    We compared COVID-19 deaths, gross domestic product (GDP) growth, and strictness of lockdown measures during the first 12 months of the pandemic for Organisation for Economic Co-operation and Development (OECD) countries that aim for elimination or mitigation (figure).


    FigureCOVID-19 deaths, GDP growth, and strictness of lockdown measures for OECD countries choosing SARS-CoV-2 elimination versus mitigation

    Caption: ” OECD countries opting for elimination are Australia, Iceland, Japan, New Zealand, and South Korea. OECD countries opting for mitigation are Austria, Belgium, Canada, Chile, Colombia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Israel, Italy, Latvia, Lithuania, Luxembourg, Mexico, the Netherlands, Norway, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, the UK, and the USA. Data on strictness of lockdown measures are from Oxford COVID-19 government response tracker. Data on COVID-19 deaths are from Our World in Data. 3 Data on GDP growth are from OECD Weekly Tracker of economic activity. 4 GDP=gross domestic product. OECD=Organisation for Economic Co-operation and Development”

    …” Although all indicators favour elimination, our analysis does not prove a causal connection between varying pandemic response strategies and the different outcome measures. COVID-19 deaths per 1 million population in OECD countries that opted for elimination (Australia, Iceland, Japan, New Zealand, and South Korea) have been about 25 times lower than in other OECD countries that favoured mitigation (figure). Mortality is a proxy for a country’s broader disease burden. For example, decision makers should also consider the increasing evidence of long-term morbidities after SARS-CoV-2 infection.”…

    “Among OECD countries, liberties were most severely impacted in those that chose mitigation, whereas swift lockdown measures—in line with elimination—were less strict and of shorter duration (figure). Importantly, elimination has been framed as a civic solidarity approach that will restore civil liberties the soonest; this focus on common purpose is frequently neglected in the political debate.

    “Evidence suggests that countries that opt for rapid action to eliminate SARS-CoV-2—with the strong support of their inhabitants—also better protect their economies and minimise restrictions on civil liberties compared with those that strive for mitigation.”…

    Published:April 28, 2021

    • paul frijters says:


      let me do you one better and sketch 1) what the medical literature sees as the gold-standard type evidence, and then 2) the three most prevalent deviations from that gold-standard that can produce you the supposed ‘finding’ that lockdowns help. You can then hopefully read this study or the many others and find the tricks they employ yourself.

      The gold standard is an RCT, meaning one wants to be able to randomly tell a large set of regions to lockdown and others not to lockdown, then comparing the outcomes over time. Now, obviously, there is no RCT. The closest we have is the Danish study looking at local lockdowns on the basis of the somewhat ‘random’ happenstance that some had a mink farm in that region (and were thus locked down) and others did not. That study finds no significant effect of lockdowns on infections (they are having, surprise surprise, a hard time getting published). Other somewhat natural randomisation studies look at North Dakota versus South Dakota (no difference found) or Florida versus California (have a guess at their findings?), or countries in the same neighbourhood at the same time doing different things (have a guess?). However, clearly, no study does the gold standard perfectly so the whole literature has to make do with more naturalistic data.

      You should of course note that that meant that all the initial lockdowns had no scientific backing and indeed went against the prevailing scientific advise. It is so easy to forget that the decision to lockdown was totally unscientific. But of course you know that and were outraged by it, since you are such a scientific purist. But let’s move on.

      Trick number one is to simply have a model in which you presume that lockdowns prevent a huge increase in infections and then you look at whether there was a huge increase in infections after lockdowns to ‘deduce’ its effect. This trick always works and several Lancet (and other top medical journals) carry such studies. Key thing is not then to include regions without lockdowns (though you can include a bit of regional variation as long as you dont have place/time comparisons that are lined up). Garbage in, garbage out.

      Trick number two is to treat covid-outcomes as a cross-country time-series whereby you select on the regions with lockdowns, comparing after with before, whilst ‘controlling’ for the surge by including the change before lockdowns. You can then even include seasonal fixed effects, country fixed effects, and population characteristics. Success guaranteed in finding a covid-reduction ‘regression estimate’, particularly if you censor the data to 2020. The key thing is not to have actual comparitors in there, like countries without lockdowns or even properly coded full-cycle time-series. I have seen this trick come by several times in the top journals.

      Trick number three is to go purely cross-regional and conveniently label the countries that experience a favourable amount of the outcomes you want (like low covid-something) by the label you want (like ‘elimination’), and then compare those regions with the other ones. One should then stay away from empirically-defined measures like the average of that Blavatnik centre ‘severity index’. You might call this your standard ‘rabbit in the hat’ trick. A version of this one is to look at what countries called their strategy in 2021 (but not early 2020!) and use that to compare. As long as you stay away from any notion of ex-ante comparability, you will get what you want with this one.

      And if you are truly sophisticated, you can manage to do all three tricks at once, calling it a hybrid model.

      Now the test to see if you have understood the above: what tricks did the articles you have been linking to employ? A level simpler: what ex-post information was used inappropriately as ex-ante information? Or, simpler still: how does this study differ from an RCT?

      • Nicholas Gruen says:

        Thanks Paul,

        When you say an RCT is the ‘gold standard’ that’s a manner of speaking. It’s a gold standard in some respects and very poor standard one in another. An RCT of whether parachutes work isn’t a gold standard because it’s ill-suited to the problem at hand – it’s costly – in money and more importantly lives – and we can work out the answer without using any gold.

        Below I’ll cite philosopher of science, Nancy Cartwright on the principles I think should be applied here, however (scandalously) little they’re taught in econometrics classes. Had they been applied here before the econometric testing got done, lots of lockdowns would have been excluded from the set of cases used for testing the proposition “do lockdowns work” for reasons lots of people have pointed out – and which I don’t think you’ve addressed at all well.

        To address a point you’ve made earlier – about my ‘amnesia’, it’s true I’ve not responded to all your points. That’s because I want you to respond to mine and lots of people here. If you want to address my point “why did all the lockdowns in Australia seem to work” you wouldn’t include lockdowns that were lifted before the virus had been effectively eradicated. In fact you’d need a convincing explanation for how Melbourne got rid of its COVID with a lockdown and a fully worked out story that people found convincing about how it wasn’t really the lockdown. Not some vague statement that doesn’t even convince you that it might have been something else. The more plausible story is that, particularly given the ongoing incompetence of the Victorian Govt – which continues btw – it needed a full lockdown to get to effective eradication.

        Here’s Nancy:

        Causal Principles
        “Will this policy work here?” The question focuses attention on the policy. But to answer it, your focus must be directed elsewhere. Where? The answer to that is supplied by considering some general facts about how causes work to produce their effects.
        Causes do not produce their effects willy-nilly but for a reason. They produce effects in some systematic way, in accord with some causal principles. A causal principle for a situation lays out all the factors that operate to bring about the outcome in question in that situation and shows how these combine to produce it. We stress three important facts about causal principles that matter to getting the right prediction about whether a proposed policy will work for you:
        1. The causal principles that underwrite policy prediction are not universal.
        2. Few causes work on their own; causal factors work together in teams.
        3. There are generally a number of distinct teams at work in any situation, each making its own contribution to the effect.
        1. Causal principles are not universal. They differ from place to place and from time to time. That means that it is not enough for you to know that the policy worked somewhere or even that it has worked at some time here. “It worked there”; it played a positive causal role there. So it was one of the factors from a causal principle that holds there. To predict that it will work here, you need to know that it is one of the factors from a causal principle that holds here. That is what ensures that it can play a positive causal role for you. You will read more about finding factors that can do so in II.B.
        2. Causes work in teams. What gets highlighted as the cause – where for you
        that means your policy – is rarely enough to produce a contribution to the effect on its own. It needs team support. If any of the essential team members is absent, the policy won’t make any contribution at all. It is like trying to make pancakes with no baking powder. So even if you know, maybe from a good RCT, that the policy worked there and that the same causal principles hold here as there, that is not enough to conclude that it will work here. That only shows that it can play a causal role here. To know that it will play a positive causal role here, you also need to know that you will have the requisite support factors here when you need them. That’s what II.A is about.
        3. Distinct teams produce distinct contributions. The overall effect achieved is usually made up of separate contributions from a number of different teams of causes, some of which can pull in different directions. The magnet, in team with the iron in the pin, contributes an upward force on the pin; the pull of the earth, in team with the pin’s mass, contributes a downward force. The overall force is a combination of the two. Social causes are just the same. Some improve the outcome in view; others contribute negatively to it. This makes predicting the actual outcome difficult. To predict that, you must take account of all the factors at work and of how they combine; that is, you need to know the full causal principle. Our concern is with less ambitious predictions than “What will the actual outcome be?”. We focus on “Will this policy make a positive contribution?” Will it make things better for some individuals than they otherwise would be? In our terminology, “Will the policy play a positive causal role here?”

        • Chris Lloyd says:

          I attended a Zoom lecture of Nancy Cartwright about two weeks back. If I had listened to her a year ago, I would have dropped out after 10 minutes. She rattled on about mid level theories and dissed RCT’s. I am a typical academic, more interested in causation within a narrow clinical framework.

          But in the current world where we are trying to understand causal effects of different public health policies within a messy real world environment, all sorts of pennies were dropping for me.

          There are lots of commonalities with David Deutsch’s musings on causality in his book Fabric of Reality which I recommend for people who are smarter than me because, in the end, I just could not understand his Theory of Everything which he claims we already have.

          • Nicholas Gruen says:

            Having read David Deutche’s Beginning of Infinity, I’m sceptical of your comments on his other book. The one I read was very interesting – at least for the first half. One thing I recall is for the first time coming across a reasonable explanation for why Lamarkian ideas seem appealing but leave holes in one’s understanding of how something could evolve that Dawinian ones don’t. (This was perhaps too soon as Lamarck is making a comeback :)

            Then it got more and more eccentric.
            Even before then, I remember his philosophy leant heavily on Popper’s falsificationism. It’s forgivable that Friedman and Samuelson fell for that stuff in the 1950s, but he doesn’t seem to have got the memo that Popper’s falsificationism hasn’t really survived the scrutiny of philosophers. Here’s quite a good article on it.

            I wish I’d known Cartwright was here – I’d like to have come.

        • Nicholas thanks for the heads up
          we are about to drive to South Australia via Mildura and did not know that despite all the QR apps etc, that we still needed passes for both vic and sa.
          Of the two the SA is the worse, I spent more than two hours on our iPad trying to get it to work, even downloaded Chrome in case it worked, eventually found a phone number for SA and was told : that it’s a SA police site that doesn’t work with iPads and some iPhones, a ‘design feature’.

          Have eventually sorted it , it’s incredible that even now after all this time the work of doing this kind of stuff is not done by those who are nominally paid to do it , rather it’s for the unpaid general public to do.

          Little wonder that Australia is through out much of the world known as ‘treasure island’.

          • Chris Lloyd says:

            What is incredible is that you need a pass to travel from one state to another, when there are ZERO cases. I am really on Paul’s side on this level of govt intrusion into our lives. I thought our constitution specifically prohibited restraint of trade between the states. Stopping people movement sure sounds like restraint to me.

            It would be good if someone organised a deliberate flouting of these rules with the MSM there when we got arrested at the border. I reckon I will be up for it. I do not accept the right of the govt to stop me driving from Melbourne to Adelaide and back again when there is not health threat.

            I had never seen internal national borders before I bussed through China in 1996. The road stops were heavily with military. You needed an internal visa. We now live in China it seems.

            However, the MSM would no doubt portray those protesting as white RW bogans as they did last July. If Paul is right about one thing it is that there is no open debate about what is necessary for public health, no outrage.

            Unless is it about Indians who happen to have darker skin (and I am very sympathetic to all of those stuck in India). My point is that Geoffrey Robinson was never outraged about inter-state travel restrictions.

        • BTW such persistent so entrenched incompetence on something as critical as this, is in itself indisputable evidence of corruption.

        • paul frijters says:

          sure, RCTs have their problems and I agree with your general critique of them (I have critiqued them many times myself). I am not claiming I see them as the gold standard, but that in the medical literature they are seen as the gold standard.

          On the amnesia, we seem to be going in circles. Your request “you’d need a convincing explanation for how Melbourne got rid of its COVID with a lockdown and a fully worked out story that people found convincing about how it wasn’t really the lockdown” is the same one each time. I suggest to you that that demand is part of your defense mechanism so as not to have to engage with many issues. Just a suggestion.

      • conrad says:

        Thanks Paul, that’s useful. I have some comments on it:

        1) I agree with Nick that RCT trials are overhyped. That’s especially the case when you have different factors with different very different effect sizes in different places. A good example of this would masks and the debate people had over them. One suggestion was they work on Japanese public transport because people are quiet, but don’t in other places because everyone yells and talks on trains. Let’s assume this is true but something most people don’t think about. If you did an RCT trial Japan, it would look like masks works, but this doesn’t generalise. So you learn nothing useful unless you live in Japan. Another example would be previous immunity. Let’s assume some groups have it and some groups don’t. Clearly any RCT trial I do them won’t be useful for populations that don’t have it.

        2) In term of trick number one, I agree – I am surprised how sloppy many fields up with baselines. But it works in reverse too, since you are clearly comparing two different conditions with essentially an orthogonal manipulation. So you don’t just need to answer why Sweden went up, but, if lockdowns don’t work, you need to answer why, say, Finland didn’t (and all the other places that didn’t get big numbers of cases but had lockdowns). What is holding their cases down?

        3) In terms of the second trick, there are ways around this that are done in physics (which I don’t know enough about unfortunately). But if you don’t have decent comparison groups, and probably never will have (which be might true given how much factors may potentially differ in different places), you can use quite finicky designs within time series. Predator-prey systems are the best and simplest example of this (where populations of one species always lag another causing two sine-like waves across time), but I don’t think I saw any covid papers using it.

        4) Whose measure is the best? Beats me. For a bit fun, I did some searching on the measure you mentioned, and one of the interesting things about that measure is that the Indian government took credit for doing such a good job on stringency and the Blavatnik guys said there measure was not good for it — note the date and imagine what they must be thinking now.

        It is also unclear to me that you could ever have a single scale measure of stringency. Presumably there are interactions left, right and center (e.g., wear masks && don’t yell vs. weak mask, yell, or weak masks and yell). So if this the case, thinking about things in terms of linear and additive scales won’t work.

        • thanks Chris. Let me respond in kind.
          On 1), of course, fully agreed, simply stating that this is how they view things in medicine. They want an RCT as proof.
          on 2) yes indeed, and I hope you have seen in my various writings how my now current response to the relative outcomes in different scandinavian countries is that i) Sweden became the most restrictive country within Scandinavia ending up with the most covid-deaths (even in just the second wave), so the intra-Scandinavian evidence goes in exactly the direction of my general observations, and ii) covid-deaths are just very low in that whole region even though a wide variety of policies was tried at different times (no lockdowns, harsh lockdowns, schools open, schools closed). So the region as a whole illustrates the lack of covid-armageddon as claimed by the SIR models.
          on 3) yes, issues get tough in time-series, but my essential point is that you can easily obscure the variation you are relying in with time-series models away from the variation you trust in an RCT (which is to start an intervention in a comparable group and then compare over time with a control group). The Bhattacharya paper I linked to earlier criticises several studies with that design.
          on 4) sure, measures have their problem. What I like about the Blavatnik one is that it takes indices of the main things important for wellbeing (schools, travel, and stay-at-home) and they have made an effort to get comparable data for all countries by day. Also, they update and backfill their measure (so their estimates for what happened previously get adjusted), but obviously their measure has problems in countries with lots of regional policy variation (like the US or China). They are trying their best on that though and I find them quite conscientious. I also note that they publish the full items separately, allowing researchers to construct their own index. Do you have a better one?

          I do want to push back on two crucial elements in your replies though. One is the notion that we should only take published studies seriously in a situation where the panic has derailed much of normal science. Essentially, many journals now have political skin in the game, such as that Lancet editor that started accusing the UK government of all kinds of things in a book. One cannot regard them as unbiased. Science will self-correct such biases, but that normally takes years. So I for instance see the papers by Gupta, Bhattacharya, Ioanidis and a few others as at least as informative as published papers in particular journals. You can disagree, but my essential point is that academia has been compromised in this panic, particularly medicine. Just as one wouldnt trust an economic paper published in a top Soviet journal to be unbiased about Soviet economics, so too do I now look very wearily at papers in the Lancet, BMJ, etc., when the editors of those journals have taken high-profile public opinions on lockdowns and other matters. Indeed, if you bind yourself to the rule that you only trust what is published in high-end journals you in effect abandon your own judgment.

          The second point is your notion of “a priori bs” on the effect of lockdowns in Australia (over and above the item I keep taking out of the usual package, which is large indoor gatherings with vulnerable people). I know that much of the media and governments in Australia keep claiming lockdowns have achieved this and are ensuring that. But so too in the UK, Spain, Italy, the US lockdown-states, etc. The governments and media there too bombard their citizens constantly with this “lockdowns are working refrain”. Consider the possibility that you have been so acclimatised to that constant bombardment that you too have started to see it as “obviously true”. Just saying: sometimes what is agreed upon by all is not obvious anymore when one looks at it. Now, you can then react like Nick and hide by demanding of a doubter “then tell me what happened in my neighbourhood and prove it” (which is basically a way of not engaging at all), or you ask yourself “yes why am I actually so certain that lockdowns work to eliminate? Is the mechanism I have in mind really the one that operated and is that mechanism the one that was supposed to operate in all those other places?”.
          Put in another way: where does your “a priori” really come from?

          • conrad says:

            I think that reply got two people :

            1) This was probably for Chris, but I agree with you about published studies. Lots of garbage ends up in the top journals — one only needs to look through issues from decades back to see what the hot topics were.

            In some fields, there is also a massive bias towards publishing stuff from authors from the prestigious universities — and not just journals. A good example of this is the guy that thought of CRISPR (Francis Mojica) who worked at some normal Spanish university no-one outside Spain has heard of and didn’t get the Nobel prize. Instead they to two people who worked at the Max Plank and Berkley. You can imagine what would have happened if the Spanish guy came from Harvard.

            I also agree that anything out of the mainstream will take forever and a day to published and will generally end up in a not so great journal. There are endless examples of this — CRISPR is one.

            Just to make things worse, the top journals also want papers that are cited a lot now (c.f., good science), so it is unsurprising that so much less than reliable covid stuff will be published.

            2) Most governments always claim they are right if at all possible. This means essentially all social science, economic, and epidemiological policies. I wouldn’t trust them as a basic rule. They also employ cronies (as you have written about), and the idiot that ran the Victorian covid response was clearly one of them. He’s still employed on some gargantuan salary.

            3) At least in Aus, I am willing to believe lockdowns worked unless you have better evidence — some of which is testable and has been mentioned here but generally ignored (e.g., it becoming endemic in pets and wild animals which then spread it again).

            In other areas, at least to me, it is clear why they never worked well and are hence not comparable — they were not designed to get the numbers down, and so you were left with lots of cases when things re-opened. This reminds of John’s analogy with bushfires, which, when fuel exists, need to be basically put out or they just re-emerge due to the exponential growth characteristics. This is a property of many systems and is well understood.

            I haven’t added the numbers up, but I imagine that this type of lockdown would have a vastly worse benefit/cost ratio than ones with other aims.

            There were also what I will deem ‘wasted lock downs’ where countries started up tourism quickly again (e.g., Greece), and simply got the virus back quickly.

            I also don’t think lockdowns can work in some places even under the assumption that they do reduce cases — basically poor countries with largely subsistence farming.

            • Hi Conrad,

              yes, I think it was you who first mentioned the pet thing to me, which is very important indeed for the overall picture. It basically makes eradication impossible. You will be happy to know this is a big point in that booklet I am writing!

              I was also intrigued by another implication of the possibility that this virus now jumps to and fro between humans and particular animals. Perhaps one reason for why this thing spreads more easily in some places than others has to do with the nature of their wildlife and how they interact with humans. Maybe Africa, East Asia, and Australia has the ‘right’ kind of wildlife and the Americas and parts of Europe have particularly ‘wrong’ wildlife. I have not seen any analysis of this possibility, but its an intriguing thought. Have you seen anything on this? Can marsupials get corona? Grey squirrels? Is there a region where all cats are forced to stay indoors? There are a lot of interesting possibilities.

          • Nicholas Gruen says:


            1) I agree with your comments about top journals and the need to retain one’s critical judgement.

            2) I suggest you go on a strict diet of not speculating on the motives of those you respect who disagree with you. It’s hard to enumerate all the reasons it’s a bad idea to do this. I don’t notice anyone else doing it other than people who render themselves peripheral to the discussion by hurling abuse around – and even then their abuse tends not to pose as psychological diagnosis.

            It certainly presents you with a fertile field for not listening to what people are saying – not putting it in its best light. You can reach for your Mandy Rice Davies trump card (“He would say that wouldn’t he”) at the drop of a hat – rather than engage with them. It’s gets worse, in speculating about their motives you imagine them to be arguing things far stronger than they are despite their protestations to the contrary.

            For instance a while back, you subjected what I was saying to a thorough psychoanalysis. It’s just wildly wrong – absurd. It doesn’t just imagine a whole psychological persona for me to inhabit, on the back of that it entirely reinvents what my argument was. Go back and read it again. If you don’t cringe with embarrassment, you’re a stronger man than me.

            Then you say that I’m not engaging because I’m not participating in your enumeration of econometric sins. That’s a fine discussion which others are participating in. I’m trying to ground-truth your claim with what I know. It’s NOT what the incompetent Victorian Government tells me I know – it’s the basic story of what happened.

            A second wave took off in Melbourne. The government imposed lockdowns which only flattened the curve and with infections rising it imposed a very strong lockdown with curfews. Nothing much happened to daily infection rates for a few weeks and then the number of cases crashed, and over a few more weeks they went to zero.

            As far as I understand it, your counter-explanation – of Melbourne, not the whole universe of countries, but of Melbourne – is that R0 was high enough to produce the rapid growth, and then fell not just below 1 but so much below 1 that numbers crashed to zero in a few weeks and that this was the product of some other much less intrusive measures – like banning super-spreader events.

            I’m not asking you to prove what happened in Melbourne, I’m calling for you to tell me some story that I and other reasonable people find even vaguely plausible in which lockdowns didn’t make a major contribution to reducing COVID spread in Melbourne.

            You are the one arguing general propositions – not me. I believe the hypothesis that some lockdowns have worked – that’s it. I’m using Melbourne as the best illustration, and have suggested it probably generalises to other sites in this part of the world, but that’s not where I’m starting. I’m starting with Melbourne.

            I just need a plausible story – not proof. Just something that is consistent with your assertion which I take to be “lockdowns don’t work, therefore Melbourne’s lockdown didn’t work”. I’m not even close yet. That’s not because of government or media propaganda. I just can’t see how banning sporting events and a few other things produces the gyrations in R0 that we observed and even if it did not even that is consistent with what we observed because we did all those things in the early phase of the lockdown and, it seems R0 remained well above 1. It then crashed to well below 1 as the lockdown tightened.

            If you think this isn’t engaging, then I suggest you volunteer to be a subject in a parachute RCT :)

            Just a plausible story.

            • Actually, I should correct my second last para

              In the first phase of lockdown R0 may have fallen to around 1 or possibly a tad below 1, but it feel markedly during the second lockdown

            • “If you think this isn’t engaging”
              I indeed do not count this as engaging. When you asked me what I meant with my assertions on lockdowns (your ‘reality check’ question) I made it clear what I was talking about and how I have now come to doubt whether lockdowns is what was responsible for the dying down of waves in Melbourne or other places in Australia. That comes with no obligation whatsoever to sound plausible to you in your internal court of plausibility that operates under your personal rules.

              What would count as engaging is for you to make an effort to generate a general story on the effects of lockdowns on, say, covid case numbers and covid deaths around the world. One that makes sense of the various things done, claimed, and experienced around the world. Why dont you make that effort and then write a post about what your thinking has produced. I do hope it turns out deeper than “I saw it work in Melbourne so it must have worked everywhere” or “where it didnt work, they clearly didnt do what Melbourne did”.

  15. Re “ elimination “
    Count no man happy until the end is known.

    China, Australia and a some other nations are down to a handful of cases per week , but none are at zero cases . None have eliminated the virus.

    Is there anywhere in the world that has eliminated covid ?

  16. Jerry Roberts says:

    Australia has postponed the onset of Covid-19 with isolationist policies. Our hospitals are poorly staffed and are not even coping at a time of minimal demand. We routinely have ambulances “ramped” up in emergency department driveways waiting for hours to discharge patients.

  17. I gather that the reduction in movement in Sweden was much the same as in Germany.

    In most of the developed world there is a significant section of the population that must continue to work if essentials are to be maintained. And many of them have little in the way of paid leave and live in fairly crowded accommodation.

    Feel that a better question is , do mandated lockdowns actually achieve that much more reduction in movement than a more voluntary approach?

    It also seems to me that the length and severity of lockdowns maps rather well to the level of entrenched incompetence ( i.e. corruption ‘rust’ )of the government of that place.

  18. I am and will always be Not Trampis says:

    Simon Wren-Lewis with a different take

    • KT2 says:

      Thanks IaawaNotbTrapis

      Paul, this paper says “In order to compute YLD, though, we must have a thorough understanding of the sequela associated with the disease, as well their prevalence. Several sequelae have been linked to COVID-19 recently 20,21 in China, but we still lack the full understanding…” – how come you can compute this, we’ve not seen you calcs, and they say more work to get data before Daly’s etc? Will you present us with your data and calcs please?

      “Years of life lost to COVID-19 in 81 countries
      . ..” This is in line with further evidence of the life-saving impacts of lockdowns and social distancing measures15.

      “As noted earlier, our analysis is limited to premature mortality. A full health impact evaluation ought to consider the burden of disability associated with the disease.Indeed, YLL are often presented jointly with years lived with disability (YLD) in a measure known as disability-adjusted life year (DALY), constructed by adding YLD to YLL19. In order to compute YLD, though, we must have a thorough understanding of the sequela associated with the disease, as well their prevalence. Several sequelae have been linked to COVID-19 recently 20,21 in China, but we still lack the full understanding of the extent that would be needed to compute reliable cross-national YLD measures at the scale of this article. We see collection of such measures as therefore of key importance in next steps in advancing our understanding of the magnitude of the COVID-19 effects on public health.”

      Scientific Reports volume 11, Article number: 3504 (2021) 
      Published: 18 February 2021
      Héctor Pifarré i Arolas, Enrique Acosta,[…]Mikko Myrskylä 

      “How most of the West got the pandemic so badly wrong?

      “But we also need, and this is a sentence I never expected to write, to give loss of life a greater weight in political calculations. The average COVID death leads to between 10 and 20 years of life lost (e.g. here), but those lost years seem to carry far too little weight among the politicians of the West. That is shocking.”

      Via I am and will always be Not Trampis says:
      May 5, 2021 at 12:31 pm
      “Simon Wren-Lewis with a different take” embedded link in post;

  19. Paul Nicholas
    Is there evidence that mandated lockdowns actually result in a greater, net reduction in contact- movement than more voluntary measures? Or do they only shift a few chairs around ?

  20. paul frijters says:

    On cue, the journal PNAS (I think this one is A* rated in Australia) just published that Harvard study on the effects of stay-at-home-orders (aka “sit-in-place” orders). Comparing places in the US with or without those orders they actually found more infections and deaths reported in the states and counties with SIPs.
    For the observant, there are several interesting aspects about this paper. The main one is that the techniques they use are the standard ones by economists the last 20 years, so no method cheating as has been rife in medical journals. Another is just how long it took to publish this, since it was written in September 2020, compared to mere weeks for the many pro-lockdown papers. A final one is how the authors fall over themselves in their piece to talk down their own results, leaving as many chicken-gates for the pro-lockdowners as possible. One can imagine the to-and-fro with referees! So the paper’s own history and write-up are indicative of the pressures in academia in this period.
    Nevertheless, its a study with a sensible design, finding that a core element of a lockdown (confinement in the home) is, if anything, counter-productive in terms of covid-outcomes in the States in the first wave. Martin’s study above essentially uses the same methods but with much more data from around the world, finding the same thing.

    • Chris Lloyd says:

      PNAS is legitimate enough for me. Give me a week to review it. If the paper contradicts 500 years of accepted wisdom as you claim, then a 6 month delay by referees is pretty admirable. The short publication time of other more orthodox papers is something we could argue about.

    • Chris Lloyd says:

      I looked at the PNAS paper. I guess the first thing I might say is that this is not a full time series causal analysis with all the unit root tests and co-integration error correction modelling stuff. But I only mention it because I think PNAS may have relaxed their standards because of the topical nature of the data.

      I do not discount their results. What they seem to demonstrate is that “lockdown” did not work in the USA. The key missing variable to my mind is on social distancing. Mobility just doesn’t cut it at all as a measure of human interaction. If you close retail then mobility will obviously go down. But people will probably have parties at their homes – as happened here in Melbourne illegally in the south east.

      So, the results are consistent with the following story: govt locks down economy, google records less travel, people do not wear masks and party hard because there is nothing else to do, and give each other Covid. So, yes this is a causal effect of the govt locking down the economy without ensuring it feeds through to social distancing – as we obviously did here in Melbourne.

      I am surprised that the authors did not present estimates for all the states separately, rather than assuming a common effect across all states. There might be a pattern to the states where the estimate comes out to be negative on deaths and infections: blue states versus red states?

      • paul frijters says:

        Hi Chris,

        good of you to have had a look and sure, the authors could do a lot more. Plenty of follow-up papers to write (and many have been written. What is stopping you from joining in?).

        And sure, one can argue that much of what ‘did the work’ might happen without lockdowns, or that lockdowns failed to change some behaviour further. Yet both of these arguments are saying ‘lockdowns had no effect because of an X that was actually of importance’. That is the right discussion to be in. What are all the things of actual importance and what effect does a policy really have on them? Welcome to doubt.

        As soon as one is in the “lockdowns is not really what matters but X is” many questions start to appear as to what really happened in Australia or anywhere else in the world. Scientists, including Conrad, have a whole list of things believed to be in that X. Voluntary restrictions, weather, social habits of interaction, prior immunity, fauna, old-age organisation, urban density, diet, health system organisation, etc.

  21. I am and will always be Not Trampis says:

    A lockdown without quarantining overseas arrivals is no lockdown. it explains why we, NZ and China have been so successful and Europe much less so.

    If people can move between regions like North and South Dakota they cannot be labelled RCTs.

    Lastly a warning, If you link an article it looks very similar to being banned when you try to post it. Get Saint Nick to change it.

  22. I am and will always be Not Trampis says:

    I forgot to add that Nancy Cartwright is impressive. More to her than the Simpsons 😊

    • Nicholas Gruen says:

      Yes, she does the voiceover instead of putting in research grants. Way more cost effective – and entertaining.

  23. I am and will always be Not Trampis says:

    On the serious side the article was very educative as I would expect from you.

  24. I am and will always be Not Trampis says:

    I have to say I have trouble taking our friendly cultist seriously on stats.
    He upheld the danish study on masks which purported to show how ineffective they were.
    Only problem was the large problem of the data involved. Master stat bloggers Thomas Lumley, Kaiser Fung and Andrew Gelman all heavily criticised it for essentially the same reason. ( Chris Lloyd makes a comment at the first!).
    Then if we are to believe his writings he clearly does not understand what a lagging variable is. Deaths and infections or even IVF and lockdowns bear this out.
    Lastly as I stated above his understanding of what has occurred in India was sadly lacking.

    I guess related to this is why he is so far out in how many people would die from Covid. Out by two decimal points.
    He relies heavily on JI of whom KT2 has shown to have problems in his figures. Indeed put this bloke’s name in Andrew Gelman’s blog and it is interesting..
    Some people may call him discredited I could not possibly comment.
    Like all cultists he appears to be unable to say he was ever wrong

  25. I am and will always be Not Trampis says:

    A little on excess mortality.
    I have said people need to dig deeper. courtesy of Scientific American ( which is on my around the traps US deaths from the flu went from 22,000 to 600. If you understand Z scores it is statistically significant.

  26. Chris Lloyd says:

    Paul, I think some of the pushback you get might simply be use of language. You say that people are saying it is not lockdown per se but X. The point is that “lockdown” is not defined by commenters, including me, regardless of there being a measure of severity. And here is the key point about discussions at Troppo. For most of us here, lockdown means “nobody can leave their effing house for months, except for 1 hour of exercise, the chemist or the grocery store.” I missed by son’s 30th birthday. I didn’t see anyone face to face for the entire winter. When I hear someone say that that doesn’t work, I think “wing nut” and so does Nick.

    We had stage 2 lockdown initially and that did not work. So definitely lockdowns need not work, even here in our community minded multi-cultural paradise. There are lockdowns and lockdowns. Here in Victoria the meaning of lockdown means stage 4. Just like football means AFL. As an ex-Queenslander, you will know it means something else up there. Perhaps a better phrase to use would be “enforced mass social isolation.” If we all agree that quarantine works, which is enforced individual isolation, it is hard to see how a mass version can fail.

    Regarding getting involved myself, I would like to see an analysis of the Victorian data but I suspect there is not much to do. There were two stages of lockdown, stage 2 for a few months where the numbers took off again, and stage 4 for four months where there was steady decline to zero. The decline started roughly at stage 4 though one could obsess about the turning point. There is just not much variation to exploit to isolate issues that folks might argue over.

    And it does rather seem to me that analysing this data is in the realm of a RCT for parachutes as Nick loves to point out. The interest is in exactly how tough the lockdown should be to maximise personal freedom but still work. What we need is for a brave government to experiment with some A-B testing on lockdown policies!

    • Nicholas Gruen says:

      Thanks Chris,

      I think we have a teachable moment here.

      The word ‘rigour’ is thrown around a lot — I’m not saying Paul does this particularly — I’ll get to that, but ‘rigour’ is associated with numbers, RCTs that kind of thing.

      Fine. But another kind of rigour is rigour in agreeing to the frame within which one is going to discuss things and being clear about what one is disagreeing on. The way I see my debate with Paul is that I think of the second kind of rigour as prior to exercising the first kind.

      Paul wants to hurry along to the question of what works at least cost (with ‘cost’ considered broadly). I agree with that instinct. But there’s work to be done in deciding how one will test claims. When Paul started on his panic attack, I tried to pin him down on a counterfactual. What exactly was he suggesting as an alternative to lockdowns?

      If he wanted not to have lockdowns (or, remarkably enough, masks) I argued that that would lead to take-off of the disease and that the government would be forced to lock down the population whatever his ministrations might be. (Sweden being the only example of a government that has resisted that that I know of.) I think he more or less agreed with me but said that his argument was therefore appealing for a ‘deeper’ counterfactual where ‘panic’ wouldn’t take over the population.

      At that point I thought Paul was doing us all a service because, although I’m usually uninterested in such arguments, it’s a worthy cause to keep asking how things might be different and better (I was also unconvinced that I knew better than people’s ‘panic’, but anyway, all good.)

      Then we got the lockdown malarky — which you’ve re-explained well. I have challenged Paul to explain what happened with the Melbourne lockdown — and in particular precisely the phenomenon you re-iterated. That is we had a lockdown that didn’t work and then one that did — and, while it’s possible we’re all wrong and there’s another explanation, it seems pretty obvious that the explanation you and I have offered is hard to resist and others don’t come close. Paul went fairly quiet at that point on my arguments (as I recall saying that S.E. Asia might have greater immunity — without explaining how this explanation of what happened in Melbourne has to take directions as to when to walk on and off-stage), but he persevered with you about the econometrics.

      I regard these propositions as self-evident from what we know so far and from our (parachutes seem to work) commonsense:

      1) Some lockdowns don’t work.
      2) Some do.
      3) If your lockdown doesn’t work and it’s a fairly ‘lightly administered’ one. Strengthening it up might help.
      4) Numerous other actions will be helpful in their own right.
      5) It may be possible to handle the problem without a lockdown (though it seems commonsensical that the sooner those applied and the less infection has spread, the more likely they are to get on top of the problem).
      5) These measures and lockdowns can complement one another.
      6) Lockdowns are likely at the costliest end of options available.

      I wonder what Paul thinks of this list. It’s not consistent with the proposition that “lockdowns” don’t work, or even that lockdowns might not work. They do work — to reduce spread — just as parachutes work to save the lives of pilots who have been shot down — but not in all circumstances and they have to be used correctly and preferably with other measures. I wonder if I can get him to agree with the proposition that, correctly applied both parachutes and lockdowns achieve their respective objectives of substantially slowing down processes that would otherwise proceed faster.

      Once we can agree on that, and not before, we can get onto the econometric work, which would need to take into account such simple and commonsensical propositions.

      • of your list of 6 I would only amend number 2. You same “some do”. I now say “some might”.

        • Nicholas Gruen says:

          OK thanks — so can you fill us in with a plausible explanation for what happened in Melbourne’s second wave if wasn’t the tightening of the lockdown?

          • and we’re back to the personal court :-)

            You are the one that is making an N=1 causal claim. On Melbourne, I am just saying I am no longer convinced. Since you make the causal claim of certainty, why dont you prove to my satisfaction that the corona wave in Melbourne would not have died down without the lockdowns?

            Flu waves have come and died down every year for decades. I look forward to see you arguing certainty for Melbourne without seeing what similar things did elsewhere.

            • Nicholas Gruen says:

              What do you mean it’s ‘personal’? I have no idea what that means. If I say that I’m confident that you shave your head, n=1 there too. But there’s a lot of plausible evidence that you shave your head. From everything I know, it’s the only plausible explanation for your bald head. Can I be sure that God doesn’t shave your head as a special miracle, just for you, each day? I can’t.

              I am completely mystified by your coyness in facing up to my claim that lockdown seems to have worked in Melbourne. Of course it could have been goblins. It could have been immunity spread by possums who were not immune when the AFL season had been delayed, but were immune by the time the second wave arrived.

              I am absolutely not claiming certainty. I’m saying I have no other plausible hypothesis. Nor am I trying to extrapolate Melbourne’s experience to anything or any one. You’re claiming that it mightn’t be lockdown. That’s an important part of your argument. It doesn’t doom your general argument, it just dooms your more extreme argument – that lockdowns don’t work or may not work.

              I’m arguing that I can’t see any way of plausibly arguing that Melbourne’s doesn’t work. I’m arguing that you should agree with that — and change your assertion from “lockdowns don’t work, or may not work”, to “many lockdowns don’t work or may not work” or propose some plausible alternative.

              I’m saying in all candour and good faith that I find that incredibly implausible. At that point you say n=1. It’s completely beyond me why you wouldn’t say “I can’t think of any plausible explanation for Melbourne’s effective eradication of its second wave of COVID except that it’s lockdown worked”.

              • paul frijters says:

                and which alternative explanations for Melbourne’s experiences have you considered? On what evidence have you dismissed them? Which alternative explanations have you not yet considered?

                My own claim was that in the world on average lockdowns have not worked to reduce covid cases and fatalities. I was not saying that there was no such effect from any policy, simply not for what is commonly viewed as ‘lockdowns’. On that basis I now look differently at the experiences in particular countries. I have become skeptical of what I initially took as true. You are the one that then immediately jumped to heavy words like ‘wing-nut’. Well, if its so obvious to you that there is no other plausible explanation, you must have indeed explored all other hypotheses. Let’s hear them.

    • Hi Chirs,

      You say

      For most of us here, lockdown means “nobody can leave their effing house for months, except for 1 hour of exercise, the chemist or the grocery store.” I missed by son’s 30th birthday. I didn’t see anyone face to face for the entire winter. When I hear someone say that that doesn’t work, I think “wing nut” and so does Nick.

      I totally agree with you that that is the heart of the matter. Just that. I invite you to reflect on the logic of that statement. I see that statement as entirely illogical and blinded, but totally accept that you and (probably) Nick see it as an obvious truth.

      Of course the word ‘lockdown’ is an abstraction of a complex changing reality of policies and circumstances. that also holds for the word ‘corona’, ‘infection’, ‘case’, ‘patient’ and everything else bandied around. People communicate in broad abstractions. One makes things (slightly) more precise in particular measurement contexts, but none of these concepts can be pinned down 100%. Each empirical paper (including the medical ones) essentially has its own measurement approach to each of these concepts, and if you look close enough its not truly consistently used even within a single paper.
      Yet, language is communication. Since I agree that what you describe is a lockdown, I dont think we really disagree on what we are trying to grasp by the term. And since we both want to speak in terms of “lockdowns worked” or “lockdowns did not work” we are in the same game of seeing causality between broad concepts. We even agree on the idea that there are ‘severities’ of lockdowns, though we disagree on whether it is clear that more severe ‘works’.

      • Chris Lloyd says:

        “We even agree on the idea that there are ‘severities’ of lockdowns, though we disagree on whether it is clear that more severe ‘works’.”

        Well, nobody has ever disagreed that there are different severities of lockdown. And also different kinds of lockdown. Curfew. Masks. Retail. Funerals. Sports. Exercise limits. I was trying to diagnose a cause of our possible failure to communicate and suggesting that Nick and I were assuming a particular strict form of lockdown when we use the word and that you were referring to … well I am not 100% sure; a spectrum of possible restrictions. Perhaps just the economic restrictions.

        “We disagree on whether it is clear that more severe ‘works’.” I really hope we don’t. You do agree don’t you that forcing every person indoors at gun-point (as they did in Wuhan) for three weeks stops the spread of Covid (assuming it does not persist in the environment which science seems to assert). So 100% enforced physical distance works, as Pasteur would have predicted two centuries ago. Is the response R0 a monotonic function of “severity of lockdown”? That might not be the case depending on how severity is measured. If severity is measured by economic cost, then probably not. If it is measured by how clearly it targets preventing humans getting close to each other, then probably yes.

        But sufficiently strict lockdown stops the spread, right? Please say yes. No post-modern navel gazing about how endpoints are defined. No angst about whether policy will translate into practice. Yes or no?

        And then we can get on to looking at the evidence for the types of lockdown that work, not to mention whether they are worth the cost.

  27. I am and will always be Not Trampis says:

    It is remarkable Nick of how many questions you ask of our resident cultist refuses to answer.
    That is an answer in itself.

    Very Catallaxy actually

  28. I am and will always be Not Trampis says:

    if he did why the need for the next question.

    Perhaps I should add somehow lockdowns don’t work but somehow IVF treatment starts after the lockdowns finish as he admits.

  29. Nicholas Gruen says:

    Hi Paul

    Relieving indent escalation and reverting to the right hand side of the page.

    Let’s leave it at your concession that some lockdowns might have worked.

    On hypotheses for how Melbourne’s experience doesn’t prove that its lockdown worked, you’ve provided one which involved our population suddenly acquiring greater immunity between the mild second wave lockdown and the heavier one that followed it. I think this is a reasonable hypothesis so long as God was the delivery vector.

    For the atheists at Troppo, I suggested in my earlier comment that the vector might have been possums — and how did they suddenly acquire immunity between the light and heavy lockdowns? I was thinking that they may have stowed away on visiting cargo ships (most planes having been grounded), but it was only a few weeks, which means it doesn’t leave enough time for them to have visited bat caves in China, got sick, got better and stowed away back to Melbourne). So we’re back to God. :(

    Perhaps teapots orbiting the sun between Earth and Mars have something to do with it? I guess we’ll have to go with that until someone proves there are no teapots out there.

    • Hi Nick,

      yes, i do still hold it possible that lockdowns worked in some places. You can call it a concession if you like but if you simply scroll to the top of the comment boxes on this post you will see in our first bout already I said “I am indeed no longer convinced that lockdowns were either necessary or even instrumental in the waves dying down. Its still possible, but no longer beyond doubt because of the failure in other places to replicate that supposed effect.”

      Your attempt at caricaturing the many possibilities that, for instance, me and Conrad have been exploring on previous posts or this one are a bit childish. I remain interested in the other hypotheses you have considered for what happened in Melbourne.

      Often one should simply start with doubt, involving the two questions “Why do I believe X caused Y in situation Z?” and “What else could have caused Y in situation Z?”. I find Chris’ answer on the first question refreshingly honest. He points to his personal sacrifice during lockdowns.

      • Chris Lloyd says:

        Paul, I find you comment about me more offensive that calling your theory that lockdowns do not work bullshit.

        You are implying that I am unable to separate my emotions from my rational mind. If I were really traumatised by the lockdown it could just as easily make me against them could it not? I was merely pointing out that the lockdown in Melbourne was a REAL lockdown that stopped people meeting face to face. Unlike the half-arsed lockdowns in other places.

        And such lockdowns “work”, if by work we mean eliminate the virus (possibly until returning travelers reinfect us). There is no new data needed to prove causality here. There is a weel established and coherent theory of how disease spreads, and no other theory on offer for how we eliminated. And if you really want more data, there is NZ whose “experiment” turned out the same so n=2.

        • Nicholas Gruen says:

          Not only that but New Zealand doesn’t have possums. QED

          (Leaving completely aside the fact that, as I said in earlier comments — ‘n’ is way above 1 or 2 when one goes city by city).

          I’m not sure why Paul couldn’t have conceded all that up front — like seriously conceded it so we could move on, not mentioned it sotto voce and then assumed that our reason for bringing it up was that it was that time of the month.

          In other words, you and I are asking ourselves, “Am I wasting my time here?” or put more boldly “Am I speaking to a wing-nut”. We’re wanting Paul to clear that up so we can proceed to discuss the issues Paul’s putting before us.

          Instead, Paul interprets our basic sense checking instincts as attempts to justify some conclusion justified by our priors. But basic sense checking is foundational — for life and for science.

          Once we’ve done that we can argue whether n=1 or n=20.

  30. Nicholas Gruen says:

    Another one of my favourite economics commentators normally given to sane and sensible advice completely flubs the lockdown test. He thinks they worked. I guess he just panicked.

  31. I am and will always be Not Trampis says:

    Nick he has noah opinion at all

  32. Just what do you all mean by ,worked? (or didn’t work for that matter)?
    Worked, until later down the track…
    Worked, if it was what you define as a lockdown…etc

    • Chris Loyd Nicholas Paul
      BTW I’m not seeking to ‘hijack’ or derail this conversation ,
      what ‘it’ is that you are arguing about is already well and truly incomprehensible.

  33. I am and will always be Not Trampis says:

    perhaps to you john but not to others.

    I thought noah spelled what worked was in his article.

  34. Nicholas Gruen says:

    I haven’t gone into this in any depth — just pursuing my ‘ground thuthing’ schtick — but it looks from the charts in this report, that just like the economic costs, the mental health costs of lockdowns look illusory if one compares Sweden with Australia — though it would be more interesting to see how other Nordics went. Sweden doesn’t stand out as doing particularly well.

    Anyway, Paul will know this stuff much better than me. It seems to me that it puts more pressure on his counterfactual (which is a kind of meta-counterfactual — assuming a world that is kind of less panicky than we know it is).

    • Nicholas
      Was recently told by a senior child psychiatrist at a major children’s hospital that admissions for suicide attempts by children have gone through the roof in the past year .
      Not that sort of thing would bother a theoretical discussion about what works.

      I’m off , best wishes, but the tenor of this whole discussion makes me want to throw up.

  35. I am and will always be Not Trampis says:

    an anecdote only illustrates evidence it isn’t evidence in itself.
    Let us assume your anecdote is close to the truth. The question then is why have suicides’ rising so much since the Lockdown?

    My guess is twofold. household incomes no longer have government payments to help them. Also lockdowns had the effect of having having families together hence more support for children.

  36. Nicholas Gruen says:

    Here’s Simon Wren-Lewis overcome by panic — rejecting the economic/health tradeoff. Cummings too.

    It’s a wonder these people who spend their whole lives trying to think straight and understand their discipline and public policy just succumb to panic.

Leave a Reply

Your email address will not be published. Required fields are marked *

Notify me of followup comments via e-mail. You can also subscribe without commenting.