Whilst there are fears surrounding workplace automation and the displacement of labour/jobs in advanced economies particularly, especially those like the UK, US and Japan, these technologies, if managed correctly, could help reverse the slump in productivity growth seen since the global financial crisis. Moreover, future job losses are likely to be offset by new jobs created by larger and wealthier economies supported by new technology as it is estimated that although 75 million jobs could be displaced, over 133 million new ones could be created, resulting a new net job creation of 58 million (World Economic Forum) and forms of automation have the potential to contribute up to $15 trillion to global GDP by 2030 (PwC 2018). However, whilst the past two centuries of automation haven’t eradicated human labour, artificial intelligence raises the issue of replacing labour on a scale not previously seen and we should bear this in mind when embracing these new forms of technology.

According to PwC (2018) there are estimated to be three waves of automation of three different kinds, occurring from the present day to the mid-2030s. Wave One will consist of algorithmic automation, which is the automation of simple computational tasks and analysis of structural data we can see already underway in many sectors and industries. Wave Two, occurring in the late 2020s, will consist of involve augmentation automation which is the automation of repeatable tasks such as filling in forms and so, will likely impact on clerical roles. The final wave, Wave Three will consist of the automation of physical labour and problem solving in ‘real-life’ situations which require a response. Whilst this is already in development, it is likely to appear in full maturity and on an economy-wide scale in the 2030s. Each wave is likely to have a different impact on the displacement of jobs; PwC estimates for Wave One a relatively low displacement of existing jobs, around 3% by the early 2020s ranging up to 30% by the mid-2030s.

Sector Impact:

Automation will vary significantly by industry sector; in the near future bigger losses may be seen in sectors like financial services whereas in the long-term bigger hits may be seen in the transport industry, where it is estimated 52% of jobs could be automated.

Over time the potential impact over the different sectors will vary; data driven services like the financial sector will be hit harder in the short term whereas in the long-run, the transportation and storage industries will face greater losses, with the invention of things like driverless cars.

As seen above, highly educated workers face a lower risk than the lower educated portion of the workforce, reflecting their greater adaptability to technological changes and the fact that they are more likely to be in senior, managerial roles that will require human judgment as well as intellectual reasoning to be applied. Males face higher risk than female as they are more like to be employed in manual tasks, which can be substituted by mechanisation robotics, however women could be more susceptible during the first and second waves due to their higher representation in clerical work. Automation seems prevalent for all age groups however younger workers are more likely to be better educated and more adaptable to deal with new digital technologies, which could indicate at their ability to flourish alongside them (PwC 2018).

Geographical Variation:

It is likely more industrialised economies and countries will have higher rates of automation as they will have a higher proportion of their labour force employed in jobs with manual or routine work that is typically more susceptible to automation (PwC 2018). Thus, countries like Italy and Germany are likely to be at a potential higher risk of automation. Whereas, in service based economies like the US and South Korea, there is a greater concentration on service sectors indicating a lower risk of automation as these seem to be less automatable than industrial sectors due to skills like inter-personal skills which AI cannot presently replicate. Countries with a higher proportion of employees situated in jobs with high educational requirements, like Singapore, are estimated to have lower potential automation rates.

Lastly, workplaces which have already embraced automation to an extent may be at a lower future risk as there is a negative correlation between potential jobs at high risk of automation against the density of industrial robots in the country, which might explain why Asian economies, which are heavily industrialised, are estimated to face smaller losses (PwC 2018).

Possible Benefits from Automation:

In the past, in the face of technological development, the work force have protested to the loss of their labour, the Luddite protest movement brought by weavers against the mechanisation of the textile industry being a notable example. However, with hind sight, the productivity gains from technological advances and mechanisation created huge new wealth which in turn, generated more jobs across the UK economy. From a macroeconomic level, automation induced job losses are likely to be largely offset by job gains arising from new technologies like artificial intelligence (AI) and robotics, thus boosting wealth and income and automation could raise productivity growth by as much as 0.8% to 1.4% annually (McKinsey). As these additional incomes are spent on goods and services, this will, in turn, generate demands for labour by growing the economy (Forbes, 2018). PwC, as well as Forbes, anticipates that they will be concentrated in non-tradable service sectors like health and education as a richer and older society is likely to demand more of these (PwC, 2018, Forbes, 2017). As robots and AI increasingly absorb more jobs, costs are predicted to decline in areas like transport and people should be able to afford better products, while at the same time, improvements could be made to infrastructure (Blackrock).

Focusing on what we will lose with automation misses a central economic mechanism by which automation affects the demand for labour, thereby raising the value of the tasks which workers uniquely supply. When automation makes steps in a process faster, or cheaper, this increases the value of the remaining human links in the production chain.

ATM Example:

ATMs were introduced in the US in the 1970s and their numbers quadrupled from 1995 to 2001 from 100,000 to 400,000. Yet bank teller employment rose from 500,000 to 550,000 during the same period. ATMs indirectly increased demand for tellers, although the number of tellers per branch decreased the number of branches increased by 40% between 1988 and 2004 as ATMs had enabled the growth of the banking industry. Secondly, as routine cash handling roles receded, this enabled a broader range of banking personnel to become involved in more meaningful relationship banking (Autor 2015).

For a more modern example, according to Forbes, Amazon has employed over 100,000 robots yet current figures suggest they are still increasing their human workforce; by enabling workers who previously stacked the bins to take courses within Amazon to become ‘bot operators’, they have ensured minimal labour displacement for their previously lowly skilled employees by training them into more highly skilled workers.

The diagnosis of health issues could also be effectively automated. An emergency room could combine triage and diagnosis leaving doctors more time to focus on unusual cases while improving accuracy for most common issues.

There are tasks however which computer programs cannot replicate as illustrated above. These are often ones which demand flexibility, judgment, creativity and common-sense as well as inter-personal skills. This highlights the importance of skills enhancing through education and training in these sectors (JP Morgan, 2018).

Furthermore, a key challenge facing many developed markets is a decline in their labour markets due to ageing populations and reduced fertility. Automation has the potential to narrow these growth differences, offsetting shrinkage in the labour force (JP Morgan 2018) as it is estimated that in ageing economies there will likely be a deficit of human labour rather than a surplus highlighting the need for automation; this productivity growth enabled by automation will ensure there is continued prosperity within nations where there are higher proportions of elderly in comparison to the working age population, alongside other measures.

Keynes- The Economic Possibilities for our Grandchildren (1930):

Interestingly, looking back, John Maynard Keynes believed that increases in technical efficiency would lead to much less human input and effort needed to achieve the same, if not better outcomes, which indeed we have witnessed. He believed this would lead to technological unemployment due to the economising the use of labour outrunning the pace at which we can find new uses for labour. This newfound lack of economic necessity, he thought, would lead to increased leisure time with citizens working a 15 hour working week or three hour shifts (Economic Possibilities for our Grandchildren, 1930).

However, it is obvious that this has not occurred to the extent he predicted and we think unlikely to occur even with these new advances in AI and robotics. As highlighted, for highly educated workers advancements could lead them more time not for leisure but to concentrate on more high value tasks. While for lower educated workers, Keynes prediction may hold true as their roles may be minimised or even eradicated which is why the government needs to act and provide further training and education for this class of workers.

Action Needed:

However, despite the aforementioned benefits, this does not mean there are not still major issues which will need to be addressed by governments and policy advisors in order to ensure these new technologies are managed in the most effective and beneficial manner. Additionally, although the scale of shifts that technology has the potential to unleash is on a similar scale to those shifts which occurred in developed economies agricultural sectors in the 20th century (McKinsey), which notably did not result in mass long-term unemployment, it is important to note that a crucial difference separating technology advances of the past from those of today is that previously machines were replacing muscle whereas new smart machines have the function to perform mental tasks, so we must bear this in mind when using historical evidence in predicting what the future holds.

For example, governments should realise less educated workers will bear the costs of automation thus they should invest in education and training to increase skills and adaptability as to ensure this group share in the gains from new technology. However, this additional expenditure in education and skills will only be fully effective if there are jobs in plentiful supply; this will require the economy running at a sufficient level of aggregate demand to maintain high employment levels (PwC, 2018). The government can help by investing in more areas like infrastructure and housing that are more beneficial to the long-term productivity of the economy but also make jobs that are impossible to fully automate (PwC, 2018).

While, JP Morgan (2018) predicts the historical pattern of little or no displacement will hold at least over a 10-15 year period as the labour force adapts, they additionally stress, as pointed out above, this depends on human skills keeping pace with technological advancement. Past technological advances have redeployed labour into other functions avoiding mass labour displacement, however the current wave has so-far been skills based and therefore it has enhanced the productivity of highly skilled workers at the expense of those who are lowly skilled- undercutting their prospects. They also note that while governments will play a role in providing this additional education, there is scope for innovation around tax structures amongst other things to push companies towards providing this, given the fact that pushing these technological reforms, ultimately optimises processes which combine human and machine labour, as Amazon has done.

Sources:

Blackrock: ‘Megatrends: A research study looking at structural shifts in the global economy and how they affect our investment thinking’,

D Autor: ‘Why Are there Still So Many Jobs? The History and Future of Workplace Automation’, 2015

Forbes: ‘Potential Winners in an Automation-Driven Economy: Production Workers, Middle Managers’, 2017

John Maynard Keynes: The Economic Possibilities for our Grandchildren’, 1930

JP Morgan: ‘Long-Term Capital Market Assumptions’, 22nd edn, 2018

McKinsey: ‘A Future that Works: Automation, Employment and Productivity’, 2017

McKinsey Quarterly: ‘Four Fundamentals of Workplace Automation’, 2015

McKibsey Quarterly: ‘The Second Economy’, 2011

PwC: ‘Will Robots Really Steal our Jobs? An international Analysis of the Potential Long Term Impact of Automation’, 2018

World Economic Forum: ‘Future of Our Jobs 2018’, 2018