2020-03-24

Coronavirus days: policy?

90343127_10157291987478869_2697016565970239488_o Before you read on, consider Flatten the Curve of Armchair Epidemiology by  Noah Haber: Vet your sources or more people will be deluded. And also Noah Kaufman insightfully TwitsNew essay: "COVID Changes Everything Except 1 Thing" which impatiently explains why everyone must immediately drop their opposition to my long-held policy prescriptions. An example of wishful thinking is Coronavirus Will Change the World Permanently. Here’s How from Politico. Leaving aside the question of why so many Noahs, we come to... (trigger warning) Trump Twatting "we cannot let the cure be worse than the problem itself" (except he said it in shouty all-caps, because he's a shouty sort of person). Naturally he got flamed for saying that, and I'm all in favour of flaming Trump, because he is a wazzock, and only a nation of idiots would put him in charge1. Anyway, I'm disappointed by the lack of cost-benefit analysis; at least, I haven't seen one.

I shall attempt one. Recall that this sequence of posts is to some extent in the nature of a diary of the plague season, so I'm recording my thinking, and what I see around me. Let's suppose a life is worth £1 million (on the low end I know, but these are mostly elderly folk dying... let's hope my mother doesn't read this...) and we (in the UK) might end up with 500 k deaths if we do nothing, then that's a cost of £500 billion. Which sounds like a large number, but you have to be careful with things that look like large numbers. UK GDP is about 2 trillion, so if we lose 20% of GDP this year that would be... £400 billion. Did I get my numbers right? That suggests that at least in ballpark, costs and benefits about balance; in which case it doesn't seem unreasonable to err on the side of humanity. That's assuming no knock-on cost to GDP in future years, and on the other side, that 500 k mostly elderly deaths wouldn't affect much GDP.

Alternatives to lock down?


The current lock down looks like a very blunt instrument. You can argue, and probably will, that even that blunt instrument is misunderstood, and that the govt-to-public interface is incapable of transmitting anything but a blunt message. But that won't stop me thinking that it is all rather badly done and could be done better. Examples abound: I can't work-from-work, even though I'm about the only person on my floor, and the risk is negligible; certainly less that the risk from going into a supermarket. Speaking of which, at the local Sainsbury's, precautions seemed minimal: no hand-san on the way in, no staff wearing masks. It feels like everywhere is suffering policies that are more tuned to large cities.

Update: would wearing masks help?

There's almost no mask usage in the UK. And pols never wear them when gathered together. This Tweet points to the Czech's making their own; see this Google docAs more and more people took to the streets and social media with masks, on March 17 at the daily government press briefing, all members were wearing masks. It seems to me this is at least worth considering. The point being that you don't need medical-grade-certified masks to make it less likely to transmit.

I'm not the only one to think of the obvious... Scarves now, masks later; and Bring on the Masks and Gloves! And Lithuania starts producing reusable N100 face masks. It took just 4 days from idea to government contract of 1 million units. Costs 1.5 EUR per unit, production capacity 10,000 / day.

FACE MASKS: MUCH MORE THAN YOU WANTED TO KNOW: slatestarcodex.

Uupdate: I wrote a post about it.


News


UK now in "lock down" by which we mean don't leave your house unless necessary; but with a fairly broad definition of necessary, including a free pass for going out for a run. In a token of madness, even single sculling is now banned, presumably in order to prove that the world is insane. On a personal level, work has now bowed to the inevitable and sent me home. I'm anticipating that the garden will be at its best this spring, especially if the good weather continues.

Updates


I had some blogs by others queued up to comment on:

* Trump Murders Someone: The weapon: stupidity: DA/QS. Meh, more like suicide by stupidity. Anyone dumb enough to follow Trump's advice deserves a Darwin award.
* New York vs Italy - EarlyWarning: New York is in serious and, I would guess, irrecoverable containment failure; or you may prefer his We Are About to Lose New York City to Covid.
* Let Decadent Airlines Go Bankrupt by Pierre Lemieux at EconLib.
* Some thoughts about science advice - ATTP
* Uncertainty in the COVID-19 model - JEB: I think one thing that can probably be concluded is that the parametric uncertainty isn't hugely important for the broad policy advice. Rather like climate science, in fact :-).
That Oxford study, in full, in brief - JEB
Harsh But True - Timmy suggests Africa may show what happens without lockdown; though a commentator points out that they skew young, which helps.
* Open Borders: Now Do You See What We're Missing? by Bryan Caplan
* A battle fought on a million fronts by Scott Sumner
* J-value assessment of how best to combat Covid-19 by Philip Thomas
What About Economists’ Expertise?
* When Free-Market Prices Are Banned by Pierre Lemieux
* How long the lockdown lasts is not just a medical question – it's a democratic one; Richard Denniss - Graun
* Economic crash could cost more lives than coronavirus, says expert - The Times - Philip Thomas, Jvalue stuff.
* The UK lockdown and the economic value of human life by some bloke called Julian Jessop; a rather informal CBA.
* [2021/04] The apparently-reasonable idea that lockdown would increase suicides appears to be false: Surprisingly, suicide has become rarer during the pandemic - Economist

Notes


1. and only a party of utter morons would have put up a candidate he could beat. It looks like they have actually learnt something this time round; I'm astonished.

30 comments:

Marco said...

I think your calculation misses a rather important aspect that many others also seem to ignore: there is a substantial percentage of people that need hospital treatment when they suffer from CoViD-19 - somewhere between 5-10 times the number of deaths. Those treatments are in many cases not just overnight stays, but literally weeks. The strain on the system is not the number of people who die, how horrible it may be, but the large number of hospitalisations overwhelming the healthcare system. This has many follow-on consequences, including people not getting treatment for their ailments and thus possibly not being able to work either. Finally, there's the group who gets ill but can stay at home - but this easily also takes a week or more.

Those loss of work hours are not free, either.

Peter said...

You're ignoring the fact that that an uncontrolled pandemic will ALSO affect GDP, because loads of people will be off sick, or attending funerals. You need to look at the economic effect of the shutdown over and above the economic effect of having huge numbers off people sick simultaneously leading to the total collapse of the health service.

Smeagol said...

on a positive note the 100 years war managed to continue largely uninterrupted despite the black death. Perhaps not the model we should be aiming for tho

William M. Connolley said...

RMG roughly agrees with my calc if not with my ideas: https://www.facebook.com/592467715/posts/10158259451202716/

William M. Connolley said...

Loss of hours and those req hosp is I think a second order effect so probably ignorable to first order if the total hosp isnt too high.

William M. Connolley said...

I forgot to add that confinement will lose some mental health and likely cost some suicides. And also loss of gdp loses quallies in general.

Nathan said...

Given there are a lot of experiments in varying degrees of 'lockdown' right now there'll be plenty of data to play with soon.

It does appear that going hard early reduces the infection rate, be interesting to see which countries experience a smaller impact to GDP...
Will there be any difference between nations?

The problem with you CBA is that it will probably find treatment for pretty much anything being of no benefit. Especially terminal illness. Perhaps you could test your CBA that way, and experiment with the numbers

William M. Connolley said...

> a lot of experiments

Agreed.

> Especially terminal illness

Perhaps the awful truth is that this is true, and we should learn from that. It is said that costs of healthcare are end-heavy. When you're not paying, that's inevitable. This is one of those cases where a global restriction might make sense: would you accept your healthcare costs halving, if the price for that was nothing but palliative care in the terminal phase?

Phil said...

Let me pull out my envelope.

In advanced countries we live about 100 years. That means about 1% of people die per year.

Covid-19 will kill about 2% with good health care, and perhaps 6% with no health care. An complete epidemic will overload heath care, resulting in basically no health care for anyone but the very lucky and the very privileged.

So that is about a two year reduction in life expectancy in advanced countries, for the privileged. And a six year reduction for everyone else.

Would a month or two long economic shutdown be better or worse than loss of two years of life? How about six years of life?

Sure, lots of quibbles.

The Hammer and the Dance.

Sam said...

cost benefit analysis is almost certainly not the right tool to be used here. Even without the usual garbage-in garbage-out critique (which I feel is pretty powerful here), there's the fact that CBA itself is not a bald decision making procedure, and the fact that it can only really be justified in analysing market failures, which this obviously isn't.

THE CLIMATE WARS said...

Applying some pandemic response money to improving third world internet speed might be a good idea, for as time is of the essence in acting to identify and isolate coronavirus hot spots, fast data communication can help contain the pandemic, witness what has so far happened in South Korea:

https://vvattsupwiththat.blogspot.com/2020/03/the-rate-of-spread-of-covid-19-is.html

[ William , can you please fix the above link-- Safari has gone infarct again ]

Smeagol said...

Your numbers seem a little high for the good healthcare scenario Phil looks from south Korea to be half that. Im unconvinced tho whether the likes of President Dunning Kruger can restart large segments of the economy until people regain confidence there not at excessive risk. Much of the shutdown (at least in the US) was the the result of local decisions the federal gov having mostly been lagging behind.

William M. Connolley said...

> Would a month or two long economic shutdown be better or worse than loss of two years of life?

You've just compared two numbers that cannot be compared in the form that you've given them, so I don't see how that helps. You have to turn them into something commensurable. I tried to turn both into money. If you don't like that, you could try turning them into time... assume that life is barely worth living during the shutdown; then you've removed say 2/12 ths of a year from everyone's life, in exchange for 24/12 for 2% of the population; and 2/12*0.98 is greater than 24/12*0.02.

> can you please fix the above link

Not possible in Blogger sadly - unlike Wordpress, comments aren't editable. But the link works cut-n-pasted. Have you considered using a real browser like Chrome ;-?

> it can only really be justified in analysing market failures

That makes no sense to me. Why do you think it is true? For example, I have £1, and the shop is selling a bunch of bananas for £1. If the benefit to me of bananas is greater than £1, I'll buy. Where's the market failure?

THE CLIMATE WARS said...

Here, through the modern miracle of Firefox, is the link :

https://vvattsupwiththat.blogspot.com/2020/03/the-rate-of-spread-of-covid-19-is.html

Phil said...

Back a few years, a fellow was leading his pig down a path, and another fellow said, "would you trade that pig for 3 sacks of flour?"
The fellow with the pig said "You've just compared two numbers that cannot be compared in the form that you've given them."
The fellow with the flour said "Say what? I've got flour, you have a pig. I'd like a pig. Do you want flour or not?"
The fellow with the pig said "You have to turn them into something commensurable. I want to turn both into money."
The fellow with the flour said "what's money. Do you want flour or not?"
...


Right. So let's compare time. And do it correctly, so one can figure out what is being calculated. And I'll give as much benefit to your opinion as I can.

Focus on the young ages. Data from Wuhan. See table 1:

http://weekly.chinacdc.cn/en/article/id/e53946e2-c6c4-41e9-9a9b-fea8db1a8f51

Death rates by age
0-9 none,
10-39 0.2%
40+ horrendous.

This is a health care system that was slightly overloaded at the peak. Probably some deaths not recorded, and probably some mild cases also not recorded. Wuhan has a population of 10.6 millions, China has a total cases of 81,000, so infection rate was 0.76%. Many doctors and nurses and much equipment and supplies came into Wuhan. This was a controlled epidemic.

An uncontrolled epidemic would have infection rates over 20% at peak. This would overload the health care system to the point where only the most basic care could be provided for almost everyone. 1918 flu type care: water, blankets, bed, soup. Death rates would be rather higher, and impossible for anyone to tell exactly. But would at least double the death rate.

So for working adults, they would face

= Total infection rate * death rate * lack of care multiplier.

Total infection rate is the total of people in the population that catch the disease before enough are immune. Models show half or more.

Death rate is 0.2%

Lack of care multiplier is between 2 and 5, let's use 2. That is supportable by looking at

= .5 * .2% * 2

Which makes it simple. 0.2% chance of dying vs 2 months of shutdown. Chance of dying is probably higher, and length of shutdown is probably lower. Life expectancy assumed to be 80, so young adults lose between 70 years and 40 years. Average is 55 years lost

Shut down is 2/12 of a year. 0.167 year, assuming total loss.

Loss of life expectancy is Chance * years lost

= .002 * 55 = .11 years

So if the shutdown is a near death experience, AND you value everyone over age 40 at nil, then the shutdown is a loss.

Yet the shutdown isn't a near death experience. One son is in a critical job, so is required to work unless sick. Doesn't affect him, other than most stores are closed, harder to spend money. Other son is working from home, as is daughter-in-law. I have been working from home for several years, and have been getting more job opportunities than ever. "Work from home only" isn't quite as unusual as it used to be. I'm not typical, the average is certainly a loss from the shutdown. Half value? Then the shutdown is a gain. Even ignoring everyone over age 40.

I've probably got typos in there, the numbers from Wuhan are not as robust as one might want, and perhaps other issues. Later.

Phil said...

"That is supportable by looking at" ... critical care numbers in table 1 of reference. 49% of critical care patients died.

William M. Connolley said...

> would you trade that pig for 3 sacks of flour?

Doesn't work. As I hope you noticed. In a free exchange, all that matters is that the two people come to agreement.

> Yet the shutdown isn't a near death experience

Indeed. So once again we come to an ambiguous result.

Sam said...

"That makes no sense to me. Why do you think it is true? For example, I have £1, and the shop is selling a bunch of bananas for £1. If the benefit to me of bananas is greater than £1, I'll buy. Where's the market failure?"

OK, so I was speaking broadly in terms of carrying out an actual CBA of the sort that the state does to make policy choices, as laid out in the green book.

The example of market exchange which you gave obviously doesn't need a CBA carrying out, because there's no market failure. Besides which, the exchange is pareto optimal, since it's win win, in which case if you were to choose to do a CBA, fine, whatever.

Firstly, it should not need pointing out that we do many things as a society which would NOT pass any rigorous CBA. For example, something like rescuing trapped miners (think of how much it cost to get those chilean miners out), or maybe those kids who got trapped in Thailand. I'm not convinved either would pass a rigorous CBA. But that is because in these instances, CBA is trumped by a duty of recsue.

Furthermore, most CBA is done under the assumption that if there are any losers, they can be sufficiently compensated to offset their loss (buying up people's homes to build a railway, say). This is blatantly not the case here, since we can't compensate the dead.

There does exist a literature on these considerations as regards the appropriateness of the application of CBA to different situations, and simply put this ain't it. All of the extremely forceful arguments against utilitarian calculus can be brought to bear in this case, and i think they're basically right. This is, rightly, going to end up being a political decision more than one driven by costs and benefits.

Phil said...

" ambiguous result" for those least affected coronavirus. Even young adults are probably better off with the shutdown, as a group. Some individuals are of course not better off with the shutdown.


Death rate for 50-59 is higher, and this age range is less impacted by the shutdown.
Death rate for 60-69 is much higher still, and this age is more likely to be retired.
Death rate for 70-79 is horrendous. Not everyone's retired, but almost everyone could be.
and so on.


Notice that this isn't two people coming to an agreement. Decision isn't bilateral, but multilateral.

Nathan said...

Other views

Both have higher values on life and lower GDP effects

https://www.statnews.com/2020/03/18/economic-rationale-strong-action-now-against-coronavirus/

https://www.nationalreview.com/corner/a-covid-cost-benefit-analysis/

It seems to be take a pick and us whatever numbers you think are correct...

Nathan said...

Also this
https://twitter.com/BetseyStevenson/status/1242180499566669828


And this

https://www.politico.com/news/magazine/2020/04/02/coronavirus-economy-reopen-deaths-balance-analysis-159248


Haven't found any that suggest not implementing social distancing

William M. Connolley said...

> Haven't found any that suggest not implementing social distancing

I think that's the case; my analysis doesn't either. However, these are all very crude either/or stuff; something more subtle would be desireable (e.g. Swedish-style not-lockdown, or some other intermediate state).

That said, https://twitter.com/BetseyStevenson/status/1242180499566669828 isn't terribly convincing, esp in desc her costs as very lowballed.

Also, I've come to wonder if valuing lives at significantly above lifetime GDP contributions, or valuing QUALYs are above annual contribution, actually makes economic sense.

> https://www.politico.com/news/magazine/2020/04/02/coronavirus-economy-reopen-deaths-balance-analysis-159248

The first offer there is reasonably interesting, including

> The authors estimated that the present value of future worker contributions ranged from about $91,000 to $1.2 million in 2007, depending on the age of the worker... In other words, the economic benefit of preventing all those potential deaths depends on which controversial measure you use: In this case, upper-bound estimates of mortality benefits associated with government interventions range between $20 trillion with the VSL approach and $828 billion if the worker production approach is extended to 2 million lives saved. Twenty trillion dollars is roughly the value of an entire year of the nation’s GDP; $828 billion is considerably less than the value of Congress’ latest economic stimulus bill.

> https://www.statnews.com/2020/03/18/economic-rationale-strong-action-now-against-coronavirus/ "If the average age of those killed by Covid-19 is 60..."

But it probably isn't; I'd expect it to be higher.

So it's good to see all these attempts at CBA; I'm glad people are thinking about it.

Phil said...

A modest proposal:

If we only have value while we work, wouldn't society be better off by killing anyone that can't work anymore? Due to age, disability or any other reason.

Of course, the idle rich should be exempted.

William M. Connolley said...

I don't think so; if you follow the link you'll find even the old have economic value, from the services they cause to exist by consuming them.

Anonymous said...

Lots of them in this article
https://www.bloomberg.com/news/articles/2020-04-01/cost-benefit-analysis-in-the-time-of-coronavirus-and-climate-change

Sorry for spamming

William M. Connolley said...

There are now too many CBA articles :-). Your first, under the text "The verdict in virtually all of them is clear: shutting down the economy to contain the spread of the virus is worth the costs", leads to a second, This Time the Numbers Show We Can’t Be Too Careful. But the first is being lazy; the second (after a loooong prologue) merely reports other research: Michael Greenstone and Vishan Nigam of the University of Chicago analyzed the likely benefits of “a moderate form of social distancing,” including a seven-day isolation for anyone showing coronavirus symptoms, a 14-day voluntary quarantine for their entire household and dramatically reduced social contact for all those over 70 years of age. They found that over a period of seven months, the result would be to prevent 1.7 million deaths. But that's rather interesting, because it's the first I've seen on the effects of "mild" measures; and given that the do-nothing toll was ~2M, that appears to say (but the article chooses not to say it) that you get 90% of the effect for these "mild" measures. Also interestingly, we get from the original Roughly 90% of the monetized benefits are projected to accrue to people age 50 or older which is expected, given the age/death profile, but again not widely mentioned.

Nathan said...

Yes too many!
They're all supporting some action, but to work out the 'optimum' is too difficult right now.
One thing that hasn't been considered is the value of all the experiments that are being conducted right now - huge value in the variety of responses and I think future CBAs for pandemics will be rooted in good data.

That being said I am not convinced that CBAs are useful for complex situations where there is a very large number of options. They're perfectly suited to choosing where a bridge should be constructed or the route of a railway line... But where you have multiple options with little hard data they're only useful for really broad 'I reckon' statements.

Phil said...

WC:"the old have economic value, from the services they cause to exist by consuming them."

So as long as their savings holds out...

William M. Connolley said...

No; they get to call on the state. However inadequate you might think it there's wealth there. Which is why crude estimates of peoples wealth are so wrong.

Phil said...

The state?
Isn't taxation theft?