Another day, another report on automation. This time it’s the National Bank of Australia axing 4000 jobs in customer service. Or how about the 66,000 jobs that were lost in retail in the US in 2017 alone. It would seem that automation continues its march through the economy. There’s no reason to think that fund managers or lawyers will be exempt. If anything their high wages make them prime targets. Human wage-labour seems to be becoming irrelevant. This is alarming but what should be done?
Before getting into the responses to automation, what’s puzzling about these reports is that automation is far more apparent in the media than in the productivity statistics. For example, before the Great Recession world productivity growth was around 1% but it’s since slowed to 0.3%.
This is the automation paradox: if it’s so revolutionary, why isn’t it having an impact on productivity? Productivity growth is lower than in the 20th century, despite our far more advanced technology. This is such a big issue that’s it’s now known as the Productivity Puzzle. According to most forecasts, care workers, not computer scientists, are the fastest growing occupation. Why aren’t the fears and potential of automation matching up to reality?
The answer is found all the way back in the Industrial Revolution in 18th century Britain: the first time machinery was applied to widespread production, triggering the modern era. Why there and then? The dominant theory is that Britain had two unique features: high wages and low capital costs. This made profitable the application of labour saving technology that replaced workers. Previously labour costs were so low, and capital so expensive that it was not worth investing in this machinery.
Applying this to today the problem can’t be the cost of capital, which, judging by interest rates, is at near historic lows. The problem must then be too low wages. The Productivity Puzzle is therefore the flip side to the well-documented stagnation of real wages in the West. As long as wages do not rise there will be less incentive to invest in automation.
This is the trap the world economy is stuck in: low or falling wages, subsequent low productivity growth. And what limited automation there is leads to increased unemployment bringing down the wages of those workers who are yet to be automated; making for a check on further automation. Incidentally this was the case for the first 50 years of the Industrial Revolution: a time of falling wages, technological unemployment for some and 12 hour days for others. This is the grim future if we do nothing about automation.
So what are the solutions? There are more or less three proposals.
The first is the so-called ‘Robot Tax’. Essentially the more robots a firm has the higher tax it pays. Perhaps ironically, Bill Gates is a supporter, but actual robot taxes are hard to find. The closest is South Korea’s reducing the subsidies it gives to automated factories. This is probably because firms already try to produce wherever costs are lowest, and any country introducing a robot tax would be providing a disincentive for firms to produce there. If the competition for corporate investment is anything to go by, no government would unilaterally bring in a robot tax, losing themselves jobs and tax revenues.
But even if a robot tax were global it would still be a bad idea. We usually tax things we want less consumption of (like alcohol or tobacco). But if done right we want more not less automation. As the robot tax is a disincentive to automate it would basically be the modern day version of the ‘Luddites’: the weavers whose response to the Industrial Revolution was to smash the machines that had replaced them. That is certainly one way to stop automation but it doesn’t move society forward.
A much more promising response is Universal Basic Income (UBI). Essentially the government gives everybody a lump sum of money. So if you’re “automated out” of a job you can carry on consuming.
By itself such a proposal is neither bad nor good. This is because we already have a form of UBI. If you add up the costs of education, the NHS, JSA, and the state pension, this is UBI (but partially in the form of services). So what’s really important is the level of the UBI. This is why nowadays UBI is divided into its left and right wing versions. A UBI of £1000 per year, subject to whatever conditions the government sets, is very different from a £20,000 guaranteed UBI a year, for example. The latter would free people from the compulsion to work to survive, from having to participate in ever more intense competition for a declining number of jobs with falling wages.
The big downside of UBI is that apart from the boost to consumption and thus demand, it doesn’t really accelerate automation. Productivity growth would carry on at it’s current, too low rate.
This is why the best response to automation is the third and least discussed of the proposals – a reduction in labour time. Imagine if, as a result of automation, the economy becomes, say, 10% more productive, why not reduce by a proportionate 10% the amount of time people have to work? If this sounds crazy, remember that people only got weekends off after the First World War. Before that it was considered normal to work 6 days a week, and for most of the 19th century 7 days a week.
Rather than get into the social, ecological, and health benefits of more free time, a reduction in labour time has one big economic advantage over UBI. Counterintuitively reducing the working week has the effect of making labour scarce and so expensive (i.e. higher wages). On a labour supply and demand diagram this would be like drawing a vertical supply line below the market equilibrium. While going against common sense, whenever this has been done on an aggregate level it has led to either constant or rising incomes.
Making labour very expensive combined with the low cost of capital would lead to massive investment to apply the labour saving technology that has already been invented. Rather than automation being a free for all of unemployment for some and overwork for others, it could be done in a way that boosts wages, gives people more free time, and accelerates productivity growth. This would shift the world economy out of the stagnation it’s currently in. As the economy gets more and more productive we could successively reduce the working week until we get to a society of free time and high wages.
If this sounds utopian, consider that since 1870 European labour productivity has increased by a factor of 15 whilst labour time has declined by just over half – and of course wages have also significantly increased. Keynes mistakenly assumed that a rise in wages and productivity would necessarily lead to a reduction in labour time. He predicted a 15-hour week by 2030. This is likely one way or another: in a negative form of unemployment as a result of automation, or by taking the positive route of labour time reduction.
