Robots threaten Asian jobs
This article was first published in Nikkei Asian Review on 6 April 2018.
Cai Fang did not mince his words. Last month the vice president of the Chinese Academy of Social Sciences called for drastic measures to slow the spread of automation and ward off a possible new era of technologically driven unemployment.
“We must cap the speed and restrict the direction of robotics’ development to avoid any bad effects on human beings,” he said.
Fang, who is also a standing committee member of China’s National People’s Congress, joined a growing line of influential figures demanding protection from machines, including Bill Gates, who last year floated a “robot tax” to retrain displaced workers. Yet while such protectionist calls are often alarmist in the industrialised world, there are good reasons to be worried about the rapid spread automation across emerging Asia, especially given its likely effects on employment in the region’s crucial manufacturing sectors.
To understand why, consider a new research report from the Paris-based Organization for Economic Cooperation and Development think tank. Released in March, it suggested that 14% of jobs across 32 industrial economies were “highly automatable,” ranging from mechanical robots replacing cleaners to artificial intelligence software taking over from translators and data analysts. That sounds scary, but the 14% figure is in fact much lower than in many other comparable studies, including a famous 2013 paper by Oxford economists Carl Frey and Michael Osborne, which examined around seven hundred jobs and concluded that 47% in the U.S. were “at risk” over the coming decades.
Rather than examining jobs alone, the OECD’s researchers also looked at particular tasks within jobs, trying to find those involving the kind of creative and social intelligence that is least susceptible to automation. Lots of seemingly routine jobs actually did include such tasks, they discovered, from dealing with customers to negotiating with colleagues. In turn, this suggested both that fears of automation hollowing out highly-skilled white collar professions may have been overdone, and that some lower skilled jobs are likely to be resistant as well.
These new findings ought to be reassuring for western employees but they remain disconcerting for workers in “factory Asia,” given the study also identified low-skill industrial roles as among the most susceptible to automation. “What they have in common is that large part of their job content is interacting with machines,” the authors suggest, pointing to everything from plant operators to metal workers.
This represents an obvious threat to employment in sectors that make up the backbone of export-focussed economies like China, Malaysia or Thailand. But it is even more worrisome for countries in an earlier stage of development, such as Cambodia and Myanmar, and perhaps most notably India — all of whom are struggling to build large-scale, labor-intensive manufacturing sectors.
Put another way, automation does not merely threaten to replace existing Asian manufacturing jobs. It also means that companies in low-income nations may decide never to create these kind of jobs in the first place, investing in robots instead. The effect could be profound, removing the first rung on a economic ladder that has long been climbed by workers leaving farms and entering factories across Asia, in turn helping their economies move from poverty and into middle income status.
Potential automation does not mean it will actually happen. Robotics studies tend merely to identify “at risk” jobs. Companies might decide to automate slowly, not least given the expense and complexity of bringing cutting-edge technologies into emerging markets. For all the hype, automation is also still at an early stage in most emerging Asian countries. China had 631 industrial robots per 10,000 employees, just a third of the level in America and a tenth of South Korea, according to the International Federation of Robotics.
Even so, there are signs the robot’s march is accelerating. Taiwanese iPhone maker Foxconn, the trading name of Hon Hai Precision Industry, has begun a drive to replace tens of thousands of factory workers with machines, disproving the idea that fiddly work on smartphone assembly lines would prove hard to mechanise. Innolux, a Foxconn subsidiary making LCD panels, announced plans in February to automate one in five of its jobs. There are then particular worries about industries like garmenting, where new equipment — including so-called “sew-bots” — may prove to be especially threatening to textile workers in countries like Bangladesh and Vietnam
Governments worried about this face few good options. Some degree of automation is inevitable. What matters is managing its advance to balance the benefits it provides to companies with the costs it imposes on workers, in particular by providing help to those most severely affected.
As Fang said, speaking to the South China Morning Post: “We have to figure out in what aspects human beings are better than robots — emotional quotient, judgment or creative thinking? Then we can make a plan to compensate the weakest group of people — workers replaced by robots and those who will never get another job.”
Flexible labor market regulations are important here, especially in countries like India where notoriously fiddly employment rules encourage companies to install machines rather than hire workers. New programmes to retrain displaced workers are needed too, as well as steps to push companies to change the nature of manufacturing jobs, allowing even routine workers to undertake tasks that involve even some kind of creative or social intelligence.
For now, few governments are likely to consider the more drastic steps suggested by Fang and Gates, not least because they fear that such draconian restrictions will encourage manufacturers to move production facilities abroad. But if Asia’s governments fail to help their workers adapt, and the onward march of automation accelerates, calls for more radical anti-robot measures will only grow louder.