Epoch-making scale of robotics
by Gary Smith
The headlines read, ‘Las Vegas self-driving bus crashes two hours after introduction of service.’ We later learned the bus was stationary at the time and that the human driver of the second vehicle was issued with a driving misdemeanour ticket.
In the US and Europe, the story behind automation is usually told from the perspective of a threat, whether to road safety, job security, wages, or the accepted way of life.
In Japan, the story is markedly different. Machines are filling a gap created by a shortage of workers. The government’s ‘Japan revitalisation strategy’ highlights a key role for robotics, and proposes a ‘robot Olympics’ to run alongside the official games, scheduled for Tokyo in 2020. The report states that robot technology will improve corporate profitability, thereby ‘helping to raise wages’.
This contrast in perception is fuelled primarily by differing demographic backdrops. Japan is aging and shrinking. The population is around 127m, and 25% are at least 65 years old. In 40 years, the population might be just 100m, with 40% of people 65 years old and over.
In the West, although industries do not need robots to address such an acute demographic problem, automation will become more widespread. Multinational companies are the key factor, and their motivation to increase profits will accelerate the rate of investment in new technologies.
In 1900, 40% of the US workforce were employed in agriculture. By 2000 that figure had fallen to 2%, but food production and overall employment rose significantly. In the 1920s cars replaced horses as the main form of transport in US cities. This was a more rapid disruption, but was still generally recognised as a job-positive story.
The risk today is that the pace of technological advancement is too fast for labour markets to absorb easily. This could have a pronounced impact on employment, wages and populist politics.
Developments in the US oil industry illustrate this effect (see Chart). Between 2015 and mid-2016, the number of rigs halved in response to the oil price decline. A recovery in the oil price and improvement in the efficiency of production – driving down the required break-even oil price – led to a rebound in the rig count to almost early 2015 levels. Oil and gas industry employment fell in line with the decline in the rig count, but has not responded to the increase over the last 12 months. Instead, automated rigs requiring a crew of only five are replacing those that required a crew of 20.
The critical question from a labour market perspective is what happens to these displaced workers. Perhaps, like horse-drawn carriage drivers in 1920s New York, they get better jobs earning more money. But then again, perhaps they take a lower-paid job as a temporary measure. Perhaps that temporary position becomes permanent. Perhaps these workers need to relocate to find a new job, made difficult by the fall in labour mobility in the US economy. For the newly unemployed rig worker, there is no evidence of a formal industry- or government-sponsored retraining programme, only self-help websites.
The days when low-skilled labourers were most at risk from automation are behind us. Studies show that mid-skilled workers are more at risk from disruption. A robot could, undoubtedly, be designed to cut hair, but developing that technology is not worthwhile when it costs only £20 for a haircut. Robots have moved up the value chain. Instead of cutting hair, they are piloting planes, trains and automobiles (although, for the time being, largely under human supervision).
While useful, the rig worker example might be said to overstate the extent of machine substitution for human labour. This is because it does not account for the complementarities between automation and labour that increase productivity, raise earnings, and boost the overall demand for labour through supply and ancillary industries. But concerns remain that the pace of technological advancement is becoming so rapid that the short-term waves created by disruption are making it more difficult to identify the longer-term benefits. Perhaps the positive effects will become clearer, but, in the meantime, people are becoming angry with these far-reaching changes.
In terms of wages, it is difficult to believe that job dislocation, and a period of unemployment, is beneficial for aggregate income growth. Dislocation is more likely to result in wage acceptance, rather than encouraging wage-demanding behaviour. As Claudio Borio of the Bank for International Settlements wrote in last month’s edition of The Bulletin, technological advances threaten labour’s pricing power. Technological change is another factor weighing on wage inflation statistics, and confounding advocates of the Phillips curve (which illustrates the inverse relationship between unemployment and inflation), including those at the US Federal Reserve.
In the longer term, robotics and artificial intelligence might be a more positive story. But, while we wait, societies may have to work harder to deal with short-term disruption. Bill Gates has raised the idea of taxing robots, on the basis that every lost worker is a lost income tax payment. That tax could be used to pay for retraining programmes.
And in the very long term, if human labour is indeed rendered redundant by automation, then the world’s economic problems will become one of wealth distribution rather than wealth scarcity. This would inspire arguments for society-changing measures such as universal basic income. The implications could be epoch-making in scale.
Gary Smith is a Member of the OMFIF Advisory Council and Member of the Strategic Relationship Management Team at Barings.