So, I’m driving around doing errands this past weekend when I hear former Secretary of Commerce Carlos Gutierrez (in GW Bush’s cabinet) interviewed about renegotiating NAFTA. He’s a big booster of the trade deal and wanted to make the point that any job loss in manufacturing was a result of faster productivity growth, not imbalanced trade. His evidence was as follows (my bold):
One of the things I go back to very often is our manufacturing as a percent of GDP. Our manufacturing output is pretty stable, pretty flat. If you go back 10, 15 years, it’s between 12 and 14 percent. But our manufacturing workforce has been declining steadily. So we’re producing the same output with fewer people. What that tells me is that technology is more of a threat to American jobs than trade.
Stable manufacturing output share of GDP?! I practically dropped the dry cleaning!
The first figure shows that, in fact, manufacturing’s share of output (blue line) has been falling since I was born in the mid-1950s. It was 11.7% in 2016, 13% in 2006, and 13.9% in the 2001. OK, that’s roughly between 12 and 14 percent, but it ain’t stable. It’s falling, and pretty steadily. In fact, Louis Uchitelle just published an important book on this long-term trend.*
The red line shows the decline in manufacturing employment, which is also steadily declining, such that the most we can determine from these trends is that we’re creating a smaller share of output with a smaller share of workers, as the scatterplot of the data in the two figures clearly reveals.
The employment share falls faster than the output share, so sure, there’s been productivity growth in manufacturing (though as I show here, not so much of late; see also Sue Houseman’s important work on this point). But this endless drumbeat about manufacturing jobs falling prey to automation seems at least woefully incomplete, if not non-economic.
First, trade and automation interact and are thus not clearly separable forces. Technological advances in communication and transportation interact with globalization to facilitate offshoring and deeper global supply chains. Nor should anyone convince themselves that it matters to workers which force is displacing them. Gutierrez implicitly argues that because it’s automation, not trade, displaced workers should somehow be assuaged. Hey, all they need to do is go from running a drill press to designing, building and programming drill-press-running robots!
Second, if productivity killed jobs, full stop, unemployment would typically be through the roof. The intervening variable has always been demand, as productivity growth generates rising real incomes that supports stronger demand and higher living standards, at least on average (if not “on median”).
We thus must ask ourselves what happened to diminish the demand for American manufactured goods such that employment in the sector fell, both in absolute terms since 2000, and as a share of total employment for decades, as shown in the figure. The answer is, of course, import substitution. American consumers have increasingly met their demand for manufactured goods through imports.
That has its pluses, as Gutierrez points out, but also its minuses: it has certainly hurt vulnerable manufacturing communities, in no small part because policymakers refused to help them adapt to the change. Moreover, this import substitution was much exacerbated, especially in the GW Bush years, by persistently large trade deficits in manufactured goods.
What does any of this hafta do with NAFTA? It’s just way too simple to brush people’s discontents about trade deals like NAFTA to automation or their failure to side with elites on how benign globalization has been (Hey, it’s reduced prices! Cheap stuff, everybody! Stop complaining!).
On what to do next, read Lori Wallach and I here for ways to make trade work better for workers, and read Dean Baker here for a highly efficient read on the costs of ignoring the downsides of globalization (of which the steepest is President Trump).
*A friend suggests a good point. Both of these shares–output and employment–fall in all advanced economies–see Germany, e.g., which at least recently runs large trade surpluses–as economies get richer and services overtake manufacturing. My point is that the decline in manufacturing output share has not been stable while employment tanked, which would imply faster productivity gains (or automation) than actually occurred.