Eakinomics: The Robots Are Coming…But Will They Make Friends?

The idea that advances in robotics, machine learning, and artificial intelligence will produce massive job displacement is currently all the rage, fed by studies like Frey and Osborne (2013) which concluded that 47 percent of U.S. jobs were at risk. Perhaps not. A new report out by the Organisation for Economic Co-operation and Development (OECD) estimates that only 14 percent of jobs in OECD countries are highly “automatable.” The key insight is that most jobs will be difficult to automate as they require the ability to engage in complex social relationships, do complex and creative problem-solving (even in manual jobs), or the ability to deal with unstructured work environments.

Nevertheless, 14 percent is a lot of jobs, and the impacts will be uneven across the economy. As the authors put it: “Automation is found to mainly affect jobs in the manufacturing industry and agriculture, although a number of service sectors, such as postal and courier services, land transport and food services are also found to be highly automatable. The occupations with the highest estimated automatability typically only require basic to low level of education. At the other end of the spectrum, the least automatable occupations almost all require professional training and/or tertiary education.” Also: “A striking novel finding is that the risk of automation is the highest among teenage jobs. The relationship between automation and age is U-shaped, but the peak in automatability among youth jobs is far more pronounced than the peak among senior workers. In this sense, automation is much more likely to result in youth unemployment, than in early retirements.”

The upshot is that the young, inexperienced, and low-skilled face the most risks. And it raises the question: what can be done about it? Certainly, it seems like a bad idea to create Robot Adjustment Assistance or some other similarly named, targeted effort. After all, the issue is that (a) a type of job has disappeared in the economy, and (b) some of the holders of those jobs are having difficulty making the transition to a new job. The poster child for getting this kind of policy wrong is Trade Adjustment Assistance (TAA), which is supposed to assist those displaced by international trade. It is virtually impossible to identify exactly which jobs are created, lost, or otherwise affected by trade deals, and the program has a poor record of helping those who participate. TAA-like efforts look like a dead end.

Unfortunately, the more general efforts at training and job placement are not much better. The policy challenge is not that the robots are coming. It will evidently take them a while to learn to make friends, get along, and work on the right problems. The policy challenge is that the United States has not been effective at job-retraining and placement in general.