Corona – what makes US different ?

Yesterday I looked at 50 US states, and how a couple of parameters, population density and GDP per capita, are correlated with the number of deaths.  That analysis revealed that for these 50 states, population density (or more correctly, the magnitude of density) is strongly correlated with the magnitude of deaths, while GDPpc a weaker, but still positive correlation.

Let’s repeat the same analysis, but this time, for 130 different countries of the world (those where I could find data on metrics such as gdp, density etc):


Now, that’s interesting! The importance of gdp and density are now reversed – while for US states density is the dominant factor (of those we’ve looked at so far)  for deaths, for the set of these countries, it’s the gdp.  Impact of density is still positive, but only very weakly so.

A couple of facts might shed some light, perhaps:

  • All 50 US states have a GDPpc way above the global mean: US GDPpc is around 60.000 USD, with a standard deviation of 11.000 USD, while the mean for these 130 countries is 15.000 USD, but with a standard deviation of whopping 18.000 USD (!!!). That is to say, that among these 130 countries are countries that are extremely poor, and probably can’t afford preventive measures nor medical care etc, leading to a larger number of deaths, regardless of density.

So, one hypothesis could be that GDP matters, but only until you reach a certain treshold, that is, once the economy of the country is large enough, GDP stops to play a major role for the number of deceased.  After the GDP treshold is reached, density seems to become dominant.

To see if that’s the case, I picked countries with a GDPpc comparable to US, and that resulted in a total of 26 of the 130 countries. Let’s run the same analysis on those 26 countries, with the expectation that now, like for US states, GDPpc should not matter, but density should:


Whoohaaa…! Neither density nor GDPpc is strongly correlated for the 26 richest countries of the world…!

What else could impact mortality (that we can measure now, that is) – let’s try with age structure, represented by median age. First, for US:


Nope. density is still the dominating driver for deaths for US states; median age has in fact a weak negative correlation with number of deceased.

What about the 26 richest countries, will they exhibit the same pattern…?


For these 26 (actually, 27, since Luxembourg, the one with worlds by far highest GDP, that was excluded as an outlier in the previous chart where GDP was measured, is now back) median age is strongly positively correlated with nr of deaths, while density is not.

So, while age seems to matter a lot for the 26/27 countries, for US states it’s density that has the strongest impact of the 3 variables we looked at.  An interesting difference, at least to me.

What have we learned by all this…? First and foremost, that the Corona outbreak seems to be a highly complex, dynamic non-linear process, where the individual outcomes between different countries can vary a lot, depending on underlying very uncertain, even totally unknown parameters.  Which in turn strengthens the argument I made in an earlier post that wrt strategies to combat the outbreak, one size doesn’t fit all – the optimal strategy is to a high degree dependent on the unique characteristics of the country.






About swdevperestroika

High tech industry veteran, avid hacker reluctantly transformed to mgmt consultant.
This entry was posted in Bayes, Complex Systems, Data Analytics, Epidemics, Non Linear Dynamic Systems, Probability, PYMC, Statistics and tagged , , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s