Automation Anxiety: Why and How to Save Work, by Cynthia Estlund

The prospect of net job losses overall would further worsen the outlook for most workers, but the ongoing hollowing out of the labour market should draw the concern of citizens, policy makers and scholars. Banner: Shutterstock/FGC
Automation Anxiety: Why and How to Save Work, by Cynthia Estlund
Adapted from the 1st, 5th, 7th and 9th chapters of Automation Anxiety: Why and How to Save Work, by Cynthia Estlund. Copyright 2021. Used with permission of the publisher, Oxford University Press. All rights reserved.
The Forum Network is a space for experts and thought leaders—from around the world and all parts of society— to discuss and develop solutions now and for the future. It aims to foster the fruitful exchange of expertise and perspectives across fields, and opinions expressed do not necessarily represent the views of the OECD.


Many trends that have unsettled the world of work and drawn the anxious attention of scholars, advocates, and policy wonks— deindustrialisation, offshoring, and the disintegration of jobs into gigs, as well as the decline of unions and the steady rise of economic inequality— are intertwined with the ongoing replacement of human workers by machines.

Advances in both hard and soft forms of technology— robots and algorithms, for example— are replicating a wider range of human capabilities and weaving together those distinct capabilities more seamlessly than ever before. The very terms “artificial intelligence” and “machine learning” hint at what is new: technology is acquiring and refining cognitive and sensory capabilities that had long been thought to be uniquely human, and is outpacing humans at increasingly complex tasks.

On the other side of the debate, many economists dismiss the latest wave of automation anxiety as misinformed scare- mongering by modern- day Luddites. They point out that predictions that machines will supplant human labour have recurred in both utopian and dystopian flavours throughout the history of industrialization. Futurists of the past have predicted that mass automation will usher in an era of human liberation from toil, or that it will immiserate all but the fortunate few who own or create the machines. Time and again, however, the economy has defied such predictions. For centuries, automation has been destroying some jobs while creating other jobs— usually better paid and less gruelling— and driving economic growth and prosperity. In short, the history of automation’s impact on the labour market has been one of “creative destruction,” a mantra to which many economists still adhere. […]

But while debate continues over whether we face a future of less work overall, there is growing evidence that automation is already destroying more decent mid- skill jobs than it is creating. Too many of those who are displaced from those jobs are crowding down into the less- skilled but still human jobs that remain instead of climbing up to the better- paid, higher- skilled jobs that require more advanced training. The prospect of net job losses overall would further worsen the outlook for most workers, but the ongoing hollowing out of the labour market should be quite enough to draw the concern of citizens, policy makers, and scholars. […]

The stakes are not only economic. Useful paid work is central to most people’s identity and to our social and political life. The experience of shared work fosters social interaction and integration, solidarity and friendship, and cooperation and compromise, and it does so among relatively diverse groups of co- workers— not as diverse as they should be but more diverse than just about any other social setting in most adults’ lives.  If a growing segment of the population finds itself detached from the world of work, our social fabric will become even more frayed and our politics even more fraught.

So we have plenty to worry about, especially but not only if job losses outpace job gains. Whether we’re facing the labour market equivalent of a tropical storm or a Category 5 hurricane, we had better start thinking about what to do about it even as we continue to keep our eyes on the forecast. […]

To find out more, consult also our dedicated website to access a wide range of resources and reports on the impact of AI on the labour market and how labour markets are changing in response to the mega-trends of technological change, globalisation and population ageing

Three Big Ideas (and Some Big Concerns)

I argue that we should judge policy proposals for a future of less work on their ability to ensure wide access to decent paid work, adequate material resources, and more time for life outside of paid work. Those are three dimensions of a good life that are at stake— either at risk or within reach— in a more automated future.

Each of those dimensions points to one of three basic public policy strategies for dealing with job losses: creating work, supplying non- work- based income or benefits, or spreading work by reducing hours per worker. Whenever and for whatever reason we face a job deficit, variations on those three options dominate the policy menu. We saw that during the Great Depression and more recently in the wake of the COVID-19 collapse: in both of those periods of mass unemployment, policy makers in the United States and elsewhere deployed all three of those strategies, sometimes in combination. For there is no incompatibility among those basic strategies of job creation, income support, and work spreading.

Each of those three strategies has also been sculpted into a high- profile Big Idea: universal basic income (UBI), a federal job guarantee (JG), and shorter working hours. However, these Three Big Ideas, unlike the underlying strategies they embody, are not all compatible with each other. That incompatibility is problematic because none of the Three Big Ideas by itself can deliver on all three goals of income, leisure, and work. Each takes aim squarely at one of those goals, and would at least incidentally advance a second one; but each falls short on a third goal. By way of preview:

  • A full-scale UBI would guarantee adequate income, and would support more free time (on the broad definition used here) for those who choose to rely on that income; yet it falls short on ensuring wide access to paid work in a future of job scarcity.
  • A job guarantee would ensure wide access to decent paid work, as well as adequate income from that work; but it would entrench the current norm of full-time work and miss the opportunity to distribute more free time in a more automated economy.
  • A shorter standard work week takes square aim at expanding free time, and would more- than- incidentally spread available work to more people; standing alone, however, shorter working hours would very likely fail to ensure adequate incomes for all.

In short, each of the Three Big Ideas meets two out of the three goals, while missing a third.

The UBI and the JG represent bold and sharply divergent strategies for averting a dystopic future of less work. The UBI would replace work, at least for some, with cash and freedom from work. But that prescription faces prohibitive political hurdles given a broad and deep cultural attachment to work; and that attachment is worth cultivating, given the social and political value of shared work. The JG, by contrast, gains normative and political appeal in its coupling of income and work, broadly accessible to all. But if automation continues to destroy mid- skill jobs, and especially if it yields net job losses, a job guarantee risks swelling to unmanageable proportions, and to put a growing number of people to work at taxpayer expense doing things that machines can do better and more cheaply. Besides, if that is where we are heading— if we can produce the goods and services that we need with less human labour— does it make sense to spend trillions of dollars to guarantee full- time, full employment? Shouldn’t we instead be aiming to distribute more time, and a better balance between work and life, as one of the dividends of automation?

Shorter working hours promises that better balance— and without any eye- popping public price tag.

Work spreading would capture a big upside of automation— more time for life outside of paid work— and alleviate one of its biggest downsides— rising unemployment and underemployment. It would stave off the dystopian worst- case scenarios for a more automated economy, and would bring real gains in well- being. Successful work spreading would also mitigate income inequality— bringing down some top incomes by reducing long hours at the top and pulling more workers up into those higher wage jobs. But the idea of shorter working hours operates on a different plane than UBI or the JG.

Far from foreclosing other strategies such as job creation or income support, the shorter hours strategy only works in tandem with some strategy for increasing incomes at the bottom of the distribution. Otherwise, shorter hours portend a decline in living standards. [...] Combining successful work spreading with an array of basic social entitlements and public goods and services and the resulting public jobs would nevertheless get us nearly all the way to a much better future of less work.

But there remains a problem of maintaining adequate incomes.

Spreading Work and Supporting Incomes

Work spreading, however cleverly it is done, is likely to leave income gaps. Ample public goods and universal basic benefits would narrow those gaps. But people will still need money in a market economy (and I’m assuming that’s what we will still have).

Higher minimum wages are unlikely to fill the whole gap. Imagine that average hours of work fell to twenty hours per week or less, and that workspreading techniques succeeded, against all odds, in keeping most people engaged in paid work for much of their adult lives at around that level. To guarantee a “living wage” at that point— that is, for twenty hours of work per week to fully support a decent living— would require huge increases in the minimum wage. That risks decimating the demand for ordinarily- skilled human labour, and making it even harder to maintain a wide distribution of paid work across the adult population. We should therefore consider other ways to bridge the income gap— that is, to supply some non-work-based income to most workers.

A negative income tax or a UBI— one that expanded as median hours of work fell— would make more sense than relying on existing means- tested welfare benefits. Those programs were designed to supply a “safety net” for an unfortunate few who fell between the cracks of a labour market that fully supported most people in some fashion. But in a future of much less work, we would be aiming to supplement the shrinking work- based income of a wide and growing share of the population. Some combination of basic income and basic benefits will be needed in that future of much less work.

As automation continues to swell capital’s share versus labour’s share of national income, an alternative strategy for ensuring that workers get a fair share of the economic gains from automation is widely distributed ownership of capital. “Who owns the robots owns the world,” says economist Richard Freeman, a leading proponent; so “[l]et us own the robots.” If workers did own a growing corpus of capital, then automation would generally increase its value and the income from that capital even as it depressed wages and demand for most labour. The worker ownership strategy, with its built- in indexing of non- work- based income to declines in the availability of work, is a smart strategy for supporting workers’ incomes in a future of less work.  [...]

Work spreading, together with expanded social benefits and investments in public goods and public jobs, can cushion the losses and spread the gains from the increasing automation of production.

The Politics of Hope and Fear in a Future of Less Work

In my book Automation Anxiety, I propose an array of work-spreading techniques, including but not beginning with a shorter standard work week. Work spreading, together with expanded social benefits and investments in public goods and public jobs, can cushion the losses and spread the gains from the increasing automation of production. […]

The three- dimensional vision of a future of less work that I argue for— a future of wide engagement in decent paid work, more time for the rest of life, and decent incomes and standards of living for all— will require big changes in law and social policy. Some readers will doubt that these changes are possible under capitalism, while others might doubt that a society thus transformed would still be capitalism. But there are varieties of capitalism, and some of them leave room for both the prodigiously innovative, energizing, and coordinating forces generated by markets for goods and services, labour, and capital, and the humanizing and rationalizing forces of regulation, redistribution, and democratization of economic life. […]

The strategy advanced here could be seen as another in a long line of progressive- minded efforts to save capitalism from itself, and from the social and political unrest that may follow if we stay the current course in the face of technological job destruction and growing economic insecurity. Randall Collins, a leading American sociologist, has argued that “the end of middle- class work” due to “technological displacement of labour . . . will generate the long- term and quite possibly terminal crisis of capitalism.” That is partly because job losses will depress purchasing power and stall the economy. But it is also because of the political consequences of widespread economic deprivation, which will lead, he posits, to the end of capitalism, the end of democracy, or both.

At least for the foreseeable future, it seems preferable and possible to pursue a middle course. In a more- automated socially democratic market economy, that will require a three- dimensional strategy that spreads work, free time, and material necessities more widely and equitably than our existing economic and political institutions will otherwise do. And in a socially democratic market economy, we will have to get there and stay there by democratic means— that is, by persuading a large majority of voters that it is the best path forward for them and their families. An increasingly vivid threat of job losses due to automation, and fears of a future of less work for themselves and their children, might help to nudge a critical mass of voters toward that conclusion.

Find out more about Automation Anxiety: Why and How to Save Work, by Cynthia Estlund (Published in October 2021, © Oxford University Press)

The 2022 International Conference on AI in Work, Innovation, Productivity and Skills took place on 21–25 February—watch the replays!

Please sign in

If you are a registered user on The OECD Forum Network, please sign in