Measuring Technological Impact On Employment

Introduction

This is the second installment of our ongoing series about the rise of automation technology and its impact on employment. In part one, we looked at previous technological revolutions to glean insights on whether this one will be different. The answer was ‘almost certainly,’ pushing us to examine how big the impact could be, who will bear the economic brunt of this revolution, and how we’ll cope with any major changes.

How many jobs are vulnerable?

Economists disagree on this (as in most things, that’s true). In this case of technology-driven displacement, we found that occupation-based assessments oriented to industries generally tend to produce fairly pessimistic results. On the other hand, task-based assessments that measure the likelihood an occupational activity will be automated and what portion of the activity can be automated generate more moderate conclusions.

Let’s look at a couple examples of these conflicting approaches and conclusions.

A widely cited occupation-based study by Frey and Osborne 1 suggests that 47 percent of US workers hold jobs that could be automated in the next 10 to 20 years. Atkinson and Wu 2 criticized that sweeping methodology and looked more specifically at the tasks that make up a job. The result? Atkinson and Wu concluded that automation puts only 10 percent of jobs at risk.

Arntz et al 3 also chose a task-based assessment over an “inevitably over-simplified” occupation-based one and met with similar results to Atkinson and Wu. Researchers in this study found that automation only poses a serious risk to nine percent of US jobs.

The most comprehensive global estimate available comes from the McKinsey Global Institute.4 McKinsey estimated that by 2030 between 15 percent (400 million) and 30 percent (800 million) of the global labor force will be displaced because of automation, and between 3 percent (75 million) and 14 percent (375 million) of workers might need to change their line of work. McKinsey also concluded that about half of today’s occupational activities could be automated, but less than five percent of jobs consist entirely of those activities.

Knoema summarized some of these current estimates in the following table and map.

“Displaced workers in agriculture and industry will likely begin shifting to the services sector, placing downward pressure on services wages.”

Which jobs are vulnerable?

Obviously, some occupations have a higher risk of displacement than others. Computers and factory-floor robots have (and will continue) to complement the activities of highly-skilled workers and displace lower-skilled workers, such those on assembly lines or operating switchboards.4

Another wrinkle: the risk of displacement is unequally distributed across industries and countries. The most common jobs in developing countries—such as agricultural work—are more vulnerable than the types of service jobs common in high-income countries, which require creative work or face-to-face interaction.5 As if that’s not enough, displaced workers in agriculture and industry will likely begin shifting to the services sector, placing downward pressure on services wages.

But, also remember that these forecasts address the possible fates of existing jobs, and new technologies will inevitably create new jobs. The question really becomes one of how many new jobs. Moreover, if new technologies increase labor productivity, per capita income and finally aggregate consumption, it might in turn increase labor demand.3 To that end, studies from the McKinsey Global Institute,4 World Economic Forum,6 and Oxford Economics in cooperation with CISCO7 agree that growth in labor demand on net will outstrip jobs displaced by automation but without certainty as to by how much.

What might limit labor automation?

It’s important to remember there’s a difference between a job being susceptible to automation and that job actually disappearing, which can contribute to overestimation of the consequences of this revolution on the labor force. Forecasts also tend to reflect technological capabilities rather than the likelihood those capabilities will be adopted, further inflating estimates.

Restraints also exist beyond any specific technology that will influence the adoption equation. A technology might work perfectly but could be subjected to economic, legal, political, and/or social limitations. (Reluctance to widely adopt nuclear energy in the United States fits this case.) Here’s a rundown of external limitations on technology that could mitigate the reverberation through labor markets.

Economics:

You need to be able to make a business case. Some companies might conclude the upfront cost of automation poses more risk than simply hiring more people. This is particularly true if labor is relatively flexible and affordable. Of course, the cost of artificial intelligence and robotics will eventually drop—perhaps quickly—but we can’t project an exact timetable.8

Legal and regulatory:

Artificial intelligence relies on data, which means businesses will wrestle with a range of legal issues—e.g., individual data and privacy rights, the consequences of incomplete and unreliable data collection, and the misuse of data-sharing platforms. This applies to the companies that develop the technology (how they optimize their machines, for example, or how they collect and use data) as well as to the industries that adopt the technology. Take the case of driverless cars, where insurance liability regulations will need to be redesigned to account for who bears or shares the responsibility for an accident: the human owner/driver, the manufacturer, the software provider, or some other player in the supply chain.8

Social constraints:

People might feel too uncomfortable to turn to robots or other smart machines for certain tasks, especially in high-stakes areas such as healthcare, aerospace, or driving. There’s also the social cost of dramatic increases in inequality should a relatively small group of tech companies and highly-educated workers gain at the expense of a large working class. New technology spreads when we understand and accept its advantages over the status quo, and history suggests this can be a long process.8

“The robots are coming, sure, but corporations, governments, and individual laborers will play major roles in determining what the robots do when they get here.”

Coping strategies

How will policymakers, investors, and individuals navigate the risks of social upheaval? Schlogl and Sumner5 identified two directions: Policies meant to slow or reverse automation trends and policies designed to help people cope with the consequences.

The first group of policies deters industries from automating and would include steps such as tax hikes and regulations. (For example, Bill Gates’ suggestion of a special tax on robots that displace human labor).8 Another way to curb technology adoption would be to incentivize businesses to use human workers, perhaps through income tax and minimum wage cuts.

The second policy route—coping policies—might include support for people in economic transition as well as the strengthening of safety nets, including unemployment insurance and wage subsidies. We might also see preparation-oriented policies, such as investment in today’s labor-intensive sectors (i.e., infrastructure and construction), which could create jobs ahead of the automation wave.

 

Conclusions

It’s clear that these new technologies have the power to reshape the global workforce, and to a pretty significant degree. And, unlike previous ones, this technological revolution will reach beyond manual labor into business, services, and even creative sectors. The impact will be unevenly distributed across countries and industries, according to wealth and the skill and education levels of the labor force, among other factors, forcing a global discussion on and response to managing the inequality gaps that threaten to follow.

What we don’t yet know is how dramatic this will all be. Though at first glance the future might seem inevitably bleak, it’s more nuanced than the chicken-little version that occupation-based analyses suggest. We’ll see some job creation, and constraints other than technology will limit how far and fast the wave travels. Above all, though, we need to remember that humans don’t just watch the world change helplessly. Societies will adapt—though probably with unevenly distributed results. The robots are coming, sure, but corporations, governments, and individual laborers will play major roles in determining what the robots do when they get here.

 

 

Looking for more data? Explore our curated dashboards on the topic in Knoema:

References

  1.     Carl Benedikt Frey, Michael A. Osborne, September 2013. “The Future of Employment”. Published by the Oxford Martin Programme on Technology and Employment. Working Paper. Retrieved from Link
  1.     Robert D. Atkinson and John Wu (May, 2017). “False Alarmism: Technological Disruption and the U.S. Labor Market, 1850–2015” Information Technology and Innovation Foundation. Working Series. Link
  1.     Arntz, M., Gregory, T., & Zierahn, U. (2016). The risk of automation for jobs in OECD countries: A comparative analysis. OECD Social, Employment and Migration Working Papers, 2(189), 47–54.
  1.     MGI, 2017 Jobs lost, jobs gained: Workforce transitions in a time of automation. Retrieved from Link
  1.     Lukas Schlogl and Andy Sumner, 2018 “The Rise of the Robot Reserve Army: Automation and the Future of Economic Development, Work, and Wages in Developing Countries.” CGD Working Paper 487. Washington, DC: Center for Global Development. Retrieved from Link
  1.     World Economic Forum, 2018. The Future of Jobs Report. Geneva: WEF. Retrieved from Link
  1.     Oxford Economics and CISCO, December 2017. “The A.I. Paradox. How Robots Will Make Work More Human”. Retrieved from Link
  1.     John Hawksworth, Richard Berriman and Saloni Goel, 2018. “Will Robots Really Steal Our Jobs”. PwC. Retrieved from Link

How Technology Affects Employment

Introduction This is the second installment of our ongoing series about the rise of automation technology and its impact on employment. In part one, we looked at previous technological revolutions to glean insights on whether this one will be different. The answer was ‘almost certainly,’ pushing us to examine how big the impact could be, who …

Evaluating U.S. Unemployment Data

Introduction This is the second installment of our ongoing series about the rise of automation technology and its impact on employment. In part one, we looked at previous technological revolutions to glean insights on whether this one will be different. The answer was ‘almost certainly,’ pushing us to examine how big the impact could be, who …