In mid-March I made a bet with my good friend and co-author Charley Hooper about the number of U.S. deaths there would be from COVID-19. The terms of the bet are here. In my post, I said why I thought he might win. Of course I hoped he would win. Unfortunately, he lost. And over 100,000 U.S. residents lost much, much more.
I waited this long because he and I both agreed that there could be a substantial number of deaths of people with the disease but not of the disease. We both agree, though, that of the 133,844 U.S. deaths so far, at least 100,000 of them are due to COVID-19.
I actually had bought much of Charley’s reasoning, which is why I titled my March 16 post “My Bet on Covid-19 and Why I Might Lose.” I asked Charley last week, when we both were becoming convinced that he lost, what he attributed his loss to. He answered that he didn’t expect various governments to be so incompetent, and he highlighted the role of New York’s governor Andrew Cuomo and some other northeast governments in making the problem much worse by insisting that nursing homes admit people with the disease.
READER COMMENTS
Mark Z
Jul 7 2020 at 10:51pm
It’s an unfortunate bet to win, I think having betting markets on this are actually likely be useful, for much the same reason as Scott’s (Sumner) and David Beckworth’s argument for NGDP futures markets. Only instead of the Fed reacting to market expectations of NGDP, it would be people in general, companies, governments, etc. can react to expectations about how bad the death toll or case load would get (and hopefully thereby prove pessimistic predictions wrong).
Anyway, I thought given the discussion about the merits of morbid bets in the comments section of your earlier post about the merits of betting on such things.
Matthias Görgens
Jul 8 2020 at 10:35pm
Even if you don’t have full blown markets, even informal bets are an entertaining mechanism to keep pundits honest and away from hyperbole.
Alan Goldhammer
Jul 8 2020 at 9:11am
I commented at the time that you were virtually certain to win the bet. A naive population exposed to a zoonotic virus that jumps species is always going to be a problematic event. We know from public health data that major influenza outbreaks caused by simple mutations in existing viruses result in high levels of morbitity and mortality.
While we don’t know the exact Case Report Fatality rate of COVID-19, it is certainly worse than past pandemics but likely not as high as what is presently observed in the US (4.4%). Most of us who have familiarity discounted the early Italian rates as way too high. My own feeling is that it would settle down in the range of 0.3-0.6%. Since herd immunity requires a likely minimum of 60% of Americans to be exposed to SARS-CoV-2 you can see easily how the mortality numbers spiral to a high level even at the low end of the CRF.
Of course Cuomo made a bad early decision but he did regroup to help out New York City. That is much more than can be said of the governors of Texas, Florida, and Arizona who had the advantage of seeing what happened in New York and ample time to put a system in place to deal with what could and as we have seen has happened. There still continues to be a large number of COVID-19 deniers which is absolutely mystifying.
Anonymous
Jul 8 2020 at 1:09pm
Those places are still doing far better than New York. Florida in particular did an excellent job of protecting its nursing homes, as I understand it.
zeke5123
Jul 8 2020 at 3:18pm
Yep — while there of course is some benefit in being “third,” what most people don’t mention is that New York was not “first.” Florida looked at Italy even before NY and devised its retirement home policy.
This is why even though the sunbelt is seeing a higher caseload right now, so far the deaths have lagged far behind NY. Hopefully, it stays that way.
robc
Jul 8 2020 at 3:28pm
I thought 60% was the high end estimate for herd immunity, with 40-60 being the generally accepted range and some radical suggestions as low as 20%?
Alan Goldhammer
Jul 8 2020 at 4:17pm
It varies all over the map for past infectious diseases for which we now have vaccines. Measles and pertussis (whooping cough) outbreaks in the US are now common because of the anti-vax movement. In both of those cases the % of vaccinated individuals is extremely high, much more than the 60% mark that I noted above. Most of my public health friends think that 60% is the minimum that needs to be achieved. Going by statistics that I read this morning we are only at the 0.9% mark so there is a huge way to go.
Matthias Görgens
Jul 8 2020 at 10:37pm
Talking about raw numbers of vaccinated or immune people is only useful when the population is homogenous.
For measles, what’s more important is probably the proportion of susceptible children, and whether they hang out without each other. IE the shape and connectivity of the social graph.
robc
Jul 9 2020 at 8:22am
Here is the article I was going off of: https://www.quantamagazine.org/the-tricky-math-of-covid-19-herd-immunity-20200630/
It also deals with Matthais point about homogeneity. In general, heterogeneity decreases the overall percent required for herd immunity.
Dylan
Jul 8 2020 at 9:17am
David,
In a comment to an unrelated post, I think in May, you wrote that you were feeling confident that you were going to lose this bet. That seemed strange to me as, IIRC, we were already around 70K confirmed deaths at the time without even taking into account what was likely many unattributed CV-19 deaths from people dying outside of the hospital setting. Do you recall what your confidence that you would lose was based on at that time?
David Henderson
Jul 8 2020 at 12:20pm
Yes, it was based on Charley’s reasoning about the Diamond Princess.
Alan Goldhammer
Jul 8 2020 at 1:11pm
The classic example of a single event with a too small sample size. The other thing missed was the mutational event that made the Euro version of the virus more infective.
Jon Murphy
Jul 8 2020 at 2:47pm
I’m not sure it’s a sampling size error. I think it’s an assumption error. The cruise liner was missing one key factor: government. A lot of us, even those of us with relatively low priors on government competency and public choice training, vastly underestimated the sheer level of incompetency of governments worldwide on this matter.
COVID has been a dark day for government.
Charley Hooper
Jul 8 2020 at 3:20pm
Further government incompetence shows up in prison outbreaks. According to the New York Times, nine of the ten largest outbreaks in this country have been in prisons/jails.
But the other reason I underestimated the spread of COVID-19 was the early data from South Korea, Taiwan, China, and Japan. The virus was far more forgiving in those countries, for whatever reason (genetics? policies? behaviors? past infections?).
In short, I properly estimated the infection fatality rate of 0.25% but I underestimated the percentage of the population that would become infected and the stupidity of governments.
Alan Goldhammer
Jul 8 2020 at 4:26pm
Government clearly had a role in screwing things up but is not solely to blame. Once a virus gets established in a naive population it is here to stay. Either it runs it course as it did in the Lombardy region in March or you try to make the best public health efforts to deal with it. The difficulty with COVID-19 is the somewhat lengthy time before clinical symptoms appear leading to super-spreader events. These can go unnoticed such as with the Biogen event that took place in Boston in late February. This was not the fault of any governmental screw up as it was sui generis in terms of East Coast exposure.
One can write a book on the number of wrong policy decisions that were and continue to be made in response to this. I am not optimistic at all about school reopening in the fall. The cost of readying schools to deal with this mess is quite high and local governments are cash strapped because of dropping tax revenues.
Jon Murphy
Jul 8 2020 at 5:00pm
One might be in the process of doing just that. Which is this one’s point: government done messed up bad. Real bad.
Charley Hooper
Jul 10 2020 at 2:05pm
🙂
Josh
Jul 8 2020 at 10:39am
We’ll never know the answer, but I wonder if you’d have lost your bet had there been literally zero government intervention. No daily pressers, no laws, no mandates, no guidelines, no FDA or CDC. Just let people watch the news and make up their own minds.
There’s overwhelming evidence that most of the social distancing we got was voluntary. And the complete disconnect between deaths and cases right now in the south shows that people are a pretty good judge of their own susceptibility and respond accordingly. So I don’t think we can give government too much credit for the few things we did right.
But the government’s failures in nursing homes probably caused 10s of thousands of deaths directly. And while I’m still not convinced masks are that strong of a factor here, to the extent they are, governments telling people that masks were useless in March was a costly mistake. The CDC requirements for testing that slowed our initial response might be the biggest mistake of the whole thing. It’s possible the politicization of ventilators in NY led to thousands of needless deaths from their overuse. Not to mention the HCQ debacle (I’ve lost track – does it work now or not?).
It seems like far from the government being a “win some lose some” entity, it just actively made many things worse and few – if any – things better.
It seems your friend bet against total government incompetence. And lost.
David Henderson
Jul 8 2020 at 12:21pm
You wrote:
I think Charley would agree with that.
Charley Hooper
Jul 10 2020 at 2:13pm
I would agree with that. But I also made some mistakes in that I was swayed by early data that didn’t hold up.
Michael
Jul 8 2020 at 12:47pm
I think it is early for this assumption, since deaths are a lagging indicator and the process of reporting of deaths (and their attribution to Covid-19) varies by jurisdiction. If deaths stay relatively low over the next few weeks, that would strongly suggest that something (or things) have changed in a good way.
This is something I’ve not heard anywhere and strikes me as extraordinarily unlikely. Putting somone on a ventilator is a big deal. Without seeing the evidence, I’m very dubious about whether lots of people are put on ventilators unnecessarily.
I think the (continued) failings of government around masks and around testing loom as the largest problems. If they actually force schools to reopen (rather than focusing on what is necessary to allow schools to open safely) that strkes me as potentially another failing. I’d be, today, totally comfortable sending my child back to school if school staff wore masks and underwent periodic testing. But I don’t expect it to happen. Masks, yes; testing, no.
Charley Hooper
Jul 8 2020 at 3:32pm
From WSJ:
Why have the outcomes been so bad on ventilators?
In other words, most patients don’t need to be put on ventilators, which are highly invasive and can cause long-term problems.
Michael
Jul 8 2020 at 5:52pm
The level of information there is insufficient to establish causation, of course.
Charley Hooper
Jul 10 2020 at 2:07pm
Yes, you are right. I was trying to find a quote that reflected how doctors were changing course after realizing that ventilators might be the problem, not the solution.
Alan Goldhammer
Jul 8 2020 at 1:23pm
there was no politicization of ventilators. Intubation was standard of care in China and Italy for COVID-19 pneumonia and that was all the doctors had back in March. They, as others, were flying by the seat of their pants trying to figure out what worked and what didn’t. It wasn’t until about a month ago that practice shifted to the use of IL-6 blockers and corticosteroid therapy. the latter has been shown effective in the large UK RECOVERY trial and there is enough observational data on tocilizumab to move that into standard of care.
Early on with no justification at all, HCQ +/- azithromycin was the standard of care which even our community hospital in Bethesda, right across the street from NIH, was using. HCQ does not work and there have been both controlled trials as well as observational studies that are conclusive. The only studies I have seen that suggest otherwise are either underpowered (too few patients) or poorly designed.
Michael wrote:
Quite right and what is being ignored by de Vos, Azar and the rest of those in the administration is what the effect on the teachers will be. It’s fine to assume that all these kids are not at risk for COVID-19 but teachers are not kids. My daughter teaches special education at the elementary school level and has students that are unruly and frequently spit or hit other students and even teachers. There is no room for social distancing in these classes and to expect students at this age to wear masks all day long is laughable. Now if de Vos and Azar step up and spend a couple of days in a class rooom, I would take them more seriously.
robc
Jul 9 2020 at 12:42pm
I know you have access, can you point me to a PROPER controlled trial? The ones I have seen have all been poorly designed.
I want one where the patient are early onset (pre hospitalization), they use (relatively)low dose HCQ and also include Zinc. With or without Z-packs, preferably testing both ways.
All the HCQ controlled studies I have seen violate 2-3 of those prerequisites.
I don’t know what the result would be, but is it that hard to actually test the right way?
TMC
Jul 9 2020 at 2:10pm
Trials, not anecdotal studies, are starting to come out.
“Five studies, including two controlled clinical trials, have demonstrated significant major outpatient treatment efficacy. Hydroxychloroquine+azithromycin has been used as standard-of-care in more than 300,000 older adults with multicomorbidities, with estimated proportion diagnosed with cardiac arrhythmias attributable to the medications 47/100,000 users, of which estimated mortality is <20%, 9/100,000 users, compared to the 10,000 Americans now dying each week. These medications need to be widely available and promoted immediately for physicians to prescribe.”
https://academic.oup.com/aje/article/doi/10.1093/aje/kwaa093/5847586
Scott Sumner
Jul 8 2020 at 2:01pm
I would have lost this bet—I didn’t expect this many deaths.
Alan Goldhammer
Jul 8 2020 at 2:46pm
Scott – all you had to do was look at the 1957-58 influenza epidemic which was not a novel zoonotic event. The CRF for that one was 0.3%. At the time the bet was made there was enough data from Wuhan pointing to this being much worse. I said at the time this was as close to a sure thing as one can get.
robc
Jul 8 2020 at 3:32pm
Or 1968-69. 100k US deaths from a population of 205MM, so you would expect about 160k from this.
But I was on the wrong side, as I thought we would end up under 100k too.
John Hare
Jul 8 2020 at 4:00pm
i guessed under 10,000 and would have bet a small amount if I could have figured out how to state the terms. I can list a number of mistakes I made.
Matthias Görgens
Jul 8 2020 at 10:42pm
Please do so.
Greg Jaxon
Jul 9 2020 at 6:39pm
In mid-March I felt the same. Then we learned that hospitals would get $39,000 for putting a C-19 positive patient on a ventilator and that there was some doubt that the anti-malarials+zinc were helping. The case for government incentives making all things worse was never clearer, so I switched sides on this bet and continue to bet that both the Fed and the CDC/NIH/FDA will maximize the damage.
Charley Hooper
Jul 10 2020 at 2:10pm
Did you quantify your estimate? What were you thinking in terms of U.S. fatalities?
Comments are closed.