'Life years' and the other side of the corona equation due to GDP drop

Not wishing to weary those sick of reading about Corona (you don’t have to!) but I link below to some very interesting modelling of the other side of the corona equation - the human cost of lockdowns, and how this balances against the human cost due directly to the virus. In my view, this aspect of the modelling has received far too little attention. The paper has gathered some discussion in the UK press recently:

http://jvalue.co.uk/papers/J-value-assessment-of-combating-Covid-19.pdf

The essential thesis is that a reduction in economic output / GDP results in a quantifiable loss in life expectancy, and thus a reduction in the total number of ‘life years’ a population enjoys. In the case of the current corona crisis, he implies that this loss in ‘life years’ due to projected GDP drop outweighs (‘dwarfs’) the number of ‘life years’ gained by the due to lockdown measures.

2 Likes

For those interested in more details from the paper, but don’t have time to read it, here is my summary of the key bits in 15 points:

#1 GDP and life expectancy are linked: “Very good correspondence between actual and predicted population-average life expectancy against GDP per head (a correlation coefficient, R, of 0.77) was found for 180 out of the 193 nations recognised by the United Nations in 2009 and excellent correspondence (R = 0.89) for 162 countries.”

#2 The estimated value of a ‘life year’ in the UK is about quarter of a million pounds - £248,209 to be precise (p6). What does this mean?

#3 He answers: “This implies that the Government and commercial organisations in the UK would be justified in spending up to just under a quarter of a million pounds on a scheme that would extend by one year the life expectancy of a citizen” (p6)

#4 However, he qualifies this in instances where the scheme may have adverse effects on a nations economic output, since a fall in GDP will reduce life expectancy thus potentially eliminating the gain you have just paid for: “protection schemes should not be put in place if their costs are large enough to cause the nation’s economic output to fall so significantly that it will cause more loss of life than if the scheme had never been implemented.” (p7)

#5 Various options he considers include:
a. Option 0 (do nothing and let the virus run its course).
b. Option 4 (a year’s lockdown and widespread vaccination achieved within 12 months).

#6 If economic output is not damaged:
a. Option 0: Costs ~16.7 million life years (=573,000 deaths). Minimal economic spend.
b. Option 4: Costs ~200k life years (i.e. saves ~16.5 million life years). This is the option most similar to what is being pursued by the UK government now. Calculations show government would be justified in spending up to a maximum of ~£4 trillion to achieve this (=16.5million life years x £250k per life year) (Table 4, p26). Again, this is assuming economic output / GDP is not reduced as this would reduce life expectancy thus costing more ‘life years’.

#7 He then factors in the economic damage in terms of how a resulting drop in GDP reduces life expectancy and so cancels out the ‘life years saved’: “However, if the cost borne by UK citizens of implementing Option 4 is so great that their total life expectancy is reduced by an even greater amount, then the option should not be pursued.” (p15)

#8 Since Option 4 saves 16.5 million life years, for a UK population of 67,000,000 this corresponds to spending no more than that which causes a maximum 3 month decrease in average population life expectancy in order to attain it (=16.5/67 years). (p15)

#9 In terms of GDP, a 3 month decrease in life expectancy is caused by a 6.4% drop in economic output per head.

#10 He concludes w.r.t Option 4: “a recession resulting in a general fall in economic output of 6.4% per person over a prolonged period would cost more life than would be restored by Option 4.” (p15)

#11 For comparison, the 2007-2009 recession led to 6% reduction in GDP and did not recover its 2007 figure until 2015. The average drop in life expectancy was seen to be at least 3 months. (p15). Thus, the economic cost of Option 4 must be limited to a recession no worse than that of 2007-2009 to be worthwhile. (p16)

#12 But that doesn’t seem likely: “The Centre for Economics and Business Research is now predicting that the coronavirus pandemic will cause global GDP to decline twice as much as during the financial crisis of 2007 – 2009. Furthermore, it raises the prospect of a 1930s-style recession” (p16)

#13 Thus Option 4 is vastly inferior to Option 0 in terms of life years: “Such an outcome, if it were to come about, would cause a loss of life years to the UK population that would dwarf the predicted toll under Option 0.“ (p16)

#14 Initial data from Germany also suggests similar conclusion: “IFO Institute of Munich has recently concluded that a partial shutdown of 2 months duration will reduce German annual GDP by between 7.2 and 11.2 percentage points, while extending the period by a further month would cause annual GDP to fall by between 10 and 20.6%.” (p16)

#15 “Such outcomes, if they were transferred to the UK, would be sufficient to cause significantly greater loss of life than would be gained by the lockdown measure, especially if the partial shutdown of the economy were to last more than 2 months in total.” (p16)

3 Likes

Does the economic modeling take into account that a recession may happen anyway for the “do nothing” option? I’ve read reports that business at restaurants had already really slowed down for days before the government officially closed them. There may be substantial changes in people’s behavior apart from what the government requires, so the “do nothing” option shouldn’t assume the economy continues humming along as before.

2 Likes

I’m glad you posted and especially glad you summarized it. I’m also curious about Joel’s questions and I thought of some others at first that I’ll have to rethink later.

Joel’s question is good, but I’m guessing it’s mostly already included in the GDP drop expected by the loss of life?

1 Like

I don’t recall that his modelling takes this into account. Personally, I’m doubtful the magnitude of potential economic dip with no lockdown would be comparable or big enough to make up the kind of difference he envisages. His scenario has everything over with by Sept 2020 for Option 0 (let the virus run its course).

He does note the following requirement about the kind of economic dip required to effect the 6.4% GDP drop:

The theory behind the calculation assumes quasi-steady conditions and it is not expected
that a temporary fall of 6.4% followed by an immediate recovery would lead to this drop.
However a prolonged recession of this magnitude would be expected to have such an
effect.

Anyhow, I’m wishing the initial Imperial modelling that instigated the lockdowns here had at least teamed up with guys like this to bring a bit of a broader perspective to their tunnel-vision numbers. Perhaps that would have allowed a more cautious response.

Edit to add: Perhaps I shouldn’t assume the government hadn’t considered what (at least with hindsight) seems like a major consideration. But looking at the range of evidence they considered it doesn’t seem like it:

Precisely what Juergen said several weeks ago in an email.

1 Like

Yes. Interesting. But there are several faults that I can see immediately with his reasoning. First, there would absolutely be a hard recession if that many people were to die. Simply removing the production capability and skills of 3% of the population (and I think that is a low estimate of the death toll were nothing to be done and everyone come down with the disease within a few months - thereby totally overwhelming the healthcare system) would cause a drop of approximately 3% of GDP. Ok. let’s say 2% since more of these might be elderly and beyond working years.

Next, let’s actually deal with the fact that a substantial portion of those that are currently being infected are medical professionals. So, if the virus simply burned through a population, you would have to assume that a large percentage of medical professionals would be exposed, many would get it and some not insignificant portion would die. This would leave the country exposed to other diseases in the future as a result of not having a proper number of medical professionals for some period of time.

Third, let’s deal with the fact that the UK, and to an even greater extent the US has as a very large percentage of its GDP and economy the service sector. If everyone that you knew was getting sick all at the same time and thousands of people in your town were sick and hundreds dying…would YOU order that pizza for delivery?!! Would you be taking your family on vacation? Would you be doing ANYTHING outside your home unless it was life and death necessary?

If we are looking at the do nothing scenario realistically, It really would look very similar to the scenario that we are currently experiencing, wouldn’t it? Everyone would be hunkered down in their homes and trying not to have any interaction with anyone. And that…would have a drastic impact on the economy. Restaurants would be suffering and closing, hotels would be vacant, the transportation industry would be all but shut down.

Yeah, you can argue that it might be that way for a few months, but then everyone would have it and it would go back to “normal.” That may be. It’s not my purpose to argue for one approach over the other. I’m simply saying that there is a massive economic impact to the do nothing approach too that isn’t being factored into this paper. So the cost of mitigation must be considered relative to the cost of do nothing. Both cost a TON economically.

1 Like

@dougummel 3% is way too high. Even the initial Imperial modelling that produced the scary 2.2 million death toll for the US if no intervention was put in place only used 0.9% IFR (infection fatality rate, the proportion of those infected who die - not to be confused with CFR - case fatality rate, which is higher and refers to fatalities among diagnosed cases only). More recent models coming from the Imperial study group seem to be using an even lower IFR=0.66% based on work by Verity et al. And this number is obviously lower than the actual %age of population that will die since not all get infected.

Add to this the very important consideration that many who die with corona were going to die within a relatively short time period anyway, and the overall ‘excess mortality’ rate gets even lower again. One of the lead modelers from Imperial (Neil Ferguson) has stated between 1/2 to 2/3 of people who die with corona would have died within a short time period anyway (even if they had not caught it). This explains why there is such a strong overlap between those who die with corona and those who already have ‘underlying health conditions’. I read recently that if you are 80+ you already have a 10% chance of dying this year anyway even without corona. So, many of the deaths we are talking about are not going to be evidenced in ‘excess mortality’ statistics - which really is the key statistic to be thinking about here.

Furthermore, the IFR=0.66% quoted above is an ‘overall’ value, when it is split up between different age groups it is much more heavily concentrated in the over-70s. Here is how the original Imperial model saw the relative breakdown of the overall IFR=0.9%:

Capture

So an even tinier proportion of the working-age population (which includes the medical professionals you mention) will die. Perhaps this is why the author doesn’t think GDP would be hit too hard with the no-lockdown approach.

Edit: I missed your point here.

I think that is only realistic if the government messaging does this. But in a no-lockdown scenario the government could quite easily instead encourage those of working-age to keep on working hard, face the danger with bravery and not let it stop work because stopping work costs lives (the GDP link) etc.

Anyhow, some of these questions are best answered by the experts and behavioral scientists. But my sense is that the paper is onto something and it is by no means a foregone conclusion that his GDP-life expectancy considerations can be left out of the equation without significantly affecting the result.

I was half-wondering if he was on here. Be interesting to hear his take on the economics of it all. According to a BBC article yesterday the full-year GDP drop predictions by various financial analysts are looking pretty grim, most of them exceed the 6.4% threshold the paper mentions, some significantly so:

I like this discussion and I appreciate the paper Henry made available. All economic modeling is based on lots of assumptions, and so is this. the general point, of course, is an important one. when the governments here in Europe started the lock down, they were unwiling to consider the cost of locking down the economy in terms of human lives. That is a dangerous mistake. Even now, European generally consider it immoral to raise economic concerns, because you cannot talk about “money” when “life” is at stake. (Sorry, Henry, I mean continental Europeans).

One common flaw in the economic argument is to correlate GDP (per capita) with life expectancy across countries and infer from there that, if GDP drops, so will life expectancy. WRONG! Cross country correlations tell us nothing about variations over time in one country. Still, recessions cost lives and very severy recessiions cost many lives. why? Because there will be an increase in domestic violence and in drug abuse, both leading to deaths. People may lose not only jobs but also homes, social connections, etc. and end up in very bad situations leading to more fatalities. Health care needs to be financed, and in a severe recession, financing drops, medical personnel lose jobs and the quality of health care declines sharply. Finally, if the lock down causes a depression as bad as the Great Depression, which is what the IMF now says, it will cause political uproar and violence, leading to more deaths. Remember, we had street wars between the extreme left and the extreme right in (continental) Europe in those days. All of this is extremely hard to quantify and I’m not going to try to do that. Just pointing in the direction in which you have to look for the economic cost.

Now, there is another source of uncertainty here. since neither Europe nor the US has had extensive testing for Corona infections, we do not actually know how many people have been infected. Johns Hopkins and others count cases of observed infections, i.e. people that have been tested positive. But that ignores the many people who are infected and not tested. My University (Bonn) just came out with a study of the city that had the first Corona outbreak in February. They did a representative sampe test and found that 15% of the general population have had the virus. Taking that into account brings down the average fatality rate to 0.37 percent. Keep in mind the uneven distribution across age groups and, not shown above, across groups with different preexisting conditions.

My personal take on it all is that Europe and the US will come to regret the panic approach of our governments. But it’s hard to predict when that will happen.

8 Likes

I’m currently doubtful that the “panic” approach was the wrong action. I understand that severe recessions cost lives, but have any of these economic/life models included the impact of overwhelmed hospitals? If not, that makes the argument much less convincing.

So far we have seen multiple instances of overwhelmed hospitals. In such a situation, not only do people not receive treatment that would otherwise help many of them live and recover, the lack of protective gear in the presence of large viral load causes a high infection and likely higher than otherwise fatality rate among medical personnel. The loss of modern medical care must also have substantial economic and life costs – has that been counted in the models? Perhaps these costs could be mitigated by denying treatment to those people likely to die soon anyway from some other cause, but is this a route we want to go down?

So far it seems that the only measures that prevent overwhelmed hospitals are shutting things down or engaging in aggressive testing, contact-tracing, and isolation. Telling healthy people to go to work and vulnerable people to stay home seems by all indications a recipe for rapid and widespread infection considering the contagiousness of COVID-19 and that it apparently can be spread by people without symptoms.

Even if the economics and overall life count work out such that it is better to have people working even when hospitals are overwhelmed, I can’t imagine that would be politically feasible. What government is going to tell the voters that they should keep working so the taxes continue to roll in but if anyone old or vulnerable comes down with COVID-19, that person will be left to die because there isn’t hospital space? Or what medical personnel will continue to risk their lives without adequate protective equipment because the government thinks it better for people to work even if more people get sick than the hospitals can handle? If someone wants to work from home because he or a family member has vulnerable health but his employer doesn’t want that due to loss of productivity, will the government say it’s okay to fire him? If a non-essential business stays open and it turns out a bunch of people catch COVID-19 there and some die, will that business be vulnerable to a negligence lawsuit because they did not take adequate precautions when an unusually contagious and fatal disease was going around? The level of uncertainty seems to be such that it would not be possible to carry on business as usual even if we tried to do so.

1 Like

We may have real world answers, at least be some point. Although some news articles have ripped them for it, Sweden has, at least so far, not shut down their country, but instead emphasized personal responsibility in social distancing, while continuing to keep the nation’s workers at their jobs. They have tried to quarantine elderly people as much as possible. Despite some saying it’s going poorly there (including President Trump), the Swedish government says that so far it’s worked well for them. Whether that’s actually the case, hard to say. They also don’t have the number of big cities the U.S. does.

I continue to think that, even years from now, no one will really be able to say with certainty whether the various lockdowns, and the resultant effects on society, will have been worth it. No one will readily admit if they think they were terribly mistaken about the cornavirus’s seriousness, and no one will want to sound callous about potential loss of life.

Edited to add:
Here’s a fairly balanced article from CNN on what steps Sweden has and hasn’t taken to combat the coronavirus.
https://www.cnn.com/2020/04/10/europe/sweden-lockdown-turmp-intl/index.html

1 Like

It’s going to be like the New Deal. We’ll be arguing in 100 years about whether it saved the country from depression or made it worse.

4 Likes

It is probably worth pointing out that one factor contributing to the softer policy pursued by Sweden is that Swedes are more socially cooperative than Americans, who have a more fragmented culture and a “don’t tell me what to do” frontier mentality.

A larger factor is that the U.S. lacked testing capability despite two months’ warning. Without knowledge of how many people were infected, especially those with mild or no symptoms, a hard shutdown was the only way to be sure of staying in control of the situation. So when it comes to allocating blame for the severe recession, first in line should be the CDC and FDA for their failure to roll out the testing capability that we needed, and on top of that, for preventing states and private labs from doing their own testing.

Good point, Joel. As a repenting libertarian, I’m skeptical that the CDC or FDA should have the amount of power they have to “protect” us the way they do. But since, right or wrong, they are in the position they are in, the buck should stop with them. Unfortunately, the one thing bureaucracies excel at is avoiding responsibility.

1 Like

It looks like the match has been lit and put to dry tinder.

4 Likes