July 16, 2020
I like primary research. I don't like to rely on journalists to interpret the world for me.
With more than 24-years of experience in global financial markets, I know very well how the financial media interprets markets, and financial and economic data. It's often: 1) wrong, 2) interpreted with a bias, or 3) presented with a shocking headline that is misleading, if not unsupported, by the article itself (most importantly, unsupported by the facts). Many times its all of the above.
That's why, when the Fed Chair speaks. I listen to the words coming out of his mouth. I don't rely on a journalist to tell me what he said. Similarly, when there is economic report, I read the original report and look at the data (and analyze it within the context of historical data, and the bigger picture). I don't wait to read about it in the Wall Street Journal.
With the above in mind, we would all be better informed on the virus, if we looked at the data ourselves.
In that vein, Harvard has compiled a database with time series data on Covid cases and deaths, by day, by country. This is the only place I've found the raw time-series data on deaths. You can see it for yourself here.
This is useful because we can look at deaths as a percent of cases in an historical chart. And it gives us a very simple big picture view of the virus.
Let's take a look.
In the above chart, the x-axis is "days since the first diagnosis of a Covid positive case." You can see that deaths as a percent of cases is 4.15% (the far right of the chart), and has been moving lower since mid-May. That May peak is right in the heart of "phase 1" reopenings of state economies.
Now, a few weeks ago, the CDC chief told us that we've likely only diagnosed about 10% of those that have, or have had, the disease in this country. That's based on surveys of blood samples taken around the country. So the CDC thinks the infection rate (i.e. cases) should be multiplied by a factor of 10.
If we do that, the death rate comes down to 0.42%. That's about 3 times the ten year average death rate of the annual flu. You can see that flu death rate in the green line in the chart below (and that's with a vaccine).
Next, the former FDA commissioner, Scott Gottlieb, has said the mutiple could be as much as 20 times higher. Remember, as we discussed back in April, there was an antibody study on 3,000 New Yorkers that showed a 14% infection rate. In New York City, the number extrapolated from that study was 21%. Wake Forest has run a large antibody study since April, which now suggests 14% of North Carolina residents have or have had the virus.
So, Gottlieb's high-end projection would put the U.S. infection rate at about 20% (more than 70 million Americans). If we use that 20x multiplier on the current infection rate, the percent of deaths-to-cases drops down to 0.21% (chart below).
That's slightly higher than the flu death rate.
So, the death rate has gone down since mid-May, for a couple of reasons: 1) treatment options are working, and 2) the large majority of people are simply defeating the disease on their own, and more of those types are being revealed with more testing (and those people are testing because they are being forced to test to return to work OR because they simply can, now that testing is, largely, open and free for asymptomatics).
Based on charts two and three, we should expect the rate-of-change in the death-to-cases ratio to decline more quickly as the rate-of-change in testing continues to increase.