Ignoring Country Cohorts

by baken1 min read8th Mar 20202 comments



I assume that many of you have been watching the "total coronavirus cases outside of China" number. If you look at them on a log chart, they appear to be growing perfectly exponentially and doubling every 4-5 days.

This perfect straight line is obscuring a very, very important underlying fact. Within countries, the second derivative of cases has been falling rapidly (please reference the "first 60 days" tab).

This is a result of a few important factors:

1. Initially, in country growth is extremely high. Prior to reaching 100-1000 cases, countries do a poor job of testing. As a result, they accumulate a backlog of cases that are discovered quickly upon commencement of testing. This makes the growth rate appear to be very large for the first 1000 cases. The aggregate data we are seeing today is incredibly skewed as a result of this effect.

2. After 1000 cases, governments are actually jumping into action. They implement large scale testing (like we are currently seeing in South Korea), quarantines (like we are currently seeing in Italy), and other aggressive policies. Additionally, some measure of social distancing goes into effect as people start to worry about their own health. The effectiveness of action can be seen in the chart above.

My prediction:

By the week of March 16th, the slope of the line on the log chart will halve and subsequently continue to fall like we've seen in China. Total cases ex-China will approximately double between March 16th and March 24th (half the rate we're seeing today), and will continue to fall.

Despite the sense you might be getting from the 100lbs of rice in my pantry, I think we've seen some very positive news over the past few days and I'm becoming much more optimistic that this will not be a generational event.