After two centuries of relentless automation, why are there more jobs than ever? Certainly, tens of millions of jobs have been lost. Whatever happened to the myriads of hostlers, blacksmiths, coopers, sucksmiths, millers, tallowmakers, wheelwrights, sicklemen, puddlers, telegraphers, stockingers, fellmongers, saddlers, ploughmen, knackers, bleacherers, weavers, thatchers, and scriveners? Most of these jobs have been either wiped out entirely or largely taken over by machines.
The advance of massively more productive machinery has clearly not led to mass unemployment. The number of people employed in advanced economies has never been higher. For example, since 1950 the number of Americans employed has nearly tripled, rising from about 58 million to nearly 149 million today. During that time the proportion of adults in the civilian workforce rose from 55 percent in 1950 to peak at 65 percent during the dot-com boom in 2000. The ratio has now dropped to 59 percent, but the lower rate is widely understood to reflect the fallout from the Great Recession, Baby Boomer cohort retirements, and younger individuals spending more time in school.
Given this history, why are so many professors, pundits, and politicians worried that this time it's different? That automation, chiefly ever more effective and productive information technologies, will soon produce vast unemployment?
In 2013, the Oxford University futurists Carl Benedikt Frey and Michael A. Osborne warned that 47 percent of all jobs in the U.S. are at risk of being automated over the next 20 years. Sounds frightening—until you consider what percentage of jobs has been automated away in the recent past. Jobs in manufacturing and agriculture, which accounted for 33 percent of American employment in 1950, are now down to 12 percent. The number of U.S. manufacturing jobs peaked at just below 20 million in 1979 and has fallen to under 12 million today. Many went offshore, but many more were automated away.
American coal production has doubled since 1950 while the number of coal miners fell from 483,000 then to 123,000 today. In 1950, about 16 percent of the U.S. labor force was employed in agriculture; that has dropped to below 2 percent today. As late as 1930, nearly 19 million horses and mules were used to plough fields, compared to fewer than 1 million tractors. In 1960, when the U.S. Census stopped collecting data on draft animals, the number had fallen to 3 million animals while the number of tractors had grown to 4.7 million. Meanwhile, farm productivity has tripled.
Despite all these jobs and more lost to automation, U.S. employment continued to steadily rise. Why? Because technological progress is a "great job-creating machine," argue Ian Stewart, Debapratim De, and Alex Cole, three economists at the business consultancy Deloitte. The trio argues that "the current discourse is biased towards the job-destroying effects of technological change due to the relative unpredictability of its creative aspects."
Analyzing technological and employment trends over the past 150 years in the United Kingdom, the three find that while machines have eliminated millions of jobs, they have also conjured into existence many more. Even better, living standards dramatically improved as the technological destruction of old jobs proceeded.
How? First, technology substitutes for labor, thus raising productivity and lowering prices. Since 1950, the percent of British incomes spent on food and clothing has fallen from 35 and 10 percent to 11 and 5 percent, respectively. In addition, the real price of automobiles has been halved. In 1948, a television in the U.S. would have cost the equivalent of $12,000 in today's money. Since then, the price of a TV has since fallen by 98 percent.
Second, the sectors that are the sources of innovation expand, boosting the demand for labor. The Bureau of Labor Statistics reports that the number of people working in computer systems design and related fields rose from 400,000 in 1990 to over 1.5 million in 2011. Similarly, the number of people employed in life sciences (biotechnology, pharmaceuticals) increased from 174,000 in 1990 to 1 million in 2012.
Third, technology improves outcomes in areas such as medicine, leading to increased demand for labor in those areas. Consider that the annual death rate for cardiovascular diseases in the United States has fallen from 805 per 100,000 in 1963 to 236 per 100,000 today. Five-year cancer survival rates have risen from 50 percent in 1970 to 70 percent today. Meanwhile, U.S. health-care employment rose from 2 percent of the workforce in 1950 to 9 percent today—that is, from 1.2 million to 13.4 million workers.
Fourth, technology lowers the cost of production and prices, enabling people to shift their spending to other goods and services, thus boosting demand for labor in those areas. For example, the demand for more personal services has greatly expanded. While the percent of their incomes Americans spent on food fell by nearly half since 1960, the percent of their food budgets spent on restaurants more than doubled from 20 percent to 43 percent. Consequently, the number of eating establishments since 1990 in the U.S. increased from 238,000 to nearly 1 million. Jobs in food service grew from 6.4 million to over 15 million now, nearly doubling as a proportion of the labor force. The number of people working as massage therapists has increased from 128,000 in 1996 to over 300,000, also nearly doubling as a fraction of the workforce. According to the Deloitte economists, since 1950 the percentage of the British workforce employed as barstaff has tripled and the percentage working as hairdressers has doubled.
As the MIT economist David Autor has shown, automation has taken over a lot of the routine physical and intellectual tasks that once were done by middle-income workers. This process has resulted in a more polarized economy, where highly skilled workers in such fields as infotech and biotech are richly rewarded while a greater proportion of the workforce toil at relatively lower-paying service jobs. Will this continue?
Autor doubts it, because he foresees a rising demand for services, involving non-routine tasks in which workers have a comparative advantage over machines—ones requiring interpersonal interaction, flexibility, adaptability, and problem-solving. Work, he argues, is evolving away from assembly-line rigidity and back toward a more pre-industrial paradigm populated by "new artisans." Perhaps more chefs will prepare fine meals in the homes of clients, dramatists devise elaborate virtual environments as entertainment, tailors create one-of-kind bespoke garments. Who the hell knows?
The Deloitte economists agree with Autor: "The work of the future is likely to be varied and have a bigger share of social interaction and empathy, thought, creativity and skill." They add, "The stock of work in the economy is not fixed; the last 200 years demonstrates that when a machine replaces a human, the result, paradoxically, is faster growth and, in time, rising employment." As Harold Bowen, chairman of the National Commission on Technology, Automation, and Economic Progress, observed way back in 1966, "The basic fact is that technology eliminates jobs, not work." That is as true today as it was then.