The Technology Trap

by Carl Benedikt Frey

  Technology at Work

  Our path to riches is best understood in terms of the adoption of a steady flow of labor-saving technologies over the centuries. As the economist Paul Krugman once quipped, “depressions, runaway inflation, or civil war can make a country poor, but only productivity can make it rich.”31 Productivity growth happens when technology allows us to produce more with less. If the adoption of machines makes labor productivity grow by 2.5 percent per year, output per person will double every twenty-eight years. The notion that the product of an hour of work can double in just about half of a working lifetime is surely sufficient justification for the disruptive force of technology, which has shrunk that timescale visibly. But while productivity is a prerequisite for growing incomes for the commoner, it is not a guarantee of such growth. And, if machines replace workers in existing functions, some people may be left worse off as technology progresses. Despite this fact, textbook economics treats technological progress as a Pareto improvement: in other words, the assumption is that when machines take workers’ jobs, new and better-paying jobs become available for everyone at the same time. As evidenced by the historical record, such models are utterly irrelevant for understanding episodes when technological progress is labor replacing. These technologies have brought higher material standards but also worker dislocation.

  The extent to which labor-saving technologies will cause dislocation depends on whether they are enabling or replacing. Replacing technologies render jobs and skills redundant. Enabling technologies, in contrast, make people more productive in existing tasks or create entirely new jobs for them. Thus, the term “labor saving” has two closely associated but not identical meanings, and the difference between the two has important implications for labor.32 As the economist Harry Jerome noted in 1934, if the 1929 tonnage of iron and steel were produced with the technology available in 1890, a million and a quarter workers would have been needed instead of four hundred thousand. Does this mean that eight hundred thousand men had lost their jobs by 1929? Surely not. At the onset of the Great Depression, employment in steel had grown.33 Better technology reduced the number of workers required to produce a given amount of steel, but the steadily growing demand for steel meant that the number of jobs in the industry grew, too. Clearly, the nature of steel production changed as the industry mechanized, but there was probably little job displacement. Unlike replacing technologies, which take over the tasks previously done by labor, augmenting technologies increase the units of a worker’s output without any displacement occurring, unless demand for a given product or service becomes saturated.34 There are many examples of enabling technologies. Computer-aided design software has made architects, engineers, and other skilled professionals more productive by helping rather than replacing them. Statistical computer programs like Stata and Matlab have made statisticians and social scientists better analysts without reducing the demand for them. And office machines like the typewriter created clerical jobs that did not previously exist.

  To see how outcomes differ for labor when a technology is labor replacing, consider the arrival of the elevator. Without elevators, there would be no skyscrapers and no elevator operators. When the first elevators arrived, more elevators meant more jobs for people with a good sense of timing, capable of stopping the elevator when it was aligned with the floor. Things changed when a replacing technology emerged: the automatic elevator, which got rid of the human operator. All of the sudden, the job of elevator operator disappeared, even though we now use elevators more than ever. The demand for elevators has evidently not become saturated, just as we still demand many manufactured goods. But in a world where the jobs of machine operators have been taken over by robots, having more automobiles leave the factories does not inevitably mean more jobs for machine operators. Thus, it stands to reason that the effects of replacing technologies on jobs and wages will be very different from those of enabling technologies. Yet until recently, economists did not make such distinctions. Since the pioneering work of Jan Tinbergen—the first winner of the Nobel Prize in Economics—economists have tended to conceptualize technological progress in a purely augmenting way. According to the augmenting view of progress, new technologies will help some workers more than others but will never replace labor, meaning that workers cannot see their wages fall as technology progresses. This was a reasonable approximation of economic reality for much of the twentieth century. Indeed, most economic theory reflects the patterns of the particular times economists observe around them. The work of Tinbergen, which was published in 1974, before the age of computerization, was no exception. For much of the twentieth century, wages rose at all levels. What makes economic analysis hard is that there are few models that apply to every time and place.

  The fact that wages have been falling for large groups in the American labor market for more than three decades has prompted economists to think differently about technological change. Pathbreaking work by the economists Daron Acemoglu and Pascual Restrepo provides a helpful formal model for understanding periods of falling wages, as well as times when wages are growing for everyone, by conceptualizing technological progress as either enabling or labor replacing. This book looks at the historical record through the lens of their theoretical framework.35 The notion of machines being capable of taking over human work is important, because it means that technology can reduce wages and employment unless it is counterbalanced by other economic forces. Even though growing productivity still raises total income—offsetting the displacement effect in part, as more spending in the economy creates other jobs elsewhere—it does not fully counterbalance the negative effects of technological displacement. In Acemoglu and Restrepo’s framework, the creation of new tasks is essential to raise the demand for labor, workers’ wages, and the share of national income going to labor rather than owners of capital. How workers fare, in other words, in large part depends on the race between task replacement and new task creation, and how easily workers can transition into emerging jobs.

  Historically, as we shall see, the extent to which technology is labor replacing or enabling has varied greatly, leading to very different outcomes for average people. When new technologies replace workers in existing tasks, those workers’ skills become obsolete. Even when technologies are replacing for some but augmenting for others, workers might suffer hardships. In recent years, the creation of new jobs for robotics engineers has provided little relief to those who lost their jobs to industrial robots on the assembly lines. The arrival of the power loom, in similar fashion, replaced the jobs of hand-loom weavers, while creating new jobs for power-loom weavers. But while hand-loom weavers’ incomes diminished almost immediately, it took decades for the wages of power-loom weavers to rise, as they had to acquire new skills and a new labor market had to develop for those skills.36 Because replacing technological progress often comes with what Schumpeter called a “perennial gale of creative destruction,” there are always winners and losers.37 The overwhelming focus of popular commentary on unanswerable questions like whether there will be enough jobs in 2050 is unfortunate. In fact, it misses the point. Even if new jobs emerge as old ones are lost to automation, that might be little reassurance for the person who loses his or her job. Modernist writers didn’t fail to take note of the automation dilemma. In Ulysses, for example, James Joyce’s hero Leopold Bloom points out that “a pointsman’s back straightened itself upright suddenly against a tramway standard by Mr. Bloom’s window. Couldn’t they invent something automatic so that the wheel itself much handier? Well but that fellow would lose his job then? Well but then another fellow would get a job making the new invention?”38

  A new job was created for someone to make the new invention. But the someone was “another fellow”: making the invention required a different breed of worker. Both the Industrial Revolution and the computer revolution primarily created jobs for another fellow, whose skills could not have been more different from those of the displaced worker. The first episode of industrializat
ion is best described by the wit of the economic historian Gavin Wright, who reckoned that “in the limit we could devise an economy in which technology is designed by geniuses and operated by idiots.”39 Early factory machines, it is true, were simple enough to be operated by young boys. And as a result, middle-income artisan craftsmen were replaced by children working for a fraction of their wages in the factories. The difference this time around is obviously that children are no longer needed to operate the machines. Computer-controlled machines can run on their own. Yet computerization has also given rise to new tasks, requiring an entirely different set of skills like those of audiovisual specialists, software engineers, database administrators, and so on. Thus, we seem to have devised an economy designed by geniuses to be operated by other geniuses. Some jobs have become automated, but computers have also led to greater demand for workers with highly developed cognitive skills. Indeed, a common misconception is that automation is an extension of mechanization. Automation has replaced precisely the semiskilled machine-tending jobs that mechanization created, which once supported a large and stable middle class. Broadly speaking, those fortunate enough to have gone to college have thrived in the age of computers. But as middle-income jobs have dried up, many semiskilled workers have struggled to find decent job. During the Industrial Revolution as well as the more recent revolution in computing, middle-aged men in middle-income jobs were the victims of progress, because their skills were unsuitable for the new jobs that emerged.

  When technological change is labor replacing, how workers fare depends on their other job options. In Henrik Ibsen’s play The Pillars of Society, written in 1877, parallels are drawn between the economic consequences of the Industrial Revolution and those of Johannes Gutenberg’s printing press. One of the characters, Konsul Bernick, assumes that the fates of artisan craftsmen in the nineteenth century were similar to those of copyists when the printing press arrived, suggesting that “when printing was discovered, many copyists had to starve.” The shipyard foreman Aune bluntly replies, “Would you have admired the art so much, Consul, if you had been a copyist?”40 Though Ibsen’s question was meant as a rhetorical one, copyists rarely opposed printing technology. As we shall see in chapter 1, unlike weavers—who suffered hardships from the mechanization of industry—copyists and scribes were more likely to benefit from Gutenberg’s invention. Many of them did not make a living producing manuscripts. To them, the movable printing press didn’t mean any loss of income. And those who copied books for a living either specialized in shorter texts that were uneconomical to produce with printing technology or became binders and designers of books. Thus, while weavers and other craftsmen, who faced worsening job options, smashed textile machines all over Europe in the eighteenth and nineteenth centuries, copyists rarely resisted the printing press in the late 1400s. Of course, the art of printing was not adopted with the same enthusiasm everywhere. Fearing that a literate population would undermine his leadership, Sultan Bayezid II issued an edict banning printing in Arabic in the Ottoman Empire in 1485, with dismal long-lasting consequences for literacy and economic growth in the region.41 But in the light of the hostility to replacing technologies that was so widespread in Europe before the twentieth century, episodes of labor unrest accompanying the adoption of the printing press were few.

  The case of the printing press illustrates a broader point: when people have good alternative job options, they are less likely to rebel against machines. Job displacement is never painless, but if people have reason to believe that they will eventually come out ahead, they are more likely to accept the endless churn in the labor market. As we shall see, the explosive growth of middle-class jobs in the mass-production industries of the twentieth century was one key reason mechanization was allowed to progress uninterrupted: an abundance of manufacturing jobs was the best unemployment insurance people could get. In this period, a wave of enabling technologies and soaring productivity growth allowed working-class people to climb the economic ladder. Automobiles and electricity spawned new gigantic industries, and with more capital tied up in machinery, firms began to raise wages to keep workers from leaving for better jobs elsewhere. People at the top and the bottom of the income distribution saw their standard of living improve enormously, and, consequently, middle-class people accepted the reshufflings in the labor market with the expectation that they would benefit too.

  Another reason people may not oppose technologies that threaten their jobs is obviously that almost everyone will benefit in their capacity as consumers. Even those who worked on Ford’s and General Motors’s assembly lines have to some extent benefited from the cheapening of automobiles as robots have taken their jobs. Yet machines only cheapen goods and services after they have been introduced, so that if a technology is labor replacing, consumer benefits will arise only after displacement has already occurred. More important, the individual costs from displacement, in terms of distress and lost income, will be much greater than any consumer benefits unless those workers have decent outside job options. The cheapening of textiles, for example, did not provide sufficient relief to the Luddites, who rioted against the introduction of machinery despite the consumer benefits brought by mechanization. The point is surely not that replacing technologies will be bad for people over the long run. The very opposite is true. But that alone does not provide much relief for those who see their jobs disappear, unless they can expect to find new work of equal pay.

  Most economists will acknowledge that technological progress can cause some adjustment problems in the short run. What is rarely noted is that the short run can be a lifetime. And ultimately, the long run depends on policy choices made in the short run. The mere existence of better machines is not sufficient for long run growth. As Daron Acemoglu and the political scientist James Robinson point out in Why Nations Fail, economic and technological development will move forward only “if not blocked by the economic losers who anticipate that their economic privileges will be lost and by the political losers who fear that their political power will be eroded.”42 Workers alone might struggle to block new technologies effectively. But the ruling elites slowed labor-replacing progress for millennia.43 Political incumbents, for the most part, had little interest in the destabilizing process of creative destruction, as groups of economic losers could challenge the political status quo. As the eminent economic historian Joel Mokyr has argued in separate accounts:

  Any change in technology leads almost inevitably to an improvement in the welfare of some and a deterioration in that of others. To be sure, it is possible to think of changes in production technology that are Pareto superior, but in practice such occurrences are extremely rare. Unless all individuals accept the verdict of the market outcome, the decision whether to adopt an innovation is likely to be resisted by losers through non-market mechanism and political activism.44

  Britain’s edge during the Industrial Revolution did not lie in the absence of resistance against technological change, but in its government’s consistently and vigorously siding with the “party” for innovation.… Resistance to technological progress in France appears to have been more successful than in Britain, and perhaps this difference offers another explanation why Britain’s Industrial Revolution was first.45

  As I will argue in a similar vein, the early decision of British governments to consistently squash any resistance to mechanization helps explain why Britain was the first to industrialize. This decision, as we shall see, was much the result of a shift in political power. As the discovery of the New World gave rise to international trade and commerce, the power of landed wealth was challenged by a new class of “chimney aristocrats,” who stood to gain from mechanization.46 And more broadly, cascading competition among nation-states made it harder to align technological conservatism with the political status quo. The outside threat of political replacement became greater than the threat of rebelling workmen from below. Even when workers managed to solve the so-called collective action problem and take to the str
eets in protest, their case was hopeless. They did not stand a chance against the British army. Many Luddites ended up being imprisoned and then sent to Australia.

  The Reform Acts of 1832 and 1867 were surely important events, but they did not turn Britain into a liberal democracy. Property rights were regarded as most important, and civil rights and political rights were still lagging behind. Few people had access to education, and property ownership remained a requirement for voting—meaning that most ordinary people were politically disenfranchised. Had Britain been a liberal democracy, the case of the Luddites would surely have been much less hopeless. As Wassily Leontief, winner of the Nobel Prize in Economics, once joked, “If horses could have joined the Democratic party and voted, what happened on farms might have been different.”47 Horses might have used their political rights to bring the spread of the tractor to a halt. In similar fashion, if the Luddites had had their way, the Industrial Revolution would not have happened in Britain. Of course, there is no way of knowing exactly what would have happened; all we know is that many citizens tried to bring progress to halt by every means they had.