It’s been a while since I last blogged about labour markets and, in particular, about the effect of automation on wages and employment. My most recent post on this topic was on the 14th of April 2018 and it was mostly a reflection on some interesting findings that had been reported by Acemoglu et al (2017). More specifically, Acemoglu and Restrepo (2017) developed a theoretical framework to evaluate the effect of AI on employment and wages. They concluded that the effect was negative and potentially sizeable (for a more detailed discussion see my blog).
Using a model in which robots compete against human labor in the production of different tasks, we show that robots may reduce employment and wages … According to our estimates, one more robot per thousand workers reduces the employment to population ratio by about 0.18–0.34 percentage points and wages by 0.25–0.5 percent.
Since then, I have seen a constant stream of news on my news feed about the development of ever more advanced industrial robots and artificial intelligence. And this was not because of some spooky coincidence (or worse). It has been merely a reflection of the speed at which technology has been progressing in this field.
There are now robots that can run, jump, hold conversations with humans, do gymnastics (and even sweat for it!) and more. It is really impressive how fast change has been happening recently in this field – and, unsurprisingly, it has stimulated the interest of labour economists!
A paper that has recently come to my attention on this subject is by Graetz and Michaels (2018). The authors put together a panel dataset on robot adoption within seventeen countries from 1993 to 2007 and use advanced econometric techniques to evaluate the effect of these technologies on employment and productivity growth. Their analysis focuses exclusively on developed economies (due to data limitations, as they explain) – but their results are nevertheless intriguing:
We study here for the first time the relationship between industrial robots and economic outcomes across much of the developed world. Using a panel of industries in seventeen countries from 1993 to 2007, we find that increased use of industrial robots is associated with increases in labor productivity. We find that the contribution of increased use of robots to productivity growth is substantial and calculate using conservative estimates that it comes to 0.36 percentage points, accounting for 15% of the aggregate economy-wide productivity growth.
The pattern that we document is robust to including various controls for country trends and changes in the composition of labor and other capital inputs. We also find that robot densification is associated with increases in both total factor productivity and wages, and reductions in output prices. We find no significant relationship between the increased use of industrial robots and overall employment, although we find that robots may be reducing the employment of low-skilled workers.
This is very positive news for most – except, of course, for low-skilled workers. Indeed, like Acemoglu and Restrepo (2017) and many others, this study shows that the effect of automation on employment and labour market outcomes is unlikely to be uniform across all types of workers. Low-skilled workers are found again to be likely to lose out and be significantly displaced by these technologies.
And if you are wondering which sectors are likely to be disrupted most/first by automation, the rankings developed by McKinsey and Company (see chart below) would give you an idea of where the disruption is likely to start. Unsurprisingly, the sectors that seem to be the most vulnerable, are the ones that use the highest share of low-skilled labour.
- “The effect of automation on wages and employment is likely to be positive overall”. Discuss.
- Using examples and anecdotal evidence, do you agree with these findings?
- Using Google Scholar, put together a list of 5 recent (i.e. 2015 or later) articles and working papers on labour markets and automation. Compare and discuss their findings.
The IMF has just published its six-monthly World Economic Outlook. This provides an assessment of trends in the global economy and gives forecasts for a range of macroeconomic indicators by country, by groups of countries and for the whole world.
This latest report is upbeat for the short term. Global economic growth is expected to be around 3.9% this year and next. This represents 2.3% this year and 2.5% next for advanced countries and 4.8% this year and 4.9% next for emerging and developing countries. For large advanced countries such rates are above potential economic growth rates of around 1.6% and thus represent a rise in the positive output gap or fall in the negative one.
But while the near future for economic growth seems positive, the IMF is less optimistic beyond that for advanced countries, where growth rates are forecast to decline to 2.2% in 2019, 1.7% in 2020 and 1.5% by 2023. Emerging and developing countries, however, are expected to see growth rates of around 5% being maintained.
For most countries, current favorable growth rates will not last. Policymakers should seize this opportunity to bolster growth, make it more durable, and equip their governments better to counter the next downturn.
By comparison with other countries, the UK’s growth prospects look poor. The IMF forecasts that its growth rate will slow from 1.8% in 2017 to 1.6% in 2018 and 1.5% in 2019, eventually rising to around 1.6% by 2023. The short-term figures are lower than in the USA, France and Germany and reflect ‘the anticipated higher barriers to trade and lower foreign direct investment following Brexit’.
The report sounds some alarm bells for the global economy.
The first is a possible growth in trade barriers as a trade war looms between the USA and China and as Russia faces growing trade sanctions. As Christine Lagarde, managing director of the IMF told an audience in Hong Kong:
Governments need to steer clear of protectionism in all its forms. …Remember: the multilateral trade system has transformed our world over the past generation. It helped reduce by half the proportion of the global population living in extreme poverty. It has reduced the cost of living, and has created millions of new jobs with higher wages. …But that system of rules and shared responsibility is now in danger of being torn apart. This would be an inexcusable, collective policy failure. So let us redouble our efforts to reduce trade barriers and resolve disagreements without using exceptional measures.
The second danger is a growth in world government and private debt levels, which at 225% of global GDP are now higher than before the financial crisis of 2007–9. With Trump’s policies of tax cuts and increased government expenditure, the resulting rise in US government debt levels could see some fiscal tightening ahead, which could act as a brake on the world economy. As Maurice Obstfeld , Economic Counsellor and Director of the Research Department, said at the Press Conference launching the latest World Economic Outlook:
Debts throughout the world are very high, and a lot of debts are denominated in dollars. And if dollar funding costs rise, this could be a strain on countries’ sovereign financial institutions.
In China, there has been a massive rise in corporate debt, which may become unsustainable if the Chinese economy slows. Other countries too have seen a surge in private-sector debt. If optimism is replaced by pessimism, there could be a ‘Minsky moment’, where people start to claw down on debt and banks become less generous in lending. This could lead to another crisis and a global recession. A trigger could be rising interest rates, with people finding it hard to service their debts and so cut down on spending.
The third danger is the slow growth in labour productivity combined with aging populations in developed countries. This acts as a brake on growth. The rise in AI and robotics (see the post Rage against the machine) could help to increase potential growth rates, but this could cost jobs in the short term and the benefits could be very unevenly distributed.
This brings us to a final issue and this is the long-term trend to greater inequality, especially in developed economies. Growth has been skewed to the top end of the income distribution. As the April 2017 WEO reported, “technological advances have contributed the most to the recent rise in inequality, but increased financial globalization – and foreign direct investment in particular – has also played a role.”
And the policy of quantitative easing has also tended to benefit the rich, as its main effect has been to push up asset prices, such as share and house prices. Although this has indirectly stimulated the economy, it has mainly benefited asset owners, many of whom have seen their wealth soar. People further down the income scale have seen little or no growth in their real incomes since the financial crisis.
- Clouds gather over global economy, casting long shadow on Europe
Politico, Pierre Briançon (18/4/18)
- IMF warns rising trade tensions threaten to derail global growth
Reuters, David Lawder (17/4/18)
- IMF outlook contains cause for celebration but a horrendous hangover is looming
The Guardian, Greg Jericho (18/4/18)
- World trade system in danger of being torn apart, warns IMF
The Guardian, Larry Elliott (17/4/18)
- IMF Warns of Rising Threats to Global Financial System
Bloomberg, Andrew Mayeda (18/4/18)
- IMF issues warning on global debt
BBC News, Andrew Walker (18/4/18)
- The IMF has a simple message: the global recovery will peter out
The Guardian, Larry Elliott (17/4/18)
- Global growth is built, alas, on shaky foundations
The Irish Times, Martin Wolf (18/4/18)
- Government debt
The Economist (19/4/18)
- This Is How Much Money the World Owes
- For what reasons may the IMF forecasts turn out to be incorrect?
- Why are emerging and developing countries likely to experience faster rates of economic growth than advanced countries?
- What are meant by a ‘positive output gap’ and a ‘negative output gap’? What are the consequences of each for various macroeconomic indicators?
- Explain what is meant by a ‘Minsky moment’. When are such moments likely to occur? Explain why or why not such a moment is likely to occur in the next two or three years?
- For every debt owed, someone is owed that debt. So does it matter if global public and/or private debts rise? Explain.
- What have been the positive and negative effects of the policy of quantitative easing?
- What are the arguments for and against using tariffs and other forms of trade restrictions as a means of boosting a country’s domestic economy?
I recently found myself talking about my favourite TV shows from my childhood. Smurfs aside, the most popular one for me (and I suppose for many other people from my generation) had to be Knight Rider. It was a story about a crime fighter (David Hasselhoff) and his a heavily modified, artificially intelligent Pontiac Firebird. ‘Kitt’ was a car that could drive itself, engage in thoughtful and articulate conversations, carry out missions and (of course) come up with solutions to complex problems! A car that was very far from what was technologically possible in the 80s – and this was part of its charm.
Today this technology is becoming reality. Google, Tesla and most major automakers are testing self-driving cars with many advanced features like Kitt’s – if not better. They may not fire rockets, but they can drive themselves; they can search the internet; they can answer questions in a language of your choice; and they can be potentially integrated with a number of other technologies (such as car sharing apps) to revolutionise the way we own and use our cars. It will take years until we are able to purchase and use a self-driving car – but it appears very likely that this technology is going to become roadworthy within our lifetime.
Artificial intelligence (AI) is already becoming part of our life. You can buy a robotic vacuum cleaner online for less than a £1000. You can get gadgets like Amazon’s Alexa, that can help you automate your supermarket shopping, for instance. If you are Saudi, you can boast that you are compatriots with a humanoid: Saudi Arabia was the first country to grant citizenship to Sophia, an impressive humanoid and apparently a notorious conversationalist who does not miss an opportunity to address a large audience – and it has done so numerous times already in technology fairs, national congresses – even the UN Assembly! Sophia is the first robot to be honoured with a UN title!
What will be the impact of such technologies on labour markets? If cars can drive themselves, what is going to happen to the taxi drivers? Or the domestic housekeepers – who may find themselves increasingly displaced by cleaning robots. Or warehouse workers who may find themselves displaced by delivery bots (did you know that Alibaba, the Chinese equivalent of eBay, owns a warehouse where most of the work is carried out by robots?). There is no doubt that labour markets are bound to change. But should we (the human labour force) be worried about it? Acemoglu et al (2017) think that we should:
Using a model in which robots compete against human labor in the production of different tasks, we show that robots may reduce employment and wages […] According to our estimates, one more robot per thousand workers reduces the employment to population ratio by about 0.18–0.34 percentage points and wages by 0.25–0.5 per cent.
Automation is likely to affect unskilled workers more than skilled ones, as unskilled jobs are the easiest ones to automate. This could have widespread social implications, as it might widen the divide between the poor (who are more likely to have unskilled jobs) and the affluent (who are more likely to own AI technologies). As mentioned in a recent Boston Consulting Group report (see below):
The future of work is likely to involve large structural changes to the labour market and potentially a net loss of jobs, mostly in routine occupations. An estimated 15 million UK jobs could be at risk of automation, with 63 per cent of all jobs impacted to a medium or large extent.
On the other hand, the adoption of automation is likely to result in higher efficiency, huge productivity gains and less waste. Automation will enable us to use the resources that we have in the most efficient way – and this is bound to result in wealth creation. It will also push human workers away from manual, routine jobs – and it will force them to acquire skills and engage in creative thinking. One thing is for certain: labour markets are changing and they are changing fast!
- What do you think is going to be the effect of automation on labour market participation in the future? Why?
- Using the Solow growth model, explain how automation is likely to affect economic growth and capital returns.
- In the context of the answer you gave to question 2, explain how human capital accumulation may affect the ability of workers to benefit from automation.
 Daron Acemoglu and Pascual Restrepo, Robots and Jobs: Evidence from US Labor Markets, NBER Working Paper No. 23285 (March 2017)