Technological unemployment is unemployment primarily caused by technological change. Since the early 1800’s, the observation of economists has been that technology has had a positive influence on employment: as technological change increased productivity, prices for commodities fell, resulting in increased demand, thereby increasing demand for labor. Machines freed workers from simple manual work but created new better paying jobs requiring more specialized skills.
However, some technologists claim that modern capabilities of pattern recognition, machine learning, and global networking are steadily eliminating the skilled work of large swaths of the middle income workforce. The warning is that technology is no longer creating jobs at the rate that it is making others obsolete. The notion of technological unemployment leading to structural unemployment (and being macroeconomically injurious) is often dismissed as the ‘Luddite fallacy.’
Labor displacing technologies can generally be classified under the headings of mechanization, automation, and process improvement. The first two fundamentally involve transferring tasks from humans to machines. The third fundamentally involves the elimination of tasks altogether. The common theme of all three is that a task is removed from a workforce, decreasing employment. In practice, the categories often overlap; for example, a process improvement can include an automating or mechanizing achievement, and the line between mechanization and automation is subjective, as sometimes the former can involve sufficient control to be viewed as part of the latter.
In principle, technological unemployment may be distinguished from unemployment caused merely by the contraction phases of business cycles. In practice, such differentiation is difficult, owing to the multivariate nature of economics. Like unemployment in general, most technological unemployment is temporary, as unemployed workers eventually find new jobs. For several centuries, the main controversy about technological unemployment has been whether it can ever lead to structural unemployment.
Historical concerns about the effects of automation date back to the very beginning of the Industrial Revolution, when a social movement of English textile textile artisans in the early 19th century known as the Luddites protested against loss of jobs, higher food prices, and lowered pay and respect for skilled workers. The original Luddites were hosiery and lace workers in 1811. In the first year, after applications in Parliament for redress had been thwarted by owners, Luddites destroyed 300 stocking frames, a technology in existence since the 1500s, but one which was being used by owners with unskilled laborers for lower pay. They publicized their actions in circulars mysteriously signed, ‘King Ludd.’
Contrary to fears, technological advancement did not ruin Britain’s economy or systemically lower standards of living throughout the following decades of the 19th century. In fact, during the 19th and 20th centuries, the opposite happened, as technology helped Britain to become much less impoverished than before. It should be noted that the Luddites were not outright technophobes. The machines were used to suppress workers’ wages at a time when the Luddites, who were artisans and skilled workers, were fearful about famine and high and rising food prices. This made the machines the most accessible targets for the Luddites’ angry expressions of this fear.
As in the preceding century, the period of 1880 to 1940 saw no underlying automation-induced structural lack of new economic opportunities for skilled workers to go to, given enough searching, although the Great Depression caused a tremendous disruption to employment. The fundamental potential for full employment had not been lost, as would later be shown by the post–World War II economic expansion. In the 1930s, John Maynard Keynes incorrectly predicted that in a century there would be a 15-hour work week as the ‘economic problem’ would be replaced by the problem of leisure. The 1950s and 1960s were optimistic times in many respects, and during this era many optimists made forecasts similar to Keynes’. While economic growth increased per capita income in the 1960s, and the average number of hours per work week declined, they have remained relatively constant in the United States since then.
Despite the depopulation of manual labour and assembly line jobs after World War II via advancing mechanization and automation, the salvation for employment rates damaged in the industrial sector (secondary sector of the economy) came from the service sector (tertiary sector), which absorbed all of the workers that automation displaced elsewhere. For example, many manufacturing jobs left the United States during the 1990s but were offset by a one-time massive increase in IT jobs at the same time. And in some cases the freeing up of the labor force allowed more people to enter higher skilled managerial jobs and technically specialized jobs, which are typically higher paying.
Therefore, fears of unemployment due to automation were generally dismissed as just another instance of the Luddite premise, which had proven fallacious time and again over many decades. Given this empirical contradiction of the premise, people who nevertheless returned to it were usually viewed by the mainstream as cranks misled by quixotic leftist political bias. For example, works by scholars including science historian David F. Noble and economist Jeremy Rifkin were often respected but discounted. They were even sometimes mocked with the disparaging label ‘neo-Luddite.’
Another effect of automation is referred to as job polarization. Automation causes sharp losses of middle class jobs, forcing a polarization of wages and greater income inequality. The phenomenon of polarization due to automation is not confined to the US, also occurring in 15 of 16 European countries for which data is available. Although there is growth in the highest skilled jobs, ecomomist Alan Blinder at Princeton University notes that these are also more ‘offshorable’ than low-wage ones. Economist Gregory Clark warns that as the increase in machine capabilities continues, basic wages could easily become so low that families will be unable to support themselves without government assistance.
Rifkin’s ‘End of Work,’ published in 1995, predicted that automation-induced unemployment would begin to be widespread within the next decade. However, the book’s concept of how IT would evolve over the next decade was incomplete (the book mentioned the Internet once in passing and the World Wide Web not at all; its IT focus was mostly on robotics); and its timeline prediction turned out to be wrong. It also did not provide much detailed explanation of any solution to the problem. The book’s subtitle called the solution a ‘post-market economy,’ but its concluding chapters did not clearly lay out how such an economy could be engineered, leaving readers to conclude that a non-market solution involving a planned economy was implied between the lines.
James S. Albus, a US government engineer and a prolific pioneering inventor of intelligent systems, automation, and robotics, was concerned for many years about the potential social impact of advanced intelligent systems. Dr. Albus was optimistic about the wealth producing capabilities of intelligent machines, but concerned about the elimination of jobs and the downward pressures advanced automation placed on human wages and incomes.
In his 1976 book titled ‘Peoples’ Capitalism: The Economics of the Robot Revolution,’ and on his websites he lays out a plan to broaden capital ownership to the point where, in his view, every citizen becomes a capitalist with a substantial income from personal ownership of capital assets, leading, in his view, to achieving a future economic system where income from ownership of capital assets supplements, and eventually supplants, wages and salaries as the primary source of income for the average citizen. Albus’s vision concerns a world without poverty, war or pollution, a world of prosperity and opportunity.
Marshall Brain and Martin Ford are IT engineers who have worried that advancing IT will displace workers faster than current economic structures can absorb them into new kinds of jobs. Ford presents an argument for why the Luddite premise, although fallacious for two centuries, might become relevant again. Both advocate pursuing some permutation of basic income or guaranteed minimum income simply to keep the recirculation of value throughout the economy from stalling due to poor employment.
Although the earliest variations of these ideas involved government (for taxation and distribution), such ideas have been evolving to include market-based analogs, such as laws (analogous to existing minimum wage requirements) that require the private sector to employ humans, but that leave the job descriptions to private innovation. In these lines of thinking, it is recognized that automation can continue to yield ever higher per capita standard of living (in contrast to classical Luddism), but the basic income is merely a way to decouple consumer purchasing power and confidence from the traditional labor market, which might suffer market failure. In its place would grow a new labor market, which by its existence would restore fallaciousness to the Luddite premise (because automation would go back to not being a threat to job creation).