History suggests robots won’t cause mass unemployment: But is this time different?
Martin Ford’s book, The Rise of the Robots: Technology and the Threat of Mass Unemployment, makes for scary reading. Ford argues that any job that is in some way routine will eventually be automated.
In Ford’s new world, intelligent algorithms will eat their way through white collar employment, making legions of occupations obsolete. And this isn’t just low-skilled white collar jobs: Ford asserts that, in the near future, doctors, journalists and computer programmers will be replaced by robots. Perhaps scariest of all, he makes the point that the ensuing mass unemployment will risk the implosion of the capitalist system itself. So is he right or wrong?
Economic history teaches that Ford is far too alarmist. The move from agriculture to manufacturing was a huge shift in employment, but nobody would suggest that it wasn’t a good one. The subsequent shift from manufacturing to services was also painful, but as a society we’re better off as a result.
We shouldn’t forget either that many have cried wolf before, predicting mass unemployment in the wake of employment losses in manufacturing, which proved incorrect. Sure, there were large job losses involved in both shifts, but the overall impact was positive. Today, in the UK, despite the information technology revolution over recent decades (the Third Industrial Revolution), total employment and the employment rate are at record levels. But is this too simplistic or complacent?
Ford’s essential point is that “this time it’s different”. He asserts that “the idea that technology will ever truly replace a large fraction of the human workforce and lead to permanent structural unemployment is, for the majority of economists, almost unthinkable.”
Ford is correct that there is an almost reflexive tendency to dismiss anyone who argues that this time might be different. But that’s for a good reason. Most economists take a dynamic approach, looking at the knock-on effects across the whole economy, as opposed to a static or partial analysis of employment losses in particular sectors or occupations.
This is the fundamental weakness in Ford’s case. He doesn’t nail the argument as to why the total impact will be negative. Instead, he cites example after example of potential automation and employment loss, but ultimately doesn’t show why this is any different from factory closures in the 1980s or the collapse of agricultural employment in the early twentieth century.
Nailing the microeconomic argument (which he does) – that employment losses will move higher up the occupational ladder – is not the same as proving the macroeconomic case for mass unemployment. There will be a lot of jobs designing the robots, writing the software, making the robots and servicing them. The productivity gains delivered by automation will also induce investment and employment creation. Of course, geographically, there could be a serious mismatch between where the jobs are lost and created, but this doesn’t mean that the net effect has to be negative.
Record high total employment in the UK contrasts with a car industry where employment has fallen from 500,000 in the early 1970s to less than 150,000 today. But factories crammed with robots and record high productivity equate to a hugely successful sector.
I remain an optimist on the relationship between technology and employment, but I have to admit, Ford’s book has got under my skin and made me wonder whether this time really is different.