Wednesday, April 8, 2020

Coronavirus

I haven't been writing about coronavirus here.  There are plenty of places to get coronavirus news. But, I have been posting daily updates on Facebook, just doing some basic data analysis, and people there seem to appreciate it, so I figured I'd add a post here.

My pathway on this subject has been thus: I basically wasn't paying too much attention, and was roughly in the "it's just a bad flu" camp in January and February.  But, I was increasingly nervous because people like Tyler Cowen and Robin Hanson were especially worried about it, in a conspicuous way that seemed unlike them.  Eventually, I looked closely enough to realize it was a potentially big problem.  By early March, I was wondering what we would need to do.  By mid-March most of the country had come to that realization, so I don't know that my path was much different than most people. By mid-March, I was watching exponential growth of the contagion, and worrying about the considerable damage that each new day's growth would bring.

By late March, though, I saw the first faint signs of a bending curve, and so I started tracking the numbers daily more carefully.  On March 30, I took to Twitter to predict that the high point in the national number of new daily cases would happen that week and that daily new cases would be below 5,000 by April 15.  I just barely made my first prediction.  It appears that Saturday, April 4 might be the high point for new cases.  My second prediction might be a little more difficult, though.

The first graph here is the US daily growth rates in the cumulative number of cases, hospitalizations, and deaths.  A lot is made of problems with the data.  Certainly they aren't perfect, but in terms of trends, I think it's more informative to work with it than it is to act like the data is extremely biased.  There are a lot of reasons to think that, which I'm not going to get into here.  But, one reason is that trends in all three of these measures are running parallel to each other in basically the fashion one would expect.  Also, the variation in outcomes limits the potential for the data to be too far off.  New York shows us what it looks like to have more cases.  Very few places look anything like that.  Also, we know that the death rate for older people is north of 10%.  If cases were much higher than what we think they are, there would be a lot more dying old people.  It is easy to come up with plausible reasons to doubt the data, but I just don't see the doubts standing up to the same level of scrutiny that the doubters are applying to the data.

Anyway, this is a lot like valuing equities.  You just have to be comfortable with quite a bit of unfalsifiable noise.  And, even with that, there are stories to discover.

The interesting thing about these growth rates is that they are declining in a pretty linear way.  The same is true generally of the individual states, too.  But with data this noisy and a time frame this short, it is difficult to see the difference between a linear trend and a convex trend.

Here is the national daily growth rate of cases since March 21, and I have fitted two trends to it, beginning on March 28. One is a linear decrease in the growth rate of 0.93% each day.  The other is a proportional change in the growth rate. Each days growth rate is the previous day x .93.  They both could describe the recent trend, and I would say they might serve as a decent estimate of the range of expectations going forward.

The last graph shows the results of each, in terms of daily new cases.  It makes a big difference. The linear change in the daily growth rate would have to be more or less right in order for my April 15 prediction to come true.  In the next few days, it should become clear what the actual trend is.

Exponential growth can come at you fast.  All in all, I think we did a pretty good job in most places of getting out in front of it.  But, the change in trend in the other direction can be just as surprising.  It could make a big difference for a lot of lives, a lot of jobs, and a lot of investments, if many states around the country can be mostly rid of new cases over the next couple of weeks. Then, the conversation can turn to how quickly we can get back to normal.  We'll see.

PS. The April 9 report shows more new cases than on April 4, so my first prediction failed also.

13 comments:

  1. I like your way of looking at things.

    And yes, in the real world, social science data is always flawed, and sometimes sketchy. One could say macroeconomics is often practiced in a fog (self-induced or otherwise).

    I sure wonder if the lockdowns are effective in reducing the new infection rates, or is it rising immunity rates? Of course, the answer may be both.

    Tyler Cowen posits perhaps multiple lockdowns are ahead, as we still have a largely naive population and a novel virus.

    In terms of employment, it took the US about 10 years to recover from the 2008 Great Recession. If the Fed sticks with its playbook, and still has a 2% inflation target in mind, I guess we have another 10 years to recover from the 2020 Depression. By 2030 we will be okay again.

    From what I see, monetary economists are not talking in terms of, "How do we get tens of millions of people re-employed in a year or two, given a sluggish global economy?" They are still talking are price level targets. I wonder if even NGDPLT makes sense when 30 million people are unemployed. Which, I guess, will be sometime in May.

    It sounds ghoulish, but I wonder if costs and benefits line up on lockdowns. My guess is not.


    ReplyDelete
    Replies
    1. The reduced infections are certainly from the lockdowns and not from immunity. Infection rates are much different in NYC than practically everywhere else, but infection rates are declining everywhere.

      It will definitely be interesting to see what the economy does over the next few months. Past recoveries from pandemics were fast, so I have hope, but there may be things about the current economy that just make recoveries slower in general. We'll see. It would be a great time to put a few million folks to work in construction! :-)

      I think the cost and benefits probably look pretty good, especially if you compare them to plausible counterfactuals, such as, waiting for tens of thousands of people to start dying every day, and people voluntarily end up going into isolation anyway. The death toll would be huge in any counterfactual. NYC has 10x the cases and deaths of the rest of the country, and that's just from having cases grow for a week or two longer before locking down. They're digging trenches to store the bodies. There is no reasonable counterfactual that doesn't end up with a major economic upheaval.

      Delete
  2. Well, maybe Sweden is a counterfactual. We will see. They say Belarus is going its own way too. Also, there are worrying clues that lockdowns do not have an exit strategy, but are a dead end. How does one leave a lockdown?

    Yes, surely all options are bad regarding COVID-19, and it is matter of taking the least-bad option. The vaccination option, which we all hope for, is said to be 18 months distant.

    Would that several millions be put to work building 5-10 million units of housing along the West Coast...outdoor work, for youngish people. A perfect match for the times.

    Now, they are talking 40 million unemployed nationally by end of May.

    And there is a legitimate point of view that with a naive population but a novel and highly infectious virus, you're fighting a losing war. The virus will inevitably get to 60%-80% of the population and then fade. So, lockdowns do suppress peak loads on hospitals, but not the final results.

    Anyway, what is done is done, and we are going down the lockdown route. Maybe multiple lockdowns.

    I hope you are right and pandemic recoveries are quick. This pandemic is on, and the global 2020 Depression deep. Are there tools in place to get 40-60 million employees back to work quickly, if ever we hear the "all clear" signal?

    What should policy look like?

    ReplyDelete
  3. http://www.igmchicago.org/covid-19/estimates-of-the-undetected-rate-among-the-sars-cov-2-infected-using-testing-data-from-iceland/

    Well, there is a Welter of studies but this one comes from Harvard MIT and suggests there is a lot more infections than we think. Ergo, old-fashioned acquired immunity might be playing a role in reducing new infections.

    ReplyDelete
  4. So, about one-third of apartment renters are not paying the rent alteady and 50 million Americans, or about one-third of the labor force, will be unemployed by the end of June, if lockdowns persist, or so predicts the St Louis branch of the Federal Reserve.

    But it has been worth it because, according to our top credentialed epidemiologists, we might have saved an unknown number of lives.

    ReplyDelete
  5. https://www.realclearmarkets.com/articles/2020/04/01/coronavirus_death_predictions_bring_new_meaning_to_hysteria_487977.html

    not long, worth reading

    ReplyDelete
    Replies
    1. On your renters stat, it would be important to know how many renters were behind on last months rent. It wasn't zero. Clear overstatements are really not going to help your case if you're trying to be the anti-hysterics poster.

      What would you have recommended for New York City and what would the outcome for it be now if it followed your recommendation?

      Delete
    2. https://www.wsj.com/articles/nearly-a-third-of-u-s-renters-didnt-pay-april-rent-11586340000

      OK, I got this from the Grumpy Economist, but evidently 18% of people are late on the rent in most months.

      So, I may have been hysterical by a factor of 2, instead of a factor of perhaps 20, as in the case of public health professionals.

      It is an open question on how effective lockdowns are. Evidently, epidemics often follow a bell-shaped curve, so New York infections would always start dropping at some point, lockdowns or no.

      There are puzzling variations, such as parts of Italy or New York City. And, I am told, the hospitals of Los Angeles County are under crowded, compared to normal.

      I think the verdict will come out in a few months that the lockdowns were an economic catastrophe and had some modest, perhaps temporary improvement in health outcomes.

      We won't do this again.

      Tyler Cowen, having assented to the economically catastrophic lockdowns, is now wondering what the epidemiologists were talking about. He looks like some guy who thought he was going to Johns Hopkins but ended up in a Baltimore bunghole house.

      So far, we have avoid 9 million dead (initial estimates based upon 3% Wuhan death rates), to 7 million dead (Luigi Zingale), to 1.9 million dead (AEI), to 250,000 dead, to 60,000 dead maybe.

      The death rate from COVID-19, if the US has 60,000 dead (and we give no credit to comorbidities) will be 0.0182%, or maybe one in 5000 people.







      Delete
    3. You didn't answer my question.

      Delete
    4. Here are death rates compared to previous years in England. People were going to go into lockdown mode in any case, Ben. The choice we made was to do it in a coordinated fashion before hundreds of thousands died instead of individually after.

      https://twitter.com/ChrisGiles_/status/1250035472493088768

      Delete
  6. OK, to answer your question I would have followed what the science community initially recommended, such as in UK, to simply go for herd immunity. We will see how things turn out in Sweden and Belarus.

    Right now, Sweden has a population of about 10 million and about 1,000 dead from COVID-19, if you give no credit to comorbidities. So, about 1 in every 10,000 Swedes has died from C19, if you give full credit to the virus and no credit to comorbidities. Obviously, for humane reasons, I hope that's the bulk of it for Sweden.

    Obviously, a totally unpleasant situation of a novel virus and a naive population. There are no good answers. And to be honest, the China policy seems to have worked, if they are telling the truth. However, China still has a naive population and a novel virus, a virus becoming endemic in the world. Perhaps a vaccine will be the cavalry.

    ReplyDelete
  7. Learning Ground learningground.com.ng is a blog that educate people both in Education, technology, news, health, business, work opportunities.

    ReplyDelete
  8. Natural herbs have cured so many illness that drugs and injection can’t cure. I've seen the great importance of natural herbs and the wonderful work they have done in people's lives. i read people's testimonies online on how they were cured of HERPES, HIV, diabetics etc by Dr Edes herbal medicine, so i decided to contact the doctor because i know nature has the power to heal anything. I was diagnosed with Herpes for the past 3 years but Dr Edes cured me with his herbs and i referred my aunt and her husband to him immediately because they were both suffering from Herpes but to God be the glory, they were cured too .I know is hard to believe but am a living testimony. There is no harm trying herbs, Thanks. Write him on WhatsApp on +2348151937428. @dr_edes_remedies deals with
    (1)Alzheimer virus
    (2)Cancer
    (3)HIV
    (4)Herpes
    (5)Genital warts
    (6)ALS
    (7)Virginal infection
    (8)HPV
    (9)Hepatitis
    (10)Asthma
    Email him on dredeshome@gmail.com
    https://dredesherbalhome.weebly.com

    ReplyDelete