Social Spending and ethnic diversity

The core of a David Goodhart's "left-wing" argument against immigration is that racial diversity makes it harder to sustain high levels of social spending. 

I don't necessarily disagree with the point, but harder does not mean impossible. 

The argument is motivated by work by Alberto Alesina and other economists at Harvard. Alberto Alesina is a smart, original, and prolific thinker who does a lot of interesting work. But this particular paper, was never actually published in a peer-reviewed journal. The chart below is the main empirical result driving their argument (from this paper). 

As David is not an economist, I'll break this down for him slowly. This is called a "scatterplot". Social welfare spending as a percent of GDP is on the vertical axis, and an index of racial fractionalization is on the horizontal axis. You can see that there is a negative relationship between the two, but there is also a lot of variation around the fitted line.

A few observations:

- All of the European countries have relatively low levels of racial fractionalization - below 0.2 - but being European actually tells you very little about levels of social spending, which are spread widely between low spending Iceland and Greece, and high spending Belgium and Luxembourg. 

- Removing the European countries would remove the negative trend. Japan (almost no diversity) has almost exactly the same social spending as high diversity Brazil or US. Low diversity Costa Rica has the same social spending as high diversity Mauritius. 

- Looking at the UK - imagine that the UK began to approach New Zealand or US levels of diversity - does that mean we would have New Zealand levels of social spending (higher) or US levels of social spending (lower).

The key point from just looking at this chart is that even if there is a relationship, which it is not even clear that there really is if you consider Europe separately to the rest of the world, diversity is not destiny. Social spending is a policy choice. Diversity might influence this policy choice, but so do a lot of other things. Even if diversity did make social spending harder, it does not make it impossible. Correlation is not causation. Etc. QED.