Adair Turner: the advance of robotics changes economics

Expounding my views on economics can be a lonely business. Though they are based on nothing more than conventional economic theory, few commentators even acknowledge the line of thought. Not long ago the FT columnist Martin Wolf wrote an article on a favourite topic of mine, the productivity “puzzle”, which went through a series of potential explanations without mentioning the Baumol effect at all. I pointed this out in the comments, but this was doubtless futile. Then the cavalry came. Adair Turner gave a lecture in Washington DC last April, which only recently seems to have been published, and which picked up many of the themes that I have been banging on about, not least about productivity and Baumol. This time Mr Wolf had to take notice, and he published a column on it without venturing to disagree. Perhaps the view will start to break out into the mainstream.

The lecture was entitled Capitalism in the age of robots: work, income and wealth in the 21st Century. It goes much further than I have, explains the logic more rigorously, and follows through more thoroughly on important implications. Though I have long known that Mr Turner has being saying similar things, I feel vindicated.

He starts with the possibility that robotics could replace pretty much all of what we now consider to be work, and asks what the implications of this are. This is a clever angle, since the rapid advance of robotics and artificial intelligence, following the development of machine learning, is giving a lot of people pause. In reality the phenomena he describes first become important more than 40 years ago, when such ideas on robotics were just science fiction (notably Arthur C Clarke’s 2001: a Space Odyssey).  This was when what I have called the Age of Light Industry started to run out of steam, even as it has continued to dominate the way people thought about economics. It’s worth describing the five sections of the lecture.

When, not if

In this Mr Turner makes the case that the complete automation of work is really just a matter of time. This is a popular view amongst techie types, though I am sceptical. The advance of intelligent machines runs in spurts of incredible speed, between periods of very little progress. In 2001 Clarke expected what is sometimes called “general AI” to be developed by, well, 2001. It remains a distant dream. The trouble is that developers persist in thinking that the human brain is analogous to a computer, programmed by an intelligent mind. That’s a bit ironic, since they are mostly ardent atheists, but they haven’t grasped the difference between evolution and design. The advance of machine learning only came when that understanding was modified; but machine learning has limits. It may take a long time before technologists make the next breakthrough.

But any disagreement I have with Mr Turner is of little importance because I with him agree about three things. Firstly that robotics will transform the world of work in the next generation and more. Secondly that it will affect some areas of work more, or more quickly, than others. And thirdly even if robots are able to do things, we may not want them to.

Explaining the Solow paradox

The Solow paradox arises from the great economist Robert Solow’s comment that “you can see the computer age everywhere but the productivity statistics”. That was in 1987, but it’s even truer now – it is the productivity puzzle. Why is it that, with the pace technology development as fast as ever, increases in average productivity are slowing down? Mr Turner points to three things.

The first is my old friend the Baumol effect, from a paper by William Baumol in 1967. Workers released from areas of high productivity tend to move into jobs with lower productivity, or a lower rate of productivity improvement, which will neutralise the effect of the original productivity advance on average statistics. So if a farmer doubles productivity, he might sack half his farm workers and employ domestic servants instead. This is a well-trodden trail for readers of this blog, but Mr Turner explains it in more detail than I have ever attempted. It is clearly true in the area of robotics which of itself it creates few jobs.

The second thing is something that I have only hinted at, and which I find interesting. As we get wealthier, we spend more on “zero-sum” activities – activities that may advance individual interests, but not society overall. Cyber crime and the security industry that counters it is an example. The search for status goods, such as high fashion is another. These activities drive arms races between rival players, including the literal arms race of deadly weapons. Not mentioned by Mr Turner you could add extracting and burning fossil fuels to the list.

The expansion of zero-sum activities creates a couple of problems. First is how do you measure measure productivity of something where the output is often negative? The answer, for statisticians, comes from the monetary income generated – which is circular: you can’t tell the difference between inflation and productivity. A second problem is that it means measured economic income, such as GDP, becomes increasingly detached from human wellbeing. What is the point of getting a higher monetary income if it simply disappears in higher property costs (land is a classic zero-sum game), security and badges of status? This is not far from the point I have made that the most dynamic bits of the British economy before the financial crash were in finance and professional services, which are (mainly) classic zero-sum activities. Discounting this and you find that “real” growth was lacklustre long before austerity kicked in.

The third reason for the Solow paradox is that a lot of the benefit to new technology is non-monetary, and doesn’t reach the economic statistics. We are living longer, for example. This is an argument often used by people on the right to suggest that things are much better than they look, and that we should not worry about stagnant or reducing incomes among the majority. This is not a debate that I have ever got into. Mr Turner acknowledges its validity, but not the conclusion that those on the right draw from it.

Meaningless measures in the hi-tech hi-touch economy

This is an attack on standard economic measures, notably GDP and productivity. Economists have always acknowledged the weaknesses of GDP as a general measure of whether an economy is delivering what people need – but those weaknesses are growing to the point of absurdity. Well, not quite. GDP (and especially the nominal measure which doesn’t try to adjust for inflation) is a useful measure for the management of money in the economy. But we cannot assume that if GDP per head is growing that so are people’s wellbeing. Likewise wellbeing may advance while GDP is stagnant.

This is an old idea. Mr Turner develops it by following through the thought experiment of what happens when most work is automated. Measured economic activity then most arises from what economists call “rents” – returns on asset such as land and intellectual property.

Developed economies: average is over

Where this is leading is to an increased gap between a lucky minority of people who are well off, and a growing body of people stuck on very low incomes. The middle ground is disappearing. Notwithstanding some on the right who shrug this off, this is a major problem. And mostly we are looking for solutions in the wrong places.

The problem will not be solved by educating everybody to a high standard so that they all have the skills to programme robots, though improving education is surely a good idea anyway. Inequality does not have its root in a skills gap, but in the nature of work. Education will simply turn into another arms race for the small number of well-paid jobs. Neither is a focus on improving productivity going to help. This simply replaces middle income jobs with lower paid ones. Meanwhile we worry about things, like the increasing proportion of elderly people, that probably won’t be such a problem after all.

But in developed countries, and many middle income ones, like China, the problem should be soluble. The economy will have the capacity to produce a good standard of living for everybody. Mr Turner suggests a number of policy responses:

  • Income support such as universal basic income. He sees the logic but is sceptical.
  • Offsetting the concentration of income, wealth and rents. Assets and higher incomes need to be taxed more. Intellectual property rights need to moderated, rather than strengthened, as now.
  • High quality urban development. To enhance areas that would otherwise be left behind.
  • Adequate wages and status for caring services. This will require some form of political intervention.
  • Celebrating craft skills.
  • Increased leisure.
  • Education for life and citizenship – breaking free from the idea of education for productivity.

Developing economies – the old ladder destroyed

The prospect for low-income countries, especially in sub-Saharan Africa, does not look good, though. The development ladder used by Asian economies from Japan to China to join the ranks of the better off, is being knocked away. There will be no demand for cheap manufactured exports, as richer countries will do more for themselves. This is a serious global problem which will require a change in development thinking. It will be important to slow population growth.

Implications for economic theory

Mr Turner concludes by pointing out that all this makes much conventional economic theory obsolete. It is too focused on maximising income to improve wellbeing. It is based on a series of idealised assumptions, such as the non-existence of zero-sum activities, whose usefulness is vanishing. Higher levels of income do not necessarily mean that wellbeing is improving.

This is easily said, but few have taken on the implications. In Britain Conservatives still talk of the virtues of an open market economy to produce a better standard of living for all. Meanwhile Labour focuses on capital investment and productivity. This is yesterday’s economics.

But more people are calling for a rethink. What Adair Turner does so well is to use conventional economic logic to show why conventional economics doesn’t work any more, and that we need fresh approaches. That’s what this blogger is trying to do in his own, much smaller way.


5 thoughts on “Adair Turner: the advance of robotics changes economics”

  1. I can’t help but think you (and possibly Adair Turner) are overcomplicating this.

    Sure we can do our washing by hand, if we want to, but I’m happy to put it all in an automatic washing machine. If at some time in the future we can have a robot to peg it out on the line and iron it, why not?

    I’m not sure if this changes economics. I don’t think so. But if it does let’s deal with it. We aren’t slaves to the subject or any economic system. The system should work for us rather than vice versa. The narrative that ‘the robots are coming to get us’ is best left to Terminator and other SciFi movies!

    I quite like Bill Mitchell’s take on all this.

    1. In the long run further automation is a good thing, but there are important issues that it throws up. Adair Turner draws a parallel with the automation of the textile industry in the early 19th Century (which gave rise to the Luddites). For a good thirty years or so the result was misery as artisanal jobs were destroyed and replaced with insecure, menial ones. The Luddites were perfectly rational. That’s the problem with so much of this airy talk that it all comes right in the end. We need to find much better ways of handling the transition to a better place.

      And yes, most of the talk about robotics is silly hype. As I say in my blog, actually I think this hype is quite a useful way of making important points about economic management that have been relevant for 30 or 40 years, which is when this trend started. And as for changes to economic thinking: I think people still think in terms of producing and consuming ever more stuff. That’s what we need to get away from.

      It actually means we can worry a lot less about a lot of things (average productivity, world trade, government debt… I plan a future post to explain), but what we can’t expect is to economic growth to fuel higher wages and more taxes.

  2. There was an interesting example of Luddite activity in my home town.

    One of the local pubs has interesting stained glass window pictures explaining the events in detail. I was quite well aware of all this at an early age. The events of 200 years ago are still remembered.

    The early to mid 1800’s were a period of rapid technological change. No doubt people felt the same then as they do now about all that. The Napoleonic Wars had an effect on the economy with National Debt soaring to some 250% of GDP to ‘pay for’ the war effort. ‘Paying it back’, of course, meant squeezing the wages of the poor whilst the better off drew steady incomes from their government ‘consols’.

    The workers of 1812 would have analysed their problems by looking at what they could see in front of them. If a machine is putting them out of a job, and/or lowering their wages, then its the fault of the machine. Just as now many workers now see it as immigrant workers putting them out of a job and/or lowering their wages.

    So maybe it does all come right in the end, but this isn’t to minimise any social problems. The process would be a lot quicker if the machines and the robots were working in the interests of everyone and not just those who owned them.

    1. The process would be a lot quicker if the machines and the robots were working in the interests of everyone and not just those who owned them.

      Indeed.I think that’s the insight that led Marx and Engles to the Communist Manifesto in 1848, when the immiseration of the working classes was still in full flow. We still have that problem: the technology is owned by “the few” who are in the process of capturing government institutions. I think it is far too complacent simply to suggest that automation is good for everybody in the long run, and we should let nature take its course. There have to be political interventions which will run against the perceived current commonsense.After 1848 the voter franchise was expanded (first in Britain, I think), civic urban infrastructure developed (sewers,water, etc.), trade unions established and social security expanded (starting in Germany). These would have been regarded as insane earlier in the century. I am suggesting similarly that political reforms are going to be needed now.

      Am I turning into a socialist?

    2. I’m glad that the Luddites are remembered with some respect in Bolton. Generally the term has become one of abuse about people blind to the benefits of progress. The episode deserves to be better understood.

Comments are closed.