Do Robots Kill jobs? In France, the Answer is, 'Non!' – and 'Oui!'
In their new paper, Competing with Robots, Daron Acemoglu (MIT), Pascual Restrepo (Boston University), and Claire LeLarge (University of Paris) survey the impact of robotics on levels of manufacturing employment in France. Examining over 55,000 firms, the researchers found that between 2010 and 2015 just 598 firms had adopted robotic technology. While these companies represent just 1 percent of manufacturing firms they account for 20 percent of French manufacturing employment. So what were the effects?
Economic theory and history tells us that while the efficiencies gained through automation cost some jobs they create even more over time by improving productivity and raising aggregate economic demand. The findings in the French economy are curious in this respect. The paper finds that among the large French manufacturing firms that adopted robots, employment actually went up, not down. Further, this increased big-firm employment tended to shift jobs away from production positions. The study does not say what kinds of jobs were created but they presumably skewed up the skills continuum toward research, development, marketing, sales, and finance.
Where negative employment impacts showed up was in “spillover” effects that shrank employment among smaller competitors who did not deploy robots. While robotics-heavy firms grew their workforces, overall manufacturing employment contracted. The study estimates that a 20 percent increase in robots creates a 3.2 percent decline in overall manufacturing employment. Moreover, the more exposed a firm is to international competitors using robots, the more likely they were to adopt robots, demonstrating a similar dynamic of robot-driven competition across national borders.
This insight resonates strongly with comments made by Eric Schmidt, former CEO of Google and Alphabet, at an October 2019 meeting of the Stanford Human-Centered Artificial Intelligence. Schmidt cautioned that the economic benefits of AI (the data side of robotics) would likely come less quickly than anticipated. To deliver their efficiencies, he said, these technologies are not simply layered on top of existing business processes but require full-scale reengineering of entire companies, a time- and resource-intensive undertaking. He suggested that while Fortune 50 firms (e.g, Walmart, Apple, Amazon, United Health, CVSHealth) probably have the capital required for comprehensive reworking of their operations — as well as the constant upgrades required to remain current— many of the Fortune 500 probably do not.
For the Fortune 50, it is easy to see how the full adoption of AI would tend to improve business performance vis-à-vis competitors allowing these firms to grow even larger at the expense of less efficient firms. The big companies would get richer, head counts among those engaged in manually-focused tasks would be reduced, and more highly-skilled jobs added to drive further expansion of market-share. External competitive pressures would also tend to accelerate the adoption of AI among the largest companies battling for an edge with competitors in China and elsewhere. All of this points toward increased pressure on less efficient firms and accelerated automation of business processes, a cycle that is virtuous for productivity, profit, and living standards and vicious for employment of less skilled workers.
If the dynamic identified by this study is correct, what ought to be our response? Attempting to limit the introduction of efficiency-increasing technology seems like a certain loser in terms of domestic and international competition as well as living standards in the long-term. As long as technology-driven efficiency is profitable, enterprising human beings will find a way to make money off it. If U.S. firms don’t do it, they will hand market share to foreign competitors able to produce higher quality and lower priced goods and services. The good news is that this technological shift has a number of bottlenecks and is likely to be a slower and more gradual than generally assumed, buying time for society, government, incumbent workers, and students to adapt and plan.
The best response would seem to be to focus on programs that help workers prepare for an uncertain and largely unknowable future. It’s worth considering strategies modeled on other highly competitive economies, like those in Scandinavia, that prioritize both economic efficiency and social solidarity, allowing uncompetitive firms to fail while relying on largely privatized retraining systems to help workers adapt to changing skill demands. By keeping our focus on workers we can avoid the competitive erosion caused by protectionism and industrial policy while incentivizing the long-term key to individual and national economic success: strong and constantly improving skills among workers.
Brent Orrell is a resident fellow at the American Enterprise Institute where he conducts research on workforce development, criminal justice reform, and social theory.