Are robots stealing our jobs? No, but they’re making them more difficult
14 December 2012
There’s a debate swirling around in the blogosphere about robots, and whether they’re stealing our jobs. Before you get visions from sci-fi movies, what we’re mostly talking about here are digital technologies – like algorithms and database tools – which automate jobs that used to be done by people. In this morning’s FT, Gillian Tett is the latest writer to argue that this digital displacement effect will make it very hard to get unemployment back down to pre-crisis levels.
This is not a new debate; it traces its roots back to the start of the industrial revolution, and the birth of economic growth, when the so-called “Luddites” smashed up the machines that were replacing their jobs. The term Luddite has come to be applied to anyone who argues that new technologies are destroying jobs, including those doing so today. The huge rises in living standards – and growth in employment – that have occurred ever since the first Luddites smashed up their looms means that history does not look too kindly on this idea.
The fundamental issue is this: long-term economic growth is based on each of us becoming more productive as workers, making more valuable outputs for each hour of work we put in. Technologies that reduce the number of people needed to do a job – from the spinning mule to the Google search engine – also make us more productive.
Over time, people displaced from one type of job have found more productive employment in another field. Just as those displaced from the fields by the agricultural revolution moved into industrial cities, many of the 4 million workers displaced from the UK manufacturing sector since the 1970s have moved into the service economy. This type of technological change can cause a lot of pain and disruption in the short term – especially when it is poorly handled, as in many Midland and Northern post-industrial cities – but over time it has proven to make us wealthier and healthier.
The same is true of today’s brand of technological change. It is causing short term pain, but it is the only way out of the economic crisis.
That might be the end of the matter, but unfortunately there’s a hitch. What if the types of technological change don’t fit with the types of workers we have? What if, as Paul Krugman has suggested in various posts, the returns from the new technologies are concentrated in the hands of too few people? And perhaps, as this must-read piece from Izabella Kaminska suggests, the companies that control these new technologies are using the intellectual property regime to extract monopoly rents from them, depriving us of the economic benefits (and jobs) that might flow from them?
These are enormous questions, and I don’t have time to go into all of them here. However, it seems to me that the most important issue here is about skills – whether the new jobs that digital technologies could create are too difficult for most people to do. Economists tend to call this “skill-biased technological change”*, which just means that new technologies favour people with certain skills.
Think about how this works in practice. When people were driven off their land by the agricultural revolution, the key thing they had to do was move into a city to find work. This presented major social and infrastructural challenges, but there was no inherent barrier to people learning to do industrial work. In this digital revolution, the jobs that tend to be created – programmers, data analysts, engineers – are often very highly skilled. Many people require much more training to do these jobs, which might take years to provide, making the technological transition very painful. But it may be that it just isn’t possible for many people to learn to do all of these new jobs. No amount of training and qualifications will turn me into a good computer programmer, for instance. If this is true, the consequences would be disastrous.
In fact, we can already see the consequences of the digital revolution on the labour market. So far, most of the jobs that have been vulnerable to digitisation – administrative work, skilled production work – have been in the mid-skilled range of the labour market. As a result we have seen a hollowing out of that middle. In his excellent report on this “hourglass” phenomenon last year, my colleague Paul Sissons showed how the economy is currently creating lots of very good, highly skilled jobs at the top, and lots of poor, low-skilled jobs at the bottom, but losing them in the middle. This is an extremely worrying trend, and it is probably set to get worse.
Is there anything we can do about this skill-bias phenomenon, without turning to the Luddite cause? Well, first we need to train as many young people as possible in the high-level skills needed by the jobs of tomorrow, particularly programming and data analysis.
But second, we should try to turn more of today’s low skill jobs into mid level jobs, making use of new technologies. Rather than accept that many jobs in retail will be low skill, why not try to make them higher skilled and more productive? For instance, developing roles which involve interpreting data effectively – rather than having to do fancy analysis on it – could benefit both businesses and employees. Making these types of changes will be an enormous challenge, but it may be the only way we can have both economic growth and decent jobs in the future.
*I’d prefer to call it “knowledge-biased technological change”, because this allows you to cover Kaminska’s point about IP monopolies as well. Some knowledge (eg the patents to Apple’s iPad) is codified, and can be controlled through IP owners. Some knowledge is tacit, and belongs to workers – it is roughly the same as skills.