Should the US Enforce a ‘Robot Tax’?

Economists Arnaud Costinot and Iván Werning from MIT delve into the potential ramifications of a proposed ‘robot tax’ on automation and employment.

A largely unnoticed facet of journalism involves understanding your readership. It’s not always as simple as it sounds, especially when reporting on technology. Balancing between over-explanation and under-explanation can be tricky. Assuming too much knowledge can alienate non-experts, while diving into intricate details risks coming off as condescending.

Recently, I reached out to LinkedIn to gather opinions on mainstream coverage of robotics, particularly from publications not specializing in the field or technology in general. For me, the recurring headline “The Robots Are Coming” has become a minor annoyance, popping up nearly every week.

The responses echoed my expectations: fears of a robopocalypse with killer robots, a lack of historical context, and an excessive focus on flashy humanoid robots and gimmicks. These criticisms are valid, and I’ll certainly consider them in my future work. I stopped using “robopocalypse” in my vocabulary a while ago, except when referring to the internet’s knee-jerk reaction to new robots.

Another common complaint concerned the discussion of jobs. Just like sensationalist robopocalypse headlines, I concur that conversations tend to veer towards sensationalism. “The Robots Are Coming” is often followed by “For Your Job.” This parallels the “AI is taking your job” narrative. Generally, AI is associated with white-collar jobs, while robots are linked with blue-collar roles. While not a one-to-one correlation, this pattern usually holds: robots in factories and AI in offices.

Sensationalism is not unique to robotics but rather a characteristic of online journalism. Our industry has been struggling for survival for a while now, longer than my tenure within it (which has been quite long). Some days, it feels like we’re all fighting for the same slivers of attention, hoping people can tear themselves away from TikTok long enough to skim a news article. When competing for shrinking attention spans amid the deluge of instantly accessible information, framing becomes a significant concern.

Such blunt force not only does a disservice to the robotics industry but also erases the subtleties required for a truly nuanced conversation. Some might prefer to skip the jobs discussion entirely, but I believe that approach is equally problematic.

So, let’s start with something we can all agree on: Robots have already and will continue to influence jobs. The presence of robots in the workforce is rapidly expanding. The more widespread and advanced automation becomes, the more significant its impact on our work.

I deliberately chose the term “impact” because it’s neutral; it’s neither inherently negative nor positive. The workforce of the future will be different, and robotics will undoubtedly be a key driver of that change.

In my writings for TechCrunch, I’ve aimed for a nuanced approach to the jobs question. Ultimately, it’s up to readers to determine if I’ve succeeded. Most people I speak to believe the impact will be positive – that robots will either replace bad jobs or, at the very least, make them better. While there’s truth in these statements, I am mindful that most of the people I speak to about robots are either roboticists or investors, roles that require optimism.

I don’t see my role as playing devil’s advocate, but I do feel a sense of responsibility to remind readers that jobs aren’t just numbers; there are humans behind each one. In a position that frequently requires me to write about layoffs involving tens of thousands, it’s easy to lose sight of this fact. I’ve certainly been guilty of leaning into abstraction. This is why I often post job listings in Actuator. For the vast majority of us, our survival depends on our ability to work. That’s the reality of our world.

It’s important to engage in conversations about the long-term impact of automation. This debate will continue into the foreseeable future, and I’m glad when people discuss it with all the context and nuance required. However, I believe we often discuss it at the expense of the short-term impact – the jobs that are immediately affected. This is where the contentious and less contentious topics of safety nets and upskilling come into play. These are subjects we’ll explore on another occasion.

However, we’re not sidestepping controversy entirely this week. In some circles, the current hot topic is even more contentious than the ones mentioned above: the robot tax. We haven’t discussed it much in Actuator, so it’s time to dive in. Given the nature of this newsletter, what follows won’t be an exhaustive examination of the subject, but it’s a good opportunity to address something that has been discussed for a long time.

Brookings defines the concept as follows:

The core idea behind a robot tax is that companies would pay a tax when they replace a human worker with a robot. Such a tax would theoretically serve two main purposes. First, it would discourage companies from replacing workers with robots, thereby preserving human employment. Second, if the replacement happened regardless, a robot tax would generate government revenues to compensate for the loss of income from payroll taxes.

Despite the Institute’s perspective, this definition generally captures the idea. When contemplating the concept, the “loss of income from payroll taxes” is secondary to the more immediate concern about the potential human toll.

Back in 2017, we published a column by Steve Cousins that concluded with:

Getting companies to pay their fair share of taxes won’t solve the broader societal challenge of automation displacing low-skilled workers, nor would a robot tax. Instead, governments should use corporate tax revenue to create free or low-cost education programs that prepare people to work alongside automation.

For those who can’t find work in the tech-driven society of the future, governments could offer universal basic income or other safety nets for the least advantaged.

To this, I say that these concepts aren’t mutually exclusive. In fact, from my perspective, funding a social safety net is perhaps the most compelling argument for a robot tax. The following statement is the closest I’ll get to a political stance in today’s newsletter: Ready? Okay. I firmly believe that providing food and shelter for those without means should be seen as a fundamental government responsibility. So, it makes logical sense to combine these two ideas.

That said, I’m not currently advocating for or against a robot tax. Honestly, I’m straddling the fence on the issue. There are valid arguments on both sides. Having discussed some of the pros earlier, I would say the primary argument against it is the fear of stifling innovation. Fundamentally, it’s the same argument against any form of business tax, though with the robot tax, slowing innovation is arguably the intended outcome.

Ultimately, the question boils down to what’s more important: maintaining the current state of employment to preserve jobs or safeguarding US competitiveness? Again, I have no illusions that you’ll find all the answers in this week’s robot newsletter. Nevertheless, if it gets more people thinking about the topic, I’ll consider it a job well done.

I hope that in the near future, I’ll have the time and capacity to delve deeper into this topic. For this week, I’m relying heavily on a study from MIT published late last year.

Published in the Review of Economic Studies, “Robots, Trade, and Luddism: A Sufficient Statistic Approach to Optimal Technology Regulation” aims to provide a “general theory of optimal technology regulation.” The MIT economists behind the study, Arnaud Costinot and Iván Werning, ultimately suggest a moderate level of taxation.

“Our findings suggest that taxes on either robots or imported goods should be relatively low,” Costinot stated to MIT at the time. “Although robots affect income inequality… they still lead to optimal taxes that are modest.”

Prominent figures, including Bill Gates and Bernie Sanders, have advocated for some form of taxation over the years. In 2017, Gates told Quartz, “You ought to be willing to raise the tax level and even slow down the speed.” He cited factors such as a broad, simultaneous job displacement across various industries.

When asked on CBS Sunday Morning about Gates’ position, Sanders responded, “That’s one way to do it. Absolutely.” His broader perspective on automation aligns with expectations: “So if we can reduce the workweek, is that a bad thing? It’s a good thing. But I don’t want to see the people on top simply be the only beneficiaries of this technological revolution.”

For a counterargument, we turn back to Brookings, which highlights the potential for automation to ultimately create more jobs:

“[E]xisting research suggests that firms adopting robots actually experience an increase in employment, undermining a key argument in favor of a robot tax,” writes senior fellow Robert Seamans. “Furthermore, a robot tax would require a clear definition of what constitutes a robot, a challenge in itself. Instead, policymakers should consider other policy changes to aid workers, potentially involving adjustments to how capital and labor are taxed, along with broader labor market reforms.”

To date, only South Korea has come close to passing legislation, although their approach primarily reduces tax credits by two percentage points rather than introducing an entirely new tax.

To gain a deeper understanding of their research, I conducted an email interview with Costinot and Werning.

TC: “Robots, Trade, and Luddism” was published late last year. Have there been any recent developments that have affected your findings?

AC/IW: Since we wrote the paper, there have been significant advances in and concerns about AI technologies. Our results can be applied to this technology.

We provide a general formula that relies on the impact of technology on wage distribution as an input. This vital input remains uncertain for AI, and there is ongoing research and speculation.

Regarding “redistribution,” is the idea that the taxes collected will directly benefit those whose jobs have been displaced by automation?

The primary focus is not the revenue from the robot tax but how the tax shapes labor demand, thus affecting wages and jobs. The idea is that the tax can mitigate potential inequalities in earnings resulting from new technologies, a form of “pre-distribution” that impacts earnings before taxes.

The effectiveness of “upskilling” campaigns has received mixed reviews. What is your perspective on such initiatives in the context of displaced blue-collar jobs?

We haven’t delved deeply into this area. However, at a broad level, the same forces apply: Skill acquisition can be analyzed similarly to our work, representing the other side of the coin. If training can improve skill distribution, there’s an argument for subsidizing it. Nonetheless, we haven’t extensively studied its efficacy or explored this topic in detail.

You suggest that a 1% to 3.7% tax on value represents the optimal range for taxing these systems. What changes beyond that threshold?

Yes, to clarify, these are the ranges our formulas produce based on available tentative evidence. However, the impact on the wage distribution resulting from automation is a crucial input with considerable uncertainty.

To your question: At the optimum, you’re balancing the improvement of the pre-tax wage distribution against the efficiency losses of the tax, finding a sweet spot. If the tax is too high, it means you’ve leaned too far into this trade-off, and the efficiency losses have become more significant. Assessing this trade-off involves considering whether other tools for redistribution are available. If not, then higher taxes may be necessary. In our benchmark, we allow for a nonlinear income tax, as present in the US and advanced countries. In our calibration, consistent with existing literature, this income tax is relatively effective, explaining why we find a relatively low tax rate.

We didn’t anticipate this outcome, and the relatively low figure did surprise us. However, the theory and evidence guided us in this direction.

Does implementing a robot tax risk stifling innovation and competition? Is it seen as a barrier to domestic manufacturing growth?

Yes, in principle, it can have both effects, unless counterbalanced by other policies. In general, these are some of the efficiency losses that are part of the trade-off we’ve considered, as mentioned earlier. That’s also why the tax rate isn’t higher.

Professor Werning told MIT, “We think it’s incorrect to discuss this tax on robots and trade as if they are our only tools for redistribution.”

What are other potentially more impactful tools for addressing inequality?

In the US, the income tax (combined with state taxes, Earned Income Tax Credit, etc.) is a crucial tool for redistribution and a key policy instrument, as evident from its size and broad reach, along with the debates surrounding it. We believe this is a pivotal point often overlooked in discussions about these issues.