On getting your facts right
Matthew Martin 1/02/2014 01:16:00 PM
In this case, Bret Stephens's biggest error was in using census data to look at incomes of the few richest individuals in the US. That's problematic because the census data is "top coded" which means they truncate all data values above a predetermined threshold. In this case, all millionaires sampled by the census bureau have their incomes recorded as $999,999. Given that nominal incomes rise with inflation, this means that over time an increasing share of the sample will be at the top coded threshold, resulting in compression of the measured income distribution regardless of whether inequality actually rose or fell. All you can really say from this data is that inequality among the poorest 99% of Americans hasn't changed much, which is something we already know--the growth in inequality has been concentrated within the wealthiest 1%, whose actual incomes are not recorded in the dataset Stephens used.
Knowing how to find and use data is a critical principles of macro skill, and Bret Stephens hasn't got it. Such a mistake might be forgivable if it only happened once; unfortunately all this has happened before, and I'm betting it will happen again.