One Reason Poverty Measures High In the U.S.

Tyler Cowen isn't his usual clear and direct self in this - otherwise highly recommended post. Let me blunt: the measured poverty rate in the U.S. is biased upward precisely because our income is distributed more evenly than many countries.

Specifically, he quotes Lane Kenworthy says:

... For each country the poverty line — the amount of income below which a household is defined as poor — is set at 50% (sometimes 60%) of that country’s median income. In a country with a high median, such as the United States, the poverty line thus will be comparatively high, making a high poverty rate more likely...

The missing piece here is what the unusual phrase "high median" means. The median is a central location in a data set, such that half the observations are larger and half are smaller.

Every country has a median income, and the U.S. has a very high one because we are a rich country.

But, every country has a distribution of income that is lopsided towards the top end. The indirect result for the case of income is that this tends to push the median towards zero.

Thus, the sense in which "high median" is used is that income distribution in the U.S. is not as lopsided as many countries. (Note that the comparison set in the Marginal Revolution post is Romania and Mexico - the U.S. will still look bad compared to developed countries.)

So, if the poverty rate is measured as a percentage of the median income (not as a percentile of all income, as the median is), and you have a high median, you will find a lot of people below that threshold - because taking the percentage of a high median yields a high threshold.

N.B. There are a huge number of other technical reasons why the poverty rate is useful but goofy, and this post doesn't touch on any of them. The comments at Marginal Revolution link to some of those.