Forget productivity growth in future?

The following are notes jotted down whilst reading a lecture (40 x A4 pages long) given by Adair Turner, Chair of INET (Institute for New Economic Thinking) in Washington DC in 2018



  • The lecture covers the possible long term impact of rapid technological progress – i.e. work automation and AI – on the nature of and need for work
  • What if all useful, versus zero-sum, human work could be automated, say 50 to 100 years on?
  • What are the implications for the distribution of income and wealth?
    • Incomes for useful work will fall to zero, but for zero-sum work will rise
    • We cannot rely on the market to determine acceptable income levels
    • Wealth will come more and more from land, brands and beauty
  • As technological advances accelerate, the national productivity growth rate will fall faster – so productivity should stop being a national priority
  • The current combination of rapid technological innovation and low measured productivity growth is exactly what we should expect


Section 1 – When, not if, almost all economic activity is automated:

  • ICT progress means, in 50 years, we’ll be able to deploy unimaginably massive quantities of computing power, and automate almost all activities we call work and for which people receive income
  • Many jobs are repetitive/ predictable, others more complex/ thoughtful/ creative, and others a mix of the two
  • Automation reduces the time needed for the first and last categories, and so employment overall (if we continue to work 35 hour weeks)
  • Accommodation and food services are far more susceptible to automation than health and social services and education
  • ICT hardware costs keep on reducing – software originals cost a lot but marginal copies cost next to nothing
  • At some stage. combinations of hard and software will equal, and then far exceed, human intelligence
  • N.B.
    • The above merely extrapolates current known technological capabilities without any possibility given to whole new ideas/ sectors unexpectedly emerging for mopping up surplus labour and offering both new productivity improvement potential but also higher wages and jobs they want versus have to do – aka unknown unknowns – e.g. the internet, search engines and social media back in the 80s
    • Re humans being overtaken by machines, what if man implants IAs (Intelligent Assistants) to upgrade his grey cells – and keeps on upgrading by up/ downloading his brain contents for updates and extra capacity – and thus keeps well ahead of any machines alone so he’s not threatened by them?


Section 2 – Explaining the Solow paradox i.e. “Computers are everywhere but in the productivity statistics”:

  • Why is measured productivity growth slowing down?
  • Because an acceleration in technological progress, which enables dramatic productivity improvement in some sectors, can be accompanied by displaced labour having no choice but to move to low-wage sectors, resulting in a decline in total measured productivity
  • How come?
    • Proliferation of low-productivity jobs – for those displaced by increasingly automated sectors:
      • In the past, labour displaced were able to move to new sectors which also had potential for productivity improvement e.g. agriculture to manufacturing, then on to some services
      • But once those with money have all they ‘must have’ plus ‘like to have’, they don’t provide the demand for more – they have ‘enough’ of what is currently on offer 
      • However, they might like a domestic servant or two, on very low wages – plus maybe the occasional painter or singer to entertain them
      • So aggregating all incomes from all (currently known) sectors arithmetically reduces GDP and national productivity because of the rapidly growing low-wage, high employment sectors
      • Thus, rapid productivity growth in one sector combined with low productivity in others results in lower overall productivity growth
      • Total productivity growth is as much driven by the productivity growth potential of the sectors into which workers move as those where they are automated away from – it’s simple arithmetic
      • The logic is that, eventually, all jobs will be automated, so all humans will be displaced to lower and lower productivity jobs – until there’s no productivity growth at all
      • Then we’ll need to ‘find something for them all to do’, albeit at lower rates of pay
    • Rise of zero-sum activities as nations get richer:
      • Zero-sum activities are those where more and more human talent can be applied (and higher and higher incomes paid) but not produce more GDP or value to humans
      • Examples include:
        • Criminals v Police – they balance each other out – they don’t add to the total sum of goods or services for increasing human welfare
        • Cyber criminals v cyber experts defending people and firms against them
        • Legal services – if divorce lawyers improve their quality and so results, the other side does the same, so soon one is back to square one
        • Corporate and IP lawyers – they secure new ventures or protect valuable IP rights, so benefit others
        • Tax accountants for minimising tax versus HMG tax officers for maximising tax-take
        • Marketing and advertising executives, and communications consultants – who seek to convince us that product A is better than B
        • Financial traders and asset managers – most add no value versus index investing
        • Financial regulators and compliance officers
        • Corporate financiers who organise M&As which rarely enhance shareholder value
        • Political campaigners and lobbyists who seek to influence votes one way or the other
      • But more education is good for all
      • And fashion design is a creative artistic process which adds to the variety and enjoyment of life
      • However, productivity improvement leads to the creation of more and more zero-sum jobs, many of which are not counted by GDP and cancel each other out anyway – another reason national productivity is seen to fall
      • If you apply AI to zero-sum jobs, you simply increase the intensity parties on both sides work at – it’s not their efficiency that matters but their effectiveness – did the lawyer win the case, or not?
    • Growth of low-cost/ free goods and services which enhance lives:
      • After automating so much and dispelling the need to work, why do we get a proliferation of low-paid jobs and zero-sum activities but no better human welfare – or do we underestimate the benefits to human welfare?
      • We only need a few very clever people plus AI to do wondrous things – e.g. to invent super drugs so we all live to 100, or forever evenand disease free – so the rest are mostly surplus to overall requirements
      • GDP accounting conventions – the methods, estimates and assumptions used – are flawed for the future viz:
        • GDP clocks salaries of the above clever inventors plus their sales – but when their patents expire, their sales revenue drops to very little yet their products and benefits continue
        • ‘GDP deflators’ used to cover price changes are suspect – all sorts of shenanigans are possible here
      • The knowledge of how to produce something can cost a lot – but the marginal cost of actually producing it can be peanuts
      • Professor Martin Feldstein -“Government statisticians are almost bound to underestimate the scale of productivity improvement – as low growth estimates fail to reflect the innovations in everything from healthcare to internet services to video entertainment that have made life better during these years”
      • Turner questions whether all innovations make life better
      • As unit prices collapse with technical advances, so do sales revenue and apparent benefits clocked by GDP – thus, GDP undercounts technological progress
      • And, as computers get more powerful, do our kids get happier/ less stressed – so is human welfare also not measured well?
      • Thus “real productivity growth fails to account for some of the most dramatic increases in productivity”
      • Note “the apparent paradox of expanding opportunities for automation combined with mediocre and declining productivity growth”
      • “With limitless potential to automate jobs, it is almost inevitable that we will observe a slowdown in measured productivity growth”
  • N.B.
    • What if we reduce average weekly work input hours to 15 (a la J M Keynes) for no loss of GDP? – productivity would gallop ahead
    • Can we really expect no more new hi-wage/ hi-employment/ hi-productivity sectors, as Turner seems to assume – just ones which have no automatable potential?
    • Where’s the huge benefit clocked of less work hours and more leisure hours – of more doing what we ‘want to do’ , not what we ‘have to do’ to earn money to pay for stuff we ‘must have’
    • Surely, the ultimate human aim is NOT to have to work at all, and instead be able to choose whether to potter in our gardens, mow our own lawns, launder our own clothes, sail boats or play golf with chums – this would not only fill available time enjoyably but allow us to develop what talents we may have enthusiastically
    • Over time, all sectors tend to peak as all potential to improve is addressed, first in quantum leaps then increasingly via marginal gains – but new sectors are new, and offer huge potential for clever people to address productivity issues therein


Section 3 – Meaningless measures:

  • GDP measures have always been imperfect but, as we get richer, they become even more so – especially with new ICT collapsing hardware costs and enjoying zero cost software replication
  • GDP fails to reflect the pace of technical progress which enables us to deliver more with less
  • There’s a limit to how many cars or washing machines we want to buy, and as we reach those limits, labour must inevitably shift to activities which cannot be automated
  • Past measures were perhaps ‘good enough’ for policy-making to reflect technological advances and GDP/ productivity growth – and each generation feeling better off than the last one
  • But no longer, when GDP counts many activities which cannot possibly improve human welfare (e.g. social networks and always-on devices) and does not count many others that do (e.g. healthcare) – and where productivity growth is rapid in some areas but more than offset by low wage/ low productivity jobs elsewhere
  • What’s the difference between consumption (tangible?) and welfare (intangible?) for different goods and services? – GDP per capita is already suspect as a measure of human welfare
  • Maximising real GDP growth can no longer be the prime objective of economic policy if it’s losing its meaning
  • Standard GDP measures are already less meaningful and less useful – they will be worse in future



  • All work will be automated – so if people still need an adequate income, employment will be dominated by low-wage jobs, many still existing because the rich like being served by people, not robots, even though those jobs could be automated
  • GDP will be dominated by property values and various forms of rent (of property and IP i.e. stuff people compete for) as all other goods and services are produced at ever falling prices so most income people have will pay for what remains either limited or is distinctly different e.g. high fashion, pop heroes, top footballers, highly talented people – or zero-sum activities attracting the very best to outdo others re winning elections, court cases, cyber defence efforts
  • According to Thomas Piketty, almost all developed economy wealth over the last 50 years has been explained by rising property values, and almost all that explained by rising land values, which is not limitless
  • Employment will be dominated by low-wage face-to-face services
  • Inequality between the rich few and the poor many will widen further
  • Measured productivity growth will be very slow


Section 4 – Average is over – Income and wealth inequality is inevitable:

  • “Where will the new jobs come from, especially the new incomes?”
  • It’s no good just blindly saying ‘give everyone better skills’
  • In the past, new ideas led to new sectors offering stuff more and more people soon found they wanted – this led to many more new jobs, often better paid
  • Productivity improvements across the board also led to more pay and so more disposable income for more people to buy other goods and services – thus did economies grow
  • If people have to work to gain income, and if there’s no minimum wage rates, then jobs will always be created to induce demand for some new service provision – an employment equilibrium will result albeit, in future, accompanied by ever-rising inequality:
    • A relative few digital companies and their clever top guys will be the big winners
    • Losers will be the rest of us, including those who previously were doing quite well but, now outplaced, are forced to accept low-wage jobs to scratch a living
  • But even if average incomes fall, that would fail to reflect everything from healthcare to internet services to video entertainment that have made most lives much better over the last few decades
  • But Turner asks: “Have always-on mobile phones, computer games and social networks made lives better?”
  • Most people will not be able to afford to live in the best areas/ cities so will migrate to where there is plentiful and cheap land with few planning restrictions or properties left unwanted and so cheap by the rich – this will enable them to live worthwhile/ acceptable lives – so social turmoil, as in Ned Ludd’s day, is unlikely
  • Meantime, over the long term, attempts to increase the productivity growth rate of developed countries are likely to be both unnecessary and ineffective:
    • We only need a few highly talented ICT experts to keep advances going in those sectors which can be automated
    • And, as some sectors get better, others less productive attract more of the outplaced labour which more than counteracts any productivity gains made so, on balance, productivity will continue to dip
  • The most important choices facing advanced rich societies in the future will be how we spend the fruits of increasing productivity and how to distribute it:
    • Forget ‘better skills’ – education is good for all, but not essential for productivity improvement – we only need a few very good IT bods for AI and super intelligence
    • Pay everyone a UBI (Universal Basic Income) to ensure they receive at least a basic minimum for a reasonable standard of living (varying only by location, property and land prices) plus enjoy high-quality public services e.g. health, education and public transport plus shared public spaces like countryside and beaches – however, UBI ignores the psychological benefit that work delivers a sense of status and self-worth
    • More people are likely to find satisfaction in becoming skilled gardeners, artists, cooks, brewers, organic farmers and beekeepers rather than software developers
    • Nations facing an ageing population problem need worry no longer – too few workers to support too many oldies will no longer be a threat
  • RCS:
    • Turner repeats many of his pearls, and not necessarily for emphasis
    • Why not push for more and more leisure and far fewer hours at work – all big benefits to ‘human welfare’?
    • With more education for all, people will be better able to decide what they want to do with their lives
    • Technical advances already mean many more low-wage people get to enjoy acceptable (to them) living standards, including cheap fun and education 


Section 5 – The old ladder destroyed – Rapid economic catch-up is no longer possible:

  • Radical automation potential combined with rapid population growth could create almost insurmountable barriers to economic catch-up:
    • Note the USA and W Europe gap with the RoW (Rest of World) over period 1800 to 1950
    • A few other nations have achieved catch-up since, partially via their manufacturer exporters mopping up surplus agricultural workers, and higher incomes from manufacturing increasing savings and investments in plant and machinery – e.g. Japan, Korea, Taiwan
    • But much of manufacturing is, or soon will be, automatable at attractive cost so, if wages keep on rising there, automation will soon take over
    • India, with a growth rate of 5% p.a., needs to create 10 to 12 million new jobs every year just to keep unemployment and underemployment stable – but they’re failing as companies start to apply state-of-the-art technology despite labour available at very low cost
    • Ditto China, with a growth rate of 7% p.a.
    • Africa has a far bigger problem – average growth rate 4.6% p.a. versus 2.7% population growth rate
  • Soon the rich world will not need cheap emerging economy labour to provide low-priced footwear, apparel and other goods – automation at home will do the jobs
  • So the key is to boost sectors unlikely to be prone to automation e.g. tourism and construction
  • And boost the quality of education for all to equalise opportunity


Section 6 – Implications for economic theory:

  • We now live in a world where:
    • Productivity improvement can be delivered with little capital investment
    • Most wealth resides in locationally desirable property/ land, IP rights and brands
    • Most wealth creation derives either from changes in the relative price of already existing assets, or from the creation of IP, brand and externality (?) effects
    • The problems of production will become unimportant
  • Hence, productivity improvement will no longer be about how to get more from less but how to resolve, in a fair and sustainable way, disputes about the distribution of those goods, services and assets, both created and natural, which automation does not make available at ever falling and close to zero prices
  • It will become a balancing act, between individual freedom versus fairness
  • According to Peter Orszag, in an article for Bloomberg Opinion commenting on Turner’s lecture:
    • The impact of new technology on total productivity growth depends crucially on who accrues the income from the new inventions – what additional consumption they choose to enjoy with that income – and the nature of productivity advances in the sectors that workers are shifted into as a result
    • And any new sectors are not expected to lend themselves to automation or significant productivity improvement
    • And, as the rich get richer, they may well choose services offered only by low-wage sectors e.g. personal care aides, cooks and servers, registered nurses and home health aides – i.e. person-to-person interactions that are, for now, difficult to automate
  • So no more big productivity gains on the horizon?
  • None foreseeable, for now




Leave a Reply

Your e-mail address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.