No, Robots aren't taking our jobs

The impact of mechanization on human employment has been a long-held concern. Long before robodoctors, drones and self-service checkouts the Luddites waged war with technology, smashing and burning the labour-saving machines they considered a threat to their livelihoods. Today, people like Tyler Cowen predict that the rise of intelligent machines will result in a society where the top 15%  are fantastically successful and wealthy, but much of the traditional work of the lower and middle-classes is performed by robots and automation. Indeed, a much-cited 2013 study by the Oxford Martin Programme on the Impacts of Future Technology found that 47% of total US employment is at ‘high risk’ of computerization and could be automated within the next few decades. ‘Threatened’ sectors include transport, logistics and office administration, but surprisingly also the service sector, which is currently responsible for many of the new jobs created in developed economies.

Technological progress tends to have two differing effects on employment. At first there is displacement, as workers are substituted for new technology. However, efficiencies gained from automation often reduce prices, increasing real income and the demand for other goods. Companies will move into industries where productivity and demand is high and create new jobs, or use new technology to create new industries. Automation also frees up displaced workers to utilize their skills in other, potentially more fulfilling and creative ways.

Setting aside something like the singularity the economic impact of robots depends on whether they destroy more jobs than they create, and which section of society gains most from the opportunities they bring. At the moment, nobody really has a clue what  future economic impact robots will have. Even a survey of nearly 2,000 experts in robotics and AI found that they were split down the middle in terms of techno-optimism (believing that robots will create more jobs than they replace) and techno-pessimism (that the rise of robots will inevitably adversely effect a significant number of blue and white collar workers).

Until now, there’s actually been very little empirical work done on the economic impact of the use of robots.  However, a new paper — ‘Robots at Work' —from the Centre of Economic Performance at the LSE makes a welcome contribution to the field.

Using data from industries in 17 developed countries between 1993-2007, the report finds that ‘robot densification’ has no statistically significant effect on total hours worked over the period. This suggests that the use of robots has not (on net) resulted in less work opportunities for humans.

It has, however, had significant, and positive, effects elsewhere. The study found that the contribution robots make to economic growth is substantial, at 0.37 of GDP growth, and accounted for one-tenth of aggregate growth over the period. They also raised annual labour productivity by 0.36 points, comparable to steam technology’s boost to British labour productivity between 1850-1910. They also found that robot densification increased both total factor productivity and wages.

Industrial robots were used in under a third of the economy during the study period, and accounted for only around 2.25 percent of capital stock even within robot-using industries. The authors suggest that the likely contribution of robots to future growth is substantial, particularly when considering their potential impact in developing countries.

There is one note of warning, though — whilst the study found no overall impact of robot densification on hours worked, the use of robots did have a negative and close to significant impact on the hours worked by low-skilled workers, and, to a very small extent, those of middle-skilled workers. Presumably time will tell whether this is trend truly worthy of concern, and whether displaced workers are able to find alternative jobs elsewhere. For now, though, this study suggests that robots are tools which assist with and complement our jobs, as opposed to threaten them.