Sweden and Mississippi

In a recent post, I mentioned Paul Krugman’s demolition of what I referred to as the “Instapundit factoid” that Sweden is poorer than Mississippi. Don Arthur gives some useful background to the dispute, but questions whether Krugman’s critique constitutes a demolition.
There are a couple of ways we might interpret the claim. The first is that there are relatively more poor people, relative to some objective poverty line, in Sweden than in Mississippi. This not true. In fact, as Krugman shows, it’s not even true when you compare the US as a whole with Sweden. There are a higher proportion of (absolutely and relatively) poor people in the US than in Sweden or other West European economies. This is not offset by ‘dynamism’ or ‘social mobility’. Poor people in the US are more likely to stay poor than poor people in European countries – this is why the notion of an ‘underclass’ originated in the US.

The second interpretation is that the ‘average’ Swede is poorer (or less rich) than the ‘average’ Mississippian or American. This raises two more definitional questions. What is ‘average’ and what does ‘poorer’ mean.

On the first point, virtually everyone who has studied the subject seriously agrees that the most useful notion of ‘average’ is the median, that is, the midpoint of the income distribution. Krugman uses this definition when he says ‘Sweden may have lower average income than the United States, that’s mainly because our rich are so much richer. The median Swedish family has a standard of living roughly comparable with that of the median U.S. family’ By contrast, Reynolds wrongly relies (implicitly rather than explicitly) on the assumption that the correct measure is the arithmetic mean.
To see the difference, suppose that Kerry Packer makes $1 billion (To simplify things suppose he makes the profit on overseas transactions and keeps the money there so no other Australians are affected). There are two possible ways you can look at this
(a) The average Australian is $50 better off
(b) This makes no difference to the average Australian
If you believe (b) you should go with Krugman and the median. If you believe (a) you should go with Reynolds and the mean. You might also start wondering why you are not producing a ton of wheat a year like the ‘average’ Australian household.

The other point raised by Krugman and others, but ignored as far as I can see by Reynolds, is that, on average, Americans work longer hours than Swedes (or anyone except people in quite poor countries). In economic terms, leisure is a good and there is no reason to ignore it in comparisons of wealth and poverty (I’ve published journal articles making just such comparisons between Australia and Japan).
The correct economic interpretation of the claim “the average Swede is poorer than the average Mississippian” is “The median person in Mississippi could work the same hours as the median person in Sweden and afford a higher standard of living” In fact, the reverse is true. As Krugman says, the average American is about as well off as the average Swede (the same goes for other West Europeans). Rich Americans are a lot richer than rich Europeans and poor Americans are a lot poorer.
Of course, in terms of the arithmetic mean, very rich people pull the average up a lot more than very poor people pull it down. But that just reinforces the point that the mean is not a useful measure of average wealth.
Update Rob Corr has a bit more on this