‘In the present chapter, I shall enter into some enquiry respecting the influence of machinery on the interests of the different classes of society, a subject of great importance’ wrote economist David Ricardo in 18171. In the two centuries that followed, the machinery changed – from the weaving machines and horses of Ricardo’s treatise, to the tractor, the ATM and now robots and artificial intelligence. But the question remains the same. What is the impact of automation on jobs?
Given machines are intended to replace human labour, the intuitive response is that automation’s impact on employment will be negative. Yet a large body of research concludes the opposite – overall, automation has a positive effect on labour demand. Yes, technology does replace jobs in specific industries over time. At the turn of the 20th century, there were around 12 million people employed in agriculture in the United States. One hundred years later, that number had plummeted to 2 million. But consistently over past two centuries, more jobs have been created overall than destroyed2.
Why is this? First, new technologies create new jobs types – think web designer or mechatronic. Second, automation creates new industry sectors and these may provide more jobs than are lost in sectors that decline as a result. Ecommerce has created sixteen times more jobs in the UK since 2010 than have been lost in retail, for example3. But the most important reason automation has maintained its track record of net job creation over so long is its impact on supply and demand. As Adam Smith noted, ‘the desire for food is limited in every man, by the narrow capacity of the human stomach, but the desire of the conveniences, and ornaments of building, dress, equipage and household furniture, seems to have no limit or certain boundary’4. No boundary except the ability to pay for the desired goods that is. This is where automation comes in, by making goods cheaper to supply. The cost of a television, for example, fell by 98% in the US between 1950 and 2017 and as a result the number of American households owning a television rose from 9% of the population in 1950 to 95% in 1970. This pattern continues across most goods and services until demand is saturated and moves on to new sources of desire. Meanwhile, increased productivity through automation results in wage increases that provide more income to spend on goods and services.
Is this time different?
Sceptics argue that recent rapid advances in technologies such as artificial intelligence will reverse this long-run trend of net job creation, with many middle- and high-skilled jobs – such as actuaries, paralegals and equity traders - at risk. An increasing body of evidence contradicts this argument, concluding that less than 10% jobs are fully automatable5. But whilst most job categories are here to stay, many will change, and at a faster rate than in the past, leading to a potential gap between the skills employers want and those employees are able to provide. We’re already seeing evidence of this. Recruitment company Manpower found that 40% of the employers it surveyed in 2017 were having difficulty filling roles – the highest level since 20076 and a study by Deloitte and The Manufacturing Institute predicted that 2 million jobs in US manufacturing will go unfilled over the next decade due to lack of skilled workers7.
Closer collaboration between industry and education needed to match skills
Addressing the skills gap will take concerted effort. Most companies are just starting to adopt new automation technologies and find it difficult to predict their future skills requirements. The technologies themselves are still evolving. Close collaboration between companies and higher education institutes will be required to ensure a supply of skills for which there is current and future demand. Most industrial economies have recognised and are responding to this. For example, the U.S., which has over recent years focused on the value of four-year bachelor degrees, is pivoting back towards apprenticeship schemes, many of them partnerships of multiple companies and higher education institutes around the industries relevant to a particular region.
In 1817 Ricardo worried whether machinery would have a negative impact on workers and their wages. Two centuries of technology innovation have not delivered evidence to support this concern. As in the past, the current wave of technological change will alter job profiles. The evidence points to this mostly being in the direction of higher-skilled, higher-paid jobs8. We need more, not less, of the machines, and our focus must be on ensuring current and future workers are equipped to work with them.
1 Introduction to Chapter 31, ‘On Machinery’ from ‘On the Principles of Political Economy and Taxation’, David Ricardo, 1817.
2 For example, this is the conclusion reached by Deloitte economists studying census records on employment in England and Wales for every decade year since 1871 together with Labour Force Survey data, from 1992. Another study on the impact of automation on employment in EU countries shows that automation drove a net increase of over 10 million jobs between 1999 and 2010. See Racing With or Against the Machine? Evidence from Europe. Discussion Paper No. 16-053, ZEW Centre for European Economic Research, 2016. Research by the Asian Development Bank showed that productivity increases due to automation led to a net increase in 33 million jobs per annum between 2005 and 2015 in 12 Asian countries - Asian Development Outlook 2018, Asian Development Bank.
3 And seven times more in the US – see The Age of Automation: Artificial Intelligence, robotics and the future of low-skilled work, RSA, September 2017
4 The Wealth of Nations, Adam Smith, 1776.
5 See A Future That Works: Automation, Employment and Productivity , McKinsey Global Institute. 2017; The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis OECD and; The A.I. Paradox, CISCO and Oxford Economics, December 2017
6 Manpower Group. 2016 Global Talent Shortage Infographic.
7 The skills gap in US Manufacturing 2015 and beyond, Deloitte and Manufacturing Institute, 2016
8 See for example: Explaining Job Polarization: Routine-Biased Technological Change and Offshoring. American Economic Review, Vol, 104, issue 8 pages 2509-26 and OECD Employment Outlook 2017
Picture © YASKAWA