Automation, Leisure And Unemployment

In my last post I demonstrated that if mainstream economic theory were applied in its purest form, as productivity increases to the point that we can produce everything that can possibly be consumed without needing everyone to work, those who are no longer needed to produce goods and services would be left either dependent on welfare or with no means of subsistence.

On the one hand, this was a thought experiment to show the theoretical limits of orthodox theory. But in fact, it is also a fast-approaching reality. In April 2018 Adair Turner presented a paper to the School of Advanced International Studies at John Hopkins University addressing the economic implications of this situation: “Capitalism in the age of robots: work, income and wealth in the 21st-century”. This post will explore the ideas he presented. It’s worth noting that Adair Turner is not some radical, left-wing idealist. He’s actually Lord Adair Turner, and is so mainstream that in May 2008 he was appointed Chair of the Financial Services Authority (FSA). The FSA was responsible for the regulation of the financial sector in the UK, so there were many questions to be asked of it following the global financial crisis of 2007/08. Taking over as chair when the crisis was already in full swing gave Turner a unique vantage point from which to witness the failings of modern capitalism.

(I also need to apologise that it’s been 7 weeks since my last post – life circumstances have prevented me from focusing on writing. After today, there may be as few as 5 posts to get to the end of this section on Distribution, and I’m going to focus on finishing these as quickly as I can . By that stage most of the economic arguments of the blog will be in place, and the final section will draw these together and ask how society can advance in the necessary direction.)

The first point to note from Turner’s paper is that the first section is titled “When, not if”. He discusses some of the features of the current technological revolution that distinguish it from earlier examples (such as the Industrial Revolution) and make it feasible to foretell the automation of almost everything. Notable among these is the development of artificial intelligence, specifically machine learning, which make it possible for computers to improve on the processes of production they were originally designed to control. Not only will computer-controlled machinery make our stuff, but computers will design the machinery making our stuff, control the machines that make those machines, and continually improve these processes of production and the design of the machinery.

While the point where we can automate almost every activity that today provides an income may be “far away”, Turner suggests that this means 50 to 100 years rather than, say, 300 to 500 years. And we don’t need to travel very much further down this road for the impact on the economy and the fabric of society to be cataclysmic (as I will demonstrate with an example next week).

Next, Turner explores what is known as the “Solow paradox” – the famous economist Robert Solow once commented that “you can see the computer age everywhere but in the productivity statistics”. The point being, despite the technology revolution of the last 40 years, productivity growth in developed economies has been much slower than in the previous 50 years.

Part of the explanation probably lies in the failure of methods of measuring GDP to respond to the changing economy, so some productivity growth simply isn’t being captured in statistics. Another factor is the increase in cost of a range of job roles and sectors that do not lead to any increase in productivity, such as lawyers, accountants, financial traders, political lobbyists and security services. This happens particularly as the individuals and corporate entities earning the lion’s share of the wealth seek to safeguard it.

But by far the most interesting factor is the increase in low productivity, low skill jobs, that are not amenable to technology. Turner explains this with a very clear thought experiment. Suppose a village of 100 farmers is able to feed itself through subsistence agriculture, but then works out how to produce food more productively, such that only 50 of them need to farm the land. They agree that the other 50 will manufacture goods so that everyone can be materially better off. The productivity of the village doubles, with everyone able to consume the same amount of food as before and also purchase the manufactured goods (for this economy to remain in equilibrium, the value of the manufactured goods would need to be the same as the value of the food, hence GDP doubles).

But then suppose that those remaining as farmers decide that they don’t want to buy these manufactured goods, they would rather employ the manufacturers as labourers extending their homes, and as servants and cooks maintaining those homes and caring for their everyday needs. Faced with starvation, the manufacturers have no choice but to accept these jobs. There is no increase in productivity or GDP and the quality of life for half the population has become significantly worse: the farmers have an increased standard of living with better housing and servants, while those servants simply have the same amount food that they had before.

This is a very good analogy for the global economy today, and indeed for the global economy since the beginning of the European empires. It’s worth reading it again to see the clear logic of how increases in productivity can end up benefiting one section of the population while leaving others worse off.

The tendency of increases in productivity to lead to increases in employment in less productive areas was first described by William Baumol in his seminal 1967 paper, “Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis”, and is known as Baumol’s Cost Disease.

To take an example in the modern economy, the proliferation of delivery drivers – be it “Deliveroo”, self-employed Amazon delivery people, or even Uber drivers (deliverers of people) – is an example of this dynamic at work. A lack of jobs in other sectors forces people into this unproductive, low-paid manual labour. Technically self-employed, they are actually beholden to a single monopolistic source of employment. Amazon’s role as employer in all but legal status is obvious, while Uber they hide behind their role as the owner of the online platform matching travellers to drivers, shielding them from all the costs of an employer but enjoying all the benefits. The ’employees’ meanwhile have no job security or employment rights. (Even these jobs will soon be automated away, with apocalyptic impact, as we shall see next week.)

This is a very brief impression (I can’t even call it a summary) of Turner’s discussion, and I would very much recommend reading his own exposition, which is extremely clear, even if you don’t read the whole paper.

Throughout this discussion, Turner highlights how distribution of income is one of the key drivers of this dynamic. The more we allow increases in income arising from increases in productivity to flow primarily to just the richest in society, the more we will see an increase in demand for low productivity, low skill jobs, and an ever-increasing supply of labour so desperate for an income that they are willing to accept them.

In the rest of the paper, Turner extrapolates the impact on the modern economy of increased automation. His conclusion is that in a world in which the production of all consumer goods is automated, the good that will become most in demand will be the only one that we cannot increase supply of through productivity: land. Those amassing wealth will be seeking the most desirable property locations, and then the most desirable holiday locations. The neighbourhoods and holiday destinations of the rich will become unattainable luxury ghettos, while those who have been reduced to the low productivity, manual labour are likely to be concentrated in over-populated urban centres.

In drawing out these conclusions, he presents the arguments against these from mainstream economics and his own rebuttals. This is too much for me to summarise in this post. Rather, I want to focus on the importance of the distribution of income, highlighted above. Turner also highlights how the transformation of the economy through increased automation will change the path to industrialisation for developing nations. He suggests that those countries who have not yet reached the middle income level will be trapped in the international equivalent of low income, unproductive labour. Again, this would take too long to summarise, but it is interesting to note that development economists have reached the same conclusion for different reasons. For example, Erik Reinert wrote more than 10 years ago (p39):

“Poor countries tend to specialize in the economic activities which rich countries can no longer mechanize or innovate further, and are then typically criticized for not innovating enough.”

Despite the brevity of this introduction to Turner’s work, it suffices to draw into focus the core concept that contributes to the ideas unfolding in this section of the blog. Increases in productivity will not lead to wealth and leisure for all. Rather, it will concentrate wealth in the hands of a few, who will both live and take holidays in luxurious ghettos, while the rest of us compete in decaying urban conditions for low-skilled, unproductive work.

Turner presents his own policy recommendations, which to my mind are sensible but don’t go far enough. We need to look much more fundamentally at the economy as a system of wealth distribution, and how highly unequal distributions actually destabilise that system. But before we get into these details, next week I want to look at one tangible example of this advance in automation that could have a shattering impact on the fabric of society within the next decade.

Leave a Reply

Your email address will not be published. Required fields are marked *