THE impact of technology on the economy is one of the most-debated issues of the moment, whether it is the potential for automation to cause unemployment, boost long-term productivity, or widen inequality. A good deal of the annual Barclays Equity-Gilt Study, published yesterday, was devoted to the subject. But one section caught my eye; the idea that technological change was making GDP a less useful measure.
The report says that
When GDP was first introduced, manufacturing accounted for a large share of the core advanced economies, and the (system of national accounts) was designed primarily to measure physical production.
But the modern economy is dominated by services and
Services cover a wide range of activities and are often customised, making their basic unit of production, as well as differences in quality and changes over time, hard to define
Furthermore, the report points out that
Digitised goods or services are often free: and without an observable market price, the (system of national accounts), by definition, excludes them entirely from GDP. But just because the consumption of a digital product does not involve a monetary transaction does not automatically mean that it is of zero value to the consumer. Thus the current treatment of digital products within the (system of national accounts) systematically underestimates the value generated by the digital economy.
This may be true. My question is how new this is. Past technological innovations have enomously boosted human welfare. But were those benefits reflected in a narrow GDP measure?
My great-aunt Amy lived in a one-up, one-down (a house with two floors and just two rooms) in a small Yorkshire town in the 1960s. That meant she had to visit an outside toilet, whatever the weather, or time of day.¹ People today have indoor plumbing. Is that benefit fully reflected in GDP in the form of the cost of toilet installation? It seems unlikely.² Great Aunt Amy also had no fridge so had to trek down cobbled streets to get her shopping every day, whatever the weather. Go back further in time and women like Amy would have had to collect the water for the house, for cooking and washing, and the firewood for heating. A third of them would have died in childbirth and they would have lost a lot of children in infancy to disease. After the 1960s, thanks to the pill and contraception, women had more control over their reproductive rights. And so on.
We are so used to these benefits that we may not fully appreciate them.³ But we would if they were gone. And I think they would be missed more greatly than the ability to check our e-mails, listen to our favourite music, or share details of our lives on Facebook. And smartphones may conceivably be part of the productivity problem because they are so distracting: users say they spend two hours a day on social networks and five hours on their smartphones. Which of us does not spend part of the working day getting sucked into Twitter debates, watching Youtube videos and the like? This is time spent not working. Indeed, there may come a point when employers start monitoring our online activity to crack down on this problem. This loss of freedom and privacy will not be measured as a loss in GDP (it may be a gain) but it will be seen as a loss of welfare.
GDP has long had its critics. It does not measure the unpaid contribution of women in the form of housework, for example. If a mob smashes all the windows in the city centre, GDP goes up when the glaziers replace the glass. We have alternative measures of welfare: longevity (and child mortality) are basic measures, and we can add human height (as an indicator of nutrition), healthy life expectancy (how old are you before infirmity takes over?) and so on. These have been heading in the right direction, dramatically so in some parts of the developing world.
But back to the Barclays study. On the subject of automation and jobs, the report argues that, initially, parts of our jobs get automated, rather than the whole thing. Take long-distance trucking. The introduction of rear-view cameras, automatic braking, cruise control etc. has made the task easier, and thus lower skilled. In nominal terms, the average salary of a trucker has grown from $38,000 in 1980 to just $46,000 today, well below inflation. So automation has increased the pool of workers who can do a given task, and thus depressed real wage growth. Technology also creates new jobs—app developers for iPhones, content moderators on websites and so on.
As for productivity, it may be too early to see the full benefits. Thomas Edison and others pioneered the electricity industry in the 1880s but more than half of US homes did not get electricity until 1925. Factories needed to be redesigned to take advantage of electrification (the old sites relied on a single steam-driven engine and were laid out accordingly). The best decade for productivity growth in the US was the 1950s, an era with few breakthroughs; technologies developed before the war were finally spreading. So it is possible that artificial intelligence, 3-D printing and the like may yet boost the growth rate and allow us to overcome the problem of demography (an ageing population means fewer workers).