COVID-19: Some Empirical Observations and Reasons for Optimism

April 14, 2020

Welcome back to a new post! I hope you all had a happy and safe Easter Weekend! Last week I published a post wondering about how the stock market can be down “only” about 20% in the first quarter when we’re facing a deadly virus, a wide-ranging shutdown of the economy and no clear idea when we can reopen again. We are expecting a deep recession and macro data significantly worse than during the Global Financial Crisis.

If you listen to the media talking heads it’s all doom and gloom, so why is the stock market holding up so well? Why isn’t the stock market down 55% as in 2009 or even 80% as in 1932? Does the stock market “know” something that even your trusted news anchor doesn’t realize yet?

I updated some of the charts I posted a few weeks ago and gathered some additional data as well. And it all looks much better than the breathless and scary headlines in the media. Maybe that’s why the stock market looks relatively solid – under the circumstances, at least.

Let’s take a look…

1: The fatality rate will likely be much lower than the early estimates

Early estimates of the fatality rate looked quite scary: 3.4%. Even now, if you look at the deaths vs. confirmed cases, the fatality rate is even higher than 3.4% in most countries, e.g., 4% in the U.S.!. But all those numbers are garbage estimates from people who don’t know statistics. When we divide the number of deaths by the number of confirmed cases, we could potentially vastly undercount the denominator because there are likely many more infected people that are asymptomatic. There are some indications that the current simple case fatality estimates wildly overestimate the lethality of the virus:

  • German scientists performed a large study in one of the very first German COVID-19 clusters, a small town in Nordrhein-Westfalen (my former home state, by the way). They found that in addition to the 2% of the population already confirmed, there were an additional 14% infected that had no idea they had the CCP critter. Therefore, the death rate was 0.37%. Even that death rate might be too high to extrapolate to the larger German population because we’re now much better prepared to deal with COVID-19 patients than back in February when the virus hit an unsuspecting 12,000-resident town unprepared for the onslaught of a global pandemic. More testing needs to be done but these early results definitely look much better than 3.4%! (Side note for the math/stats geeks: Some critics of the study have pointed out that the 16% estimate is too high due to the less than 100% specificity of the test. But that issue is quantitatively not relevant as this paper points out.)
  • Iceland has been aggressive with testing (easy to do in a country of 300,000): more than 10% of the entire population has been tested so far (compared to 0.5%-2.0% in most of the large developed countries). The fatality rate so far: Just under 0.5% (8 deaths out of 1,700 confirmed cases). Even that death rate might be too high if you consider that in the remaining 90% of the population not tested yet there will be many more asymptomatic people. Source:
  • Updated 4/14/2020 8:30pm PDT: At two NYC hospitals (New York Presbyterian Allen Hospital and Columbia University Irving Medical Center), 215 pregnant women who came in for delivery were routine-tested upon admission. 33 of them tested positive for COVID-19. Most of them asymptomatic. That’s 15.3%, almost exactly the same figure as in the German study above and much higher than the 1% confirmed infection rate. It would also lower the current New York fatality rate from 5+% to 0.36% if we extrapolate the infection rate of 15.3% to the entire population. Link: NEJM
  • As pointed out in the excellent Wall Street Journal op-ed piece by two Stanford Medical School professors, 0.9% of the people evacuated from Wuhan on January 31 eventually tested positive for COVID-19. If you extrapolate that 0.9% prevalence to the entire Wuhan metro area you had a 30-times higher infection rate than reported and confirmed at that time. And thus a much lower death rate! Of course, there could be self-selection, and thus, the result is not as robust as the German scientific study involving a large group of randomly sampled subjects.
  • Updated 4/17/2020: A sample of 3,300 subjects in Santa Clara County California found that the true infection is 50-85 times larger than the confirmed cases. That means that the true fatality rate is only between 1/85 and 1/50 of the current naive case fatality rate: between 0.12% and 0.20%. About in line with the seasonal flu. Vie WSJ: “New Data Suggest the Coronavirus Isn’t as Deadly as We Thought

So, the lesson here is: without trying to minimize how scary and deadly the disease is, the death rates floating around in the media are complete junk. Of course, journalists not only get away with exaggerating the lethality, but they are even incentivized to do so: “if it bleeds it leads!” In contrast, when people put their money where their mouth is, i.e., the stock market, cooler heads prevail. I think the stock market holding steady foreshadows large downward revisions of the fatality rates in the future.

2: The age distribution of cases and deaths is heavily tilted toward old people

Just to be sure, a 0.37% death rate is still substantial. If I faced a 25% chance of catching the virus and then a 0.37% conditional probability of dying from it, that’s still about a one-in-one-thousand probability of dying over the next few weeks and months. I could certainly rationalize staying at home then! Why take a chance, right?

Well, that 0.37% number is still complete junk science if we were to apply it to the whole population equally. Let’s look at the following completely unrelated example to drive home this principle: In the U.S., more than 5,000 motorcyclists die every year. Does that mean I have a 0.0016% probability (5,000/320,000,000) of dying in a motorcycle accident this year? No, I have a 0% chance because I don’t ride a motorcycle. On the other hand, an average motorcyclist has a 0.059% fatality rate, much higher than the average American!

The same principle is at work with COVID-19. The fatality rates differ wildly by age. Unfortunately, it’s not that easy to get your hands on high-quality data. I found data on the Swiss Health Ministry site that details the number of cases, hospitalizations and deaths by age group. Here’s what I found when I processed the data:

  • Out of 884 deaths up to April 13, there were 860 in the 60+ age group and only 24 in the 0-59 age group.
  • Thus, unconditionally, Swiss residents age 80+ have a COVID-19 fatality rate 366 times higher(!!!) than those aged 0-59: 1,352 per million vs. 3.7 per million residents.
  • Swiss residents aged 0-59 have only 1/27.9 the death rate per million residents of the overall population: 3.7 vs. 103
  • On April 13, the (naive) case fatality rate (deaths divided by confirmed cases) among all Swiss residents was 3.47%, coincidentally close to the early Wuhan figure (and likely far overestimated for the reasons laid out above). But notice that for people aged 0-59, the case fatality rate is 0.15%, only 1/23.9 of the overall rate.
  • And conditional on being a confirmed COVID-19 patient, people aged 80+ have a 129.2-times higher death rate to 0-59-year-olds: 18.78% vs. 0.15%.
  • If you assume that, just like in Germany, the true infection prevalence is 8-times higher than confirmed, then the death rate for people aged 0-59, conditional on being infected is not even 0.15% but under 0.02%.
Swiss Fatality Data 2020-04-13
Fatality rates in Switzerland as of 4/13/2020. Source: Bundesamt für Gesundheit

And just to be sure: I’d never argue that old people’s lives count less. Quite the opposite, we should likely be much more stringent in quarantining and protecting older citizens. But we can also be a lot more lenient in opening businesses again and letting younger folks out of their homes and go back to work (but please wear a face mask!!!).

By the way, I tried to find comparable data in the U.S., but the CDC publishes very incomplete data that covers only a fraction of the fatalities (less than one-third). Qualitatively, you had similar patterns though quantitatively not quite as stark as in Switzerland. But then again, there could be a selection bias in the data set if it covers only a small fraction of the total fatalities.

Update 5/5/2020: On the CDC page, there’s no more data on the age distribution anymore. I wonder why…

Summary so far

The 3.4% fatality rate that was floating around initially and is still probably stuck in people’s heads is far too high. Once we make two adjustments 1) account for the bias due to testing only people with bad enough symptoms and 2) accounting for the age distribution of fatalities, then the fatality rate looks a lot less menacing, see the chart below. If you apply the German fatality rate of 0.37% and then assume that young people have only 1/23 of the overall fatality rate we’re down to 0.016% for people aged 0-59. The actual figure might be higher or lower depending on what kind of age distribution you have in the under-counting of asymptomatic people. For example, if you assume that the asymptomatic patients are primarily young and healthy people their true fatality rate might be even much lower than 0.016%.

Fatality Rates Media v Reality

This is all great news for the restart of the economy. It debunks that insane narrative that nobody will even want to go back to work or go out to the shopping malls because you will die with a probability of 3.4%. Nonsense! Be careful and wear a facemask and minimize the chance of even contracting the virus and then even if you get the virus there’s a minuscule 0.016% chance of dying from it. All that sounds much better than the media’s narrative of “3.4% of the people who catch it will die”

In fact, if you take your motorcycle to work or to the mall then you should be 3.5-times more scared of dying in a traffic accident (0.059%) than COVID-19 (0.016%)!

3: Most countries were able to “bend the curve” already

A while ago, I shared some thoughts on the state of the world, including some charts. Well, we now have more data on cases and fatalities and I thought I’ll post an update on some of the charts. Because I added more countries and I like to automate the process I switched from Excel to Octave (a free version of Matlab). So I updated the charts from a while ago and also added a few additional countries, 26 total:

  • North America: Canada, USA
  • South America: Brazil
  • Europe: Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Poland, Portugal, Spain, Sweden, Switzerland, UK
  • Asia-Pacific (incl. Middle East): Australia, Israel, South Korea, Taiwan

Notably absent: China and Iran because I don’t trust their data.

Since I can post only a small selection of charts here, I’ve set up a Google Drive folder if you want to check the other charts not displayed here:

Google Drive Shared Folder: “Big ERN’s COVID-19 Charts”

All charts are copyrighted but, of course, just like everything I publish here they are totally legal to use and reuse for free as long as the source is properly acknowledged.

Let’s start with the U.S. up to April 13:

  • In the top panel, we notice how the growth rate has already slowed down. In the chart with a log-scale on the y-axis, you see that as a flattening of the curve.
  • In the middle and bottom panels, I also plot the daily new cases and deaths (this time on a linear scale, not in logs) and I also include a Hodrick-Prescott-filtered series (lambda=10) to smooth out some of the daily noise and potentially weekly seasonality apparent in some of the places. Both new cases and new deaths have certainly slowed down over the weekend.
Total cumulative cases and deaths (top), new cases (middle) and deaths (bottom) for the USA. Notice, the top chart has a log-y-scale, the other two are linear scales. Source:

Other countries are already ahead of the U.S. For example, Germany seems to be over the hump. New cases peaked in late March already and have since trended down from close to 6,000 to under 3,000. Deaths are lagging behind a bit (as expected) but even they are probably at or even beyond the peak.

Total cumulative cases and deaths (top), new cases (middle) and deaths (bottom) for Germany. Notice, the top chart has a log-y-scale, the other two are linear scales. Source:

Germany’s small neighbor to the South, Austria, is in even better shape. New confirmed cases have been plummeting since the March 26 peak. Austria has also just opened schools and small shops again and plans to open more businesses, including shopping malls in May.

Total cumulative cases and deaths (top), new cases (middle) and deaths (bottom) for Austria. Notice, the top chart has a log-y-scale, the other two are linear scales. Source:

I have a lot more charts but can’t display them all. See the charts I posted on my Google Drive with the name “C19_Summary_XYZ”” where XYZ is the 3-letter country code.

From just eyeballing the country charts I would put the 26 countries in the following bins:

  • Significantly past the hump: Australia, Iceland, South Korea, Spain, Taiwan, Italy
  • New cases definitely declining, deaths still near a peak: Austria, Switzerland, Czechia, Denmark, Finland, Germany, Luxembourg, Norway
  • New cases near the top, maybe slightly declining: Canada, France, Israel, Poland, Portugal, Sweden (hard to tell – too much daily volatility, potentially even weekly seasonality), USA
  • New cases still rising (though no longer exponentially): Belgium, Brazil, Ireland, Netherlands, UK
  • I tried to be conservative; I could have put some of them in a higher category…

4: Even countries with no draconian shutdown measures bent their curves!

OK, critics may argue that as soon as we loosen the shutdown restrictions, the cases and deaths will revert back to geometric growth and the situation will get out of control again. Overcrowded hospitals, dead people lying in the streets, you name it. We shall see how it works out in Austria and Denmark. But I doubt we even have to wait that long because there are at least two countries without a full-scale shutdown to watch carefully:

Sweden opted to just let the virus do its thing while keeping businesses mostly open. Apart from prohibiting gatherings with more than 50 people, the government only gives recommendations to residents. So far at least that has not had any outrageously negative effects on cases and/or deaths. Sweden’s curve has certainly flattened and daily new cases and deaths look like they are at or even slightly past the peak. Sweden’s government has so far resisted calls to shut down the economy. A lot of doubter’s arguments, e.g. in Forbes’ “Sweden Continues With Controversial Coronavirus Strategy: Is It A Big Mistake?” sound less like a well-reasoned argument and more like an appeal “hey, we shot ourselves in the foot with our complete economic shutdown, so it would be nice if Sweden could do the same – shared misery is half the misery, you know?!”

Total cumulative cases and deaths (top), new cases (middle) and deaths (bottom) for Sweden. Notice, the top chart has a log-y-scale, the other two are linear scales. Source:

Update (4/14/2020, 4pm PDT):

As some people pointed out, Sweden may not exactly be over the hump (yet). There’s certainly a weekly seasonality pattern here. Cases and deaths usually jump Tuesdays to Thursdays. I updated the chart and included a 7-day rolling average in the chart for Sweden, see below. New cases are only flattening if you look at the 7-day rolling average. Deaths are bending down just slightly. It’s too early to call that a victory, but the situation is certainly not getting out of control in Sweden either. I will monitor this carefully!


Update 4/28/2020: The brain behind the Swedish policy, Prof. Johan Giesecke:


Iceland is another country without too much of heavyhanded government response. As mentioned above, it has one of the highest numbers of confirmed cases as a percentage of the population (the second-highest in my 26-country sample, after Luxembourg). But that’s due to testing people at a rate 5-10 times higher than in other countries. But new cases and deaths have slowed down significantly. Iceland is open for business, even welcoming tourists – if you can get there. And to the shutdown-fetishists’ chagrin, people are not dying in the streets: 8 fatalities so far. Out of 1,700+ total confirmed cases, 993 have recovered and out of the 770 remaining active cases, 10 are critical (Source:, accessed on 4/13). So, the hospitals are not exactly overflowing either. It’s a great success story and it should be celebrated. Wouldn’t that deserve at least a 1-hour special on CNN?

Total cumulative cases and deaths (top), new cases (middle) and deaths (bottom) for Iceland. Notice, the top chart has a log-y-scale, the other two are linear scales. Source:

So, both Iceland and Sweden were able to flatten the curve without flattening (killing?) the economy.  If and when we loosen our shutdown measures there is a good chance that the situation will not get completely out of control from there if we follow some of the same less-restrictive prescriptions from those two Scandinavian pioneers: be reasonable, keep your distance, but go about your business as usual in every other way. If the rest of the world can get off their butt and acknowledge that Sweden and Iceland probably succeeded then there’s hope for a quick recovery!

By the way, as I was writing this I found this very nice 60 Minutes Australia episode on YouTube talking about some of the issues that are on my mind. It was the best and most rational reporting I’ve seen so far. Why don’t we have journalists like that in the U.S.?

  • At the 5.45 mark: Prof. Bjorn Lomborg about the difficult tradeoffs: “Going too hard on lockdown is a strategy that will come back to bite us in the long-run”
  • At the 8:35 mark: The Swedish experience. Looks a lot nicer than the Australian “police state” right now.
  • At the 13:36 mark: Infectious disease modeler Prof. Emma McBride on the benefit of letting the disease spread in “low-risk” demographic groups. They must have looked at the same data as I did here!

Update 4/15/2020: I should mention that I never intend to create the impression that Sweden and Iceland will go through this episode completely unscathed. Quite the opposite, they will likely also experience a recession, just like the rest of the world. But it will likely be a lot less severe than in the countries that completely shut down for an extended time.

5: U.S. States Data

I can also apply the same analysis to the 50 U.S. States plus D.C. The data come from the New York Times GitHub page.

Again, I can’t post all 51 charts here so I keep the state-level data in my Google Drive folder for your enjoyment. But here are a few observations: Within the U.S., we have just as much, probably even more variation than across countries in my 26-country sample. On the one hand, you have NY and NJ with more cases per population than even Iceland and Luxembourg. On the other hand, you have dozens of states with lower infection rates than even some of the less-impacted countries. (though no U.S. state can beat Taiwan).



The same format, but this time I plot the fatalities per 1 million residents, see the chart below. Again, NY State tops even Spain and Italy (though I doubt that NY will top the Lombardy region of Italy). But there are plenty of states with even lower fatality rates than some of the less-impacted countries. That’s really good news!


Critics will certainly argue that all the states with lower infection rates are just waiting to become the next New York State. But I doubt that! Here are some time series charts of cases per 1 million residents. I can’t fit them all in one chart, so I bunched them together by groups of 10 to 11 states. Here are the 31 states with the highest case rates so far, see the three charts below. All states are bending their curves already. I doubt that any other state will reach the NY state 10,000 cases per million residents anytime soon! (well, South Dakota is still in exponential growth but from a very low level right now, so let’s monitor that!!!).

Part 1: Cases per 1m residents. The 11 highest-ranking states.
Part 2: Cases per 1m residents. States ranked 12th the 21st.

Also, notice how the nation’s largest state by population, California, is only ranked 31st among the 51 in cases per capita. And the curve there is bending really nicely:

Part 3: Cases per 1m residents. States ranked 22nd to 31st.

Some of the states that were floated as “The Next New York” such as Michigan, Louisiana and Florida also don’t look like they are going down the tubes. Here’s Michigan: new cases peaked in early April, deaths are probably near the peak or even declining.

Michigan: new cases reached a peak in early April. Deaths might have peaked or are close to peaking.

Here’s Florida: daily new cases seemed to have peaked:

Florida: new cases started a slight downward trend. Deaths might also be close to the peak.

And here’s Louisiana:  daily new cases have plummeted, deaths are also probably past their peak already:

Louisiana: it looks like cases peaked in early April. Deaths could be at or even beyond the peak.

So, to the doom and gloom fetishists’ great disappointment, MI, FL and LA will not be another New York State and certainly not a New York City. Thank God for that!

Also, back to California: new cases have already started to bend down. Deaths are probably near their peak:

New cases in California peaked in early April. New deaths look like they might be leveling off.

And finally, here’s my home state of Washington, the state with the first confirmed case and first confirmed fatality: No more out of control exponential growth (see top panel), cases have slowed down and even deaths seem to be on a downward trajectory:

Washington State: new cases are now declining. Deaths have also dropped.


So, overall, these are hopeful developments from the COVID-19 front! No wonder the stock market is holding up better than during the Great Depression. COVID-19 will not wipe out civilization, that was clear from the beginning. But investors have become hopeful that it will not wipe out the economy either!

Update 4/27/2020: The academic blessing of opening up again also came from one of the most impressive researchers in the field, Dr. John Ioannidis from Stanford University:

And this here is the full interview (more than 1 hour, but worth every second!!!)


A caveat…

Of course, here’s a scenario that could completely derail the stock market’s optimistic outlook: politicians and unelected officials are unwilling/unable to admit that we might have somewhat overreacted during the whole shutdown. Nobody should feel ashamed because back in March we had no idea how bad the virus will turn out. The shutdown was 100% defensible back then and I certainly had my worries about Sweden and Iceland (and thus kudos to them for serving as the guinea-pigs for the rest of us).

But if people are not willing to admit that it’s about time to relax the economic shutdown soon, out of some misguided bureaucratic pi$$ing contest – excuse my language – we’re in for a lot of trouble. Expect the stock market to completely tank to 2009 levels (-55%) and maybe even to 1932 levels (-80%) in that case. 

Sorry for ranting!

Update 4/28/2020: I’ve decided to close the comments for this post, the first time I ever had to do that. Nice discussion, but I realize that some people disagree with me and there’s not going to be common ground, ever. Let’s just move on to different topics…

Title Picture Credit:


211 thoughts on “COVID-19: Some Empirical Observations and Reasons for Optimism

  1. Sorry Karsten but I disagree.

    1) Bending the curve is meaningless since the preliminary data indicates many of the infected do not have sufficient antibodies after recovery. So they could not only be reinfected but it also reduces the possibility of vaccine efficacy.

    2) Korea is seeing cases where “cured” individuals have virus reactivating; so there is a lot we do not yet know about the virus.

    3) Higher survival rate sounds good until you realize most of the infected have permanent damage. Look at their chest X-rays. Most wont lead a normal life.

    4) We could have a second more deadly wave in fall (like the Spanish flu). TBD based on data.

    Every medical paper I read and discuss with my virologist, RN and ER friends; makes me nervous. And I consider myself prepared enough that I wrote several blog posts including getting a O2 machine at home.

    The only reason the market rallied because the Fed extended their charter; created a SPV to buy junk bonds in conjunction with the Treasury. Having the Congress, Fed, Treasury and POTUS focusing only on stock markets might mean higher prices for now. ECB bailing out Italy and Spain again helps as well.

    TLDR : Markets higher only due to CB puts. Virus is still deadly

    1. “most of the infected have permanent damage.” False. You think most of the asymptomatic cases have permanent damage? *Perhaps* most of the hospitalized cases have permanent damage, but even that is very questionable at this point.

        1. Actually the data out of Hong Kong showed 25% of post virus cases had permanent lung damage to the extent they could not walk fast and climbing stairs was out of the question. In addition there is an extremely high incidence of post virus folks needing dialysis for kidney failure. So much dialysis there is a shortage of sterile fluids. There is also an incidence of neurological problems especially in the cranial nerves. Hong Kong managed this virus almost perfectly from a non pharmacologic perspective.

          IMHO this virus is not a natural virus but engineered. It hits way too many systems. So far lungs liver kidney heart brain blood gut and immunity. Too many people are presuming convalescence will restore people to full health. With that many systems involved there will be wide spread chronic disability.

          Restrictions were lifted in Singapore and Wuhan and both places are once again 100% locked down. I don’t care how many graphs you post, the only thing that matters is herd immunity and that requires 80% of the people to have been infected PERIOD. That process will take minimum 2 years unless there is a vaccine. If restrictions are lifted here in 3-6 weeks we will be locked down once again. On a per 1M population basis the administration has actually done a pretty good job of flattening the curve. The US is way down the list in terms of infections/1M population

          1. The side effects are scary. But again: 25% of what? People in ICU or all infected? Well, it can’t be all infected because apparently 90+% of the infected are completely asymptomaticc, see WSJ article:

            The quarantine on/off/on measures in Wuhan show that the current government approach isn’t working. We don’t protect old people enough and young folks (who might eventually all get it!!!) might as well go about their day as usual.

          2. “IMHO”? And on what educated basis do you have that opinion please? To the best of my knowledge the experts in genetics, epidemiology, bioinformatics do not share your opinion. Nor would William Ockham.


            The authors of The Lancet statement note that scientists from several countries who have studied SARS-CoV-2 “overwhelmingly conclude that this coronavirus originated in wildlife,” just like many other viruses that have recently emerged in humans. “Conspiracy theories do nothing but create fear, rumours, and prejudice that jeopardise our global collaboration in the fight against this virus,” the statement says.

            1. The question is: could this virus have been escaped from the Wuhan Lab? It’s enough if a natural virus was being studied at that lab, then jumped over from bats to humans due to shoddy lab procedures and the rest is history.

              The #1 indicator that people are lying? They’re answering a question I haven’t asked. I didn’t ask “was the virus engineered?” but simply “did the virus escape from the Wuhan Lab?” and I think that there is circumstantial evidence for that.

              1. Yes, but need to have that circumstantial evidence discussion over a beer sometime. Whether that happened or not does not change the behavior of the Chinese gov that occurred in Wuhan after Dr. Li Wenliang spoke up about this disease. They have done far worse for far less provocation. Those exotic meat markets are filthy enough and dangerous enough to be a sufficient explanation, but yes, it is impossible to prove a negative (in regards to the lab you refer to). Hence, this discussion is best done with the clarity of some hops.

    2. I agree with points above except I think those with more permanent damage are likely among those hospitalized.

      Also, Karsten its not just the death rate denominator that is off (I.e those with mild symptoms not included) – the numerator (deaths from covid) is also understated. Why aren’t you trying to adjust for that as well – by cherry picking and saying denominator is off but not also try to adjust for the numerator as well it seems incomplete. WSJ ran recent article on severely understated deaths from COVID in Lombardi region.

      And you can reopen economy – doesn’t mean people will want to run out and be in a crowd. I can tell you my family of 5 won’t be going on subway, to restaurant, stadiums, plane,cruise, etc anytime soon. And how do you reopen schools without putting staff and those with immune issues in harms way? 50 dead educators from Covid in NYC already. Do you see elementary school kids or severe special needs kids wearing masks and social distancing effectively? Do you see parents not sending kids in sick as they do now (parents treat like daycare and don’t want to miss work)? Do you tell those over 60 they can’t teach?

      Iceland is one of the most sparsely populated countries in the world (a little over 3 people per sq km- over 10x less than US). A lot easier to social distance when there are few people around. Not a great example for rest of world.

      1. I notice Karsten has yet to reply to your comment- so I will echo the key question– why only look at the mistakes with the denominator?
        Forbes published an analysis of all deaths in Italy, Portugal, UK and France and suggested that the deaths from Covid 19 are likely double the reported numbers (
        I will add a question- Re Sweden- They have twice the number of cases per 1Million compared to Finland and 10 x the number of deaths per 1Million. Sweden has similar case rates as Denmark and Norway, and yet again multiples higher death per 1Million, 2-4 x higher in fact. That requires some ‘splainin’..

        1. It’s a serious problem, for sure. For example USA and Switzerland are on a similar trend in deaths per 1m residents (USA just a few days behind, due to later takeoff).
          It’s really hard to understand how Switzerland, Germany and Austria with very similar health systems, similar demographics, similar development levels, similar culture, etc. can have very different mortality rates. It’s possible that there are different methods for counting deaths.

          I’d wager that in the U.S. we likely overestimate the deaths because there’s an incentive for hospitals to inflate their numbers. There was a WSJ op-ed where an MD insinuated that.

      2. I most definitely think that the deaths from COVID-19 are mismanaged in most countries. For exmaple, in France they initially underestimated the deaths because only people dying in hospitals were counted. Then they came across a ton of additional deaths and thus fatality stats rose a lot in April.
        In the U.S. I suspect that deaths are overestimated. Anybody who dies from any cause will count as a COVID-19 death if the test is positive. Heart attack, stroke, traffic accident, etc.

        1. I would be shocked if the deaths in the US are OVERestimated. First of all the story that is peddling that narrative is from highly questionable sources. But there is nothing like data– and death is the best data point as there is no judgement involved in counting the dead when it comes to totals,
          So far -to April 16- there have been 569403 deaths from all causes. Barring huge seasonal differences in death rates given the death all causes for a year over 206-2018 and prorated for the first 105 days of the year one would have expected deaths to be around 815,000. So when there are Covid areas reporting an overall increase in their total deaths that and the increase falls well shy of that towns usual deaths, that is a very worrisome number suggesting under counting, not over.

          NVSS – Mortality Data

          1. Sorry typos made that above a jumble…
            the overall deaths expected refers to data from 2016-2018– and the point is that nationwide total deaths this year are WAAAY down– around 2000 fewer a day for the year so far…so when NYC or London, or any place with a spike in COVID19 reports they have MORE death than years past, it is likely the difference is even bigger than the guess, because their deaths should be down as they are for the whole country.

          2. For the US aggregate we’ve had FEWER deaths total, while for NYC and NY State we had more deaths than normal.
            So, the impression I’m getting here is that we may have undercounted the true COVID-deaths in NYC. The point Andrew made.
            But there are probably a lot of deaths in the REST OF THE COUNTRY where it’s the opposite.

        2. “Anybody who dies from any cause will count as a COVID-19 death if the test is positive. Heart attack, stroke, traffic accident, etc. ”

          A person testing positive for Covid-19 virus admitted for some other primary diagnosis resulting in death would not likely have Covid-19 listed as the primary cause of death. If the patient has significant Covid-19 signs on admission or after admission and those symptoms contributed significantly to deterioration and death… yes, it will be listed as THE cause of death, or a significant contribution to the death of that patient. Sometimes it is not all that clear cut, because there may be several contribution causes of death…all of which should be listed.

          The known complications of Covid-19 (and there are many) only count if they occur.
          Filling out a death certificate requires good medical judgement, and I don’t think Mds have a any real incentive to not be honest in filling one out. (I could not access the WSJ article, since I do not subscribe.)

          1. “A person testing positive for Covid-19 virus admitted for some other primary diagnosis resulting in death would not likely have Covid-19 listed as the primary cause of death.”
            That’s what I would hope. But I suspect there’s a monetary incentive to classify a death as COVID and that could sway doctors to go one way in borderline cases. I certainly HOPE it isn’t so, but I’ve heard rumors…
            Also, there have been deaths in NYC that are being treated as “presumed COVID” without any test where people died alone at home. I’d be surprised if there isn’t some over-counting there.

    3. These cold statistics(with an obvious political bent)miss so many important issues, that I as a physician am aware of. Much of what is written here is hogwash-yeah, maybe the numbers make sense, but we aren’t living in a lab. It’s not like A-asmptomatic, B-infected, C-dead . All those tired arguments about MVAs and flu are just BS. There are plenty of young people spending weeks in ICUs, some with no pre-exisitng conditions, and many others who are not old, but have conditions compromising their immunity. Everyone with half a brain knew that there would be more zoonotic pandemics, and not everyone is in a luxurious silo, or out in the woods. The ineptitude, denial and scapegoating by Dear Leader cannot be wished away by his apologists.

      1. Show me the stats of how many young people are in the ICU, as % of their population. Right now your own comment is guilty of what you’re accusing me of: hogwash, anecdotes over data, ineptitude, denial, scapegoating. Ever heard of projection?

        1. (apologies in advance for the snarky tone, but I’ll respond in kind to ERN’s comment!)

          Hey Karsten, ever heard of garbage-in, garbage-out???

            1. Your so-called data analysis relies on numbers we know, with near certainty, are not accurate. And these data are not point estimates that could be wrong in either direction— they are biased TOO LOW. We know death counts are too low due to lack of testing.

              If your analysis were neutral and based on accurate data it would not reveal so much about your politics.

              1. It’s not a “so-called” data analysis. It’s data analysis. Take it from someone who has a PhD in the field an who’s worked in this field all his life: data analysis is always subject to uncertainties and limitations. That doesn’t mean we can throw up our hands and give up and ignore the data.

    4. Vaccines use adjuvants (like aluminum) to confer a larger immunologic response than even a naturally acquired infections, thus leading to greater chance of immunity and thus for the herd as well.

  2. I liked the comparison to motorbike deaths. Last year, in the UK, we were apparently in the grip of a knife crime “epidemic” with daily figures of the deaths so far that year from knife crime. My daughter and her friend were going by train to London and the friends mum said I am going to drive them to the concert because of all this knife crime. I replied she was far more likely to kill them in the car ride than the minuscule tiny risk of them being knifed. Strangely knife crime is not making the news this year!

    There is a problem here with the data. Most of it is total garbage!
    On deaths: The most important statistic is how many people are dying in total from all causes. All we get is how many people are dying with some (possibly tenuous link) to Covid 19 (the definition of which may vary from country to country.) It is certain that some of the deaths would have happened anyway in short order and Covid 19 either slightly accelerated this or actually had no impact at all. Without the comparison to how many people usually die each day in March/April the number of deaths is actually pretty useless.
    On number of cases: On top of the issue of how many people are being tested which varies from country to country not all tests are created equal. In the UK we keep rejecting tests we have bought because they do not work. You get too many false positives and false negatives. If our tests are not good enough then how reliable are tests elsewhere? And if that is the case it is impossible to know the true mortality rate of this thing.

    The thing I do not quite get is the bending of the curve. Sure if you stop doing stuff the curve will bend. Even in Sweden they have stopped doing stuff. But when you restart doing stuff then why will the curve not simply start to unbend. Are we, instead of having one big lockdown, in for a series of mini lockdowns?

    1. NY Times published an article on deaths from all causes this year compared to last year.

      3350 official Covid deaths at the time of writing the artlce, but 5330 more deaths than expected.

      “These numbers contradict the notion that many people who are dying from the new virus would have died shortly anyway. And they suggest that the current coronavirus death figures understate the real toll of the virus, either because of undercounting of coronavirus deaths, increases in deaths that are normally preventable, or both.”

      1. Thanks. That is far more useful information. New York and London are certainly hotspots for this virus.

        1. Judging from the timeseries in the article the stdev is more like 200 than 2000

          1. You can’t judge the standard deviation of “deaths from all causes” from any number displayed in my article. My guess is that +/-2000 deaths is the normal variation of different flu seasons alone, not counting all the other causes.

            1. I think It’s pretty clear from the second plot that the stdev is no where close to 2000.

              1. OK, let’s explain my point again:
                The hypothesis is that “all deaths” has increased by 5330, but COVID accounted for only 3350. This lead the original commenter to the conclusion:
                “suggest that the current coronavirus death figures understate the real toll of the virus, either because of undercounting of coronavirus deaths, increases in deaths that are normally preventable, or both”

                And I’m saying that statement cannot be made unless you show me that the extra 1980 non-corona deaths are significant. You can only determine that if you show me the figures of “all deaths” over the 1/1-current period over the last few calendar years and relative to that variation, the 1980 is significant.
                The daily COVID-19 fatality standard deviation has nothing to do with the number I’m after.

                Hope this was clear enough.

                1. Seems highly unlikely that the NON COVID deaths in NYC are way up this year when it appears total deaths-all causes- for the whole USA appear to be way down for the year.

                2. It is very clear that you are not understanding my point!

                  I agree that the faily stdev of corona deaths is irrelevant.

                  I am looking at the second plot in the nyt article referenced by the previous commentor (which seems like the natural place to look to see if these claims are statistically significant since it’s the subject of the article!?) If you look at the second plot in that article its very clear that the 1980 excess deaths is very significant (tstat is probably more than 10). In the last 20 years, only two months have an excess or deficit of more than 300, sep 11th and the most recent one.

                  It doesn’t prove that corona is the cause of the excess deaths but it’s clearly not a statistical fluctuation.

      1. I wish I was as optimistic about people. On my last trip to a grocery store, there was maybe 2 masks wearers for ~40 people I ran into. No staff wore masks, gloves, or goggles. Half the country is in deep denial.

          1. “harmless?’ That is as least as irresponsible as the death death death cries. Italy- particularly in the northern part -clearly shows that this is far from “harmless.”

              1. It is also possible (perhaps likely) that the 80 year olds in USA nursing homes are being cared for by 28 year olds who are over represented as shoppers at Walmart vs for example investment bankers, or architectural engineers), and that Walmart shoppers (and checkout persons) who have Sars-2 are mostly asymptomatic…and not wearing masks. It would be reasonable to assume that the people who do most of the care at nursing homes preferentially shop at Walmart vs Tiffany’s… or even Trader Joes. We do need adequate epidemiological studies to make rational decisions as to when, and how, to start opening up the public business economy again, as to how to proceed. This can and should be done… with statistically sound studies. Germany is in the process of doing something along that lines… in the USA … we have squandered our early opportunities to do so…and need to start getting serious about making headway on such a study. With a serious attempt to do surveillance and find out what our R(0) numbers are… and seeing how it changes as we begin to get people back to work… judiciously, while monitoring how that affects this pandemic so we can adapt as needed. There is a possibility that we can do that very effectively (and fruitfully), but we need to get the data that allows up to know when and how to do that MOST effectively… saving the most lives at the least possible cost.

                These studies need to be properly designed. We need to be careful about making big decisions in regards to “opening up” based on the current data coming from Sweden. Their case fatality rate of over 12% suggests that there is a lot happening there that we do not wish to emulate.





        1. Yeah, right. My answer is the same:
          “Another reason to follow Sweden’s example. If you’re waiting for the virus to completely disappear and never return and never return from people from other countries you have to shut down your economy for the next 5 years.
          Let the virus spread slowly!”

          1. We will probably have a vaccine well before then, or at least a better understanding of how to treat the virus in the future (For example, we are realizing now that ventilators may be making things worse for many patients).

            I do think that aggressive testing + contact tracing + better public awareness of personal hygiene (masks, hand washing, maintaining personal space, many people working from home) can likely keep the virus at bay with much less cost than the current quarantine measures. It feels premature to give up and go for the herd immunity strategy


            Is Sweden really doing things much differently than other countries in Europe or the USA? This Swedish journalist seems to think not… but the article is not sufficiently addressing the nuances to make that determination. Classes for grade and hs students are canceled, sports stadiums are closed, people are being asked to “socially distance” themselves, people who can work at home are asked to do so. One can go to a restaurant, but cannot buy a drink or eat at the bar itself. Who knows how far apart the tables are, right? Testing for virus, testing for antibodies at the appropriate times, and tracking down contacts for quarantining can make a huge difference in the epidemiological outcomes. Sweden claims to have a more trusting relationship with their governmental agencies than some parts of our USA population, where individualism/”don’t tread on me” attitudes in some components of the population don’t seem willing to look at the big picture or accept that this covid-19 is not a hoax, nor is it benign.

            Should we try to get people back to work? Yes… with careful monitoring and a robust system to track and isolate contacts of infected people. This requires reliable testing protocols for both active infection (virus), and for evidence of immunity/recovery post infection (anti-body testing), a great many more people to follow up on contacts and quarantine of the infected persons.

            Getting back to work without causing a huge spike in hospitalizations does require a very robust and well funded public health service, retention of the experts in epidemiology and public health infrastructure and yes, a federal government that is willing to take the advice of epidemiological experts whether from the USA or elsewhere. If we gear up properly we can judiciously “open” up again if all of us cooperate. If we do not gear up for doing so, we pay the same price (or more) we have already paid for having trashed our public health infrastructure.

            1. Not all schools are closed:
              “elementary schools and middle schools are all open.”
              High schools and universities are working remotely.
              Denmark opened schools again as well.

              I also disagree with your conditions for reopening. By that standard, we’ll never reopen. At the current rate of testing 300k people a day it will take 1000+ days to test everyone. And after 14 days, your negative test result is likely moot again.
              Other countries (Denmark, Austria, Germany) are now slowly reopening without that crazy “evidence of immunity” idea.
              So, I think that the people who propose this whole Immunity Certificate B.S. (Bill Gates, etc.) haven’t really thought through the thing. We shouldn’t listen to people like that.

          3. From the Johns Hopkins map site today….

   (For reference.. not my favorite medical journal)


            Total Confirmed

            Total Deaths

            Total Recovered

            case-fatality ratio: 12.25%

            I did not try to confirm these numbers anywhere else… but … why would their case fatality rate be so high?
            Is there something wrong with this data or it is explained by something in Sweden. Maybe they are only testing and hospitalizing extremely ill people or are they simply doing far less testing than the others? Oregon and Washington’s rates were already thought to be due to testing being limited to people with illness and the associated symptoms consistant with Covid-19

            Case fatality in Germany on the JH site: 3.70%
            In Oregon: 3.95%
            Washington: 5.58%
            New York: 7.86%

            1. Good point! My theory is that Sweden doesn’t test very aggressively and will likely only capture those who are ill enough to require medical care.
              Also, I haven’t seen any results of random antibody testing (similar to the Santa Clara or Gangelt/Germany studies) for Sweden yet. I wouldn’t be surprised if Sweden already has 2x the spread of the virus in the overall population (mostly asymptomatic). It’s the declared target of the policymakers there, right?
              So, with 2x spread and much less testing per resident it would explain the 3x or 2x case fatality rate.

            1. True, though death is a lagging indicator (adding up time in incubation period if they were tested before developing symptoms, the early symptomatic period, and hospitalized days before death), and over 90% of the total cases in Singapore are from April.

      2. “People will be much more cautious than before.” I wouldn’t be so sure. I live in Brooklyn and today, 4/23, when we are required to wear a face covering when we can’t distance 6+ feet, I see ~50% of people wearing masks. None of the sidewalks are more than 6 feet wide, and there are a lot of people here — if you’re out in public, you cannot practically always be 6+ feet from people. And yet half of people are not wearing masks. There is no city in the US harder hit than NYC, I am a couple blocks from a hospital and hear the sirens all day…still not enough masks.

        1. I’d like my state to reopen. I go for a walk every day and encounter other people doing the same. We are actually smart enough to stay far enough away from each other.
          If people in Brooklyn want to be locked inside their apartments until the sidewalks are widened, I with you best of luck. It’s a federal system and every region has to decide for itself. But please don’t ask for a federal bailout.

        1. I’ve seen the footage from the newly opened places. They seem to be really cautious (face shields, face masks, temp check at the entrance) so I think it’s a worthwhile experiment.

          It’s what the the “hammer and the dance” paper proposed. We’ve reached the linear growth in Georgia, now it’s time to slowly relax.

  3. A very good read. Thanks. Please do keep producing these graphics; the next two or three weeks are going to be very interesting and will hopefully confirm that we have, globally, turned an important corner. Hopefully re-infection won’t be a big factor, the virus won’t evolve before we have a vaccine and releasing lockdown won’t provoke a surge in new cases. As has been said, there is a lot we don’t know.

    I agree with your comments about needing to protect the old since they are by far the most vulnerable. We need to value them (I include myself in this) since they have striven all their lives to add value. However, they are not adding value now having retired. That is, I think, another (admittedly, probably small) part of the calculation the stock markets may be making. It may be cold hearted but the fact is that the unproductive will be hit harder by this crisis than the productive so, when the lockdown is relieved, the economy will spring back with less unproductive mouths to feed. Its horrible logic but since when did the markets ever rely on sentiment?

    What is even more sinister is that, as usual, the crisis is hitting the poor more than the rich. That is unforgivable. The UK is behind the pace of recovery from the outbreak (as your analysis shows) but I am hoping that the way the crisis has highlighted here our reliance on underpaid ‘key workers’ in health, care, food processing and home delivery will be responded to when the crisis has passed. I also hope that the massive state intervention in providing support for unemployed and underpaid is built upon not rolled back.

    One concern with the numbers (building on the previous comment from PJ): on deaths the issue in the UK (and its similar elsewhere I think) is that only deaths in UK hospitals are counted in the week they occur. There are also the distorting factors of number and quality of tests in each country PJ mentions.

    1. I haven’t seen hard data on poor vs. rich. My personal suspicion is that this hits rich old people in rich states (NY, NJ, MA) more than older folks in other states.
      But we’ll need better data to answer that conclusively.

    1. I’ve also seen R0 values as low as 2.5 for this virus. Still much higher than the flu but not as bad as 5.7.

    2. R0 is not a fixed constant. It was 5.7 when on March 10, Bill De Blasio told everyone to go out and about and not worry about a virus that’s “only as bad as a common cold” (his words, not mine)
      People’s behavior has changed since then!
      Look at Sweden, their cases don’t grow like a R0=5.7 anymore. All without a shutdown

      1. Sweden
        Total Confirmed
        Total Deaths
        Total Recovered

        case-fatality ratio: 12.25%

        Sweden’s case fatality ratio suggests to me that they are grossly underestimating their total cases. This data is from the Johns Hopkins site, as of yesterday. Sweden has a good hospital system, and it seems unlikely that their ability to treat ill people is less than ours, or less than Germany’s. IF that is true, their denominator may be 2-3x less than they are reporting.

  4. I sure hope you are correct to be optimistic! We are following Sweden closely as my wife is from there and her parents luckily got on the last flight home before the borders closed. If you extrapolate out their cases and fatalities for the population they are doing worse than the USA, and they are testing less. Since they are behind the USA and Italy by 2-3 weeks the jury is definitely still out.

    I agree we have to start the engine up again soon, what I worry about is the second wave after everyone goes back to work. There were only 15 cases in the USA 2 months ago and we couldn’t contain this, it’s crazy to think the curve will just go to zero before there is herd immunity or a vaccine. Oh and it’s been proven that some can get reinfected again.

    So yep open it up and let it spread, or enter the Great Depression II by staying in lockdown. If the death rate is as low as the data outlined suggests then only 200,000 will die. It gets a little ugly though if the rate is closer to 3%. I’d be petrified if I was in the high risk category, I mean would you get on a airplane if it had a 15/1000 chance of crashing? That’s what the high risk people will be doing every time they go out in public until there is a vaccine.

    1. I hope you’re wrong. Sweden is certainly not 2-3 weeks behind the U.S. Sweden had a very similar path and takeoff point and bend point as most other EU countries (CHE, FRA, AUT, GER), much earlier than U.S.

  5. Hi,

    I agree with you.

    No need to create totalitarian states and cause a economic depression of global proportions to fight this disease.

    It is a severe virosis, but the treatment is causing extreme side effects.


  6. Thank you, Karsten, for your unfailingly thorough and thoughtful treatment of available data.

    Might you be willing to apply your methods to the question of what would the USA (country and state level) case and death projections be given the current course (continued lockdown) vs Sweden’s approach (high-risk individuals isolate while others carry-on with common sense precautions)? The resulting projections would be useful for estimating what a second wave (post reopening) might look like. It would also permit constructive conversation about the efficacy of the lockdown when weighed against the human costs of a recession or depression.

    Rudimentary calculations using IHME data as of April 10 for COVID-19 deaths and Worldometer population data suggested that Sweden’s estimated wave-one death rate (0.1314% would be 7x that of the US wave-one death rate (0.0186%). These estimates fail to incorporate real-world issues like a wave-two or differences in age demographics and population density. Nevertheless, I am curious to see how BigERN could improve our knowledge base in considering these questions.

    Keep up the fantastic work, my friend!

    1. I’m not running my own projections. That’s mostly guessing anyway. I am perfectly fine looking at past data.
      I think that the IHME projections will be as wrong as they were in the past.

  7. Nice work to compile all of this data in one place. One caveat with the “let it rip” in the “younger” age group is that although the death rate is low for the 18-49 year old group, they are still 30% of the hospitalizations so you still risk overwhelming the health care resources.
    Boris Johnson, although he is 55, is a good example – was in intensive care for 3 days needing supplemental oxygen, but ended up not dying. Also as alluded to by “Financial Freedom Countdown” there is the possibility of a medium term effect on lung function which has been observed after the 2003 SARS outbreak though it may resolve in the longer term –

    Though I consider myself to be young (55) and healthy, I’m definitely not signing up to get COVID-19.

    1. That’s a concern. Will a wider spread overwhelm the hospitals. Well, again, we can monitor Iceland and Sweden. Some of the models assume countries without a shutdown will keep their R0=5.7 as before. But Sweden/Iceland show that by just being cautious we likely push the R0 close enough to 1 or even below.

  8. The choice to shut down or not is not exogenous– a place is going to be more likely to not shut down if they have enough protective and medical equipment (and ideally enough tests, both for current infections and antibodies). They can handle more cases without them resulting in preventable deaths. When a country does not, then shutting down is the only option to decrease the death rate. Correlation is not causation.

    1. Great point highlighting the danger of extrapolating data from small homogenous populations like Sweden/Iceland and applying to places like the US!

    2. Possible. Just like Sweden, we now have a lot of excess capacity in the health system, even in NYC: Javits Center and the Comfort are basically empty. So, let’s not be exogenous, but let’s open the U.S. economy again. Perfectly endogenously! 🙂

  9. Counterpoint on the doom n gloom: if we didn’t have doom and gloom from the media no one would have listened and taken this seriously! Perhaps me included… Our collective staying-at-home and preventive measures and precautions all led to the curve flattening that we are seeing and enjoying now.

    Though I like your analysis because I also wonder when have we passed the hardest point and when can we move on to the steady state part of the curve where we try to keep severity and deaths to a minimum while also living life (no way we can survive 18+ months of today’s lock downs IMHO). Austria, Sweden, etc will be good canaries in the coal mine so to speak.

    1. Possible. You gotta scare everybody’s pants off to get them moving.
      Status Quo was never an option, but the Swedish approach might get you to 90% of the reduction in spread without the damage of the shutdown. Time will tell! 🙂

      1. Sweden is not a “ideal” as you think. They have closed senior high schools and universities, and banned gatherings of more than 50 people. They have “asked” people to avoid non-essential travel, work from home and stay indoors when they can.

        About half the Swedish workforce is now working from home. Public transport usage has fallen by 50% in Stockholm. Its streets are about 70% less busy than usual.

        Let’s not give the impression that they are a “let it ride” situation.

        ((not to mention that since they are getting hammered in number of deaths, there is a huge national outcry to impose more restrictions))

        1. Straw-man argument. I never said that Sweden will be ideal. Everybody will be in awful shape. Chances are Sweden will go trough a GDP drop roughly in line with the Global Financial Crisis. But the rest of the world will go through a 3x GFC or worse experience!
          But you’re right, I will mention this in the post so people don’t think everything is A-OK over in Sweden.

  10. Another angle that might be interesting to consider relates to two factors:

    1) Other lives lost during the pandemic due to people not seeking help or unable to seek help for other conditions due to closure of facilities and/or fear of facilities.
    2) Lives NOT LOST due to a reduction in risk factors such as motor vehicle accidents/homicides/general stupidity.

    This may not be known until we can compare baseline annual death data within various categories to 2020. It would not surprise me in the least to see a net gain of lives during 2020. When one considers global MVA alone at an estimated 1.35 million annually. Even cutting this number by 20% would result in a net savings of 270,000 lives.

    1. Motor vehicle accidents? I thought they kill 40k a year.

      Also: Let’s see how poverty/stupidity/crime will work out when the economy is in depression territory. I have the suspicion that this will all be a massive reduction in life expectancy, much worse than the virus itself.

  11. Well done, Karsten, but let me throw one more monkey wrench in that isn’t considered- the case fatality rate (whenever it is known) is NOT absolute. It is dependent on the adequacy of medical resources. An area not “medically overwhelmed” may have a lower case fatality rate than one that is, as some patients “on the edge” are saved with appropriate intervention.
    We normally treat case fatality as an absolute ratio or number – but only at some medical treatment constant.

    1. So far, no area has been overwhelmed. Javits Center and Comfort in NYC are both essentially empty. And we’re supposed to be past the peak. So, a cautious opening should be feasible and will not overwhelm the already underutilized resources

      1. I would argue that both China and Italy were overwhelmed! At least in terms of ICU beds. But I applaud your hope that things are slowing down.

  12. Thanks for your analysis!

    I am also pleasantly surprised at how the curves are bending worldwide, but I do think it’s too early to declare that we’re coming out of this completely. There are a lot of unknowns about the possibility for reinfection, capability of hospitals to withstand a 2nd potential case surge when we begin to open up, and what this will mean economically for travel, large conferences, cruises, sporting events, hotels, concerts, theme parks, etc. over the next 1-2 years.

    We need to continue to learn as much as possible about the virus via antibody tests, pharmaceutical research, vaccine research, etc. and make smart decisions based on that. There’s still a vast amount of the world’s population with absolutely no protection from this virus and it’s reasonable to assume it’s not going to completely go away until a vaccine is developed or a high % have contracted and recovered from it. Reports of reinfection are concerning, potential for mutation is concerning, and lack of sufficient data is concerning.

    I am trying to be optimistic, but even the biggest pessimists shouldn’t be surprised to see the curve bending when most of the world is shutdown. As far as the Sweden question, I think the jury is still out. After we get through this week, we will see if they continued to grow (weekend reports have consistently shown huge dips). Also, even though the government has only issued “recommendations”, many Swedes are pretty compliant with the Government. From what I’ve read, daily life there is quite a bit different than it was a few months ago although they are obviously less restricted than here in the states.

    I think the biggest key moving forward is going to be test & trace. We need to be able to separate sick people from the rest of the population until they recover without the drastic measure of “assuming everyone is a potential carrier”. A bunch of countries have avoided shutdowns by incorporating this in one way or another. I’m discouraged that about 20-25% of current tests are coming back positive. That’s far too high to implement test and trace effectively.

    1. We’re in an economic death zone. We are way past the point where we can wait for years so scientists can find 99% significant results. By that time we’ve destroyed 50% of the jobs, 50% of GDP and 80% of the stock market. And probably 20% of life expetancy.
      Even the initial estimates of 2 million dead in the U.S. are actually not as bad as a shutdown for 18 more months which will kill many more people indirectly (poverty, lower life expectancy)

      1. I don’t disagree that we’re in the economic death zone. I’m also don’t think we need 1-2 years for 99% significant results. Early antibody tests are very promising and they will probably be vastly scaled up over the coming month and provide a lot of extra data (even if it’s not 99% significant) that can aid decision making and lead to opening large portions of the economy sooner than originally projected.

        I disagree that economic depression or recession will cost more lives, or even cost lives in general. There’s a lack of scientific evidence supporting that claim, and in fact, most scientific evidence points to an inverse correlation between short-run economic expansion and life expectancy. If the depression or recession were to significantly stall growth for the next 10+ years, we would probably see that data fall more in line with your prediction of increased death, but most predictions I’ve read aren’t that extreme.

        Sources for life expectancy and Economic cycles:

        1. Intriguing studies. They have to be wrong and miss some other effect, but I don’t have time to debunk this now. 🙂
          I also don’t like to see aggregate numbers because mortatlity is on a general downward trend. I like to see mortality of individual cohorts: a) those who kept their jobs vs. b) the likely 20-30 million that will lose their jobs in this episode.

          1. I think almost certainly those who become unemployed will see increased mortality. There’s also numerous studies backing that up. It’s somewhat baffling, but research seems to bear out that for 1st world countries economic recession decreases mortality, even overcoming increases in mortality caused by becoming unemployed. Pretty interesting stuff, and not what I expected either when I was researching it.

            The article states that unemployment increases chance of death by 73% for those who experience it, but economic recession decreases it by 9% for everyone else. Therefore, if more than 1/8 ( approximately 9/73) of the population is unemployed at once, mortality will increase and if less than 1/8 is unemployed it will decrease. The workfore is ~165 million, and 1/8 of the US population is 41.25 Million. Therefore unemployment would need to be sustained at >25% for morality to increase in the entire population.


            1. Scanning through this very briefly, I think it’s not that convincing.
              You may not increase the risk of immediate death DURING the depression. A long stretch of unemployment and poverty now will likely lower your life expectancy later. We will only notice the effect of a depression many years after the event when poor people die significantly younger. How would the regression with contemporaneous hazard ratios capture that?

  13. You might consider making a moving average of 7 days for these cases. Sweden in particular has a very strong weekly signal, and the “over the hump” conclusion is contradicted by the moving average.

  14. The Swedish data have a very strong 7-day cycle. I don’t think you can conclude that Sweden is past the peak based on the downside of the cycle.

  15. Enjoyed the post. I also very much enjoy the comments. Thank you. I would be interested to see some of the south american countries added to the analysis. Off the top of my head Ecuador and Brasil would be good additions if there is good data available (I haven’t looked).

    1. Brazil is in the 26 countries I consider. Very low levels (as % of population) and still increasing rapidly. We shall see if this starts bending soon.
      Definitely a country to watch due to slow response and half-hearted shutdown.

      1. Well so far there is very low testing level (tests per 1M population) in Brazil, not worth any analyzing so far…

  16. Very elegant analysis Ern. However, by most “classic” metrics (Buffett ratio, Shiller CAPE, etc) the market was significantly overvalued GOING IN to this Virus. Covid may yet prove to be the proverbial “black swan” that takes the stock market down by 80%. Further, as you well know, EVERY majorbear market decline in the past was usually accompanied by the 50%+ “retrace” that we just witnessed over the last couple of weeks. I fear more pain is in store.
    Even Bob Shiller fears the same…

  17. The way I see it, there are two possibilities. One optimistic, one pessimistic.

    The pessimistic case is that “social distancing” had a very strong effect, much stronger than anyone expected. The whole world went into lockdown in mid-late march, and stopped the virus in its tracks. Job well done!

    …Except that’s also kind of bad. It means that as soon as we lift the lockdown measures, the virus will go right back to fast exponential growth. We can’t ever lift it until it’s completely eradicated or we have a vaccine (or some large fraction of us get it). Expect two more years of lockdowns, with maybe brief interludes where we lift it and it starts spreading again.

    The optimistic case is that a lot of our date is wildly wrong, particularly the case counts. We know that there aren’t nearly enough tests to go around, and that was especially true at the beginning. From, a large fraction of tests are still coming back positive, something like 20% for the US overall and 50% in New York.

    What if the virus actually did spread to about 20% of the world’s population, and the overall fatality rate was much lower than expected (even lower than your 0.54% estimate)? It could peak simply from reaching herd immunity (combined with some mild social distancing effectiveness). We just wouldn’t know because we don’t have nearly enough testing, especially on people who aren’t showing symptoms.

  18. I wish you are right, but I believe you are wrong. Couple major issues with your analysis:

    1) Not enough time has passed to proclaim lower mortality. The cases are not closed. Closed cases mortality rates is atrocious 21% (which we can adjust for people who never counted, even adjusting x10, still 2.1%).

    2) Testing (and even death) data is unreliable in most of the world because of extremely limited testing and death attribution.

    3) You assume this will be over when these peaks pass after months of social distancing. They will return with the same speed once restrictions are lifted.

    4) Extrapolating to the stock market – stock market has been extremely overvalued even before the pandemic. You yourself said so as Fed been pushing on a string to stimulate inflation. How much more overvalued can the stock market get even if you are right?

    1. 1: The closed case stats are likely complete junk data. A lot of people who stay home with mild symptoms are probably never added to the recovered ranks
      2: If data are so unreliable, then that’s another reason to open the economy. Thanks!
      3: Sweden and Iceland are doing their best to disprove that
      4: You’re making my case even stronger. That means out of the 15% drop since the peak, 10% came from deflating the bubble and only 5% from COVID-19. The stock market seems to believe even less in any serious damage from this critter.

    2. 4) Stocks will not be overvalued if they get to the levels where they used to be before this covid crisis. You need to understand that valuation of stocks depends of money in-flows to the economy, FED is pumping money into economy so stocks will probably follow, check MZM chart at Fred.

      And no, inflation will not skyrocket!

  19. I love your optimism, nicely backed by data! I hope you’re right!

    But doesn’t the real recovery depend on consumers going out and spending again? Where will that consumer confidence come from?
    – The now-unemployed? not gonna be spending (and employers may be slow to rehire until consumers are back);
    – Those who have had to dip into their savings to get through this? also likely spending less;
    – Certainly not the retired and nearing-retirement folks, who have seen a scary dip in their portfolios;
    – Probably not everyone else, if the media reports are creating an atmosphere of fear, as you point out.

    And the next 12 months (at least) will be dominated by media attention to testing, contact tracing, and the imminent threat of renewed lockdowns (likely with uninterrupted lockdowns for older people). So fear is going to have the dominant mindshare until a vaccine is confirmed. And confidence might even take a second dip, if early consumer optimism is thwarted by the messy realities of reopening.

    So shouldn’t we be following consumer confidence indices along side the data you’ve got?

    1. My hope is that people are swimming in cash with their $1,000/week unemployment checks and stimulus checks. There will be plenty of people with money burning holes in their pockets waiting to spend. You watch it! 🙂

  20. Argggh. One of my favorite retirement finance-related sites has began to replicate LTCM where they started with a simple and very profitable on-the-run/off-the-run arb and wound up short ooodles of volatility (lost focus of their core mission/strength)… we know how that ended. I’l be back when the site re-focuses on retirement-related finance. Until then best wishes to everyone regardless of how serious or not they think the current situation is.

      1. One must always look at what’s going on in ALL times. It’s not the looking, it’s the discussing or better still the arguing. I miss you cold hard objective undeniably quantified contributions. And btw, you already got your bailout to the tune of $TRs. Hell, they are even gonna buy HY now. 😉

      2. ERN f*** that sh**. Your comments on current events are very beneficial. Unfortunately there is a lot of very pesimistic people out there… 🙂

  21. Thanks for the good work analyzing this and not disputing your numbers. I also agree that things are looking way better than people thought they might. I still don’t think there was an overreaction though, at least not for dense areas. To give a way to think about it in contrast to the motorbike one, close to 1 in 1000 NYC residents are currently dead of Coronavirus (I think we should hit 1/1000 in about 2 days). That is without taking guesses at how many of the extra die-at-homes are connected to Coronavirus. So pretty big deal if you live in NYC like me, however you think about it.

    In NYC, I think the shutdown did not come nearly fast enough and even once our governor started taking action, was illogically ratcheted up over (a pretty short amount of) time instead of going straight to shutdown, when you wouldn’t have data to support whether the pace was fast enough for weeks. I don’t know if the right term is underreaction, but way too slow to react appropriately in NYC.

    1. Agree 100%. In NYC it was an “underreaction” because as of March 10, Bill De Blasio still touted how this virus is just like the common cold.
      I also think that NYC will likely be one of the last places to reopen again.

  22. Nice work, Karsten! The research, the analysis, the graphs, and your discussion and overall organization of this article are all world class.
    I did a back of the envelope calculation for Age 65 to 74 similar to what you did for 0 to 59 based on US data. CDC (, Fig 3) shows a 1.8% overall mortality rate for Age 65 to 74 in 2018.
    Using the table you produced of Swiss data, and combining 2 cohorts (Age 60- 69 and Age 70 to 79) I get an estimate for Death as % of cases as (2.05% + 7.50%)/2 = 4.77% to adjust in a simple fashion for cohort Age 65 to 74 to match CDC data, and then dividing by 8 as in German example in small town in Nordrhein-Westfalen, = 0.6% when Covid-19 has gone thru the whole cohort (let’s say in 1 year to make calculations simple) This will add to the overall mortality (1.8%) about 1/3 more deaths in 2020 in this cohort Age 65 to 74 due to Covid-19. This is not small increase in mortality nor is it huge, but numerically similar to what heart disease causes in this Age 65 to 74 cohort (~ 30% of the deaths per year, heart disease is leading cause of death).
    My take on the effect on this Age 65 to 74 cohort (of which I am a card carrying member 😊 ) is that a combined approach based on the Tomas Pueyo paper on the “hammer and the dance” ( ) mentioned by you above, and these 2 articles by David L Katz, MD ( and
    (Katz suggests an approach similar to what Sweden is doing) will keep mortality within this 0.6% range or less for this 65 to 74 cohort since the ICU’s etc. will not be over run.
    Of course, 0.6% additional risk of death over the next year, is not something I relish. On the other hand, it is not a cancer suffering and death or an Alzheimer’s suffering and death or a stroke suffering and death, which I personally fear a lot more. Therefore, a well-planned but rather quick reopening of the economy at some point in May timeframe would seem reasonable balance to me of mortality risk vs. economic harm (which in turn has mortality risk and other poor consequences).

  23. The Ayn Rand Institute, in Santa Ana, California, with its promotion of objectivist ideas and perspectives, has some valuable insights to support all Karsten’s posts, based on: cool-headed non-emotional analysis, objective reality, only the evidence, only the facts, reason, logic, the scientific method, individual rights and individual freedom, free markets, – controversial on some things – you may not agree with everything – but worth checking out –
    …..those who do, and who value intellectual precision and clarity, and who are open to ideas that might be unconventional in today’s culture: their lives will never be the same……in a good way…..

  24. I had multiple issues with your last post on Covid mostly related to the health component since its what I do daily. I felt about your last post the way you feel about the media (apparently) re egregious misrepresentation even in the context of generally agreeing with much of the underlying.
    This post while obviously similar in tone/thesis was very well done and I notice and appreciate the difference.

    This is a shitty complement. But still, I appreciate your ongoing efforts and quality. Your blog and its data has substantially aided me in financial decision making. I deeply appreciate what you are doing and hope you will persist.

  25. I tried to leave a reply but it got deleted/not posted. So now a short version.
    I thought the health stuff on the last covid post was garbage even if in places where I agreed with underlying.

    This post is very well done.

    Thank you for the ongoing content and your continued efforts and success at producing quality.

  26. My vegemite-eating patriotism comes out when you mention the Australian reporting. To be honest, Australian reporting has a lot of problems, so if it’s superior to American reporting, there must be quite a bit of poor journalism going on in the US.

    Love the Aussie shout out!

  27. Ern – Thanks for this and all your columns, and please keep them coming! The field of economics is fascinating for it its spectacular breadth. I can think of no subject that is not ripe for analysis under its lens.

    Tell us – Have you implemented any defensive measures to your portfolio as a result of this pandemic? For example, in though your options accounts are well suited for generating returns in a down market, did you or will you sell equities now that the market has recovered some?

    Take care, and stay safe.

    1. No adjustments except for some tax-loss-harvesting.
      Option writing was very lucrative in March due to insane implied vol.

      I had contemplated selling some equities in late February but didn’t do so in the end. After that, I realized it’s too late. I hope that the recent bounce is justified because I can see a decent way out of this mess. Let’s hope it’s not a bear trap! 🙂

  28. We should follow the Scandinavians on this one! Where’s Bernie Sander when you need him?

  29. ERN- I wonder how it would look to do comparisons of some of the varying approaches different US states have taken. For example it sounds like Utah is much more in line with Sweden and Iceland in its approach. Utah has closed schools, restaurants, and prohibits gatherings of more than 10 but most of this without the force of law. Salt Lake the largest city has an official legal order but with relative lax enforcement. It is also an interesting that Utah is one of the most urbanized states that is an overwhelming majority of the population lives in the urban areas, something like 80 or 90%

  30. The number of deaths in US continues to increase every day. It is true that the rate of increase is slowing from day to day but that is true of any trend otherwise all of humanity will be dead. Still the daily number of deaths in all the hot spots, NY, Italy, Spain, France, and UK is very high and has not peaked in some of those hot spots, let alone rest of the world. So the total damage is not known yet.

    The 2nd issue I have is on the effectiveness of social distancing. CA did it early, esp. Bay Area, and came out much better (deaths/million) than other states. Remember that CA has very high exposure due to overall wealth, lots of travel between US and China again more so for Bay area.

    But your overall conclusion is well supported by data that the virus is not as deadly as originally thought and media is certainly making it look worse because “bad news sells” and their goal is to make money. Even ignoring another aspect of media being more left leaning and the desire to make current administration look worse.

    1. Thanks!
      Agree: CA did this much better than NY and NYC.
      Daily new cases have certainly peaked in ITA, ESP. Probably even deaths. I think both cases and deaths are past the peak in NY also, but it takes a few more days of data to confirm it (weekly seasonality issue).

  31. ERN, I’m a little surprised by your optimism and the reasons for it. I don’t think the deaths are going to be a big killer of the economy, but I think the shutdown will be. You’ve been so focused on charts and numbers (after all, that -is- the big ERN way!) that I think you’re missing the behavioral basics.

    There are so many dominoes to fall from all of this that I’m utterly shocked the stock market is as high as it is. Between people losing jobs, companies going out of business, schools being closed, and practically all events being shut down I think there are going to be a TON of unintended consequences. I know all the closures and social distancing is self-imposed, but results are likely to be disastrous in the near to mid-term. And everyone seems to be forgetting that the world in many ways is still going on. The next bubble that may (or may not) have been building over the last decade could pop at any second.

    And I’m an eternal optimist!

    1. I never claimed that the DEATHS will be the reason for the economy to tank. Quite the opposite, I believe the politics and the overreaction is what will potentially kill the economy.

  32. As always thanks for the high quality of your articles. You have a minor typo “Out of 884 deaths up to April 13, there were 860 in the 60+ age group and only 24 in the 0-59 age group.” I think you mean “80+ age group” not 60+? also the age group would be 0-79?

  33. I’m a long term fan of this blog, but you’re out of your area of expertise here.

    The two big things that you’ve missed out are morbidity and the death rate without hospitalization.

    Firstly morbidity: non lethal side effects of covid are pneumonia and multiple organ failure. These are serious problems with life changing implications.

    Secondly, and more seriously is the death rate without hospitalization. Most people who go to hospital with covid get better. Most people who go to hospital with covid would have died without supplementary oxygen and antibiotics. (The antibiotics are for secondary bacterial infections). The death rate without hospitalization could be as high as 17%. This was observed in Wuhan. Italy is getting 10% death rates in its overwhelmed hospitals.

    If you just let the epidemic run it’s course then almost all cases will be in the peak and most cases >90% will be unable to go to hospital because the hospitals will be full. This will lead to an order of magnitude more deaths.

    We’re starting to see what happens in a population with inadequate control measures. Look at Ecuador with it’s 1000 confirmed deaths and dead bodies piling up in the streets. (They’ve had way more than 1000 deaths)

    1. I didn’t miss those things at all.

      1: As with any disease, the deaths are just the tip of the iceberg. I never argued otherwise. It still doesn’t justify closing the economy until next year. Just like the deaths will be minuscule, the lasting effects will be minuscule also, both compared to the population and compared to the effects of the shutdown.

      2: More seriously, means that you’re more seriously confused about the second point. There were no overwhelmed hospitals in the US:
      WA State: “Army field hospital for COVID-19 surge leaves Seattle after 9 days. It never saw a patient”
      Almost equally underutilized hospitals even in NYC (Javits Center, Hospital Ship)
      Swedish Hospitals are not overwhelmed either.

      The 17% dead outside of hospitals in Wuhan? I’m surprised it’s not 100%! People were locked up in their apartments, doors bolted/welded shut. No wonder people died inside. People were not killed by COVID but by communism.

      1. Sweden is a terrible model for comparison to the rest of Europe or the USA for several reasons-
        1) The economic changes taking place is Sweden even without strict government action remain to be seen.
        2) In Sweden half of all households are ONE PERSON
        3) in a 2017 survey of Sweden – 55% of those age 16-24 reported no socializing at all with any close relative.
        They were already social distancing as part of their day to day life before this hit. And as previously mentioned, they are more rule followers than many, certainly more than Italians (and Americans). I would be very cautious drawing comparisons from Sweden.
        What happened in Italy?

        I also second the notion that only looking at death rates or even only looking at hospitalization misses a lot. Those who are not asymptomatic are not just getting better in a day or two like the flu. The illness lasts on average 11-14 days, twice the typical flu, which also can be shortened with Tamiflu. Longer illness is not just a problem for hospitalized patients. For those who do not end up hospitalized, but are symptomatic, how many days of work do they miss? What is the effect on more and more of those absences on an economy? Death is not the only thing that can cripple an economy.

        1. 1: Strawman argument. Never claimed Sweden will escape a recession. But they’ll avoid a depression
          2: Meaning what? Half of NYC households are one-person. Hasn’t helped there
          3: Just like the U.S. Most grandparents see their kids/grandkids only very occasionally because they live across the country. In that sense, USA is very similar to Sweden and very different to the multi-generational households in Italy and Spain.

          1. First of all- let’s stick to the facts and not make up statistics.. Half of NYC is one person household?
            No – it is 32% (about the same share as the EU average)
            Sweden single person households is a big outlier– ( the housing in Sweden has basically been designed to meet this unique Swedish demand– small single apartments dominate.

            The Swedes have an old saying that I am not sure even has an equivalent in English or many other cultures– ““ensam är stark” (Alone is strong.)

            With regard to the comment about grandparents – it misses the very unique experience in Sweden. We are not talking about rarely seeing mormor (grandma)..Swedish young people move out at a younger age than just about anywhere – between 18-19 compared to an EU average of 26.
            They have a different approach and it makes them different from us.
            “It is special in Sweden – and the Nordics – that there is much less variation in leaving age than other countries,” explains Gunnar Andersson, professor of demography at Stockholm University.

            “In other parts of Europe it’s not considered problematic to depend on your family and in southern Europe it should even be considered a goal – if you don’t, it would be like rejecting your family,” he says. “In Sweden…it’s the goal to create an independent individual…there’s seen to be something wrong if the child stays at home.” (

            .Also that 55% did not say they had rare or infrequent social interactions with grandma– it said that 55% of them had NO SOCIALIZATION at all with ANY close relatives. Not just grandma, ANY RELATIVE. I don’t have data on other countries, but do you think we would see numbers that high for no relations with any close relatives at all on such a survey in the US?

            1. In the U.S. we have car culture like nowhere else (and everyone just riding with one person per car). Not much public transportation. A lot less interaction with your neighbors than in most other countries. People saying “excuse me” if they get even within 3 feet of you. Very, very different from how things work in Europe.
              So, I can spin the differences in societal norms in the USA’s favor where Sweden’s policies will work here just as well for the exact same reasons.

  34. Statistics for the win! Thanks Karsten for a thorough analysis to bring some optimism to us all. I love your brain!

  35. There’s certainly an above-average sophistication to this analysis, but at the end of the day it’s coming from an armchair epidemiologist. He’s good with a spreadsheet, but what he doesn’t know about COVID-19 can literally kill you and thousands of others when this type of misinformation is peddled. Please do yourself a favor and listen to the experts, who spend their waking hours buried in the data and the science.

      1. The 2 million deaths is what could happen with no lockdown. We’ve already had some lockdown so we won’t see 2 million deaths.

        A common problem that many experts have is that they are blind to their lack of knowledge in areas outside of their expertise. It’s why doctors usually make terrible investment decisions.

        You’re not an epidemiologist and you have made some basic mistakes that I’ve outlined in my other comment.

        1. 2.2 million deaths in the USA would translate into 66,000 deaths in Sweden. Current death toll: 1,540. Days ago, the IHME forecasted 18,000 deaths for Sweden by August 4, now less than 6,000. If you revise your estimate down by 2/3 over a span of three days, you’re a louse epidemiologist.
          So, I’m not being mean to the epidemiologists. There are some that actually know what they are doing (a lot of them with Scandinavian last names). Unfortunately, too many people, you included, are listening to the charlatans.

          1. I call straw man argument.-
            1) The 2 million deaths was predicted by Imperial College based on (the VERY UNLIKELY) NO MITIGATION AT ALL in the USA. Sweden has closed all universities and secondary schools and banned any gathering of more than 50 people. They did not do an analysis of Sweden and it’s particular geography and demographics so you cannot just extrapolate the total numbers for Sweden from the USA or UK.

            Also the Imperial College laid out all the basis and parameters used based on what they knew at the time.. Which part of what they had do you think they got wrong?

            1. You’re making exactly my point: The shutdown was based on a faulty study with faulty assumptions. Even if we had done no shutdown, we’d seen fewer than 2.2m deaths because people would have changed their behavior, just like they do in Sweden right now.

              Also, the (still) reputable IHME (U of WA) has a model for Sweden. The total estimated eventual deaths were >18,000 as of April 13. Nowhere close to the 0.66% of the population as the Imperial model. But even that April 13 model was way too pessimistic for Sweden. Only four days later they now predict <6,000 deaths. A downward revision of ~70% in four days. Applied junk science.

  36. ERN I appreciate the skeptical view and hope your findings of lower mortality rate / higher prior infection total is right.

    But the Santa Clara number you say implies 0.2% mortality is confusing when you compare to NYC totals

    In NYC itself there are…

    8.3 million residents
    12,000 or so deaths so far

    If mortality of all who get infected is 0.2%, that implies 72% of NYC population was already infected

    I don’t have a good theory but you can see how conflicting datapoints are!

    1. Personally, I found the 0.2% number a little too low.

      If we assume a 0.5% fatality rate in NYC then 12,000 deaths would imply 2,4000,000 infected so far. That’s 28.6% of the population. Note that 28.6% means currently or previously infected. In addition to the roughly 15% infected (NYC hospital study mentioned above) there could be another 15% who used to be infected and asymptomatic and are now cured not even testing positive to the simple test anymore (but likely would test positive to the more advanced antibody test).
      So, I don’t find the numbers logically inconsistent.

      And then in addition: a small difference in the age distribution can make a huge difference if older folks have a 100x fatality rate.

  37. Thanks for this, ERN. In the ocean of “experts” with myriad agendas or single-factor focus it is refreshing to look at the numbers (as they are and with their limitations) for some insights.

  38. You mentioned death over forty times in an article about optimism. I connected the dots!

    I agree our actions will effect the economy more than the virus itself. I think the only difference in how you and I are thinking is this: If we didn’t have the shutdown (which I generally despise but understand), it likely would have been disastrous. Take NY for example. They came close to overwhelming their medical system. If they had waited two more weeks to institute the shutdown, with the exponential grown during those two weeks it would have exploded. Never mind comparing NY to Sweden or Canada or whatever. How about comparing NY to NY? Us playing “armchair epidemiologists”, what do -you- think would have happened in NY? You have my best guess.

    1. What do you think is the difference between NYC and SF? The variation of outcomes in cities who all shut down is larger than the variation in outcomes between USA and Sweden, where the former shut down and the latter didn’t.

        1. SF deaths: 20. That’s 22.7 deaths per million
          NYC deaths: 14,604. That’s 1739 per million (using 8.4m residents, please correct me if I’m wrong).
          So you’re saying that a shutdown delay by a few days has raised the fatality per resident by a factor of almost 100?
          It doesn’t make any sense. Please try again. It’s a serious question:
          *What explains the vast difference in fatality rates between NYC and SF?*

          1. What explains the vast difference in fatality rates between NYC and SF? My guess (and it’s just a guess) is that NYC has 10x the population, and more importantly is a crowded, busy and dynamic travel hub with the virus coming from all over the world, into that one tiny spot.

        2. Or add another observation:
          King County, WA (Seattle): 360 deaths. That’s 159.8 deaths per million. King County shut down earlier than everyone else. So, why does it have a fatality rate much higher than SF?
          The date of the shutdown can’t explain the vast range of fatality rates. Isn’t it odd that death rates vary so wildly in large prosperous cities similar in demographics?

          1. There is going to be some randomness on when the first cases arrive in each region. The first known case of coronavirus in the US was in Seattle, and the first known death was in a nearby retirement home. It seems to have spread quite rapidly in that area before there was much public awareness, testing, or social distancing. There were a couple of cases of individuals who died from covid in a retirement home up there and it went completely undetected for quite a while (a couple weeks I believe) because no one bothered testing.

            Looking back at old news articles, 35 people had died in that one retirement home alone as of march 21st, and several other retirement homes were also hit hard (I can imagine that there are probably lots of medical professionals/ others going back and forth between different homes spreading the disease). A number of retirement homes were hit really hard in that area.


            I agree that there is probably a lot of variation in cases that we don’t understand but I think this one is largely explainable. They quarantined early, but they also got hit very early, possibly the first region in the US, at a time when people were barely starting to pay attention.

  39. Many people will chose to stop working and live off the government as long as they can. Unemployment benefits looking pretty good for a large percentage of low waged workers vs going back to work.

  40. ERN, I found it’s hard to compare different cities, states, and countries for a variety of reasons. Eliminate all the noise for a moment.

    I said “Take NY for example. They came close to overwhelming their medical system. If they had waited two more weeks to institute the shutdown, with the exponential growth during those two weeks it would have exploded.” Talking about NY, and not anywhere else: What difference (regarding infection) would it have made if they hadn’t shut down?

Comments are closed.