Here’s a curious phenomenon that America’s political leaders and media need to focus on more: Practitioners of what I call fakeonomics keep prattling on about how the nation’s trade policy has almost nothing to do with the woes of domestic manufacturing and its workers, and that more automation and other forms of productivity gains are almost entirely to blame. At the same time, the official (and most authoritative) data keep proving them wrong. Just look at the latest figures, from this morning’s government report on America’s labor productivity.
These figures – from the Labor Department’s Bureau of Labor Statistics (BLS) – measure how many hours of human work it takes the U.S. economy to produce a given quantity of output. So although they don’t directly gauge the role being played by machines and software of all kinds in American industries (including manufacturing), they’re no doubt an excellent proxy for automation levels and how they’re changing. In other words, if labor productivity is rising at a respectable rate, then it’s reasonable to conclude that at least lots of job loss is indeed attributable to the adoption of labor-saving technologies – as opposed to more net imports or more job offshoring. And the opposite holds logically, too,
As I’ve written, BLS itself admits that its published numbers likely overstate labor productivity growth for the whole economy – and therefore, it would seem, the automation rate. But even leaving aside this flaw, the new numbers once again show that “respectable” is the last adjective that’s appropriate to describe improvement lately in this gauge of efficiency.
BLS’ new figures for the fourth quarter of 2016 and for that full calendar year incorporate comprehensive revisions going back to 2012. Given how many quarterly and annual figures have been updated, the most efficient way to show how miserable productivity growth has been recently, and how weakly it’s correlated with manufacturing employment, is to show how the previous figures for the last three economic recoveries before the new adjustment and after it. (As discussed before, comparing economic performance during similar phases of a business cycle – e.g., recoveries versus recoveries, or recessions versus recessions – is the best way to get apples-to-apples data.)
Here are the manufacturing employment and manufacturing labor productivity changes for the expansion of the 1990s (as presented in this previous post):
manufacturing employment: -0.38 percent
manufacturing labor productivity: +46.78 percent
And here are the results for the shorter expansion of the 2000s (taken from the same post):
manufacturing employment: -12.44 percent
manufacturing labor productivity: +41.08 percent
And here are the results reported in that post for the current expansion – which has been longer than that of the previous decades, but not quite as long as its 1990s predecessor:
manufacturing employment: +4.68 percent
manufacturing labor productivity: +22.88 percent.
Where do the new revisions leave us? Manufacturing employment is now up 5.24 percent. And the sector’s labor productivity is now up 22.70 percent. In other words, it’s productivity growth has been slightly weaker than previously estimated – when it was already historically lousy.
It’s hard to escape the obvious conclusion: It’s high time for globalization cheerleaders to stop letting trade off the hook for major manufacturing job loss – and transition to work that’s actually productive.