EconoMatters, Globalist Perspective

Robots Are Not the Problem

Less “rise of the machines” — and more “rise of the incomes.”

Industrial welding robot. (Credit: Phasmatisnox - Wikipedia)


  • Since the dawn of the industrial revolution 200 years ago, automation has been intensely debated around the world.
  • Ever since England’s Luddites smashed looms replacing weavers, worries about robots have been with us.
  • Productivity is seen as a problem because it helps fewer people do more work.
  • The reason why poor countries are poor is because productivity is so low.
  • An economy can only consume as much as it produces. Developing countries cannot get richer without more productivity.
  • All nations must encourage creative destruction and productivity or risk future economic stagnation.

Since the dawn of the industrial revolution 200 years ago, automation has been the subject of both exultation and trepidation around the world. Then, England’s Luddites smashed looms that were replacing jobs for weavers. Worries about robots replacing workers have persisted ever since.

Despite these worries, there has been a consensus, at least in the developed world amongst academics, policymakers, and the public that technological progress is a good thing: providing much of the impetus for economic growth.

Several years into a plodding recovery from the great recession, however, Luddite ideas have once again captured the attention of not only the popular media but also a number of academics.

Outlets from 60 Minutes to MIT’s Technology Review to the Associated Press have featured major stories on the perils of automation, and many intellectuals, including Andrew McAfee and Eric Brynjolfsson are blaming automation for labor market problems.

While these ideas have only recently regained traction in the U.S., Luddite logic has long been a fixture of economic policy in developing countries. Because of the large “oversupply” of labor in many of these nations, jobs are often treated as a scarce resource to be spread as widely—and as thinly—as possible.

Can’t afford productivity?

Productivity is therefore seen as a problem because it helps fewer people do more work. As one CEO of a large Indian manufacturing company once said to me, “we can’t afford productivity in India; we need the jobs.” Indeed, this is a common view among policy makers in developed nations, and so they intentionally avoid pro-productivity policies.

Of course, this is exactly the “Luddite Fallacy” that economists have long condemned. In fact, the very reason poor countries are poor is because productivity is so low.

So is productivity and automation the answer? Is it really that simple? What about all those other people that get replaced by automation?

As it turns out, it really is that simple—most of the time. The problem with the Luddite Fallacy is that it assumes that employers make a simple trade: higher productivity for fewer workers. But the truth is not nearly so simple.

Economic research has shown that when productivity increases, employers make all kinds of different decisions about employment. These decisions depend on the market for their products, their competitors, the nature of the labor market, and the type of innovation that has occurred.

More jobs from automation, not less

In fact, a number of microeconomic studies across developing and developed countries alike find that, after productivity increases, roughly half of employers decrease workers—but the other half add workers.

And finally, from the macro perspective, studies by the International Labor Organization (PDF) and World Bank (PDF) have shown that productivity is much more likely to accompany net employment gains in an economy than job losses. In other words, even if productivity gains do eliminate some jobs, they are creating other ones.

Think of it this way. If your company produces widgets and you suddenly find a way to produce twice as many widgets with the same amount of workers, how will you decide what to do?

Since your widgets only require half the labor costs they did before, you can produce them for less money. This means that you might be able to sell more of them—maybe twice as much, maybe three times. Then you might want to hire more workers.

These benefits from productivity are not only theoretical. They are not only seen in developed countries, but they have been observed in developed nations around the world, from Bangladesh to Costa Rica to Taiwan.

Cheaper goods, more purchasing power

But even if you decide to lay workers off, the productivity gains will still benefit the economy. This is because productivity enables cheaper widgets. As a result, widget purchasers now have more money to spend on other things, stimulating demand for other goods and services in the economy and creating jobs in those sectors.

This effect is more obvious from a macroeconomic perspective. Trade imbalances aside, any economy can only consume as much as it produces. This means that developing countries cannot get richer without increasing productivity.

Instead of shrinking from automation, therefore, all nations must encourage creative destruction and productivity – or risk future economic stagnation.

Automation is the path forward. To be sure, policymakers need to understand its benefits and drawbacks. We need policies that help maximize the former and minimize the latter. Such policies exist, but to bring them into existence we need to overcome the simplistic fears of robots stealing our jobs.

Tags: , , , ,

About Robert D. Atkinson

Robert Atkinson is president of the Information Technology and Innovation Foundation.

Responses to “Robots Are Not the Problem”

Archived Comments.

  1. On February 5, 2014 at 10:49 am PeterBurgess responded with... #

    I like this piece, but it really begs the question of why, given our amazing technology, we have a global society and economy that seems to be in the doldrums, and for most people in the middle class, seeming to offer a rather dismal future.

    I have some academic training in economics, but it is my background in engineering and corporate management accountancy that informs most of my analysis, and especially what I understand of complex systems.

    I do not get much excited by macroeconomic analysis. In the end this analysis shows big results, but as some sort of an average that really means not very much. Worse, making reliable predictions based on this sort of analysis is mathematically impossible given the complexity of the underlying social and economic system.

    On the other hand the micro analysis can be very powerful and very effective as a basis for socio-economic performance improvement … and in my view this is where there are huge opportunities for a game change. The corporate world uses cutting edge big data analytics to sell potential customers their products, but society as a whole does nothing like this to optimize the performance of society.

    Worse, the whole structure of business and economic metrics is designed on the assumption that more is always better. This may have been a good proxy for progress in the dim and distant past, but for the past 50 years or so, this has been a pretty silly assumption, especially in rich developed economies.

    I am excited and optimistic about the future, but only if there is substantial behavior change. I see radical reform of metrics as part of what needs to be done to get the behavior change we need.

    Peter Burgess … TrueValueMetrics
    Multi Dimension Impact Accounting

  2. On February 5, 2014 at 11:22 am Raymond Millsaps responded with... #

    lots to think about here. humans become redundant. the task will be for them to deal with meaningful leisure, cyborgs will be equally as adapt at high end professions like doctors, lawyers and how easy it is to replace the entire optical stream to get a pair of glasses. these machines will take care of us until consciousness is moved from the bioplatform to the digital platform. viva evolution…

  3. On February 10, 2014 at 10:05 pm Wayne Caswell responded with... #

    There’s a lot of truth here, under the right circumstances, but is that where we are now? Strong unions and collective bargaining once drove higher wages, better benefits, and profit-sharing incentive programs, and whenever productivity increased, so did incomes. But companies today use high unemployment and a glut of able but desperately unemployed workers to pay less, deny benefits, avoid collective bargaining, and eliminate profit-sharing. That way, all of the profits from outsourcing, robots, automation, productivity increases, and HR expense cuts flows up to the investors who pay just 17% or less on their capital gains. It’s a myopic short-term strategy that’s not sustainable in the long term.

    During the Industrial Revolution, disruptive change obsoleted jobs too, but people were able to retrain for the new Information Age. Now, however, disruptive change is happening at the exponential rate of Moore’s Law, and where companies once retrained their employees, giving them time to take classes, the workers are now left on their own, and if they don’t have the right skills, they’re laid off. Try retooling your skills while working full time and raising a family, or during the gaps between jobs, and while having to pay for it all yourself with yet another student loan.

  4. On February 11, 2014 at 5:41 pm Lewis J. Perelman responded with... #

    Atkinson’s continue repitition of this argument rests on the dubious assumption that the past is prologue. It denies the possibility of systemic phase change — which in fact has numerous historical precedents. Ultimately it expresses merely wishful thinking rather than any kind of proof, a variation on the Micawber principle: “Something will turn up.”

    The scenario portrayed by Kurt Vonnegut in his 1952 novel “Player Piano,” while admittedly ahead of its time, today seems far more credible.

  5. On February 11, 2014 at 8:07 pm robatkinsonitif responded with... #

    thanks for the comments. A few comments in return. First, we are a long, long way from robots being smart enough to do most jobs. Many many decades at minimum. But even if they could do mos jobs they can’t do all, and because human needs are largely unlimited that will get translated into more demand. I would venture to say that if somehow the average American made 10 times more money than they make today (which at an unheard of 10% productivity growth a year would take 50 years at least to acheive) that they would easily find things to spend it on: first class airline seats, massages, more vacations, really nice cars, more restaurant meals, more live concerts and shows, better urban design, eliminating pollution, etc. etc. And all of these things employ people. Not everything can be automated. So Lewis, I dont need proof, only logic: productivity gains always lead to 2nd order economic benefits (e.g. lower prices) and those benefits always get translated into additional economic demand. The world has never experienced faster than 4 percent productivity growth per year. To somehow assume that growth will be an order of magnitude higher is simply beyond reality.