Automation Nation? Exaggeration!

No, robots won’t cause mass unemployment. But our failure to prepare people for the high-end service economy could be a real disaster.

By Stephen Rose Ruy Teixeira

Tagged economyEducationinnovationJobsManufacturing

What are we to make of the rise of the new service economy that is radically changing how Americans make their living? Interestingly, commentators on both the left and the right advance a common—and very gloomy—narrative about what’s happening to American jobs. They see an economy out of balance, with low-skill, low-pay, dead-end service jobs replacing the good, high-paying manufacturing jobs of the past. They wonder where the middle-class jobs of the future will come from—or indeed, if there will be any such jobs at all.

For the Trumpian Republicans, this is the “American carnage” that is destroying the middle class and everything it holds dear. Meanwhile, many on the left are scarcely more optimistic: Jeff Faux, the former president of the Economic Policy Institute, writes in The Servant Economy that our failure to protect manufacturing jobs in the 1980s doomed the middle class, and predicts that by the end of the 2020s there will be a 20 percent drop in the real wages of the average American, including a large contraction in professional jobs due to continued outsourcing.

This view is mistaken: It confuses current problems like high-income inequality and stagnant wages, which have many causes, with structural economic shifts that have profoundly and irreversibly transformed how everything in our economy is produced. Where we work, how we work, what we consume, and how we consume it have all been radically altered by the relentless march of technological change and educational upgrading. These shifts reflect our ability to produce more commodities with fewer but more highly skilled workers and to produce a wider range of goods that more consumers can purchase. This is both inevitable and, on balance, a good thing.

Consider the massive changes that have taken place in our economy since World War II. In 1947, nearly half of American workers were employed in the goods-producing industries of manufacturing, mining, agriculture, and construction. By February 2018, this share had dropped to 14 percent, while 32 percent of the labor force had moved into service industries. In terms of the raw number of workers, overall employment grew by 97 million during those decades, despite employment in goods-producing industries declining, meaning that all job growth was indeed due to service employment. Much of this growth happened because of the large influx of women workers, but the share of male workers in services grew tremendously as well.

At the same time, the educational attainment of America’s workforce has skyrocketed. In the 50-year time period between 1967 and 2017, the proportion of high-school dropouts in the workforce fell from 37 percent of adults to just over 8 percent, while those with a four-year college degree or more rose from 13 percent to 37 percent. Looking at postsecondary education as a whole, those with at least some college went from just one quarter of adults all the way up to 63 percent.

This is a remarkable upgrading in the skills of the workforce. And it was repeated in all advanced industrialized countries, as new production processes demanded more skills for workers across all occupations. Even as the share of people going to college was growing, the college wage premium (the difference between average college graduate and high-school graduate wages) in the United States grew as well: The earnings premium for workers with a bachelor’s degree over workers with just a high-school diploma jumped from 40 percent in 1980 to about 73 percent today. This huge increase in demand for the highly educated was fundamentally driven by the massive shift out of goods production and into the service industries.

That means the popular portrait of a service economy replete with low-skill, dead-end “McJobs” makes no sense. The swelling ranks of the college-educated have been accommodated by the new service economy; they are mostly employed in relatively high-paying jobs. Recent research traces in detail what these educated workers do, showing that the increased education levels of the workforce correspond to a dramatic “upskilling” of our job structure over time. Few people realize that nearly half of all detectives and policemen have a four-year bachelor’s degree, usually in criminal justice. Further, registered nurses at most large hospitals are now required to have an undergraduate degree as well, and many even have a master’s. Finally, those with graduate degrees who were once highly concentrated in positions at universities and K-12 schools, are now managers, analysts, and researchers for private firms, and they
typically receive higher pay.

The real culprit is not lost manufacturing jobs. It is stagnating pay for the non- college educated, for those in manufacturing and all jobs.

To show how the occupational distribution has changed over time, co-author Stephen Rose, in his 2010 book Rebound, sorted workers into three occupational tiers: The top tier included all managers and professionals; the middle tier consisted of the middle-skill jobs of supervisors, the nonprofessionally self-employed, skilled blue-collar workers (including carpenters, electricians, plumbers, and machinists), police and firefighters, technicians, and clerical workers; while the bottom tier included factory operatives, laborers, truck drivers, longshoremen, retail sales clerks, and food and personal service workers like security guards, janitors, and waitresses. Between 1967 and 2015, the share of top tier managers and professionals grew from 21 to 37 percent. In contrast, middle-skill jobs declined from 40 to 34 percent and the low-skill share declined by 10 percentage points, from 39 to 29 percent. These shifts are consistent with the higher educational attainment of today’s workers and, again, completely contradict the narrative about the increasing number of dead-end, low-paying McJobs.

In Rebound, Rose also developed a new “functional” way of looking at employment by focusing on the workplace and the type of work being carried out. This approach highlighted the importance of office workers in a way the traditional divisions of industries and occupations haven’t been able to. In particular, the front offices of manufacturing, transportation, health care, and other industries were combined with finance, government administration, and business services into one common office category. By combining front office workers in other sectors with workers in finance, business services, and government, the true size of office work in our economy—nearly half of the workforce—can be discerned.

The other four sectors included the nonoffice personnel in the high-skill services of health care, education, and communications; low-skill services (retail, personal, and food services); manual labor in industry; and primary production (mining, farming, fishing). In 2015, this approach showed that 44 percent of U.S. jobs were in offices, while another 20 percent were in high-skill services. Just 15 percent of jobs were manual labor, mainly manufacturing, construction, and transportation industries. Finally, 19 percent were in low-skill services and 2 percent in agriculture, mining, and timber. Since 1967, the big change has been the rise in office work and high-skill services (up 14 points as a share of jobs), while the big decline has been in industrial manual labor, down 13 points. Interestingly, the share of low-skill service jobs is just about the same today as it was back in 1967.

So, instead of the standard lament that manufacturing jobs are being replaced mainly by low-skill service jobs, these numbers help us see that that they are, in reality, being “replaced” by rising white-collar jobs in offices, as well as high-skill services. Even among non-college-educated men, where economic trends have hit hardest, a newer study by Rose from 2018 finds that just 23 percent of the decline in good-paying jobs was due to the movement away from manufacturing and into low-paying service industries. These facts contradict the traditional left narrative that fingers deindustrialization as the major cause of male labor force problems.

The Problem Is Declining Wages

The real culprit is, instead, stagnating pay for the non-college-educated within industries. There is certainly a problem here that needs fixing, even if it is not the one we’ve long pointed our fingers at. It is instead a problem of declining wages, for those in manufacturing and all jobs filled by those without college degrees. For example, in 1960 (the heyday of high blue-collar pay), men in professional and managerial jobs earned 46 percent more than men in manual jobs; by 2015, this premium rose to 94 percent. Since relatively few men without a college degree work in managerial and professional jobs today, this pay gap based on occupations is the real basis for the widely discussed rise in male earnings inequality between educational tiers.

Moreover, to fairly assess the significance of this trend in manufacturing jobs, we should be asking ourselves: Why is manufacturing declining and, further, what do actual manufacturing jobs look like today? First off, the overwhelming reason for the decline in manufacturing’s share of employment, from 36 percent 50 years ago to its current share of 11 percent, is high productivity growth, combined with shifts in demand toward services, reflecting a rise in consumer affluence. Half of the loss of manufacturing’s employment share occurred between 1953 and 1976, years in which America had a trade surplus. In another 2018 study, Stephen Rose found that only about one in ten lost manufacturing jobs could be attributed to the oft-cited culprits of trade and globalization. In this sense, declining manufacturing employment is most properly viewed as an inevitable product of capitalism’s evolution over the last half-century.

Finally, manufacturing industries, as currently configured, simply no longer provide as many blue-collar jobs as they once did, even controlling for the size of the sector. In manufacturing today, only about half of jobs are blue-collar, compared to 70 percent in 1960; with the decline of clerical staff, these white-collar manufacturing workers—nearly half of all manufacturing employees—are managers, sales representatives, accountants, human resource professionals, and accountants. Indeed, the bulk of noncollege blue-collar jobs for men—including those that pay well—are now provided by the nonmanufacturing-heavy industries of construction, transportation, utilities, warehousing, wholesaling, and trade, and not by traditional manufacturing. Thus, not only are manufacturing jobs unlikely to “come back,” as many on the left (and right!) wish, but such a development would not be the boon to manual blue-collar jobs they suppose.

What is the future of American jobs?

Technological change is disruptive and always has been—new products and companies can lead to a massive decline of competing companies and industries. Citing “Moore’s Law,” which says that the speed of processing will double every two years, many observers predict that machines will become more adept at performing both manual and intellectual tasks. Some of the examples of new IT and AI technology already in place include Baxter robots in advanced manufacturing, IBM’s “Watson” platform in health care, and even the thousands of touch screens at McDonald’s and other restaurants. Those that may soon be on the way include eldercare robots and self-driving automobiles and trucks. Looking at these current and projected changes, jobs pessimists see a massive displacement and impoverishment of American workers coming in the future.

The prediction that such technological advances will devastate American workers has been made many times before. In the 1950s, there was a similar fear that automation would lead to massive unemployment. And in fact, automation in manufacturing has proceeded non-stop since then as manufacturing output grew six times larger by 2015, using fewer employees than in 1960. But during these years, we have actually expanded the broader labor force greatly as waves of boomers, women, and immigrants have easily found employment without driving up the unemployment rate.

The trap that jobs pessimists fall into is that they view the economy through the lens of goods production, particularly in low-skill sectors like manufacturing. But this is woefully off-base when we consider the economic trends of the last 70 years and the fact that they are likely to continue. We’ve already discussed the changes in occupations, the rise in educational requirements, and the shift from manual production to office work, various types of health care, and education. All of these changes are accompanied by complementary modifications in what we consume that have drastically reduced the significance of goods production in the economy. Nothing shows this better than the declining shares of food and clothing being purchased between 1947 and 2016. In 1947, 47 percent of consumer expenditures went to these categories of goods, while in 2016 the corresponding figure was just 19 percent. This means that 28 percent of final consumption was freed up to spend in other areas of the increasingly services-oriented economy—areas that typically require a much different and more educated workforce than the one needed to produce food and clothing. And of course, that spending created a huge number of jobs.

Table 1: Consumer Expenditures, 1947 and 2016

Due to rounding, figures may not exactly compute.

In the future, the distribution will continue to change as, for example, recreation and health care increase in importance. Moreover, and very importantly, within categories there will be a shift to fancier furniture and appliances (included in the housing category) and eating out (included in the food category). John Maynard Keynes in the 1930s, and John Kenneth Galbraith in the 1950s, both thought that the compounding of innovation would lead to the satisfaction of all of our needs by the advent of the twenty-first century. Obviously, they underestimated our ability to “desire” more and more consumer goods and services. As new IT/AI advances make some goods cheaper, people will move on to the latest and greatest versions of these consumer goods. This process involves transforming the “luxuries” of the past into the mass consumption of the middle and upper-middle classes—e.g., as high-priced coffee, the latest smartphones, memberships in fitness clubs, and world travelling have already done. These shifts also mean new jobs to meet these new desires for consumer spending.

It’s not that easy to replace people

Jobs pessimists are also wrong in thinking that we can displace most of the workers who interface with customers. First, people often require a human touch. In health care, there is clear evidence that people’s attitudes affect their recovery and patients are unlikely to want to give up the personal comfort and advice that real-life doctors, nurses, and other health-care workers can provide.

Second, the growth of online retailing will never completely replace personal shopping, which for many people is an enjoyable pastime. The human touch is also essential to retail, where salespeople can encourage shoppers to buy, or provide consumers with expert advice. Even in restaurants with automated ordering, food and drinks are still delivered by the wait staff, who often are asked for recommendations. And in terms of automated checkout at grocery stores and other retail outlets, it is limited by customer confusion and fear of theft.

Third, even if all these advances do come to pass, they may not be adopted as quickly as jobs pessimists predict. For example, due to their limitations, self-driving cars will take a long time to be adopted. While their advanced sensors and GPS-based understanding of the roads provide many capabilities, there remain many unknowns that can’t be programmed easily. For example, these vehicles perform badly at four-way stops because they continue to wait for the intersection to be cleared, among other things. And even before the Arizona accident that killed a pedestrian crossing a road, these cars “were having trouble driving through construction zones and next to tall vehicles, like big rigs.”

Reducing the College-Wage Premium

In other words, contrary to what is so often believed, the jobs of the future will have much in common with the jobs of today, even as we slowly move away from manual labor to skilled labor and more business professionals. For example, the office economy, which employed 34 percent of workers in the past, grew steadily and did not lose its employment share—in fact, it gained strongly—even when the flexibility of computers reduced the employment of clerical workers. Currently, the vast majority of office workers have more skills and at least some college experience, and often a four-year degree.

While there may be some professions that see reduced employment, overall demand for workers in business decision-making and sales should remain high, given the trends we have outlined here. Further, a lot of the new technology requires programmers and technical assistants. For example, the rise of gaming, even where it replaces jobs in physical recreation, is creating many jobs for highly skilled programmers, designers, and videographers.

As for the decline of good-paying jobs for less-skilled—noncollege—workers, this is not primarily a question of the disappearance of manufacturing jobs, but rather how well these and other noncollege jobs will pay, as noted earlier. Indeed, nearly three-quarters of the decline in good-paying jobs for noncollege men in manufacturing is attributable to worsening relative wages for such workers within manufacturing and other industries.

What is to be done?

There will be jobs in the future, and plenty of them. But a reasonable case can be made that highly educated workers are being rewarded too much relative to their less-educated counterparts. Today’s approximately 73 percent premium for a college education (over high school) is too high, both because it is a signal that the new service economy is demanding more college graduates and because of its contribution to wage inequality. A balanced and healthy economy should see a wage premium closer to 50 percent—which represents a bit more than 10 percent per year of schooling attendance (compounded)—high enough to still make borrowing money to attend college a good investment. For one thing, this was the average premium rate for a bachelor’s degree over a high-school diploma between 1950 and 1970 in the United States (and that was higher than it had been in the 1930s and ’40s). Finally, such a premium is consistent with the premium level in many other industrialized countries; the European Union’s average is 52 percent with Sweden having the lowest premium; the French premium is 40 percent, the United Kingdom’s is 48 percent, Japan’s is 52 percent, and Germany, with the smallest share of college graduates, has a 58 percent premium.

We can reduce this college-wage premium by using the laws of supply and demand. Goldin and Katz, in The Race between Technology and Education, found that, from 1980 to 2010, the premium rose because the supply of college-educated workers increased more slowly than the demand for these workers. Consequently, to decrease the premium, we need to increase the supply of college-educated workers by around 20 million over the next 15 years. (Right now, we are on track to add only about 8 million.) Such an increase in the college-educated workforce would raise total economic output by $500 billion by 2025 (about 3.5 percent of GDP). And by increasing the relative supply of college-educated workers and decreasing the number of workers with at most a high-school diploma, the relative earnings of workers with different levels of education would narrow and take a big bite out of inequality, rolling back about three-quarters of the education-based increase in inequality we’ve seen since 1979.

We could get additional boosts to our economy if we substantially upgrade our tottering infrastructure over the next 15 years. Right now, American transportation, energy, and communications infrastructures all lag behind those of other advanced nations and are not appropriate for the needs of our high-skill service economy and the educated workers that drive it forward. This is a foolish oversight that makes our economy less productive and needs to be rectified.

Our analysis has important political implications. First, the common tendency among progressives to lionize manufacturing jobs and disparage service-sector jobs is profoundly counterproductive. Worst of all, such an attitude is at odds with the way the economy is actually evolving. The dominant and growing segment of the economy today is in high-end services; again, we are not exchanging manufacturing jobs for low-skill, low-pay “McJobs.” The central question therefore is not how we can get more manufacturing jobs but rather how we can build on the growth of high-end services by upgrading our workforce and creating an environment where these jobs can flourish, deliver high wages, and generate more rapid increases in our standard of living. This will be a challenge—and it won’t be met by a quixotic quest to reinvent America’s economic past.

Second, it is time for the left to stop worrying about an overproduction of college graduates. It is not the case that the economy is rife with overqualified college graduates filling low-skill jobs. This staple of news stories is yet another myth based on flawed studies that have almost half of all college grads overqualified for their jobs; in contrast, a 2017 study by Rose found that only one-quarter of college graduates were “mismatched” with their level of education at the current time, a figure that has been stable over time. In reality, given the high bachelor’s premium, we are actually under-producing college graduates; we need more than we have now or are projected to have in the future to meet the demands of our postindustrial service economy. Therefore, a central focus of the left should be on dramatically expanding the number of college-educated workers in our workforce.

A key problem to solve here is that too many of our high-school graduates are not adequately prepared to succeed in a four-year college. Students with a low “B” average or worse in high school have a very low probability of obtaining a four-year degree. This means that we also need a national drive to improve elementary and high-school education by building on lessons, positive and negative, from recent campaigns like race to the top, common core curriculum, and increased testing. One easy place to start is with free, universal pre-K education to close the gaps that open before children even enter the K-12 system.

Of course, we should also accept the reality that many young people will still not want to, or be able to, get a four-year degree. To help these individuals, we need to, instead, improve the pay of middle- and low-skill workers through apprenticeships, more technical education leading to certificates, and labor reforms that help worker bargaining.

Finally, without in any way making light of the economy’s current problems, or the very real struggles many families have faced in past decades, the left would be well advised to adopt a more positive attitude toward today’s postindustrial service economy. This economy is, by and large, the natural outcome of technological advances and ongoing productivity growth; it represents progress, not decline, for the country and its economic potential. When a society moves from rotary-dial phones to the amazing devices we all carry around now that put a world of information at our fingertips; when it advances from a society of high-school graduates and dropouts to one where most have attended college; when the proportion of consumer spending devoted to food and clothing drops from almost half to less than a fifth; when homes have doubled in size and foreign travel is commonplace, that’s progress.

Thus, the left should optimistically seek to further this economic transition and adapt it better to the needs of the country, not pine nostalgically for the decades immediately after World War II. American jobs will continue to evolve toward higher skills over time; most assuredly, they will not disappear. The left’s challenge is to make sure those jobs pay as well as possible and provide solid economic security for the workers the left aspires to represent. That, in and of itself, is hard enough without chasing the chimera of the manufacturing jobs of a different time.

Read more about economyEducationinnovationJobsManufacturing

Stephen Rose is a research professor at George Washington University and an affiliated scholar at the Urban Institute.

Also by this author

Is Growth Over?

Ruy Teixeira is a senior fellow at the Center for American Progress and author of The Optimistic Leftist: Why the 21st Century Will Be Better Than You Think.

Also by this author

Demography and Its Discontents

Click to

View Comments

blog comments powered by Disqus