A key claim made by economist Tyler Cowen in his excellent book Average Is Over: Powering American Beyond the Age of the Great Stagnation is that the US middle class is thinning out due to job polarization and that the trend will continue in the coming decades as routine middle-wage jobs are automated. Cowen:
Of the jobs lost during the recession about 60 percent of them were in what are called “mid-wage” occupations. What about the jobs added since the end of the recession? Seventy-three percent of them have been in lower-wage occupations defined at $13.52 or less. This general trend, namely more rapid job growth in low–paying jobs, can be seen in the numbers from 1999-2007 so we can’t blame it on the financial crisis or the particular problems of today.
I believe Cowen here is using 2011 data from the National Employment Law Project. The NELP has since updated its numbers through the first quarter of last year, finding that things have improved a bit with lower third of jobs now accounting for 58% of the new jobs versus 73% in Cowen’s book. But still, middle-wage occupations — which, again, constituted 60% of the recession losses — accounted for just 22% of the recovery job gains.
More recent numbers from a new research note from Goldman Sachs also support Cowen. Using Department of Labor data, the bank first ranked 100 industries by their prerecession wages and then divided them by fifths. GS finds that although the “hollowing out in the middle is real, it is not unique to the post-crisis period.” Each recession has sharp drops in the middle with no accompanying sharp rebound during the recovery. GS:
First, the decline in employment during the recession was sharpest in middle-wage industries.
Second, employment growth has been considerably more even during the recovery, with slightly faster growth in low-wage industries.
Third, the net effect since the recession began–a moderately greater decline in employment in middle-wage industries than in low- and high-wage industries–is quite similar to the longer-term trend prior to the recession (that is, the group of bars on the left follow the same pattern as the group of bars on the right).
This occupation-level view sends a similar but slightly stronger message. Once again, growth at the extremes–service employment at the low end of the wage distribution, professional and management employment at the upper end–outpaced growth in the middle in the decades prior to the recession.
No surprise, the construction industry is especially weak. But two other big losers, production and office administration, I would think are vulnerable to both to offshoring and technology. And another losing occupation, transportation, would seem vulnerable in the future to driverless cars and trucks. I would urge US policymakers to read Average is Over, as well as the upcoming The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies by MIT’s Erik Brynjolfsson and Andrew McAfee.