John Maynard Keynes once famously said that the difficulty lies not so much in developing new ideas but in escaping from old ones. And one of the oldest and most pervasive and pernicious economic ideas is that technology kills jobs.
Economist David Ricardo worried about this in the early 1800s, when he wrote, "I am convinced, that the substitution of machinery for human labour, is often very injurious to the interests of the class of labourers." A century later, U.S. Secretary of Labor James Davis, who served three presidents, asked, "Is the machine that turns out wealth ... giving us a permanent jobless class?" And now, as we emerge from the ashes of the Great Recession, there seems to be no escaping the technology-kills-jobs meme. It's everywhere: covers of magazines like The Atlantic and The Economist; a seemingly endless stream of books with titles like The Rise of the Robots and The Second Machine Age; and even in statements by leading economists who you would think might know better.
Paul Krugman, in a column titled "Sympathy for the Luddites," warns darkly that technology is displacing and devaluing even highly educated workers. In a blog post titled, "Rise of the Machines, Downfall of the Economy," financial pundit Nouriel Roubini writes that "a massive technological revolution will sharply reduce jobs over time." And even Larry Summers has stated that he no longer believes automation will always create new jobs: "This isn't some hypothetical future possibility.... This is something that's emerging before us right now."
Like the Luddites of early 19th century Britain, who smashed powered looms, today's Luddites argue an accelerated pace of technological change is leading to widespread automation, which in turn will lead to mass unemployment. Some of today's Luddites are unabashed in their desire to throw a monkey wrench in the machine. Brian Merchant asks in Motherboard, "Are you a fierce labor activist, willing to risk life and limb to destroy the mechanical implements of your impending poverty, to protest the loss of your livelihood? Then you might actually be a Luddite." But most, wanting to avoid being labeled a "Luddite," couch their warnings with the obligatory, "It's not that I am opposed to technology," but then go on to predict it will decimate work. Case in point: Erik Brynjolfsson and Andrew McAfee's The Second Machine Age opens with the requisite an "unprecedented boost to mental power [through digital technologies] should be a great boost to humanity," but then goes on to plan for "when the androids are coming" to take away most of our jobs and most of us will need to be on the state dole.
But whether it is hardcore or soft, this Neo-Luddism is dangerous. It leads companies to slow the pace of adopting new technologies to automate work as they fear political and consumer backlash. Even more importantly, it leads policymakers to throttle back on policies that foster and enable automation. The recent soft-Luddite "Open Letter on the Digital Economy," organized by Brynjolfsson and other tech luminaries, calls for more work to identify business models in which technology is a complement to--not a substitute for--labor. Hod Lipson, an engineering professor at Cornell University states, "The evidence is that technology is destroying jobs.... It is something we as technologists need to start thinking about." These kinds of calls to action have consequences. One example: the goal of the National Science Foundation's (NSF) National Robotics Initiative is to accelerate the development and use of robots in the United States that work beside or cooperatively with people. In other words NSF is favoring research into technology that will do relatively less to boost productivity than other technology. But without increased productivity, it's impossible to raise living standards, something that will become more challenging every year as more and more retiring baby boomers stop being producers while remaining consumers.
Luddites offer up a number of arguments to support their claims of employment doom and despair. All are wrong. The first is that the pace of technological change is speeding up so fast that it will inevitably induce jobless vertigo. Among those warning that we are accelerating out of control are Brynjolfsson and McAfee, who tout that "technical progress is improving exponentially," and Singularity University's Ray Kurzweil, who writes that "an analysis of the history of technology shows that technological change is exponential, contrary to the common-sense 'intuitive linear' view. So we won't experience 100 years of progress in the 21st century--it will be more like 20,000 years of progress."
This notion of techno-exponentialism is based on Moore's Law--the observation that semiconductor processing speeds will double every two years. But after 50 years of progress, Moore's Law is actually poised to slow down. Possibly as soon as 2020, the dominant silicon-based semiconductor architecture will likely hit physical limits (particularly pertaining to heat dissipation) that threaten to compromise the trend unless a major leap can be made to radically new chip architectures, something not likely to happen at a commercial scale anytime soon.
A second argument they make is that if technology is advancing exponentially, then it follows that the destruction of jobs also will be exponential. For proof, Neo-Luddites cite a study by researchers at Oxford University, who reviewed occupational categories and estimated which ones technology will replace. They claimed that 50 percent of jobs will be eliminated by technology in the next two decades.
But this pessimistic scenario (or optimistic, depending on your view) is highly unlikely, not only because the researchers assumed a much more rapid pace of innovation than is likely (see above), but also because they wrongly assume that many occupations, such as fashion models, manicurists, and aircraft mechanics, are in fact easy to replace with machines (hard to see how a robot will be walking down the runway wearing the latest Donna Karan dress). The Information Technology and Innovation Foundation conducted a similar analysis using more realistic assumptions and estimated that approximately 80 percent of jobs were either difficult to automate (such as custodial and sales jobs) or extremely difficult to replace (think of surgeons and top managers).
But even if the Oxford prediction is right, it would mean that productivity would grow at 3.75 percent per year, which would double output every two decades--only modestly faster than the average productivity growth that America has experienced over the last century.
How can this be? Are we actually destroying almost half the jobs in America every few decades? Well, yes and no. When businesses use machines to boost productivity, they are able to cut prices, which increases demand for their products, which in turn requires more workers. This pattern was at work throughout the 2000s, according to data from the U.S. Bureau of Labor Statistics (BLS). Industries where prices fell usually were the ones with productivity increases, while industries where prices rose were more likely to have seen productivity declines. Moreover, BLS found that in most industries these price declines led to relatively faster increases in demand for goods and services, and in some of these industries that increased demand actually led to more employment, even after accounting for increased productivity. For example, as the retail industry adopted automated ordering systems and other advanced IT systems in the last decade, productivity in "other general merchandise stores" (such as warehouse clubs) increased more than twice as fast as economy-wide productivity, and that enabled big price declines that spurred increased sales. The result was that the industry added jobs faster than almost any other sector. Around the world, this positive dynamic explains why a recent World Bank study of 26,000 firms showed that the majority of firms that increased productivity also created jobs, while firms that decreased productivity shrank their workforce.
But even if lower prices don't lead to enough increased demand for that particular good or service (for example, higher productivity in funeral homes won't increase the demand for funeral services), lower prices allow consumers to use those savings to buy other goods or services, leading to increased demand in those industries. The same is true with wages. The BLS found that on average industries that boosted productivity more than other industries increased wages faster. As these workers earn more money, they spend it, and that spending create jobs. When this process takes place throughout the economy, you get the cycle we have seen for centuries: increases in spending create jobs in most industries in a virtuous cycle.
This in fact gets to the nub of the argument and why Luddites are so wrong about this critical issue: they overlook the fact that when an organization uses technology to replace jobs, it does so to lower prices (or wages). These savings or added wages are not buried under a mattress. They are used to buy other things, in a wide array of industries. That is why higher productivity always has and always will lead to increases in demand, not decreases, and why it creates employment instead of undercutting it.
But Neo-Luddites argue this time is different because today's new industries aren't creating enough jobs. Tech is killing old jobs, they complain, but tech companies aren't creating enough new jobs to fill the void, giving examples like the fact that Instagram sold for $715 million dollars but only employed 13 workers. In Rise of the Robots: Technology and the Threat of a Jobless Future, Martin Ford predicts that new industries will "rarely, if ever, be highly labor-intensive." In a cover story on automation and jobs for The Atlantic, Derek Thompson agrees, writing "our newest industries ... just don't require many people." But these Neo-Luddites are taking the side of the government official in Asia who Milton Friedman took to task for having workers build a road with shovels instead of earth movers. When the official explained they were using shovels because they needed the jobs, Friedman quipped, "Why not use spoons?" Do we really want Google and Facebook workers to be using the equivalent of digital spoons? Can you imagine how much poorer we would all be if Google, Facebook, Twitter, and others required 10 times the number of workers to provide us with the same services. Not only would we be paying for services that are now free, but the millions of people now selling insurance policies, serving meals in restaurants, and building houses would instead be working at tech companies who would not be producing anything more than they currently do.
Finally, we are told a tale of woe in which widespread automation allegedly will make the masses miserable. Robots will be competing with workers and driving down wages while driving up profits for a few lucky capitalists. Again, Thompson writes that "technology could exert a slow but continued downward pressure on ... wages." Harvard economist Richard Freeman argues that far more people need to "own the robots," because he fears the world will evolve to a class divide where a few own robots and the rest are impoverished. But both notions are wrong. Higher productivity either means lower prices for consumers--which means higher wages in terms of purchasing power, or higher wages for the workers in the organization. And the notion that owners of "robots" will attain monopoly profits ignores the fact that competition is robust in most industries and that competition limits profits. All industries and firms will use "robots," not a select few.
To be sure, this doesn't mean that robots and automation won't lead some people to lose their jobs, although the lower prices will end up leading to other jobs being created. However, we can and should do a better job of helping workers who do lose their jobs to technology, especially as concern about technology-based job loss is so widespread. In 1962 as the United States was leading the charge for global free trade, Congress passed the Trade Adjustment Assistance Act (TAA), which provided help for workers hurt by trade, including reemployment services such as training and job-searching assistance. John Kennedy eloquently summed up the reason when he stated, "When considerations of national policy make it desirable to avoid higher tariffs, those injured by that competition should not be required to bear the full brunt of the impact. Rather, the burden of economic adjustment should be borne in part by the Federal Government." Fifty years later it's time to transform TAA into TTAA (Trade and Technology Adjustment Assistance) so that workers who are displaced by either trade or technological innovation receive meaningful help to transition to other jobs or careers.
John Stuart Mill once wrote that one person with a belief is equal to 99 who have only interests. Today we all have interests in accelerating technological innovation and automation, but a few people have the belief that we should forestall that progress because they hold to a discredited 19th century idea that tech kills jobs. It's time to replace that idea with this one: tech creates jobs, wealth, and opportunity.