What is the right way to tax a robot?
Automation raises important questions about the prospect of widespread technological unemployment. How will people make a living? What will happen to people’s self esteem? What will happen to income equality? And now, a new question: how will we fund government?
Eighty percent of United States federal tax revenues currently come from individual incomes — 33% via payroll taxes and 47% through income taxes. Of those income taxes, about two-thirds is wage related (as opposed to investment income). That means that roughly 65% of total US federal tax dollars depend on wage-based income.
That’s a huge dependency. Imagine what might happen to our ability to fund government if we experience widespread technological unemployment.
[Tweet “Automation threatens the 65% of total US federal tax dollars that depend on wage-based income.”]
How About a Robot Tax?
That’s part of why Bill Gates is thinking about this problem:
Certainly there will be taxes that relate to automation. Right now, the human worker who does, say, $50,000 worth of work in a factory, that income is taxed and you get income tax, social security tax, all those things. If a robot comes in to do the same thing, you’d think that we’d tax the robot at a similar level.
Bill Gates is not a dumb guy, and so he’s no doubt simplifying to get a point across, but the idea that we can tax a robot like we tax a human is a vast oversimplification. Even ignoring the potential disincentives for innovation that a robot tax might raise, there are a handful of problems that need to be considered.
Sales-Based Robot Tax
The first is that robots don’t earn wages. Robots are capital investments; outlays of cash intended to raise the productivity and profitability of the firm. In the case of robots, that productivity means automating business processes, which means reducing or eliminating labor-related costs. So, to be clear, a robot is not like a person whose income can be taxed. A robot doesn’t have income. The company that sells the robot can generate income, but not the robot itself.
Automation is Complicated
A “robot tax” also conjures up an overly simplistic view of what’s actually going on. It’s like Homer Simpson walking out the door on Tuesday night, and Bender showing up Wednesday morning to replace him.
That’s not the way actual automation works. Automation doesn’t replace jobs, it replaces tasks. Sure, sometimes we automate clusters of tasks in just such a way that they neatly replace this or that specific job. But it’s usually not that neat or tidy. Automated tasks can just as easily eat into small portions of many people’s jobs. Automating expense report submissions is a perfect example of that. No, automation doesn’t just waltz in like a robot showing up for work one day. It frequently happens in fits and starts over the course of many years, and through the patching together of various solutions from many different technology suppliers, often with teams within the automating company doing much of the work themselves.
Value Creation and the Robot Tax
The point here is that it’s easy to anthropomorphize automation in ways that distort the actual challenges that lie ahead. Robots won’t earn an income like humans do, and so we’re not going to have a robot income tax.
What we might have is some sort of tax on the companies that make robots and automation technologies more generally. So perhaps when Salesforce or Microsoft or Rethink Robotics sells an automation solution that displaces human labor there should be some special automation sales tax?
Perhaps, but a simple automation sales tax is also problematic because it fails to capture the actual value created through automation. Let’s say that an online commerce company, like Amazon, replaces its 40,000 warehouse workers with robots and artificial intelligence.
Let’s assume that it amortizes the cost of its automation technology over eight years at $100 million a year. The company has really smart human management, so this investment greatly increases its profitability well beyond what it might in some other company. Let’s assume, for example, that it results in $200 million in annual labor savings and an additional $100 million in increased revenues thanks to better customer satisfaction. Using this simple example, a simple automation sales tax would have a tax basis of $100 million base, even though there was $300 million of actual value created by this investment in automation.
That’s a lot of value created by labor-displacing automation technology that’s just not captured in a simple sales tax.
AVAT or Corporate Income Tax?
One solution is what most developed countries outside of the United States address with a Value-added tax (VAT). So perhaps it is worth looking at the idea of some special automation VAT, or “AVAT” based on the value generated by automation technologies.
But even the AVAT is still fraught with difficulties. How, exactly, is the value calculated? And how does one determine which kinds of investment actually count as labor-cutting automation technologies?
So maybe we’e better off assuming all corporations will automate and simply rely on a good old corporate income tax. Right now corporate income taxes only account for 11% of federal tax revenues. To offset wage-based taxes, that would need to increase significantly. One could imagine a gradual shifting of the tax burden from individuals to corporations as evidence of accelerating technological unemployment showed up in our employment statistics over time.
A Corporate State
There are real dangers with a big jump in corporate income taxes, of course. What kind of society would we create if corporations funded the bulk of our government spending? When one pays taxes one assumes that one should get a certain amount of say in how those taxes are spent. While far from perfect, we have a democratic system for dealing with those expectations when they’re spread across a few hundred million humans, but it’s unclear how our government would function with that power of the purse concentrated in the hands of just a few thousand corporations.
No Clear Answers
I’m still scratching my head that I’ve actually written this article about taxes. I hate thinking about taxes and don’t pretend to be an expert on them. But I do care about highlighting the consequences of large-scale automation, and I think taxes and government funding are going to be subject to some very big transformation thanks to automation technologies.
Whether we increase corporate income taxes or develop a more targeted method for shifting the tax burden directly onto automation, if technological unemployment becomes a real problem, then our heavy dependence on wage-based taxes guarantees that we are also going to have a problem with funding government. We won’t be able to tax the robots themselves (at least for the foreseeable future). But we will be able to put a robot tax on the owners of the robots.
The question is how.