UK policy wonks following Troppo in saying the lock downs were a mistake (but hiding the message a bit)

Here at Clubtroppo, we have been saying for well over a month now that a quick look at the economic damage and the health damage of the responses to the corona virus tells you they dwarf the possible benefits of suppressing the virus, anywhere in the West. This has lead to the prediction that the narrative will flip very soon such that the lock downs will be openly recognised as extremely unhealthy for the population, far worse than the virus. That shift is indeed occurring right now in Australia.

A team of befriended policy wonks in the UK – Richard Layard, Gus O Donnell, Nancy Hey, and others – have now openly adopted the same methodology we have used here at Troppo and that I have co-developed the last three years in the UK. Their paper is here. If you read it properly you will find it says exactly the same as I have done, but tries to soften the message a bit to make it palatable to the politicians. So they dont quite say that the UK politicians have been total buffoons for having instigated the lock downs and that the damage of the lock downs is far greater than their potential benefits, but they sort of say exactly that. As Yes Minister could have said, a very courageous stance. Well done!

Let’s take the paper apart and see how they apply the WELLBY approach and the tricks they have had to resort to, to soften the message. Walking through their paper and their assumptions is a great way of learning the logic of wellbeing cost-benefit analyses and the real tradeoffs the UK and our world have faced.

The basic idea of using the WELLBY.

Firstly, they buy into the notion that we should pick our policies on the basis of how many WELLBYs are generated at what costs, whereby a WELLBY is one point in life satisfaction on a 0-10 scale for one person for one year. I was the first to coin the WELLBY and it was first seen in a peer-reviewed paper recently (Frijters et al. 2020) though I have been teaching it for two years now at LSE. The first mention on national television was by our very own Gigi Foster on ABC Q&A this week! Much of the methodology followed is in a Handbook I co-wrote with one of these authors (who did the numbers on this paper) for the UK government, which these authors have read and in essence follow. [1]

So the name of their exercise is to try and track for every month in the near future, starting in May 2020, whether or not to lift the lock downs in the UK based on the stream of WELLBY of the two scenarios (continued lock down versus no lock down).

Of course the crucial thing here is the assumptions on what would happen in these two scenarios: continued lock down or no lock down. As is usual in this kind of exercise, they do not fully flesh out either scenario because that is somewhat impossible, but they implicitly assume many things by how they say things would progress in them. Indeed, they give a lot of “flesh on the bone” for their scenarios.

Lock Down versus no Lock Down, what do they mean?

In their “continued lock down scenario”, they envisage sustaining another 100 thousand new unemployed every month. In the “no lock down” scenario they say people will get out of unemployment at the same rate they went in (several hundreds of thousands per month). You might say that is rather generous towards “no lock down” since we know the bounce back in recessions is nowhere near as fast as how quickly the problems mounted, but this in actuality lets the government off the hook because it pretends the economic disaster can be fully remedied in only a few months. They assume something similar for the income losses. If you think it through, this reduces the costs of the implied recession by a factor of about 10 versus what the IMF and other economic forecasters are now saying (which I first anticipated and then simply followed in my later posts). If only!

Similarly, they assume each month of lock downs saved 35,000 lives as opposed to lifting the lock downs. They motivate this by saying that preventing a full spread of the virus saves about 0.2% of the population that would otherwise die of it (which is the number I have been using too, but they quote UK modelling). So they say 4 months of lock downs should count as saving 150,000 lives. We will dissect this later in greater depth, but for now just note how they thus populate their two scenarios with actual numbers on the crucial issues.

They add a whole bunch of other assumptions on various additional issues. They include the issues I have talked about, such as mental health and worsening health of the elderly if you keep them locked up. They also add stuff I did not talk about, like confidence in the government, traffic deaths, and air pollution, which were extensively flagged in the underlying Handbook and are nice to see in such calculations, though, to be honest, they are small fry relative to the main tradeoff: the income and unemployment effects of the lock downs versus the lives saved from the lock downs.

Their bottom line versus mine

Now, when I did these calculations it became immediately clear to me that the UK should never have entered lock downs and the damage was easily 10 times worse than the most optimistic scenario one could pick on the lives saved from lock downs. This was true whether one looked at lives-versus-lives or allowed for quality of life to be valuable as well. These authors have almost the same underlying assumption on the actual death rates of this virus, and similar assumptions to me on levels of income loss and wellbeing effects of unemployment. You would think they should thus get the same brutal conclusion, particularly as at least three of these authors have read our clubtroppo posts.

Yet, they don’t get the answer that lock downs should end immediately, hence in May, but they say their calculations show they should continue in May and then be lifted in June. How convenient! That gives the UK politicians the impression these authors agree with the previous lock down decisions and that they can go on with the lock downs in May, which they had just announced anyway.

Where did the rabbits go into the hat though? How did they pick the assumptions so as to avoid the embarrassment of saying the UK government had blundered, and nevertheless say exactly that if you read carefully, allowing the headline to be the advice that the lock downs should end in June, which is at the earliest politically possible date?

They employ three tricks where they differ from what I assumed. Let me walk you through their tricks by walking you through the essential truths they had to circumvent to avoid embarrassment.

Valuing the reduction in GDP: public services or personal income?

Firstly, they faced the difficulty that the income shock is so enormous that reduced public services via reduced taxation in the future is easily going to kill ten times more people than the 150,000 corona deaths, even in the UK.  Any usual number, even using a high-end number of the cost per QALY in the NHS (which would be 30k per QALY), would get them that. How did they get round this “inconvenient truth”? [2]

What they did was not to count reduced public services at all and simply take GDP to mean person income, as if the income losses have no consequences for tax revenue. So no deaths of despair or austerity deaths. Even then, that would not be enough to avoid the conclusion though that the recession will kill far more than it saves because individuals too buy health and have a value for their life that is no higher than their own income. So the next step was not to look at the true benefits of personal consumption either, but to pretended that one can value the wellbeing effects of the income loss as if these losses had the same marginal effects as small regular changes in personal income have on life satisfaction (keeping health constant!).

So they basically chose one of the lowest wellbeing effects of income one can find in the literature, which is the effect of hardly noticed changes in income on life satisfaction –  easily a factor 10 times less than normal for how people have been found to react to noticed losses of income – and used that to get a wellbeing value for the economic collapse, now seen solely as a loss of personal consumption. Their implied value of a QALY thus becomes near 800k, some 50 times higher than the marginal production costs and 27 times higher than what the NHS high-end number is.

Their assumptions on this thus differ totally from what they have themselves advocated in the past, what anyone else assumes, and of course willfully ignores public spending effects of the GDP loss which would inevitably have lead them to the head line that these lock downs were a bad idea. This is what allows them to scale down the wellbeing benefits of easing lock downs so as to be in the same ball park as the anticipated deaths from easing lock downs. So they artificially reduced the importance of the recession by about a factor of 20.

Quite a rabbit, hej? The things you have to do to keep the politicians and medical establishment on board! Anyhow, moving on.

Corona deaths from lock downs versus no lock downs.

The second problem they faced is this business of just how many lives are lost by easing lock downs versus no lock downs. What is their story for what “no lock down” would look like? They in the text advocate a kind of Sweden scenario for easing lock downs. Very sensible and politically palatable. Yet, that assumption comes with obligations as to the death rate you should then assume would happen with “no lock down”. You’d think they’d then look at Sweden and pick a reasonable number for how much more death a reduced lock down in the Swedish variety would cost versus the UK variety.

The problem they would have faced in that kind of calculation is that Sweden has a lower corona mortality rate than the UK (213 per million versus 287 per million).

That of course wont do at all. So whilst they say the UK should adopt the policies of Sweden, they are desperate to avoid assuming that would come with the death toll of Sweden, because then again the UK government looks ridiculous in hindsight. They had to come up with some big number as to lives saved by UK lock downs versus Sweden policies.

So they made up the numbers in the fairly weird notion I described, ie they presume that without lock downs you’d have all the maximum 150,000 corona deaths in just 4 months time, worth 35,000 per month if you ease the lock downs, a rate they say you’d get to immediately per month if you’d ease the lock downs.

Weird, right? And there is then the open question of just what “continued lock downs” then mean. You cant be saving 35,000 lives every month forever if you say that the virus could never kill more than 150,000 anyway. So one needs to think about this. What are you really then assuming about the whole business of flattening the curve, herd immunity, and all these other ways of thinking about what to do in the next 2 years to avoid deaths by opening up at some point?

Mainly, by assuming you save 35,000 deaths per next 4 months of continuing the lock downs, you  are assuming that you wouldn’t get the 150,000 deaths at any point after that. This means you are saying there is some future scenario that involves a lock down now that can prevent you those 150,000 deaths. That needs a “deus ex machina” which no-one in the world yet has, a magical solution to the question of how to have no lock downs and yet also no flaring up of the virus.

They indeed put such a rabbit into the hat by saying that if there were continued lock downs, by September, the UK could do a South Korea, which they interpret as saying one can avoid excess deaths by track-and-trace from then on, as well as avoid further economic damage.

Well, how convenient (and somewhat dangerous. The politicians might believe them that waiting till September makes some sense!). You’d think they should then also calculate the ongoing economic and mental health costs that South Korea is experiencing, and of course factor in that this would have to be kept up until a vaccine (which might well be never). Yet, of course, doing any of that would get them back to the immediate conclusion that the lock downs were a monstrous mistake at the very outset, not saving much at all that wont happen later whilst doing huge damage now. So this is another thing they needed to shove below the carpet.

Its a mightily funny rabbit hence that they put into the hat on the death rates, seemingly specifically designed to make the lock downs look good for another month. Also, you are then assuming a rather immediate and rapid loss of life from easing lock downs that could not also be ameliorated by smarter technologies.

Unemployment effects on mental health in the recession.

Finally, the effects of unemployment on mental health. As you might recall, in my own calculations, I minimally said unemployment costs 0.7 WELLBY per year of unemployment, which is so large so quickly thatmillions of unemployed dwarfs the effect of these corona virus deaths. They use the same 0.7 figure as me (which is a long-established figure). So how don’t they get that the lock downs were a monstrous mistake on this point alone?

Here they simply refuse to calculate both the cumulative and ongoing damage of the lock downs, only looking at the margin of that another month of lock downs would do on top of the lock downs we had. Then , they plug in a number on additional unemployment that is easily 5 times lower than the increases we have so far seen, which of course is a bit of a cheat. But even then, within my calculations I would have had to conclude that the increase in unemployment far outweighed the extra deaths because the extra unemployed would have extended the time in recession and increased the time everyone expects to be in unemployment. That is why these deep recessions are so detrimental and lead to “deaths of despair”. How did they get round that?

This is where their V-shaped assumption on the recession actually helps them. By assuming a bounce-back at the same speed as the in-flow of unemployment, they essentially say the recession is over within 4 months of lifting the lock downs. No self-respecting economist thinks that is reasonable (at least not anymore. They were such to be found in the first month of the reactions, but reality has caught up with those).

Mainly, this reduces the effect of those 100,000 additional unemployed rather spectacularly, though there are two conflicting effects. By saying the recovery is super-fact, they of course then count a large quick reduction in the mental health of the unemployed in the benefits of removing the lock-downs (and easing the anxiety of the politicians by pretending things can get back to normal quickly). Yet that is not truly the proper calculation because one needs to think of the effects over the next 5 years of another month in lock down. If you’d have a normal recession employment is not back till within many years, meaning the additional 100,000 makes things worse for a whole 5 years for the whole stock of unemployed (which is by now millions in the UK). The discounted flow of that would look very different. A calculation in the same methodology as my previous one on the shape of recessions would certainly have meant the lock down should never have happened and should cease at the earliest moment (because because the loss of the whole stock of unemployed then becomes paramount).

In conclusion

I applaud these authors for their work in using the WELLBY to argue for a speedy lifting of the lock downs in the UK. I totally agree it should be as quick as possible and hope policy makers and academics in Australia and elsewhere take note of the tricks these authors had to employ to avoid the conclusion that the lock downs were a monstrous mistake to begin with. Politically unavoidable, perhaps, but a monstrous mistake nevertheless and hence to be avoided in the future.

I understand why these authors couldn’t follow accepted practices on cost-benefit or reasonable assumptions on recessions because of the political sensitivity of telling the UK politicians and their health advisers that they have inflicted huge net health damage to the UK population. They have shown themselves prepared to bend scientific plausibility so as to coax the UK politicians into a more sensible stance, as well as more promising way to look at policy making in the future.

Courageous, but there is the question of conscience. They have made the choice depicted in the Yes Minister skit linked to. Among others.

 

Notes

[1] It is somewhat strange that in the first version of their paper, neither references the originating WELLBY paper nor the handbook fleshing out the methodology, which has been noticed by others. However, given how much we’ve criticized the UK government on this blog, maybe its for the best they didnt mention my name.

[2] On top of taking the government expenditure route, I also took the “historical relation between GDP and health” route, which got me very similar numbers. So they are ignoring that as well. So no “deaths of despair”.

This entry was posted in Coronavirus crisis, Death and taxes, Health, History, Methodology, Social Policy, Society. Bookmark the permalink.

12 Responses to UK policy wonks following Troppo in saying the lock downs were a mistake (but hiding the message a bit)

  1. Robert Banks says:

    Paul – is there something going on that is loosely like hyperbolic discounting: namely, that deaths, or risk of death, close to us in time and space, are “weighed” much more heavily than those some time in the future, or affecting people socially or spatially distant from us?

    Maybe it doesn’t have to be hyperbolic – just a high discount rate.

    If one had sufficiently precise estimates of current and future risks, one could calculate the discount factor being applied.

    I think this would mean that one could think of PV(Wellbys).

    Rob

    • paul frijters says:

      one can try and rationalise it that way, sure, but its not really like that if you consider how quickly people’s lives were sacrificed, such as the IVF children and their parents, and the hundreds of millions of unemployed thrown under the bus.
      There is no making sense of the choices made if one presumes people were aware of the various consequences and simply had a weird weighing in time. They were just truly not aware or interested in many of the consequences, even though these included huge loss of life.

      So its more like a wave of panic that blinded people to other threats and other issues. The logic of the sacrifice. We are really quite strange animals.

      • Kien Choong says:

        “Availability bias” seems a simpler explanation than “hyperbolic discounting”. I wouldn’t say that decision-makers are uninterested in the consequences of their decisions. Rather, our decisions are biased towards the proximate vs the distant.

        I think Kahneman thinks that organisations/institutions are better at making rational decisions, and so some kind of aggregate wellbeing analysis should help with decision-making.

        But aggregation necessarily entail value judgments, and so need public discussion on aggregation (and other matters).

    • Paul
      While multiple view points , polyphonic narratives and the like have by now , been around for quite a long time in Art, movies , novels etc . Most people still love stories etc that are told from a simple single perspective .
      Simple ,easy to understand falsehoods can be reassuring in a way that , ‘ it’s complex, ironic and depends on how you look at it’ doesn’t reassure.

  2. Kien Choong says:

    Wellby (nice sound!) seems to be a good metric.

    Is it possible to group people by their Wellbys (e.g., high, medium, low) and look at the effect of lifting a lock down early on each group? Personally, I would give more weight to the low Wellby group, as I think the high Wellby group would be more resilient.

    I understand Wellby combines objective and subjective measures of wellbeing, correct? Is it possible to separate out the two, and report the results for an “objective Wellby” vs a “subjective Wellby”?

    I think presenting results by group (high, medium, low) and by objective vs subjective would help stimulate public discussion and democratic decision-making on the way forward on the timing of relaxing (or completely lifting) a lock down.

    • Kien Choong says:

      PS,could we also report Wellby by age groups? Some people in the community might be interested in how it affects the elderly vs the young vs the working adults.

      • paul frijters says:

        absolutely, and distributional analyses have been done with the WELLBY (though not in the linked papers but in that Handbook).
        Still, as said above in another response, there is no weighting in time or over people that makes the choices made seem rational.

  3. Paul it’s funny , Covid19 is overwhelmingly an threat to old males( and I am one) just when I thought I was at best irrelevant ,or annoyingly deaf , they’ve turned the whole world upside down just as to protect me.

    It will be interesting to see the figures for nations that have life expectancies that are roughly where Australia was in the early sixties : around 70 years , particularly for males, nations that have clean water, ok but not super modern health care, reasonable diets and low obesity etc. My guess is if a nation doesn’t have a lot of males over 75 then the impacts of Covid19 will be fairly slight regardless of ,policy .

    • PS feel that the situation for the UK could actually be easier than it is for Australia and NZ. For the UK facing up to the fact that short of a vaccine or other treatment the virus will always be with us ,will take time , but the need for acceptance is unavoidable.
      However because Australia is an island and we have done such a job of suppressing the virus, for now, we are faced with the possibility of being stuck in in the denial phase of grief for possibly years and years to come.

    • derrida derider says:

      Morbidity tends to occur in the last decade of life, whether that is in peoples’ 60s to 70s as in the countries you posit, or in their 70s to 80s as in modern Australia. And it is age related morbities, not chronological age, that puts you in danger from covid.

      In many middle income countries smoking is still fairly universal among older people. Not helpful if you are exposed to an infectious lung disease.

  4. Paul from the AFR
    “The government’s epidemiology modelling shows the virus’ effective reproduction rate (Reff) was already below the crucial threshold target of 1 before the level three restrictions were put in place in late March, and before some states adopted even stricter social distancing rules.”

    I thought that might be the case…

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.