The problem of herd non-immunity

I listened to Russ Roberts and Tyler Cowen discuss the coronavirus. If you missed it, maybe you can find it at the Mercatus video archive at some point. Here is rough paraphrase (caricature?) of part of the dialogue.

Russ: Think of this as a three-week vacation. Why can’t the economy recover from a 3-week vacation?

Tyler: A lot of organizational capital will be lost.

Russ: Wha???

Tyler: Employers and employees know how to work together. When those relationships need to change, it takes a lot of time for matching and training to work out.

Russ: You seem to think that the virus will be a factor for a long time. Why?

Tyler: Suppose that in the very short run we get it under control through social distancing. That means that a lot of people will not have had it yet. What is likely to happen is that there will be a series of outbreaks, and that means a series of shutdowns. It means that it will be a long time before people feel comfortable going to locations where they will encounter crowds.

Think of this as the problem of herd non-immunity. It means that we could go a long time during which people change their behavior to avoid catching/spreading the virus.

To me, this makes two things important. One is the process of testing and approving a treatment. If a treatment can be shown to work, then we can be much more relaxed about allowing people to get the virus.

The other is having a random testing program. John Iaoannidis is getting a lot of flak, some of it deserved, for what he wrote yesterday. But I very much agree with this:

The most valuable piece of information for answering those questions would be to know the current prevalence of the infection in a random sample of a population and to repeat this exercise at regular time intervals to estimate the incidence of new infections. Sadly, that’s information we don’t have.

If the government won’t do this (and I have little confidence that they will), then I hope some corporation or non-profit will take it on.

6 thoughts on “The problem of herd non-immunity

  1. Random sampling = quality assurance. It doesnt care about your underlying assumptions, unless “random” has a bias.

  2. Ionnaidas is completely correct, I don’t know why he’d take flak for this. Unless we allot some fraction of tests to random population regularly we have no way of knowing what is or isn’t working. If current strategies are failing, we have no idea; on the other hand, once infections are receding, we will not know until weeks or even months later how much, and may be unnecessarily restrictive, since as the number of tests goes up, the number of people testing positive will keep going up even after the infection rate starts to decline.

    Hopefully some other institution starts doing this since the government refuses to.

  3. What we actually need is a reliable immunoassay, not a test for whether or not you have the virus the day you took the test.

    People persist in believing this started in the last month, but I strongly suspect the numbers are telling a different story- the broad geographic spread tells me that the virus was in the US since last Fall, and we are now probably at the peak numbers of infected, and that number is probably a million or two. The problem is that we only knew to start looking for this disease in early February. The numbers of new confirmed cases is closely tracking the increase in the numbers of tests being run, but people are misinterpreting this as the infection spreading rapidly in the last week- it probably isn’t spreading rapidly now at all due to all the measures taken in the last week, but the pool of infected was and is much larger than we thought when testing started.

    If I am right, the US will see “new confirmed cases” peak by the end of next week at around 10-15 thousand/day. Deaths will continue upwards and probably reach 3-4 hundred/day by the end of next week, too, and then start the same decline seen in China, and starting to be seen in Italy. I think the tail will be long and annoying, with new cases showing up in dribs and drabs all Summer long, but deaths will fall to less than 10/day by the beginning June.

    I will put it right here- the US will have 150,000 “confirmed cases” and a total of 7500 deaths by the end of the year.

  4. Also, when statistical sampling of the population is done with immunoassays, it will be discovered that less than 5% of the infected were ever identified by the tests being run today.

  5. Right now, we are behaving like the homeowner who discovers a termite husk next to an electrical outlet, opens up a 1 foot section of the wall next to the outlet and discovers termites eating on that particular stud and thinks to himself, “Thank God I caught this early- I now have a chance to stop it from spreading to all the other studs and joists in the house!” Additionally, we are like that homeowner opening up the wall on ten other studs and finding termites eating on them, too, and exclaiming, “Good grief, the termites are spreading faster than I can find them!”

Comments are closed.