Modeling Poverty

Radar

Several years ago I moved to Brazil and, after some time, started thinking about reasons that would explain why Brazil is poor while Canada (my home country) is rich. I looked at the man my wife had hired as security guard/gardener: he can’t read or write, has a large family, barely earns enough to make ends meet. How can he survive? Then I asked myself how such a man could survive in Canada. After some thought, I surmised that temperature (Canada is cold!) could be part of the explanation: in Brazil, close to the equator, he can sleep under a mango tree if he wishes and eat from the tree itself; of course he doesn’t actually do that: I’m just giving a concrete example to make a point: he doesn’t have to spend much on clothing, food, shelter, heating, transportation, etc. A lot easier to survive here than it would be in a cold country.

I then asked myself a couple of questions:

  • How many hot countries do I know that are rich? Very few.
  • How many cold countries do I know that are poor? Very few.

It seems that the correlation between temperature and poverty is high. At the time, I thought the correlation could be as high as 70% or 80%! I’m not claiming that I invented this explanation: I had never heard of temperature explaining poverty, but I suppose this must be an old theory. Anyway, I recently decided to put this theory to the test and also include factors that humans cannot control other than temperature and that could also explain some of the variation in a country’s wealth.

Here is an app that lets you try out a few models to see how much the following three predictors can account for a country’s wealth/poverty level:

Of course, the average rent obtained from natural resources is dependent on human actions, but it’s the closest I got to getting values for natural resources for this little exercise. The wealth level is measured using Per-Capita Gross Domestic Product (GDP, data from the world bank).

Look at the app or examine the following outputs:

Variation Explained:
Percent Variation Explained
Average Temperature 22.1
log(Natural.Resources) 17.9
Landlocked 10.2
Residuals 49.8

poverty.png

Surprise! Fully half of wealth/poverty can be explained by these 3 factors! I expected more as I said above, but still … 50 percent! And essentially nothing politicians, entrepreneurs, citizens, can do about it. Somewhat depressing, isn’t it? Some countries simply have a (much) lower ceiling on achievable wealth.

The code is available here.

Leave a Reply

Your email address will not be published. Required fields are marked *