We already have robot taxes. They aren’t good policy.
Our accidental robot tax.
Should we tax robots? Past hype about automation leading to mass unemployment has consistently underestimated the ability of entrepreneurs to find new uses for idle labour. If anything, today’s problems can be blamed on too little, not too much automation (the UK and US have sluggish productivity growth and full employment). But many economists believe the future could be different. The AI-enabled ‘robots’ of the future might not complement human workers enabling them to spend more time on more important tasks, but eliminate the need for human workers altogether.
In a recent op-ed for The New York Times, Eduardo Porter echoes Microsoft founder Bill Gates and makes the case for a ‘Robot Tax’. Porter’s argument is subtle. He isn’t necessarily against automation, rather he wants “to level the playing field, to ensure that investments in automation raise productivity.” He has a secondary concern too, automation could erode the tax base. Workers pay payroll taxes (e.g. National Insurance Contributions or Social Security payments), while robots don’t.
Referencing research by Daron Acemoglu, he distinguishes between efficient and inefficient (or premature) automation. To understand the latter concept, imagine if the government handed out massive grants to any company that replaced a worker with a machine. You’d probably see a lot of shops and bars bring in the ‘giant iPads’ and automatic checkouts you see in McDonalds and most supermarkets. This wouldn’t boost productivity by much (at least not more than the cost of the grant) and would leave workers worse off.
A paper by Ryan Abbott and Bret Bogenschneider from the University of Surrey argues that the status quo is somewhat equivalent to the robot subsidy scenario.
Writing for The Conversation they state:
“[E]xisting tax policies encourage automation, even when a person would be more efficient than a machine. That’s because automation allows firms to avoid wage taxes, which fund social benefit programmes such as Medicare, Medicaid, and Social Security in the US, or National Insurance contributions in the UK.
“In the US at least, there is a further incentive to automate because firms can claim accelerated tax deductions for automation equipment, but not human wages. Wage taxes are generally only deductible as paid. This structure allows firms to generate a significant financial benefit from claiming significant tax deductions sooner for robots.”
At first glance, it looks as if they have a point — workers are taxed, while machines aren’t. And I commend them for arguing from the premise that the tax code should be neutral between capital and labour.
But once you submit their claims to greater scrutiny, I don’t think they hold up.
Let’s start with the point about payroll taxes, such as NICs and Social Security. Do payroll taxes actually incentivise firms to lay off workers and replace them with machines? Economic theory and most empirical evidence suggests not. A meta-analysis of 52 studies found that workers bear most of the burden through lower wages and that there are “no statistically or economically significant effects” on employment. A more recent study from Canada found “no impact on employment, productivity and profits, but significant impacts on wages”. Some studies have found negative employment effects but the largest effects come from studies that exploit unusual policy changes and therefore may be less generalisable.
If the incidence is mostly borne by workers and not by firms as the above research suggests. then higher payroll taxes don’t encourage excessive automation.
They also suggest the ability to write-off investments at a rate faster than they depreciate gives capital (robots) an unfair advantage over labour. They offer an example:
“…assume a robot has a total capital cost of $100,000 and seven years of useful life, while an employee has a total wage cost of $100,000 over seven years. If accelerated depreciation for capital is available, the firm may be able to claim a large portion of the $100,000 depreciation as a tax deduction in year one rather than pro-rata over seven years. For instance, the firm might claim tax depreciation for an automated worker of $50,000 in year one, $30,000 in year two, $10,000 in year three, and in diminishing amounts to year seven. By contrast, wage taxes must be deducted as paid (i.e., 1/7th in each year).”
However, there is a flaw in Abbott and Bogenschneider’s comparison. Seven years worth of workers’ wages aren’t paid up front, but capital investments are. This changes the picture rather substantially. Money has a time value. All things being equal, it’s better to have £100 today rather than £100 in 5 years time. This is for two reasons. First, inflation will erode the value of that £100. Assuming a 2% inflation rate, your £100 will only buy £89.14 worth of goods in five years. Second, you could invest that £100 in new productive ventures and earn a return greater than inflation.
If you take the time value of money into account and a company can’t immediately deduct the cost of an investment into a new robot, then a robot would have to earn additional returns on top of the market rate of interest to be preferred to a human worker.
As a result, anything other than allowing firms to write-off capital investments immediately is an implicit robot tax. Though unlike a ‘perfect’ robot tax, it doesn’t just hit robots that replace human workers but also machines that make human workers more productive.
Over the past decade, the UK has tried to reduce the burden on business by cutting corporation tax. However, in order to pay for the cuts in the headline rate, they slowed the rate at which new investments in machinery can be depreciated. By reducing the value of capital allowances, the impact of corporation tax cuts on investment and growth was blunted.
In the last budget, Chancellor Philip Hammond announced that he was increasing the Annual Investment Allowance to £1m. This is the equivalent of abolishing the UK’s implicit robot tax for firms investing under £1m each year. There is a strong case for extending that treatment, transforming Corporation Tax into a Cashflow Tax which allows all investments to be deducted immediately as business expenses. This would create a tax system that would be truly neutral between investments in workers and machines. That would help workers too. A study by Eric Ohrn found that when US states implemented a system of instant depreciation they saw a 17.5% increase in capital investment, which led to 2.5% higher wage growth.
Instead of discussing the case for an ideal robot tax in a hypothetical scenario where automation is causing mass unemployment, we should look at the one we currently have; evidence suggests it depresses investment and is one of the causes of Britain’s sluggish productivity growth.
