The current U.S. election cycle has seen every major candidate promise to bring manufacturing jobs back to the United States from overseas. Donald Trump has loudly pledged to “take our jobs back from China,” and Bernie Sanders has extensively criticized American trade policies. Even Hillary Clinton, who previously supported trade agreements like the Trans-Pacific Partnership, rails against the very same deal she helped to create. No matter what the candidates say, manufacturing jobs aren’t coming back to America. These election year promises are based purely on nostalgia and a total misunderstanding of economics. What’s worse, the U.S.’s presidential candidates are apparently blind to the newest wave of structural changes that are already impacting employment in the United States.
U.S. manufacturing employment peaked at just under 40% of the workforce during the Second World War. During World War II, massive government spending on military equipment created a great deal of demand for factory workers. From the end of the war up until the present day, manufacturing employment as a percentage of the workforce has been in a virtually uninterrupted decline. Today, only a little over 8% of the U.S. workforce is employed in manufacturing. In absolute terms, U.S. manufacturing jobs peaked at just under 20 million in 1979. Today, just 12 million Americans work in manufacturing.
The story behind this remarkable decline is not, as is commonly believed, the offshoring of American factories. The real story is automation. U.S. manufacturing output – the total value of goods manufactured in the United States – grew throughout the entire period described above. Even while factories in key industries like steel and automobiles moved overseas, the U.S. produced more goods than ever before. It simply did so with less workers. As industrial economies develop, firms substitute machines, robotics, and computers for laborers, and they develop more efficient ways of operating. This is an absolute necessity if they want to compete with other firms who are doing exactly the same thing.
Even while Japanese automakers were supposedly “stealing American jobs,” Japanese manufacturing employment declined continuously as a percentage from 1973 onward. China, once the world’s preeminent source of cheap manufacturing labor, is now the world’s largest market for industrial robotics. Manufacturing employment in China peaked absolutely in 1996, and it has been declining gradually ever since. When one examines these statistics, the absurdity of the proposal to “bring jobs back from China” becomes perfectly clear. The manufacturing jobs that the U.S. lost in the preceding decades did not move on to greener pastures – they no longer exist. There is no way to bring back these jobs as machines can do them better, and cheaper, than any American worker.
Some commentators have (incorrectly) pointed to “reshoring” as a potential source of resurgence in U.S. manufacturing. In reshoring, companies bring factories back into the United States, typically in low-wage, non-unionized areas like the American South. But as some authors have pointed out, these factories are highly automated and do not generate nearly as many jobs as their counterparts from the 1950’s and 1960’s. We can thus expect manufacturing employment to follow the prevailing trends of the past 70 years: output will rise, even as jobs vanish.
This war of words around U.S. manufacturing is a battle about trends and events that have already come and gone. As early as the 1960’s, James Lee Boggs wrote that cybernetics and machinery were causing U.S. automotive production jobs to dry up. It is politically convenient for present-day candidates to appeal to manufacturing’s heyday. Many Americans look back on the decades following the Second World War as a golden era of high growth, low unemployment, and rising living standards. Unfortunately, nostalgia seldom makes for good economic policy. Even more troublingly, the misplaced focus on manufacturing has blinded many Americans to coming changes that may be just as radical as those that struck U.S. manufacturing workers in the late 20th century.
As manufacturing jobs started vanishing in the United States, most displaced workers found employment in the service sector. Services – comprising every non-agricultural, non-manufacturing job – have always employed that majority of the U.S. labor force. Around the beginning of the 1990’s, it started becoming apparent that many service industry jobs might be vulnerable to automation as well. Office automation and corporate restructuring laid off secretaries and mid-level managers, igniting fears of a new wave of structural unemployment.
Today, the imminent possibilities for automation in the service industries are readily apparent. The transportation industry, which is the world’s largest in terms of employment, will be completely revolutionized by the invention of self-driving cars. Massive companies like Amazon are experimenting with totally automated warehouses. Not to be left behind, firms in the food service industry have invented fully automated burger-producing machines and baristas. Massive strides are currently being made in machine learning, which will allow computer programs to gradually overtake the roles of many present-day office workers. A famous study by Carl Benedikt Frey and Michael Osborne suggests that 47% of U.S. jobs are highly susceptible to automation in the next couple of decades.
We might imagine that the impacts of this coming revolution will be felt many years from now, but they are already upon us. Employment for prime-aged U.S. workers – those between the ages of 25 and 54 – peaked in 2000. After both the recessions of 2001 and 2008, the percentage of these employed workers never recovered to pre-crisis heights. From 2000 to the present, this employment ratio dropped 4% – a total of 5 million workers. In this light, it’s easy to understand the anxiety that many American workers currently feel. Just like their European counterparts, U.S. workers are suffering from a gradual rise in structural unemployment, and the technological trends identified above are only going to make this problem much worse.
It’s time to start talking about how the developed economies are going to deal with automation and structural unemployment without falling back on nostalgic gibberish about manufacturing jobs. The growing obsolescence of human labor is a general long-term trend, not a short-term problem. During the Industrial Revolution in the 19th century industrial productivity rose tremendously, but long-term structural unemployment was relatively rare. Working people fought for shorter hours, and their efforts paid off. From 1840 to 1940, the workweek in U.S. manufacturing fell from 70 hours to 40 hours. Today we still have 40-hour workweeks, even though productivity is much higher.
It is absurd to think that a society with self-driving cars, robotized warehouses, automated restaurants, and highly computerized offices will still be working 40-hour weeks. This is a recipe for massive structural unemployment. When the demand for human labor falls, the only sensible thing to do is reduce the length of the workweek. Instead of adding to the ranks of the unemployed, we can all benefit from rising productivity. Yet contemporary politicians have said nothing about reducing the length of the workweek, and some are even proposing that we work longer hours. Just as was the case in the Industrial Revolution, it falls to ordinary people – both the employed and those who are now excluded from the labor force – to demand an end to long hours and unemployment. The solution to our present dilemma certainly isn’t a manufacturing revival, but it may be a 20-hour workweek.
 I suggest examining prime-aged employment data to exclude the possible impacts of changes in the population of retirees and students. Many economists use these two groups to explain structural shifts in the U.S. labor force, but as I show above they cannot fully explain the decline in U.S. employment as a percentage of the available workforce.