These people should know better!

Note: I’m not going to have my usual Numbers section today, since I want to make sure I’m using the best possible growth rate to project total deaths. It will be back tomorrow!

I’ve been reading the Wall Street Journal since the late 1970s. I’ve always thought their reporting was top notch, although I have disagreed with a lot of the contents of the editorial page (but certainly not all, especially when it comes to foreign affairs and international trade). However, two items on the editorial page in today’s Journal seemed to me to completely miss the point about the coronavirus crisis. Unfortunately, missing the point on this topic can mean lots of lives lost. So I wrote two letters to the editor, which I reproduce below (with a little explanatory text for each one, since you may not be able to find these items online unless you’re a subscriber – but try anyway!).

RE: “New Data Suggest the Coronavirus Isn’t as Deadly as We Thought”

(This op-ed, by Andrew Bogan who’s a molecular biologist, makes an argument that several other people have made to me about my death projections: That I should be using a mortality rate that’s way lower than the one I’m using. This was when I was projecting deaths by multiplying reported total cases by a 4-7% mortality rate. I’ve stopped doing that altogether this week, because the reported cases number is ridiculously low and getting more ridiculous every day – in fact, it clearly only measures the availability of tests. The only good number we have to measure the scope of the pandemic in the US (and even then it’s not perfect) is the daily deaths numbers. And I’m projecting deaths by simply applying the current growth rate in that number

I have had at least 4 or 5 people – two in the last two days, both very rational people! – make an argument very similar to Mr. Bogan’s)

I found Andrew Bogan’s analysis to be logically correct but to completely miss the point. He argues that any projections of total pandemic deaths based on a presumed case fatality rate (i.e. total deaths / total cases) is going to be ridiculously large, if the standard estimates of 1-3% are used for the rate. This is because the reported cases number is without doubt a vast underestimate of actual cases.

I don’t dispute any of this. But the question is, how can we ever measure the true rate (and he doesn’t even try to do that), when one of the two components of the calculation is totally unknown? Even at the end of the pandemic, by definition we’ll only know reported cases and reported deaths. But let’s say at that time there will be some good way to estimate how much the true case number differs from the reported one – the number of deaths will still be the same!

The real point is we want to project total pandemic deaths now, using whatever information we have. How can we do that without having to guess at the fatality rate? In fact, we do have a “hard” measure of the true rate: We can get that by dividing total deaths so far by total closed cases – i.e. the sum of deaths and certified recoveries. The Worldometers.info site has been publishing total recoveries since March 26. On that day, the “actual” case fatality rate stood at 41%. It has gone up and down since then, but as of yesterday it was 37% - meaning that 37% of all resolved coronavirus cases have resulted in death.

Of course, there are good reasons to believe this number will go down over time. But the fact is it hasn’t gone down in three weeks, and other countries including Italy and France have similar numbers (37% and 35% respectively). Given this, it’s very hard to believe the final case fatality rate will be something less than 1%, and most likely a lot more than that.

RE: “Trump Rewrites the Book on Emergencies”

(This was an op-ed by Christopher DeMuth of the Heartland Institute)

Mr. DeMuth praises Mr. Trump for managing the coronavirus response through “vigorous localism, private enterprise and professional dedication, with the federal government providing essential national leadership but staying within its constitutional rails.” This is a wonderful example of making a virtue of necessity, since on Monday Mr. Trump asserted that he had absolute power to control the response  – and therefore the governors wouldn’t have any say in the matter if they didn’t agree with him (and a few weeks ago he had assured the governors that the Feds would take care of all their needs). Of course, by Friday he’d gone back to his previous position that the responsibility for the coronavirus response lay totally with the governors, which is the position that Mr. Demuth praises.

If the pandemic in the US is in fact under control now, and if the loosening up that Mr. Trump is advocating actually happens without a serious resurgence of the virus, this will seem like quibbling over small details; I for one will be glad to admit I was wrong about this. However, the data point in exactly the opposite direction. We will know in the near future which was correct, Mr. DeMuth or the data.