I grew up in the 1980s and 1990s, a time that saw the end of the Cold War and a new wave of prosperity. An age when the modern industrialised capitalist economy reigned supreme and globalisation, then still in its infancy, seemed to be the answer to global poverty. We clung to the hope that it would raise billions out of subsistence.
I was sold on the centre-left policies of the Clinton and Blair governments that left Big Business and the financial world largely to its own devices to thrive, while using a sensible and fair tax regime to pump money into well-run public services.
But recently, and certainly since the GFC (Global Financial Crisis in 2008-2010) and the rise of Trump and Brexit, like many of my peers, I have begun to question the fundamentals of that system.
In a previous blog post I evoked ways in which the modern capitalist industrial society undermines democratic values (as a riposte to Fukuyama’s seminal ‘The End of the World and the Last Man’ and ‘Identity’). But here, in Daniel Susskind’s recent book ‘A World Without Work’ I found a plausible framework that explains so much more about the system that is creaking at its very foundations. Capitalism and technology are engaged in a sort of self-reinforcing positive feedback loop. A vicious circle that has the end result of super-charging inequality and creating economic insecurity for large swathes of people.
Susskind makes a compelling argument to demonstrate that the increasing failures of the market economy today are caused to a large extent by technological progress already in play. He hypothesises that we are on a path that will eventually lead to a world with very little work. This is a journey that may happen in fits and starts and take unexpected turns along the way and that may even witness sudden bursts of demand for labour. It is a path nonetheless that can have only one destination.
Rather than the old cliché of ‘robots will take our jobs’, Susskind’s book is far more nuanced. He explains why there have been so many false dawns along the way too (the idea that humans would become obsolete was made way back for example at the turn of the twentieth century when cars replaced horses as the main source of transportation) and makes a good case for the fact that the future will be different. Indeed all the current economic indicators point to this fact. That we have entered the last gasp stages of a world that will be able to provide plentiful employment for its working population.
There is currently an abundance of research and commentary to explain rising wage and wealth inequality and the spread of middle-class angst, largely driven by economic insecurity. This is a class of people who have traditionally been the main driving force of prosperity in advanced economies and are essential to its stability – not just in economic terms but socially and politically as well. And indeed the data bears this out. Since the 1980s there has been a curious ‘hollowing out’ of jobs at the middle-income level. Lower paid and higher paid wage packets have grown while those in the middle have faltered and in some cases fallen.
In 2003, the ALM hypothesis (named after the three MIT economists who proposed it: Autor, Levy & Murname) was put forward to explain this strange phenomenon of income polarisation. It proposed that the level of skill (defined by economists in strict and rather simplistic terms as the level of education) was a poor indicator of whether automation would replace a certain job. What was a better indicator was whether a job was ‘routine’ or ‘non-routine’. In other words, if you could sit down and tell a software engineer how you did a certain task, he or she could then write an algorithm for a machine to do it.
This came to be known as ‘explicit knowledge’. The tasks where a human being struggled to explain in layman’s terms came to be known as ‘tacit knowledge’. Most jobs requiring explicit knowledge, in other words, those that are routine, fell in the middle-income bracket. Lower paid work was usually manual in nature (cleaners, truck drivers etc) and therefore ‘non-routine’ in nature and remained unaffected. The theory was so convincing that today it is considered standard trope for bodies even such as the OECD, IMF, World Bank and many other prestigious think-tanks worldwide.
What Susskind argues though is that the ALM hypothesis is breaking down and automation is already encroaching on those jobs considered ‘non-routine’. As he states ‘The ALM hypothesis has encouraged us to believe that there is a wide range of tasks that can never be automated…The Age of Labour to which we have become accustomed to, will carry on.’
The main reason for this, he posits is that following the mid-1990s, there has been a revolution in AI (artificial intelligence) in what has come to be known as the ‘pragmatic revolution’. Machines no longer try and mimic human intelligence – they are instead slowly building their own logic base (or computer rationality), devoid of human input.
A good example is Google’s AlphaGo Zero. In 2016, its predecessor, AlphaGo beat the best Go player at the time in a 5-match game. Go is a Chinese board game and considered devilishly hard to gain competency in. Its rules are simple enough compared to say chess but the number of potential moves rises exponentially compared to the latter. After 3 moves in chess, there are just over 70,000 potential moves, in Go that figure is closer to 1 billion. AlphaGo won by reviewing games played by the best human experts that had gone before—involving over 30 million individual moves.
AlphaGo Zero though did not have any human intellectual input, it simply learnt the rules of the game and played itself millions of times over a three-day period. When AlphaGo Zero played its predecessor, it thrashed AlphaGo comprehensively.
The implication of advances such as this implies that machines – using their own logic base – will be able to learn even those ‘tacit’ tasks that under the ALM hypothesis machines could not take over. The answer to this problem by most economists has been to argue that the hypothesis essentially stands, it is just the list of jobs classified as ‘routine’ and ‘non-routine’ that has to be adjusted. Susskind though categorically believes this to be wrong. ‘This is not a seismic quibble, but a serious shift. Machines are no longer riding on the coat-tails of human intelligence.’
Of course, the switch to automation won’t happen overnight. Something else the ALM hypothesis highlighted was the fact that many jobs, when broken down, are made up of micro-tasks – not all that can be done by machines. Susskind quotes a 2017 review by McKinsey and Company that reviewed 820 different occupations. It found that while fewer than 5% of all jobs could be fully automated, more than 60% of all the occupations were made up of tasks of which at least 30% could be. A machine could never give a moving oratory to a captivated jury for example, but machines today can retrieve, assemble and review a wide range of legal documents. These are tasks that make up a big part of lawyers’ jobs and almost the entirety of junior lawyers’ day-to-day work. In fact, JP Morgan now uses a system that can review more commercial loan agreements in a few seconds that would require 360,000 lawyer hours to achieve normally.
This fact has actually been widely used by economists as an optimistic explanation of why automation ultimately won’t take all our jobs and why in the past we have all prospered while technology has raced ahead. Historically, technology has largely complemented our labour skills rather than substituted them. This goes to the crux of the great debate between optimists and pessimists when it comes to the automation of jobs.
The ‘complementary force’ can be broken down into three simple factors. During the Age of Labour, tech helped take boring repetitive tasks out of the hands of humans that then allowed them to become more productive. It also helped humans earn more, making the economic pie bigger for us all. Finally, it totally transformed the economic pie, creating opportunities that were completely unforeseen not fifty years ago.
Think ATMs or even sat-navs. Rather than replace bank tellers, automated cash machines in reality freed the tellers from the simple yet time-consuming task of handing out money. It sowed the seeds of the advent of the inter-personal relationship banking culture that sprouted in the late ’80s and early ’90s that provided more all-encompassing banking services to clientele.
The reduced costs and better banking services went on to increase the number of customers. So while the number of tellers required per bank decreased, the number of overall banks increased, meaning that between 1980 and 2015 in US the number of banking staff employed actually shot up by 20%. This is a classic example of the complementary force of technology. The same went with sat-navs. They made taxi drivers more efficient in navigating complex urban areas, increasing their share of rides and with it improved revenue.
Many economists argue that such positive effects of tech will just continue into the future – they may very well displace some jobs, but will create new better paid ones in an ever-expanding economic pie.
Even before covid-19, Susskind was far from convinced of such an optimistic future. He argues that better productivity and the bigger pie effect is useless to the majority of people if most of those jobs have been automated.
Taking the ATM and sat-nav examples to their ultimate conclusions, today online banking has gone on to decimate high-street banks rather than encourage more of them. In the UK, the number of branches has fallen from a high of over 20,000 in the mid ’80s to around 10,000 today according to a 2020 House of Commons Briefing Paper (Bank Branches: why are they closing and what is the impact?’), directly as a result of online banking. On the other hand, driverless vehicles could make taxi drivers and truck drivers – employing around 3.5 million people in the US alone obsolete in the near future.
Susskind argues that we can already see the general effect of tech on the large inequalities today in wages and the increasing returns on capital relative to labour. Up to the 1970s there was a strong correlation between wage and productivity (ie profit margins). Since then though, there has been a great decoupling that has slowly been getting worse. As he states ‘In the two decades since 1995, across twenty-four countries, productivity rose on average by 30 per cent, but pay by only 16 per cent. Instead of going to workers, the extra income, has increasingly gone to owners of traditional capital.’
Going back again to the banking example, the closure of branches in UK brought on mainly by the move to online services lead to job losses but also precipitated a fall of 6% in banking costs for the six major retail banks in a relatively short period of time between 2014-2017. Capital is currently winning out over labour in the age of technological progress.
In the economy as a whole, who has not seen with their own eyes the astronomic rise of a small group of tech companies that make eye-boggling profits while employing a relative small amount of people?
Susskind points out that Apple today employs 132,000 people. If this sounds pretty labour-intensive, if you compare it to the biggest company of the 1960’s, AT&T, which employed over 750,000 people at its peak, you will clearly put the figure into perspective.
The future of labour in tech industries right now also looks bleak. In a riposte to the changing pie effect argument as well (the idea that tech will bring changes to the economy that will require a new wave of labour), the newest social media companies ominously employ even fewer people.
When WhatsApp was bought to the tune of US$19 billion in 2014 by Facebook, it’s whole payroll consisted of 55 people. Instagram, bought by Facebook as well for US$1 billion, employed a mere 13!
Even the OECD (Organisation of Economic Co-operation And Development: A Paris-based economic think tank) states that 80% of the increasing disparity between productivity and wages can be directly attributed to technology. The IMF puts it closer to 50%, but states that globalisation makes up another 25%, which as Susskind correctly states, is mainly due to transportation and communication technological progress in itself.
This all may change. Future tech companies may suddenly require massive pools of labour, but all the signs are that the general trend will be towards a world with increasingly less and less work. And certainly one where the quality of jobs, and therefore the level of income that can be gained, will generally decrease. Think Amazon and the well-documented labour quality issues or Uber drivers’ complaints of zero-hour contracts and problems of contractual work in general across the board.
What has been our Response?
The widely-accepted panacea today to creeping task-encroachment by machines has been better and more targeted education programs: Teach people skills that are not routine and machines will conceivably not be able to do in the future. But as Susskind has shown, there is an exponential growth in what machines can and will do. For now though, it remains our best response.
The education systems in most advanced economies leave their students woefully unprepared for the tech challenges of even today. You only have to consider how little importance is afforded to computer sciences in the UK curriculum to have an insight.
And not only that, they saddle many students with a lifetime of debt. The astronomical costs of higher education against its perceived usefulness is very much in question, even now. Well, especially now in this age of covid, where many education systems have been super-charged in their move to online services. This trend may eventually prove more permanent than temporary, especially given the reduced costs entailed, and incidentally also largely decimating another decent pool of unionised labour in the form of teachers.
Ultimately, as Susskind points out, education itself will inevitably lose out to technological progress —however comprehensive it may be. The one stark fact that stood out for me, amongst all others in his book, was an OECD study he quoted of adult, literacy, numeracy and problem-solving skills. This determined that only 13% of all workers in OECD countries (advance economies mainly) use skills on a daily basis that are higher than computers in proficiency – at today’s technological levels. Never mind what the machines of tomorrow will be capable of!
OECD stats show that between 1998 and 2018, tertiary education levels increased in and amongst the 25-54 age group from 23.79% to 44.48% respectively in advanced economies. But according to a 2018 IMF Working Paper (Drivers of Labour Force Participation in Advanced Economies: Macro and Micro Evidence) the participatory rates of employment – that is to say the percentage of the total working age population in employment or actively seeking work — actually fell. And it fell sharply in countries such as the USA. That is to say that while more people than ever before have a better education, those who have given up even bothering to look for work all together has risen. And the fall was broad-based amongst all socioeconomic groups and skill levels.
The Problem with Unemployment – Participation Rates
Much has been touted recently – in the pre-covid era – about US employment figures. It has stood below 4% in the last few years, its lowest level since at least 1969. So what’s all this gloom and doom about technological unemployment then? Look a little closer though and the figures hide some disturbing facts.
Apart from the already-mentioned issues with the decline in the nature and quality of work available across the board and the pitiful wage increases associated with it, there is another major concern : the aforementioned participation rates.
An analysis of these figures over time reinforces the general trends that Susskind mentions above, and helps vindicate his arguments. Male working-age participatory rates have been falling across the board, with an average decline of about 6% in a basket of advanced economies between 1985-2016.
So, while unemployment has been falling, the rate of those engaged in any way with the job market is in fact around the lowest level it has been since 1977. According to the IMF paper, this is particularly worrisome as even a small decrease in the participatory rates of this group can have out-sized effects on the economy ‘…since prime-age men are still the largest segment of the labour force in AE’s (advanced economies) and have traditionally been the main income-earners for their families.’
At the same time, however, corresponding female participation rates have shot up on average by around 10%.
This is all very much in line with Susskind’s argument. We are witnessing increases in supply of jobs that require social intelligence in the so-called ‘pink-collar’ sector mainly because these jobs are still considered outside the reach of machines at this very moment in time. These jobs include nurses and personal care workers, as well as food prep and retails salespeople. And of course, in line with his reasoning, nearly all these jobs pay well below the national average wage.
Lastly and most strikingly for me has been the phenomenal increase in participation rates amongst old people (classified as those over 55 years of age). This has increased significantly since 1990s (following decades of steady decline). According to the IMF paper ‘..the increase is particularly pronounced for the 55-64 age group, but in the past decade, even individuals older than 65 remained in the labour force longer.’
Again, this matches Susskind’s ideas of a dwindling labour income pie where people are forced to work for longer as the majority of wealth flows exponentially to an increasing share of traditional capital rather than being accumulated by the labour force over a lifetime. Even the IMF paper touches on this, explaining the increase in retirement age participation rates, being due to ‘suppressed returns on retirement savings as global interest rates fell, losses in financial wealth and potentially higher indebtedness.’
The Rise of Neo-Socialism
Left unfettered, this trend towards rising inequality and a world geared towards employment that has no actual decent work left in it can only be transformed, Susskind argues, by apocalyptic cataclysms such as revolution or war. ‘If we are to find a way to narrow the inequalities by a less cataclysmic route than in the past, it is clear that tinkering and tweaking as the State has tried before, will not be enough.’
In his drive to seek solutions, he doesn’t delve too deeply in the solutions he proffers but does offer a solid framework of drastic measures. The drive behind all these solutions are deeply socialist in nature – but not the traditional socialist policies of the past which have mainly focused on the production side (think nationalisation), but rather on the distribution side. The State will have to expand drastically and become what he calls the ‘Big State’. It will have to be income-sharing in nature, taxing the few left in work – usually highly-skilled well-paid jobs – at rates as high as 70%. There will also be a need for the Big State to capital-share – in other words – tax the profits of traditional capital more efficiently and widely share those dividends. This could be done through other methods apart from taxation such as state capital ownership, for example as is currently practised in Norway. There, profits from oil have been re-invested in markets through the Citizens’ Wealth Fund, that – pre-covid – was worth over US$1 trillion and would have given each citizen on average around US$190,000.
Ultimately, the main tool of re-distribution he puts forward is the much-talked about UBI (Universal Basic Income). Susskind though points to issues with UBI such as the animosity that will be created deciding who is in and who is out (think immigrants and rich people claiming UBI that could cause social and political upheaval). He proposes a Conditional Basic Income (CBI) where those who claim are in turn required to perform some non-economic duty for the State, a form of National Service. In a world with less work this might be a good idea and help the State give a purposeless population that has lost the ability to enter gainful employment meaning.
I am not so comfortable with this proposal. While I’m sure it was meant innocently, CBI can (and will be) easily politicised. It would give the state a ready and attentive income-hungry force that can equally be used for malevolent as well as benevolent ends. Think Modi’s India, Orban’s Hungary, Putin’s Russia or even Trump’s America with a massive swathe of the population receiving an income from state on condition of some type of uneconomic work that their respective governments deem worthy.
The most fascinating remedy that Susskind provides is that of a Big State that will have to involve itself in helping its population find constructive ways to fill all the spare time available. What he calls the coming Age of Leisure. ‘As we approach a world with less work, a traditional source of purpose for many people will fall away and a gap will appear. New sources of purpose will emerge, not all of them benevolent. We may want a meaning-creating state to step in and, through interventions like leisure policies and the CBI, guide whatever floods in to fill work’s place.’
Even though the solutions Susskind makes are a logical progression borne out of solid data, it is hard for many of us to swallow them. But consider this. During lockdown that many countries find themselves in now due of course to covid, much of the social media traffic has been about finding (or boasting) of innovative ways to pass the time locked away indoors. In South Africa, a small innocuous paragraph in the official government lockdown literature really caught my eye. It gives clear advice on what people should do with their spare time, including:
- Keep doing enjoyable and relaxing activities such as reading, pc, board or card games, social networking or watching television.
- Engage children in your care in creative ways; create fun learning activities, play games and try to keep their daily routine going.
- Stay active by doing simple exercises within your home and garden.
- Or create an exercise plan specifically to suit your environment.
(Talking Points: 21 Day National Lockdown Covid-19 – 24th March)
Is it such a stretch to imagine a Big State – with a large population out of work guiding us all in our leisure pursuits as well?
I had begun to grow uneasy with our modern capitalist system some years ago. And generally had a vague idea of the coming threat of automation. But what Susskind’s book has done is put down in a captivating argument the potential for serious instability that both capitalism and technological progress – locked in a kind of Dance of the Devil – could bring in the near future. The potential destruction this might reap on our collective societies is enormous.
Massive changes to the very fabric of our lives will be required to face up to it – that on top of global warming – it is reasonable to assume will bring about a perfect storm of serious economic and socio-political upheaval and maybe even war and revolution if the challenge remains uncontested.