Are we going to see much more support for restricting immigration in an era of mass automation?

Those like Reich and Goolsbee-who think manufacturing is healthy and the employment decline is due to productivity increases- point to the fact that the change in real manufacturing value added, relative to GDP, is stable. But what's obscured is that, in 2010, thirteen of the nineteen manufacturing sectors were actually producing less than in 2000. But, more importantly, when corrected for the problems identified by Mishel-overestimation of output in the computers and electronics sector, and problems with how inputs are measured-manufacturing output actually fell over the last decade, while GDP increased by 17 percent. Employment in manufacturing is declining mainly because of reduced output.

Productivity growth, rather than being the cause of declining manufacturing employment, is the prerequisite for manufacturing employment growth. This should not be surprising. Only highly efficient factories can survive in todays global economy. William Nordhaus has shown that, within each manufacturing industry, increases in the rate of productivity growth were associated with increases in the rate of job growth from 1948 to 2003. A Brookings Institution study extended the Nordhaus study until 2009 and concluded that "there is no evidence that productivity increases were significantly correlated with job loss." It is not inevitable that manufacturing will decline. Many nations with higher manufacturing wages than the United States- including Germany, the Netherlands, and Norway- have seen either stable or increasing manufacturing output as a share of GDP. Other countries- Korea, Austria, Poland, Finland, the Czech Republic, and the Slovak Republic- have actually seen their manufacturing sectors grow as a share of their economy.

This question of what happens next is thus central to the economy’s trajectory. Without the productivity gain, nothing happens. Workers able to produce more than before, for whatever reason, is the sine qua non of economic progress. But only if accompanied by rising output are the effects for workers undeniably positive.

Historically, that has been the dynamic. From 1947 to 1972, for instance, economy-wide productivity roughly doubled. But output surged as well and, at the end of the period, the same share of the population was working and men’s wages were up 86%. In the manufacturing sector, productivity rose by 3.4% annually, but real value added rose by 4.2% annually; employment during the period rose by more than three million.

Compare that period to the 21st century, when America has lost nearly five million manufacturing jobs. Was any of this because of extraordinary technological breakthroughs that caused productivity to surge, allowing firms to do much more with many fewer workers? No. In fact, the average rate of productivity growth in manufacturing this century has been 3.1%—lower than 1947–72 and no different than 1972–2000. But output growth has been only 1.3%, less than a third the rate of the earlier period. We’ve gone from the world where firms use a doubling of productivity to double output, to one where they use it to lay off half their workers. Had output growth this century equaled that of 1950–2000, manufacturing employment today would be near an all-time high.

So when policymakers blame automation for job losses, they are looking in the wrong place. Productivity gains have always been with us—in fact, they used to come faster. If anything, the American economy is suffering from insufficient automation—as reflected in declining productivity growth, stagnant wages, and remarkably little use of robots. American manufacturers use only 200 industrial robots per 10,000 workers, the standard measure of adoption. In both Germany and Japan, that level exceeds 300. In South Korea, it exceeds 700. With greater automation and higher productivity, American firms would likely be more competitive in the international economy.