Oei, Hemel On 'Robot Taxes'

Daniel Hemel and I just published a couple of blog posts — at Surly Subgroup and Whatever Source — about the taxation of robots (or, "Robotaxation," as I like to call it). You can read them here:

Shuyi Oei (Boston College), So About That Robot Tax ...

I came across a couple of news stories recently about how South Korea is introducing the world’s first robot tax. But based on the press reports, it sounds like the so-called robot tax is actually just a reduction of the tax deductions available to businesses that invest in productivity-boosting automation. The news sources themselves concede that this “robot tax” not exactly a tax on robots but rather a tax benefit reduction for automation investment. ...

A lot of what seems to be driving the tax conversation is the fear that robots are taking over jobs, though there’s some uncertainty about the extent to which robots are to blame.

Personally, I’ve been having a hard time squaring the newly ascendant tax conversation about the robot tax with the broader legal scholarship on robots. In some of the news and other commentary discussing Robotaxation, my reaction has been something to the effect of “I’m not sure that word means what you think it means.” Turns out, there is something of an existing conversation about what constitutes a robot in the first place — see, for example, Richards and Smart (2013) for a nice discussion of some of the definitional issues. See also this “What is a Robot?” piece in The Atlantic. In defining “robot,” it might matter how a robot moves in the physical world, what kind of quasi-independent agency it seems to exercise (autonomous vs. semi-autonomous), how humans interact with it, and even what sorts of emotions it triggers in us mere humans. We might understand some automated machines to be robots but others to just be automated equipment. And these distinctions make sense, from the viewpoint of areas like tort law, privacy law, the law of principals and agents, and the more general regulation of robots (and of artificial intelligence as a subcategory of robots).

But in some of the tax discussions about robots that I’ve seen on the interwebs, it’s quite clear that the authors don’t necessarily mean Robot when they say Robot. ...

Daniel Hemel (Chicago), Should Robots be Subsidized?:

Proposals for a tax on robots have attracted increasing attention since Bill Gates floated the idea in a February 2017 interview with Quartz’s Kevin Delaney. (Shu-Yi Oei has an excellent overview of the debate over at the Surly Subgroup blog.) The concern that motivates most of these robot tax proposals, as I understand it, is that robots will replace human labor as an input into the production process, leading to higher unemployment and economic inequality. Missing from the debate is the argument that robot taxes should be negative  —  in other words, that robots should be subsidized. Yet in my view, the argument for a robot subsidy is much more persuasive than the argument for a robot tax.

First, a word about the word “robot.” Defining a robot is devilishly difficult. If the concern is about technological innovations that reduce the ratio of human labor inputs to economic outputs, then presumably the proposed robot tax should be targeted at those innovations. As Shu-Yi notes, not all such innovations resemble R2-D2 (and, I might add, not all innovations that resemble a robot reduce the ratio of human labor inputs to economic outputs  —  see, e.g., Ellie, the World’s First Robotic Educational Elephant, brought to you by People for the Ethical Treatment of Animals)....

Let’s set the line-drawing question aside for now and assume that a robot tax would apply to all or some innovations that significantly reduce the ratio of human labor inputs to economic outputs. But wait: An innovation that significantly reduces the ratio of human labor inputs to economic outputs is, tautologically, an innovation that significantly raises the ratio of economic outputs to labor inputs. And the ratio of economic outputs (in dollars) to human labor inputs (in hours) is the very definition of labor productivity. And gross domestic product is simply hours worked times labor productivity. If we want GDP growth, then we either have to work longer hours (which doesn’t sound fun) or raise labor productivity. As Paul Krugman puts it: “Productivity isn’t everything, but in the long run it is almost everything. A country’s ability to improve its standard of living over time depends almost entirely on its ability to raise output per worker.” ...