Scott Winship has an expanded and annotated version of his recent National Review article (“Men’s Rising Earnings“) at the Brookings Institution website. He begins by citing an influential study by Michael Greenstone and Adam Looney of the Hamilton Project:
Greenstone and Looney, whose work I generally admire, argue that the rise in labor-force dropout among working-age men masks frightful trends. People who spend the entire year jobless are not usually included in earnings statistics, but if these men, most of whom are less skilled, did work, they would be factored into the Census Bureau’s Current Population Survey (CPS). The result would be not the stagnation in male earnings seen in Census Bureau figures, but sharp decline. Taking this into account, the Hamilton Project reports that median earnings among men ages 25 to 64 declined by an astonishing 28 percent from 1969 to 2009.
Scott suggests that the finding is misleading for the following reasons:
1. He suggests that analysts should rely on the CPI-U-RS, which the Census Bureau and the Bureau of Labor Statistics use to analyze income trends, rather than the CPI-U, as the CPI-U is widely believed to overstate inflation. Using the CPI-U yields a 12 percent decline in median earnings among men ages 25 to 64 while using the CPI-U-RS yields a decline of 1 percent.
2. Rather than count all non-working men ages 25 to 64 as below-median earners, Scott offers an alternative assessment that only counts non-working men who report not being able to find work and those who report being sick or disabled, leaving retirees, students, stay-at-home dads, armed-forces members who live in barracks, and men living in mental institutions or residential-care facilities out of the analysis:
If we accept the Hamilton Project’s correction but follow the Census Bureau’s practice of using the CPI-U-RS, the 28 percent decline through 2011 shrinks to 19 percent. If we then include as below-median earners only those non-working men who report being sick or disabled and those who report not being able to find work, median earnings fall just 12 percent. Finally, if we compare the peak year of 1969 with 2007, another peak year, rather than with 2011, a year in which the economy was recovering from the worst downturn since the Great Depression, there is no decline at all.
3. Scott then makes an effort to account for how immigration has changed the composition of the U.S. labor force. Unfortunately, it impossible to identify immigrants in the data before 1993, but Scott uses the Hispanic population as a crude proxy:
Assuming the 1969-to-1970 earnings change was the same as that for men in general, median earnings among non-Hispanic men declined from 1969 to 2011 not by 12 percent but by 8 percent. From 1969 to 2007, they rose by 2 percent. Among Hispanic men, earnings fell by 24 percent through 2011, but this decline is simply a more dramatic demonstration of how rising immigration pulls the median downward. Among non-Hispanics, the earnings increase for blacks was stronger than for whites, so this is not simply a story about non-Hispanic whites’ doing better than everyone else.
4. He also argues that any assessment of how compensation has changed over time ought to take into account non-wage benefits, which represented a much larger share of total compensation in 2011 than in 1969.
Scott raises a number of other issues as well, e.g., he argues that analysts ought to use the PCE (personal consumption expenditure) deflator rather than the CPI-U-RS, as it does a better job of accounting for substitution bias. One thing should be clear: Scott is certainly not arguing that income trends in recent decades tell us that all is right with the U.S. labor market. A number of demographic, cultural, and economic changes that have shaped labor market outcomes for U.S. men in recent decades, including a large influx of less-skilled immigrants, early retirement, rising incarceration rates, and rising female labor force participation. The overall picture really is alarming, in my view, particularly as it relates to institutionalized men. But Scott’s done us a great service by clarifying some of the underlying data.