Productivity paradox

Despite automation, productivity growth has slowed in all the advanced economies. Why?

Computers are “everywhere but in the productivity statistics,” wrote Nobel laureate economist Robert Solow in 1987. What we call the “productivity paradox” has since become more and more apparent. Many jobs have been taken over by automation. And although robots and artificial intelligence continue to promise ever more radical change, productivity has been slowing in all advanced economies, writes Adair Turner, former chairman of the UK’s Financial Services Authority, in a blog on the Institute for New Economic Thinking site.

As Turner points out, many explanations have been suggested for the slowdown: low business investment, poor worker skills, deteriorating infrastructure, excessive regulation; some even question whether information technology is as powerful as it is said to be. But perhaps is it time to rethink our notion of productivity, he says.

Our view of productivity is inherited from the time when we moved from agriculture to industry, says Turner. “We start with 100 farmers who produce 100 units of food,” he writes. Then because of progress, 50 farmers can produce the same amount. The other 50 move to plants where they manufacture cars and washing machines. Using this model, productivity doubles.

But what if the more productive farmers don’t want to buy cars or washing machines and, instead, hire the other 50 as domestic servants and artists? Then, even if agricultural output never slows, the productivity growth rate would tend toward zero.

Our economy is full of “zero-sum” activities: legal services, financial regulators, cybercriminals and those who fight them, advertising agents, hairdressers, and so on. Increasingly, our economies are divided between automated, high-productivity-growth activities and low-productivity, low-wage jobs.

As a result, “measured GDP and gains in human welfare eventually may become entirely divorced,” writes Turner. In 2100, when solar-powered robots will manufacture just about everything, all their activity will probably account for a trivial proportion of GDP, because it is so inexpensive. On the other hand, all measured GDP will focus on zero-sum or impossible-to-automate activities: housing rent, sports prizes, artistic events. Productivity will probably shrink almost to zero, but at the same time become almost irrelevant to human welfare.

Turner concludes, “Computers are not in the productivity statistics precisely because they are so powerful.”