The Ecology of Technology: Should robots pay taxes?

How Tax Policy Favours Robots over Workers and What to do About it

There is a longstanding and growing public debate about the costs and benefits of automation. Earlier this year, Bill Gates argued that robots who take human jobs should pay taxes (link to: Mark Zuckerberg recently warned the graduating class at Harvard University that, “Our generation will have to deal with tens of millions of jobs replaced by automation like self-driving cars and trucks.” Elon Musk has joined the ranks of industrialists in favor of guaranteed minimum income, which he says will be necessitated by automation.

It isn’t just industry leaders who are aware of the problems automation poses. This year the EU Parliament considered a proposal to tax robot owners to pay for retraining displaced workers, although that legislation was rejected.

Neglected in this debate has been the critical relationship between tax policy and automation. In an article forthcoming in the Harvard Law and Policy Review (Volume 12.1) (link to:, we argue that existing tax policies encourage automation, even when a human worker would otherwise be more efficient than a machine. That’s because automating lets firms avoid wage taxes, which fund social benefit programs like Medicare, Medicaid, and Social Security. Only with human workers are firms responsible for paying taxes on wages to the federal government (and sometimes additional wage taxes at the state or local level) based on an employee’s salary.

Automating also allows firms to claim accelerated tax depreciation on capital costs for machines. This allows firms to claim early tax deductions relative to the actual economic depreciation of an asset over time (like a robotic worker).  Wage taxes, on the other hand, are only deductible as they are paid. This system allows firms to alter the timing of their deductions, which can result in a present value benefit and a substantial impact on firm finances.

Finally, automation results in indirect tax incentives for firms. For example, human workers are also consumers who are responsible for paying consumption taxes such as retail sales tax (RST) or value added tax (VAT). However, employers are thought to bear some of the incidence of these indirect taxes, as they may ultimately have to increase worker salaries and benefits to offset these taxes. Because robotic workers are not consumers, they are not subject to these indirect taxes and thus allow firms to avoid the burden of indirect taxes.

Not only do existing tax policies unintentionally encourage automation, they also result in dramatically reduced tax remittances. That’s because workers account for most of the tax base. Workers are taxed according to their wages as well as their consumption habits. Corporate taxation now represents less than 9 percent of the overall tax base. When firms replace human workers with machines (or elect not to hire people initially), the government loses the ability to tax those workers, and this is not compensated for in the form of higher taxes on corporate earnings. In the aggregate, this could potentially amount to hundreds of billions or even trillions of dollars a year in lost tax revenues if robots replace workers to the extent some experts predict.

The automation incentives and effects on tax remittances stem from the fact that tax policies are designed to tax labor rather than capital. This creates unintended consequences when the labor is capital.

We propose changes to existing policies to remove the tax incentives in favor of automation.  There are several mechanisms for achieving this goal, and the optimal solution may be a combination of mechanisms.

Corporate tax deductions could be disallowed for machines that replace human workers. This would eliminate the breaks firms receive on wage taxes, deduction timing, and indirect taxes from automation.

An “automation tax” could be based on existing unemployment systems. Firms could be required to pay additional amounts into an insurance plan if they automate at the expense of jobs.

Tax preferences could be granted for human workers to counter preferences for machines. This would in a sense mirror the proposed disallowance of corporate tax deductions for automated workers, but would create offsetting incentives for human workers.

A corporate self-employment tax could be created for firms that rely on machine labor. This would substitute for what the firm would have paid if it were staffed with people for whom it owed wage taxes. This would be similar to individual self-employment taxes that are paid to approximate the social security taxes that would otherwise be owed. This tax could be based on a ratio of corporate profits to gross employee compensation expenses.

Finally, the corporate tax rate could be increased in combination with the above mechanisms.

Ultimately, tax neutrality between human workers and robots will improve efficiency by allowing firms to select the more efficient worker, robot or human, without tax-based distortions. Because tax policies currently encourage automation, it incentivizes firms to replace workers with robots for tax purposes even when people might otherwise be more efficient. However, tax neutrality will not solve the problem of reduced tax remittances. That might require a fundamental rethinking of the ways in which we tax labor vs. capital, and companies vs. workers.

As a first step, creating a tax neutral system that allows the marketplace to choose the most efficient option would be a promising development. However, our proposals could be adopted in stronger forms to actively discourage automation. This would be a way to practically implement Bill Gates’ suggestion to slow the spread of automation, and the enhanced tax remittances could be used to fund other types of employment and social benefits.

There are advantages to automation besides efficiency, as one of us has argued, such as improving safety and promoting innovation. Whether or not discouraging automation is desirable, it is, as Gates said,  “interesting for people to start talking about now.” (link to:]

This post comes to us from Professor Ryan Abbott and Dr. Bret Bogenschneider at the University of Surrey School of Law. It is based on their forthcoming paper, “Should Robots Pay Taxes? Tax Policy in the Age of Automation,” which will be published in Volume 12.1 of the Harvard Law and Policy Review. The working paper is available here. The post originally appeared on Columbia Law School’s Blue Sky Blog.



Share this Post!

About the Author : Kris Henley

Communications and Outreach for Surrey Business School's Centre for the Digital Economy, a newly-founded research centre to explore the implications of the Digital Economy for business, government and society.

Skip to toolbar