A Robot Tax? No. But Let’s Talk about Sharing the Wealth Robots Create.
This first appeared in SIIA’s Digital Discourse Blog.
Bill Gates made headlines recently by advocating a tax on robots.
“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.”
Gates wasn’t clear how this would work. Would the company using the robot pay the tax? The company producing robots? Would the tax be like an income tax based on an imputed income to the robot? And how would “robot” be defined? Is a distributed cognitive computing system like IBM’s Watson a robot because it helps doctors make a diagnosis. Are ATM machines robots because they replace the transactional tasks of bank tellers?
Even asking these questions reveals that the idea has a long way to go before it is ready for prime-time consideration by policy makers.
But we should not try to develop the proposal in any detail because a tax on productivity gains is a terrible idea. We tax the things we want to discourage. Sin taxes are designed to lower consumption of cigarettes and alcohol. President Trump wants a 20 percent import tax because he wants people to buy American
Why do we want to discourage productivity enhancing robots by taxing them?
As the Financial Times put it, there’s no more basis for a robot tax “than there is for taxing the use of Excel spreadsheets, or electric toasters, or any other labour-saving device. It makes no sense to penalise technological innovation that raises productivity and creates wealth.”
Economist Robert Gordon has emphasized that we are in the midst of an extended productivity slowdown. Computers had their focal point of impact in the decade 1994– 2004, when productivity grew at a rate of 1.03 percent per year, little better than half that of 1920– 70 but substantially faster than the rates of 0.57 percent per year for 1970– 1994. But for 2004- 2014 productivity grew at an anemic 0.40 percent per year. In his view, the personal computer and the Internet have had their impact through a one-time-only revolution in business practices in the 1990s.
The New York Times pointed out that robots are not killing the American dream of high paying jobs — we are living through a time of decelerating productivity, not robot-driven unemployment.
In the absence of dramatic AI-driven productivity gains, we are likely to stay at the sluggish rate that has prevailed in the last decade. That’s why former Chairman of the Council of Economic Advisors Jason Furman said last year that his greatest worry about AI is that “we do not have enough of it” to pull us out of our productivity doldrums.
Policymakers should seek ways to encourage productivity growth, not ways to slow it still further.
Nevertheless, we need to share the bounty created by productivity increases more equitably. In the years after World War II, wages increased along with productivity gains, but around 1970 wages and productivity diverged — productivity continued to grow, sometimes faster, sometimes slower, but wages stalled. After thirty years of this wage lag, labor’s share of the national income declined from its 1947–2000 average of 64.3 to 57.8 in 2010.
We need to share the wealth more equitably. One way to share the wealth is to make sure that people have the right skills to help create it, and that their wages are commensurate with the contribution they make to improved output. We need training and retraining programs designed for a 21St Century economy, and substantial government support through a fully-funding reauthorization of the Career and Technical Education Act.
But some will not make a transition to the new skills needed and for them a social safety net is needed. A thoughtful extension of our current income support systems is needed to accommodate those caught in the transition.
We might be at the cusp of a new AI-driven industrial revolution, but as Tyler Cowen says, that is not exactly comforting. Think of Blake’s “Dark Satanic Mills,” Dickens depictions of industrial and urban squalor and the popularity of the revolutionary philosophy of Karl Marx and Friedrich Engels to remind yourself of the extent of the dislocation and political unrest created by the first industrial revolution in the 1800s. Surely, we can do better this time.
It makes sense to assess long term approaches, as Finland is doing with their experiment in universal basic income in which “2,000 unemployed people between the ages of 25 and 58 will receive a guaranteed sum — a “basic income” — of €560 a month for two years.”
Benoit Hamon, the Socialist Party candidate for the French Presidency, is also promoting a universal basic income of €750 ($803) a month for all citizens over 18. While Hamon describes his funding mechanism as a “tax on robots” it turns out to be an extension of the “social charges” that employers pay for workers to all “value added.” As the Financial Times notes this idea is “well worth considering, given the effect high employment taxes have had on France’s labour market.”
Gates had the right idea to use the extra money from his proposed robot tax to retrain workers robots replace, and to fund social safety net programs that support displaced workers. But he got the funding mechanism wrong.
The best way to pay for these transition programs is through a revised tax system that spreads the burden fairly. How that should be designed is a thorny problem. But picking out just the source of our increased wealth to pay for the transition is self-defeating.