Sunday, April 30, 2017
Photo of Robert J. Shiller

A moderate tax on robots, even a temporary tax that merely slows the adoption of disruptive technology, seems a natural component of a policy to address rising inequality.

Visit article 79

Robotization Without Taxation?

NEW HAVEN – The idea of a tax on robots was raised last May in a draft report to the European Parliament prepared by MEP Mady Delvaux from the Committee on Legal Affairs. Emphasizing how robots could boost inequality, the report proposed that there might be a “need to introduce corporate reporting requirements on the extent and proportion of the contribution of robotics and AI to the economic results of a company for the purpose of taxation and social security contributions.”

The public reaction to Delvaux’s proposal has been overwhelmingly negative, with the notable exception of Bill Gates, who endorsed it. But we should not dismiss the idea out of hand. In just the past year, we have seen the proliferation of devices such as Google Home and Amazon Echo Dot (Alexa), which replace some aspects of household help. Likewise, the Delphi and nuTonomy driverless taxi services in Singapore have started to replace taxi drivers. And Doordash, which uses Starship Technologies miniature self-driving vehicles, is replacing restaurant delivery people.

If these and other labor-displacing innovations succeed, surely calls to tax them will grow more frequent, owing to the human problems that arise when people lose their jobs – often jobs with which they closely identify, and for which they may have spent years preparing. Optimists point out that there have always been new jobs for people replaced by technology; but, as the robot revolution accelerates, doubts about how well this will work out continue to grow. A tax on robots, its advocates hope, might slow down the process, at least temporarily, and provide revenues to finance adjustment, like retraining programs for displaced workers.

Such programs may be as essential as our work is to healthy human life as we know it. In his book Rewarding Work, Edmund S. Phelps emphasized the fundamental importance of maintaining a “place in society – a calling.” When many people are no longer able to find work to support a family, troubling consequences ensue, and, as Phelps stresses, “the functioning of the entire community may be impaired.” In other words, there are externalities to robotization that justify some government intervention.

Critics of a robot tax have emphasized that the ambiguity of the term “robot” makes defining the tax base difficult. The critics also stress the new robotics’ enormous, undeniable benefits to productivity growth.

But let’s not rule out so quickly at least modest robot taxes during the transition to a different world of work. Such a tax should be part of a broader plan to manage the consequences of the robotics revolution.

All taxes, except a “lump-sum tax,” introduce distortions in the economy. But no government can impose a lump-sum tax – the same amount for everyone regardless of their income or expenditures – because it would fall heaviest on those with less income, and it would grind the poor, who might be unable to pay it at all. So taxes have to be related to some activity indicative of ability to pay taxes, and whatever activity it is will be discouraged as a result.

Frank Ramsey published a classic paper in 1927 arguing that to minimize taxation-induced distortions, one should tax all activities, and he proposed how to set tax rates. His abstract theory has never been a fully operational principle for actual tax rates, but it provides a powerful argument against presuming that the tax should be zero for all but a few activities, or that all activities should be taxed at the same rate.

Activities that create externalities might have a higher tax rate than Ramsey would have proposed. For example, taxes on alcoholic beverages are widespread. Alcoholism is a major social problem. It destroys marriages, families, and lives. From 1920 to 1933, the United States tried a much harsher market intervention: outright prohibition of alcoholic beverages. But it turned out to be impossible to eliminate alcohol consumption. The alcohol tax that accompanied the end of Prohibition was a milder form of discouragement.

Discussion of a robot tax should consider what alternative we have to deal with rising inequality. It would be natural to consider a more progressive income tax and a “basic income.” But, these measures do not have widespread popular support. If support is not widespread, the tax, even if imposed, will not last.

When taxes on high incomes are raised, usually in wartime, it turns out to be only temporary. Ultimately, it seems natural to most people that taxing successful people to benefit unsuccessful people is demeaning to the latter, and even the recipients of the handout often do not really want it. Politicians know that: they usually do not campaign on proposals to confiscate high incomes and pad low incomes.

So, taxes must be reframed to remedy income inequality induced by robotization. It may be more politically acceptable, and thus sustainable, to tax the robots rather than just the high-income people. And while this would not tax individual human success, as income taxes do, it might in fact imply somewhat higher taxes on higher incomes, if high incomes are earned in activities that involve replacing humans with robots.

A moderate tax on robots, even a temporary tax that merely slows the adoption of disruptive technology, seems a natural component of a policy to address rising inequality. Revenue could be targeted toward wage insurance, to help people replaced by new technology make the transition to a different career. This would accord with our natural sense of justice, and thus be likely to endure.


Next opinion

Photo of Yanis Varoufakis

Today’s technological revolution is marked by the increasing socialization of the production of capital. A practical response would be to socialize the property rights over the large income streams capital is now generating.

Visit article 61

A Tax on Robots?

ATHENS – Ken makes a decent living operating a large harvester on behalf of farmer Luke. Ken’s salary generates income tax and social security payments that help finance government programs for less fortunate members of his community. Alas, Luke is about to replace Ken with Nexus, a robot that can operate the harvester longer, more safely, in any weather, and without lunch breaks, holidays, or sick pay.

Bill Gates thinks that, to ease the inequality and offset the social costs implied by automation’s displacement effects, either Nexus should pay income tax, or Luke should pay a hefty tax for replacing Ken with a robot. And this “robot tax” should be used to finance something like a universal basic income (UBI). Gates’s proposal, one of many variants on the UBI theme, allows us to glimpse fascinating aspects of capitalism and human nature that rich societies have neglected for too long.

The whole point of automation is that, unlike Ken, Nexus will never negotiate a labor contract with Luke. Indeed, it will receive no income. The only way to simulate an income tax on behalf of Nexus is to use Ken’s last annual income as a reference salary and extract from Luke’s revenues income tax and social security charges equivalent to what Ken paid.

There are three problems with this approach. For starters, whereas Ken’s income would have changed over time had he not been fired, the reference salary cannot change, except arbitrarily and in a manner setting the tax authorities against business. The tax office and Luke would end up clashing over impossible estimates of the extent to which Ken’s salary would have risen, or fallen, had he still been employed.

Second, the advent of robot-operated machines that have never been operated by humans means there will be no prior human income to act as a reference salary for calculating the taxes these robots must pay.

Finally, it is hard philosophically to justify forcing Luke to pay “income” tax for Nexus but not for the harvester that Nexus operates. After all, both are machines, and the harvester has displaced far more human labor than Nexus has. The only defensible justification for treating them differently is that Nexus has greater autonomy.

But to what extent is Nexus genuinely autonomous in a manner that the harvester is not? However advanced Nexus might be, it can be thought of as autonomous if and only if it develops consciousness, whether spontaneously or with the help of its makers.

Only if Nexus (like the Nexus-6 replicants in the 1982 film Blade Runner) achieves that leap will “he” have earned the “right” to be thought of as distinct from the harvester he operates. But then humanity will have spawned a new species and a new civil rights movement (which I would gladly join) demanding freedom for Nexus and equal rights with Ken – including a living wage, minimum benefits, and enfranchisement.

Assuming that robots cannot be made to pay income tax without creating new potential for conflict between the tax authorities and business (accompanied by tax arbitrage and corruption), what about taxing Nexus at the point of sale to Luke? That would of course be possible: the state would collect a lump-sum tax from Luke the moment he replaces Ken with Nexus.

Gates supports this second-best alternative to making robots “pay” income tax. He thinks that slowing down automation and creating tax disincentives to counter technology’s displacement effect is, overall, a sensible policy.

But a lump-sum tax on robots would merely lead robot producers to bundle artificial intelligence within other machinery. Nexus will increasingly be incorporated within the harvester, making it impossible to tax the robotic element separately from the dumb parts that do the harvesting.

Either the robot sales tax should be dropped or it should be generalized into a capital goods sales tax. But imagine the uproar against a tax on all capital goods: Woe betide those who would diminish domestic productivity and competitiveness!

Ever since the emergence of industrial capitalism, we have been terrible at differentiating between property and capital, and thus between wealth, rent, and profits. This is why a wealth tax is so difficult to design. The conceptual problem of differentiating between Nexus and the harvester “he” operates would make it impossible to agree on how a robot tax should work.

But why make life under capitalism more complicated than it already is? There is an alternative to a robot tax that is easy to implement and simple to justify: a universal basic dividend (UBD), financed from the returns on all capital.

Imagine that a fixed portion of new equity issues (IPOs) goes into a public trust that, in turn, generates an income stream from which a UBD is paid. Effectively, society becomes a shareholder in every corporation, and the dividends are distributed evenly to all citizens.

To the extent that automation improves productivity and corporate profitability, the whole of society would begin to share the benefits. No new tax, no complications in the tax code, and no effect on the existing funding of the welfare state. Indeed, as higher profits and their automatic redistribution via the UBD boosted incomes, more funds would become available for the welfare state. Coupled with stronger labor rights and a decent living wage, the ideal of shared prosperity would receive a new lease on life.

The first two industrial revolutions were built on machines produced by great inventors in glorified barns and bought by cunning entrepreneurs who demanded property rights over the income stream “their” machines generated. Today’s technological revolution is marked by the increasing socialization of the production of capital. A practical response would be to socialize the property rights over the large income streams capital is now generating.

In short, forget about taxing either Nexus or Luke. Instead, place a portion of Luke’s equity in the farm in a public trust, which then provides a universal payment to everyone. In addition, we must legislate to improve the wages and conditions of every human still in employment, while our taxes provide Ken unemployment benefits, a guaranteed paid job in his community, or retraining.


A Robots Tax?

Read Comments (3)
  1. Comment Commented

    If the cost of research for making robots is partially taken from the same kitty that would have created employment for those who are going to be unemployed by the robots, the tax makes enormous sense. Unfortunately the social capital loses identity once it is channelized into social or private gains; either way there is usefulness. Read more

  2. Comment Commented

    Taxing robots might make sense if there was any evidence that robots were actually having a significant impact on the economy. However, the actual data shows almost the opposite. Productivity growth has crashed (to zero or near zero) throughout the Western world. So much for Tech (including robots) transforming the economy.

    Of course, if robots really destroy jobs, wouldn't it make a lot more sense to stop immigration first. Immigration has huge negatives even with job losses. Immigrants taking jobs are just one more reason to close borders. Read more

  3. Comment Commented

    What is a robot? A printer collating and stapling copies autonomously? An automated lift gate in a parking garage? Isn't income tax a tax on productivity already? Shall we reward inefficiency by punishing efficiency? I don't have an answer but I dread a resurrection of the Luddites.
    Read more