Atif Mian and Amir Sufi write,
what we want to focus on today is the remarkable separation in productivity and median real income since 1980. While the United States is producing twice as much per hour of work today compared to 1980, a small part of the gain in real income has gone to the bottom half of the income distribution. The gap between productivity and median real income is at an historic all-time high today.
Pointer from Mark Thoma.
They are comparing average output per hour to real median income per family. Rhetorically, they attribute the divergence in the two ratios to the numerators. That is, average output has grown faster than median income. However, there are also two denominators at work. The first denominator is “hours of work.” The second denominator is “family.” I suspect that a fair amount of the divergence in the two series is due to divergence in the denominators. That is, I suspect that the ratio of “hours of work” to the number of family units fell markedly between 1980 and today. There is a downward trend in hours worked, particularly for men. In addition, there is an upward trend in the number of family units, due to divorce and lower propensity to marry.
It would seem that this would be an easy issue to check.
The “debate” over the great productivity-income gap is another disappointing example of how even the masters of data analysis simply refuse to strip data down to comparables.
This analysis http://www.heritage.org/research/reports/2013/07/productivity-and-compensation-growing-together (cited today in e21) shows how the alleged gap can be closed to levels that would render the controversy moot.
Incidentally, accusations — I would not call them debates — about invidious gender pay gaps, long-term stagnating wages , economic inmobility, all share an equally blatant disregard for rigor.)
Tyler Cowen certainly thinks household composition won’t do the trick:
http://marginalrevolution.com/marginalrevolution/2013/09/does-changing-household-size-resurrect-american-economic-performance.html
It’s not an issue I know a great deal a lot, but there is a certain grasping-at-straws vibe to many of these arguments, from people like Scott Winship. I think a good rule of thumb for looking at these issues is “does this argument for why observed changes aren’t real also hold for pre-1973?” If so, it’s probably not a good one.
For example, if you trot out “changing households” as why median income stagnation isn’t real, ask yourself “Would household composition changes not have had an effect when we saw observed income changes?” I don’t think that’s the case, although I could be wrong.
Other examples of these people use are “CPI bias” and “consumer surplus from the internet” and “new goods”, as if there wasn’t surplus from a new good like, say, penicillin.