The quickly settled International Longshoremen’s Association strike takes us one more step toward the Great Inversion: a future in which people in the skilled and semi-skilled trades boast higher average wages than most college graduates. If the justification for most college degrees—only 10 percent of them in STEM fields—reduces to their boosting future earning power, then those graduates will have a problem. (Granted: a college education should be about more than just maximizing earnings.)
It has escaped no one’s attention that the settlement with the longshoremen’s union will bump up annual starting salaries to about $80,000, with mid-career salaries over $150,000. Both benchmarks are higher than those of 90 percent of college grads. The longshoremen’s victory will likely set a market norm, not just for union members, who account for just 10 percent of the workforce, but for the skilled workforce at large. The explanation is found at the intersection of technology and demography.
It’s no small irony that, while the longshoremen’s union railed against automation, it is automation that will accelerate the Great Inversion. Its wage-enhancing role arises from the inexorable logic of demographics.
The most consequential demographic development is not just that the U.S. workforce is aging, but that it’s approaching a historical first: there has never been a society, anywhere, with fewer young people than old people. With skilled trades, some of the implications of this unprecedented inversion are obvious and can be summed up in a simple logic chain:
Rising wealth, regardless of slowing population growth, leads to greater consumption of physical things, which entails more mining, manufacturing, energy production, movement of goods and materials, and maintenance. Even the booming services sector requires physical machines and products; iPhone apps and Zoom conferences alike are dependent on hardware.
The workforce is shrinking fast, because the average age of employees in the trades is older than the already-aging population, and the share of students entering those fields has been in decline for decades.
It’s far easier to use information tools to automate information tasks than physical ones. Thus, software and artificial intelligence (AI) more easily and reliably replace “knowledge workers,” not only those responsible for rote tasks but also those who generate much of what passes for creativity—for example, advertising copywriting, boilerplate television scriptwriting, or basic legal tasks, not to mention myriad aspects of documentation.
AI is accelerating the decline in demand for many “knowledge” jobs—and downward pressure on wages for them—as it shifts knowledge-centric tasks away from the “back office” and toward people on the frontlines, where non-college skilled workers can easily use AI tools.
Economists would say that all of the above would naturally induce a migration into trade-related training and occupations. Indeed, that shift is underway: recent data point to enrollments falling for colleges and rising for trade schools. Ordinarily, this would suggest a looming oversupply in the skills market, which, in turn, would in due course exert downward pressure on wages. Add to this the fact that higher wages generally induce employers to speed up adoption of automation.
But this time, it really is different. Why? A trifecta of counterforces.
First, there’s the demographic cliff. The older-trades workforce is shrinking faster than the population as a whole, but on top of that, a birth dearth reduces the future labor pool. These two datapoints combine with a third factor—the newfound political enthusiasm for reshoring U.S. manufacturing—that amplifies the demand for skilled labor. This is where automation and especially robots play a new role.
The vast majority of jobs in the skilled trades remain unautomated because most physical tasks, even those as simple as carrying heavy objects, are devilishly difficult for autonomous machines. In fact, until now, some 90 percent of all industrial robots are used by just 10 percent of manufacturing firms. Unsurprisingly, the largest firms are where one finds most robots because the tasks are in the easy-to-automate, high-volume production lines, where robots can be physically isolated for safety. Only in the past decade have innovators finally developed the kinds of robots that can operate safely alongside people in complex and unconstrained environments. A combination of powerful on-board AI and sensors, advances in materials and motors, and lithium batteries have made such robots possible.
Now, hundreds of companies are seeking to commercialize the kinds of automatons that were once the sole province of science fiction. At least a dozen firms have machines on the cusp of viability, and a few have already brought them to market. Amazon has deployed several types of autonomous wheeled and walking automatons in warehouses to aid its workforce. And Elon Musk’s anthropomorphic Tesla robot program may in time become more valuable than its automobile lines. For now, most of the leading-edge companies are in the U.S. and Europe, but several impressive Chinese start-ups are chasing the robotic Holy Grail for precisely the same reasons. (Indeed, China’s looming labor problem is far more severe than ours.)
The dirty little secret of automation, especially where autonomous robots are concerned, is that it still requires people, typically higher-skilled and more highly-paid, especially in complex, high-consequence physical work environments. Thus, it would benefit longshoremen and others in similar trades to shift from opposing automation to embracing it, with this caveat: they need a bigger say in its implementation. That’s something management should also embrace.
One of the biggest impediments to effective adoption of automation, especially robots for physical tasks, is that design engineers too often fail to appreciate the operational, practical, and human-centric challenges of integrating new classes of machinery onto the frontlines. In the end, the rapidity of adoption will be determined not just by the price of robots (and thus the equivalent hourly robot wage) but also the ease with which they can be integrated, especially working alongside those skilled-trade employees. More rapid adoption of productivity-enhancing automation is just what’s needed to generate the profits that enable higher wages and, in turn, sustain businesses and lower costs for consumers.
This virtuous circle of automation and labor rests on that fact that demand for skilled labor will only grow. This reality is often overlooked by observers confused by the oft-used farm-factory analogy. It’s a version, less politely put, of advising workers in the trades to “get over it,” since America has more food and far fewer farmers today, thanks to automation. This is a fallacy arising from a category error.
How so? When it comes to potential consumption, there’s a world of difference between things and food. Not only can demand for the first grow far faster than for the second, but there is also no limit to the former and a clear limit to the latter. Setting aside starvation-level economies, where food demand is by definition unmet, in mature economies food consumption/production only rises with population growth. But demand for manufactured items grows as fast as wealth does, and as fast as innovators can invent new kinds of products that people want to buy. Innovators, in effect, create new demands for goods; farmers cannot create new net demand for food. (This has profound implications for the energy sector, too—another skilled domain.)
The profound difference between the categories of food production and fabricated things is clear in the data. Over the past half-century, U.S. agricultural consumption has essentially tracked population growth, with both rising by about 80 percent, while industrial goods consumption has risen about 300 percent.
Thus, history does show that productivity growth lessens the need for farm labor faster than the need for agricultural output. But, that’s not the case with manufacturing—unless, of course, labor shifts overseas. Here again, the data show that, over the last half of the twentieth century, even as manufacturing productivity rose (that is, fewer labor-hours per output), the U.S. manufacturing workforce remained surprisingly unchanged and really started to shrink only when an increasing share of manufactured goods were imported—namely, when the production and labor were exported.
The assumption that “globalization” and cheap overseas labor has driven the offshoring trend is only partially true. Possibly the bigger factor over the past few decades in the U.S. has been the ever-expanding regulatory state that has discouraged manufacturing in the U.S.
As a National Association of Manufacturers analysis shows, the average large manufacturing firm spends $25,000 per year per employee in regulatory compliance costs, while small firms spend $50,000 per year per employee. It’s useful to note that the NAM survey shows that the biggest share of U.S. regulatory compliance costs come from the one-two punch of complex “economic rules” and “environmental” compliance. While similar studies for Chinese firms are not available, China’s government more likely provides subsidies rather than regulatory burdens.
Thus, the success of reshoring will depend on three factors.
Congress and the next administration will need to look for ways to make the U.S. friendlier to widespread industrial expansion. It will take more than targeted and strings-attached government subsidies and inducements to repatriate industries.
Policymakers must also avoid trying to fix what’s not broken. The U.S. has the most affordable and reliable energy infrastructure in the world. Industrial production is inherently energy-intensive. As it stands, far too many state and federal energy policies are increasing costs and degrading reliability. Europe has already run that experiment for us; it is de-industrializing before our very eyes, in large measure because of high-cost energy.
Policymakers, unions, and manufacturers need to embrace automation and robots. There is no other way to resolve the skilled labor shortage in a meaningful timeframe.
We can’t slow aging (though we may be able to extend our working years), and the birthrate won’t accelerate anytime soon. Rather than indulging in comic-book-style apocalyptic narratives, we should be exploring the opportunities that AI creates. In general, if labor-saving technologies were net job destroyers, unemployment would have risen continually over all modern history.
No doubt some will shout “bravo” for the longshoremen’s victory—not only those in similar occupations, but also consumers worried about another supply-chain disruption. But for those who lament the inflationary spiral that higher wages may induce, a better option exists: embrace a more automated future.
Photo by Spencer Platt/Getty Images