Is fear of automation merely neo-Luddite sentiment or legitimate concern?

By Stefan M. Kløvning

Insight – Pew Research published new findings on Thursday about public expectations for automation by computers and robots on the future job market in ten countries across the world. The countries investigated were Greece, Italy, Poland and Hungary in Europe; Canada and the U.S. in North America; Brazil and Argentina in South America; and South Africa and Japan in Africa and Asia respectively.

On the question of whether they expected robots and computers to take over many jobs during the next fifty years, US citizens were the most optimistic, with only 65% saying it would “probably” or “definitively” happen. On the top, we find Greece, Japan and Canada with 91%, 89% and 84% respectively. Greeks and South Africans were the most certain, at 52% and 45% of their populations respectively who claimed that such a transformation would definitely happen.

For most asked during the polling in almost all countries, the downsides significantly outweighed the upsides. They were presented with two potential downsides and two upsides: (1) people would have a harder time finding jobs, (2) the [wealth] inequality between rich and poor would be much worse than it is today, (3) the economy would be more efficient and (4) there would be new, better-paying jobs.

An average of 80% thought the first scenario was likely, with only 17 percentage points between the highest (Greece – 91%) and the lowest (Japan and Hungary – 74%). The second had a bit more differentiation, with only 63% of Italians and up to 87% of Greeks believing wealth inequality would worsen with automation. Still, almost 77% of citizens across the ten countries thought on average that this would be an important factor in the trend.

Far less had confidence in the propositions in (3) and (4). 74% and 61% of Japanese people and Poles, respectively, thought it was likely that the economy would become more efficient with automation, but this confidence wasn’t commonplace in other countries. 6 countries had less than 50% of its citizens thinking this was likely to happen, whereof 3 had less than 40%. Least confident were Italians, with only 33% thinking it was probable. People were even more pessimistic about the last scenario, with no country having more than 47% believing it was likely that automation would bring new and better-paying jobs to the market. Here, again, Italians are the least confident, with only 24%, but the Americans aren’t much different, with 25%. The only country with more confidence than Brazil and Poland, both with 37%, is Canada, at 47%.

Richard Wike and Bruce Stokes, the authors of the report, also showed that whether the person asked believes the current economic situation is good or bad is a statistically significant factor in the results.

The findings, overall, however, appears to be that people generally believe the downsides outweigh the upsides with automation, which begs the question: Will it?

To answer this key question, we must first look at whether this threat has at any time appeared before. Indeed, it did quite much so during the peak of the industrial revolution in England.

The latter half of the 18th century saw great technological innovation: the water wheel, the spinning jenny, the steam engine, etc. This was the beginning of what would eventually become part of the cause of increased economic prosperity and life expectancy rates. But this did not occur instantly. The English working class suffered under the transition, though, according to some historians, it was largely due to the Napoleonic wars, with food getting scarcer and more costly (Sidenote: just the conditions needed for Marx to get people to buy into his doctrine). Even under these poor conditions, some people still worried that machines could make it even worse as they would put them out of a job. The group known to be most serious about this threat were the Luddites. The Luddites was a radical group of textile workers and weavers, most active between 1811 and 1817, who destroyed weaving machinery as a form of protest. They had to fight against the British Army several times, but one time it got so far that there were more soldiers fighting the Luddites and other domestic protesters than Napoleon at the Iberian peninsula. The time they used to learn the skills necessary for their jobs, they thought, would go to waste as machinery would replace their role.

They were right, of course, but I don’t think many today would with hindsight say the downsides of the industrial revolution exceeded the upsides. After all, the life expectancy rate in the United Kingdom has doubled from 40 in 1800 to 81.2 today, a trend with most other countries following suit.

The key to understanding this phenomenon is what is often termed as “creative destruction.” Creative destruction, first described by Joseph Schumpeter in 1942, is the “process of industrial mutation that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one.” This happens when an innovation changes the traditional arrangements and frees resources to be used elsewhere. The labour and resources would no longer be useful in one arena and thus transferred to another where they are more needed. However, this redeployment will not happen immediately after they’re put out of work, which makes rapid automation a legitimate concern.

David Rotman delineates the parallel for Technology Review:

At least since the Industrial Revolution began in the 1700s, improvements in technology have changed the nature of work and destroyed some types of jobs in the process. In 1900, 41 percent of Americans worked in agriculture; by 2000, it was only 2 percent. Likewise, the proportion of Americans employed in manufacturing has dropped from 30 percent in the post–World War II years to around 10 percent today—partly because of increasing automation, especially during the 1980s.

However, he quotes Lawrence Katz, a Harvard economist who has done extensive research into how technological advances have affected jobs over the last decades, claims that there is no historical pattern which suggests that it leads to a net decrease in jobs over an extended period of time. “While it can take decades for workers to acquire the expertise needed for new types of employment,” he says, “we never have run out of jobs. There is no long-term trend of eliminating work for people. Over the long term, employment rates are fairly stable. People have always been able to create new jobs. People come up with new things to do.” Katz expects this trend to follow in the future but concedes that there is something different about today’s digital technologies that could potentially affect a broader range of work. “If technology disrupts enough, who knows what will happen?” Rotman rhetorically asks in his reportage, “Will the job disruptions caused by technology be temporary as the workforce adapts, or will we see a science-fiction scenario in which automated processes and robots with superhuman skills take over a broad swath of human tasks?” Thus we’re back to where we started. Will it?

Erik Brynjolfsson and Andrew McAfee claim to have proof that the current technological advance does, in fact, destroy jobs faster than it is creating them. The two infographics pages to the right illustrate the problem. The first graph shows that U.S. productivity and employment have been more or less congruent during the 20th century, but diverged in the 2000s, what they call “the great decoupling.” Brynjolfsson is certain that this is because of technological change, but there’s no consensus among economists as to the cause of this great decoupling.

We can see an even greater divergence between U.S. GDP per capita and household income. For those who don’t know, GDP per capita means total output (the value of all finished goods and services) divided by the number of citizens in the country. Its divergence from average household income can easily be explained by technological advance making the economy more efficient but doesn’t say as much as the former graph about the degree to which jobs are being replaced. This should, however, disprove the claim that automation doesn’t lead to the economy being more efficient, point (3) in the Pew poll.

The second page tells us a bit more. Workers in the first and last percentile of skill have increased most in the change in share of employment, with .2% and .3%, respectively. The middle, however, doesn’t appear to have changed much. Those from 50 to 80 in the skill percentile have barely increased their shares at all, and those from 10 to 50 have seen their share decreased.

On the right on the page, we also see an overview of the vulnerable and fastest growing jobs. 8/10 of the fastest-growing jobs have to do with computer systems, networks, and software, showing the bright side of creative destruction, but again, we have to recall their argument that jobs are being depleted faster than they are created.

We’ll conclude by looking over the legitimacy of the four propositions laid out in the Pew poll but in opposite order.

1. There would be new, better-paying jobs.
   • On the table with the fastest-growing jobs, we see that most of them are caused by technological advance, working the wonders of creative destruction. To figure out whether these are better-paying jobs than the ones who are being replaced would require its own thorough analysis, which is beyond the scope of this study. In other words, there will be new jobs, but whether these are better-paying is uncertain.
2. The economy would be more efficient.
   • Brynjolffson and McAfee’s statistics indicate that technological advances increase productivity, especially in the 2000s, though the latest decoupling has split the line between that and employment. An increase in productivity is a strong indicator of the economy as a whole getting more efficient.
3. The [wealth] inequality between rich and poor would be much worse than it is today.
   • There’s been a great amount of talk about inequality lately, especially since Piketty’s publication of Capital in the 21st Century in 2011. His research had great influence on the attention of the entire economist community on the trend, but he also received a lot of critiques (see: Anti-Piketty). If we accept his proposition of growing inequality, however, how does technology work as a factor? Rotman writes in another Technology Review article connecting the lines between Piketty’s and Brynjolfsson’s work:
     

     Brynjolfsson lists several ways that technological changes can contribute to inequality: robots and automation, for example, are eliminating some routine jobs while requiring new skills in others (see “How Technology is Destroying Jobs”). But the biggest factor, he says, is that the technology-driven economy greatly favors a small group of successful individuals by amplifying their talent and luck, and dramatically increasing their rewards.

     Brynjolfsson argues that these people are benefiting from a winner-take-all effect originally described by Sherwin Rosen in a 1981 paper called “The Economics of Superstars.” Rosen said that such breakthroughs as motion pictures, radio, and TV had greatly broadened the audiences—and hence the rewards—for those in show business and sports. Thirty years later, Brynjolfsson sees a similar effect for high-tech entrepreneurs, whose ideas and products can be widely distributed and produced thanks to software and other digital technologies. Why hire a local tax consultant when you can use a cheap, state-of-the-art program that is constantly being updated and refined? Likewise, why buy a second-best program or app? The ability to copy software and distribute digital products anywhere means customers will buy the top one. Why use a search engine that is almost as good as Google? Such economic logic now rules a growing share of the marketplace; it is, according to Brynjolfsson, an increasingly important reason why a few entrepreneurs, including the founders of such startups as Instagram, are growing rich at a staggering rate.

     The distinction between Piketty’s supermanagers and Brynjolfsson’s superstars is critical: the latter derive their high incomes directly from the effects of technology. As machines increasingly substitute for labor and building a business becomes less capital-intensive—you don’t need a printing plant to produce an online news site, or large investments to create an app—the biggest economic winners will not be those owning conventional capital but, instead, those with the ideas behind innovative new products and successful business models.

     This appears to indicate an increase in inequality, but the exact impact technology has on it calls for its own study, and is not certain based on the citations from Brynjolffson. Daron Acemoglu wrote a paper for the NBER in 2003, saying “This consensus is built on the notion of technology-skill complementarity: technical change favors more skilled (educated) workers, replaces tasks previously performed by the unskilled, and increases the demand for skills. Consequently, many commentators see a direct causal relationship between technological changes and these radical shifts in the distribution of wages taking place in the U.S. economy. … These considerations imply that technical change that increases the demand for skills can have much amplified effects on inequality, because it also will change labor market institutions and preferences towards redistribution.” As Acemoglu argues against other factors, such as globalization, I’ll judge this fear legitimate, though I think fiscal policy will have an influence on its significance (P.S. The degree to which inequality as such is bad de facto is also up for debate).

4.  People would have a harder time finding jobs:
   • This is the million dollar question. In making such a forecast we can not say anything for certain, we can only think in probabilities. According to Brynjolfsson, the jobs are running out faster than they are creating because of the technological impact delineated in his and Acemoglu’s work. On the other hand, Katz expects things to go alright as it has before. Brynjolfsson and Acemoglu make a strong case for why creative destruction, as a result of technological advance, might, in this case, be leading to more jobs being replaced than created. High-skill work and creativity is in high demand and appears to only get more so as manufacturing jobs get automized. A net decrease in work will make it more difficult to find a job, but the different kinds of work demanded also calls for a reform in what is taught and how in the education system, though that’s a topic of study on its own. Anyway, as the job market changes, the education system must adapt accordingly.

We’ve here seen that much of the concern about the downsides of automation have proven more or less legitimate, that the current situation is significantly different from that during the industrial revolution, and that the economic upsides have been underestimated. Creative destruction has produced great benefits in the aftermath of the first industrial revolution, but today jobs seem to disappear faster than they are created. The technological change has increased economic output but at expense of an increase in wealth inequality by especially targetting low-skilled jobs, and increases the demand for high-skill labour. Between 1980 and 2005, the share of employment by the most low-skilled employees still increased, indicating a feud between theory and data, but the statistics still show that much of low-skilled labour is on the decline. If there were only the market involved in the process, it would be simpler to make a forecast on what would happen in the future, but as the government intervenes, for the better or worse, the trend could go a lot of different ways depending on what incentives are put in place and which legislation is passed to address the issue. Should we trust Congress to act fairly and effectively? Despite being constantly pushed by lobbyists for this and that, we should hope so, and have a debate about what suggestions could work and should be set in place, and advocate them to be instated.