Editors Note: This Forum is available as our spring 2017 print issue. We are pleased to make it freely available online thanks to the generous support of the Cameron Schrier Foundation, the William and Flora Hewlett Foundation, the McCoy Family Center for Ethics in Society at Stanford University, and the National Endowment for the Arts. Please help us continue to make content like this open access by becoming a member!


About every quarter century, as Brishen Rogers points out, accelerating automation seems to bring a wave of anxiety about the disappearance of jobs and a corresponding wave of enthusiasm for basic income. It happened in the 1960s and the early 1990s. It is happening again now. The machines will take all the jobs, the argument goes, and something must be done about it.

But Rogers is correct to emphasize that fears about the end of work have proven to be overstated. In both of those recent episodes, machines did kill some jobs, but created more as the economy grew and new occupations emerged. The arrival of automatic teller machines, James Bessen points out in his book Learning by Doing (2015), led to an increase, rather than a decrease, in the number of jobs for humans. Demand for banking services increased, and people carried out tasks other than mechanically handing cash to customers. The impact of automation for some individuals—those unable to retrain or find as work that paid as well—was terrible. But in the aggregate, the number of jobs increased. Since the anticipated mass unemployment simply did not occur, the introduction of a basic income proved unnecessary.

Basic income addresses a hypothetical future problem of fewer jobs—but market economies can adjust to create new kinds of work.

It is of course possible that this time is different, that the cleverness of artificial intelligence (AI) will really destroy jobs and that the economy will not display its normal adaptability. A much-cited study by Oxford University academics, for instance, claims that about half of all jobs in the United States and UK are at risk from AI during the next twenty years. But economists at the Organisation for Economic Co-operation and Development did a careful assessment of that research and argue that the proportion of jobs affected will be fewer than one in ten—bad enough but not apocalyptic.

This is not a call for complacency, however. Like Rogers, I see the changing economy as an enormous policy challenge. Past episodes indicate that governments have not yet figured out how to minimize the disruption costs when there is structural change in the economy from automation. I agree with Rogers that helping individuals make the transition in their own lives—through public sector investment in education and stronger worker protections—is a vital part of the puzzle. But I also contend that we should embrace new technologies to ensure the economy as a whole is growing and therefore creating demand for new activities.

The latter part actually requires more robots, installed faster. We need businesses to improve their productivity and sell innovative services and products attractive to consumers. The low productivity growth of the past ten years suggests there has been too little automation, not too much. Productivity is an abstract economists’ term, but it is the driver of rising living standards over the long run; it means getting more services and products that people value out of the available resources. Machines make it possible by taking over routine tasks and saving time.

As Rogers illustrates, however, new technologies have changed the realities of business, and there are specific concerns about the behavior of the companies doing the automation. The big multinationals setting the pace on automation have gained a good deal of market power in several sectors of the economy. They use their economic heft to political ends, to ensure the regulatory framework and weak antitrust policy continue to support their position in the market and the economic rents they gain from that. They are adept at arbitraging different tax and regulatory systems to minimize the amount of tax they pay.

The existing framework of employment regulation and public policy is not structured to keep up with these new demands. Most importantly, as Rogers emphasizes, it is not designed to support individuals in a labor market consisting increasingly of big global corporations and contingent local work. For example, a minimum wage assumes the policy can be delivered through stable employers rooted in their community. But the old nation- and community-based social contract between businesses and citizens has crumbled.

We should embrace new technologies to ensure the economy is growing and creating demand for new activities.

We lack even the minimal data to understand how many people are working in this gradually emerging robot economy. For all the talk of the “gig” economy, there are no statistics on how many people rely on gig work, what they earn, and what they would like to be doing. Recent work by University of Maryland and Census Bureau economists points out that two sets of official U.S. figures do not even agree on the trend in self-employment. Data based on tax returns suggest it is rising, but data based on household surveys suggests that it has been falling. The former is probably more accurate, but there is unhelpful statistical fog (in all countries, not just the United States) in part because existing surveys are built around the conventional idea of the job and that structure is steadily disappearing. This makes it similarly difficult to gauge the effects of the new economy on benefits, pensions, and health insurance.

Looking at other countries’ labor markets might help, however. Driving for Uber in the United States might be described as contingent: a few hours a week around one’s domestic tasks, or as a second source of income. But in the UK, Uber is licensed as a taxi company and many of its drivers have switched from other cab companies. In Indian or African cities—or for that matter in the high unemployment suburbs of Paris—being an Uber driver is a pathway into the formal economy from a far more contingent existence as a day laborer, or being unemployed.

It is certainly not impossible to devise practical policies to tackle these challenges, but the task is immense and one governments have failed before. The legacy of failing the last generation of workers in industrialized states and regions is still with us—in declining towns, chronic ill health and drug abuse, substandard housing, disappointing educational outcomes, and low incomes. It is hard for adults to retrain at all, and especially when the fundamental skills needed in the workplace have changed so much. The education system (on both sides of the Atlantic) is failing miserably to equip children for the jobs of the future, so fundamental education is a big task ahead and one that is not on the political radar.

Rogers’s call for a more robust public sector is spot on. We need measures for taxpayer-funded education and retraining; increased provision of public services; and perhaps even job guarantees for some redundant workers. Above all, though, we need constraints on the market power of corporations and their tax avoidance.

Perhaps these measures seem politically unrealistic, but they are certainly more realistic than a basic income. After all, many of them have been put into practice in the past.

Besides, a basic income is an answer to the wrong question. It addresses a hypothetical future problem of no jobs in place of jobs of the kind many people still do—such as driving trucks, or nursing patients. Market economies can always adjust to create new kinds of work. We have very few horse-and-carriage drivers or dock hands these days, and many social media consultants and special effects designers for video games. If the robots take over those jobs in turn, humans will define other activities as work, impossible to foresee now. This is the kind of transition that has been occurring for a quarter of a millennium now, and it will occur again.

Given the uncertainty, however, basic income experiments in places such as Finland and the city of Utrecht in the Netherlands are to be welcomed. These will give economists some evidence to chew over, especially regarding the effect on work incentives, or the cost of such a large-scale program. But in the United States, at least for the foreseeable future, it is quite hard to envision enough voters being won over to a something-for-nothing option. 

Skeptics of a basic income are right to point out its utopianism. What is needed now is less of the beautiful dream and more of the dirty realism. Get the data to understand people’s job market alternatives and choices, work out how to regulate the actual labor market more effectively, and make a start on specific policies to re-tether the automating businesses to the societies out of which they are making their profits. And most important of all, continue to grow the economy to ensure new jobs—whatever form they take—are created.