Jeremy Corbyn has called for a new tax on robots.
I’m delighted to hear a politician engaging with the coming challenges from automation. But I don’t think this is the right idea. Because we already have a tax on robots. It’s called Corporation Tax.
Corporation Tax is a tax on profits. The more you make, the more you pay.
Companies invest in robots primarily because they increase profits. They do jobs more quickly, more cheaply, and more reliably than humans. Hence companies can produce more at lower cost. In theory then, those companies investing heavily in automation ought to be making disproportionately greater profits, and paying more tax.
Unfortunately, it doesn’t always work like that.
Analysis by the House of Commons Library (with the caveat that this was commissioned by Labour) suggests tax avoidance costs the state around £2.2bn per year. This does not include the re-routing of profits internationally into lower (or zero) tax destinations.
If it did, the figure would likely be much, much higher.
The global companies investing the most in automation are also those with the scale and capability to avoid tax most effectively.
A tax on robots sounds good. It plays to our deep-seated fears of being displaced by machines. Fears that go back centuries: the Luddites weren’t against technology, they were against losing their jobs to it.
But this is the problem with so much of modern politics. The answer to everything has to be a new policy, because that sounds good in speeches and looks good in headlines.
Taxing robots also sounds a lot more palatable to half the electorate than cutting down on tax avoidance. Talking about tax avoidance —as Labour has done — leads half the population to think that they will end up paying more tax. Rather fewer people think they will be affected by a robot tax. And it may well be the wrong people.
A brake on progress
Taxing robots specifically targets those companies that are driving innovation. They may well be the same companies investing the most in tax avoidance, but that doesn’t mean that automation is inherently wrong. Automation may cost a lot of jobs, but many of them will be ones people would rather not do. The question is perhaps not whether we should be protecting bad jobs but working out what people are going to do instead.
Every listed company is obliged by its commitment to shareholders to operate to its best potential. Sometimes this means their leaders prioritise short term success over sustainability — a mistake that hurts us all. But over the long term, without a fundamental change in our economic system, we have to accept that the duties of leadership to shareholders means that companies are going invest in automation.
If that switch is inevitable, do we want the companies here to be the last to make that switch, because we make it expensive? Do we want companies in the UK to be the last reap the benefits of automation, or the first?
Software and hardware
Think about the types of robots that are being employed. Some of them are bound by location. These are primarily physical robots: drones for delivery, self-driving cars and trucks, warehouse operations. Still, many of the physical robots in manufacturing and production can easily be moved anywhere offshore — likely close to a port in a country with cheap electricity.
The real challenge is how you tax software robots. These will outnumber the physical robots by an order of magnitude. The digital systems that replace accountants, solicitors, call centre workers, retail staff, administrators. How do you tax those?
You can’t do it explicitly because counting them would be near impossible. The closest analogy today is the investment that large software companies make in counting the licences for their tools used by corporations. This investment is enormous and the technical expertise considerable.
I don’t think it is practical for any state tax collector to tackle this challenge across all of the possible types of software robot, many of which will never be identifiable as having captured part or all of one human role.
The answer is international
What we have to return to then is a tax on a principle, rather than a tax on a particular aspect of business. If you make profits, it is because you have access to the infrastructure and assets of a nation — including its people. There is a cost to that access, and it is corporation tax.
What we might consider is creating bands of corporation tax based on the relative profits a company makes compared to its employment base. Huge profits but few employees might see you pay a higher rate than a company with low profits but many people. This might have the negative drag effect I decried earlier, but set at the right level and introduced progressively it could start to offset the losses from personal tax and employer contributions (PAYE and NI).
None of this is feasible without international co-operation though. As I have said, the companies investing the most in automation are usually those with the scale and scope to move profits around the world to minimise their tax bill. Only greater international co-operation can make this harder and close the tax gap for each nation.
Robots not androids
When we talk about robots, particularly in the context of work, we naturally think of machines that look like us. If they look like us, then they can be counted and taxed like us. This is a fallacy.
The robots that replace many of us at work will look nothing like us. In fact they might not look like anything at all. Just a few million lines of code in a server somewhere in the world. They might never be bought or sold, but created in house from a collection of open-source components, so they can’t even be taxed at point of sale.
We can’t tax robots. Which leads us to the difficult but obvious conclusion: we have to get better at taxing profits.