The age of automation will be shaped by class struggle, not demographic destiny
Some see automation as the solution to economic malaise, but Nick O'Donovan's work explains how it is instead symptomatic of a failed economic paradigm, and heralds a wave of industrial unrest
We can expect the recent rise in industrial disputes in the UK to continue for the foreseeable future. Wages have barely risen for decades, and sharp rises in inflation mean the second quarter of 2022 saw a record drop in real earnings.
But pay is not the whole story. We can understand growing industrial unrest in the context of automation too. As Nick O’Donovan explains (see the tweet below; the whole thread is must-read), recent and upcoming strikes by railway workers have also centred concerns about the dilution of human inputs into traditional labour processes in the transport industry.
In this post, I attempt to situate this dynamic in a wider context, in the hope that it might clarify what to do about it.
Automation to the rescue?
Charles Goodhart and Manoj Pradhan’s recent book The Great Demographic Reversal explains the rise of automation — which they welcome — with reference to the impact of demographic change on the economy.
They see the productivity slowdown (in the UK and elsewhere) as caused by population ageing. A declining proportion of working-age people in society means that (other things being equal) workers can demand higher pay without necessarily becoming more productive. Low growth and high inflation — i.e. ‘stagflation’ — inevitably follows.
Yet this scenario did not actually come to pass as the West aged. The entry of Chinese workers into the global labour force served to discipline Western workers’ pay demands. But now that China is ageing rapidly too, Goodhart and Pradhan warn that the reckoning cannot be avoided.
Goodhart is a major influence in central banking, having worked for nearly 20 years at the Bank of England; likewise, the influence of a central banking lens on his wider economic analysis is quite apparent. Like the Bank’s current leaders, he views the recent rise in inflation as driven by wage demands, overlooking excessive profiteering, and has endorsed raising interest rates to address it:
“the present stance of monetary policy is still way below what is the appropriate level, particularly in the US and UK, given the emergence of a wage/price spiral”
This is where automation comes in. Goodhart and Pradhan see automation as a way to mitigate the impact of demographic change. Inflation would be tempered, since robots don’t need pay rises. The remaining human workforce would be, on the one hand, reallocated to more productive roles (i.e. designing and maintaining automated labour processes) but also, on the other hand, disciplined by the emergence of a replacement labour force. Growth can therefore safely resume.
A more sceptical view
Let us be clear: The Great Demographic Reversal is a gloomy book. But Goodhart and Pradhan are hopeful that accelerating automation will at least mitigate the myriad macroeconomic impacts of population ageing. Alas, I think full-on gloom might be more justified, albeit for different reasons. Automation is not the answer to low productivity, but rather another symptom of the same underlying problem.
To explore this, we need to welcome Nick O’Donovan back to the discussion.
O’Donovan’s outstanding book Pursuing the Knowledge Economy chronicles the failure of the ‘knowledge economy’ paradigm, in terms of both the economic reality of knowledge-led growth, and its promotion by a generation of policy-makers in the 1990s and early 2000s.
It feels rather surreal from our present vantage point, but it was not that long ago that the champions of the growth of knowledge-intense firms and industries were heralding a new epoch in which the struggle between capital and labour had finally been resolved. Everyone could prosper; all countries like the UK really needed to do was provide their citizens with the means to succeed through ‘education, education, education’. Knowledge had become the world’s core commodity, and nobody was excluded from acquiring it.
Sadly, as O’Donovan shows, things did not quite work out as anticipated. Knowledge was never freely available, but instead increasingly captured by tech companies to support rent-seeking rather than genuine innovation.
Furthermore, highly-skilled knowledge workers did not acquire more power or autonomy within production processes. This is partly because they were competing against many others with equivalent skills. But it is also because they had little control over the intellectual property (IP) they created (and data they generated) for their employers, so had limited labour market leverage.
Firms in knowledge-intense industries had few incentives to contribute to economy-wide productivity growth, and focused instead on protecting their own profit-attracting assets and customer base. The business model of platform companies, in particular, depends upon near-monopoly status, thereby inhibiting tech diffusion. O’Donovan’s work can be read alongside Mariana Mazzucato’s analysis of ‘unproductive entrepreneurialism’, and Cecilia Rikap’s account of ‘intellectual monopoly capitalism’, which demonstrate aggressive IP strategies, and domination of wider knowledge-creation spaces, by a handful of tech giants.
Automation, power and politics
The trend towards automation and the growing role of artificial intelligence were inherent in the positive vision for the knowledge economy from its inception; humans would be employed higher up the value chain. But these trends are now being propelled forwards by the same structural forces that destroyed the knowledge economy’s progressive potential.
A 2019 inquiry by the Business, Energy and Industrial Strategy Select Committee (then chaired by Rachel Reeves, the current shadow chancellor) made headlines by declaring:
“The problem for the UK labour market and our economy is not that we have too many robots in the workplace, but that we have too few.”
However, the Committee qualified its positive view of automation with an important warning:
“The ownership and use of new technologies, processes and robots is likely to make businesses richer and more powerful. For a fairer society it is important that we consider ownership models of technologies whilst being cautious not to stifle innovation. We have seen from previous inquiries that the practices of businesses such as Amazon and Uber can lead to workers being exploited by increasingly monopolistic firms who earn huge returns that do not flow back to the workers who help create that wealth.”
This points us towards the real risk of automation. It may not destroy jobs — it certainly will in some industries and occupations, although these are likely to be replaced by new jobs in others — but it promises to intensify the disempowerment of workers which has accompanied the knowledge-based economic model.
As David Spencer and Gary Slater argue, the impact of automation in the UK will probably materialise as a deterioration in job quality, not a reduction in overall numbers. This is partly a story about the kind of industries that will create jobs (for humans) in the future, that is, largely not high-tech industries. But it is also partly about how automation will intensify the labour processes still requiring human input, policed by greater monitoring and surveillance (itself an increasingly automated management function).
It may be that automation does help to solve the economic problems envisaged by Goodhart and Pradhan, principally the revival of inflation. O’Donovan, and many other commentators, agree that it is likely to increase productivity (again, other things being equal). But the social consequences are very likely to include more inequality. And we are already seeing the political consequences in terms of the rise of right-wing populism (the link between automation and populism is charted in Pursuing the Knowledge Economy).
In my view, these political developments pose a far greater macroeconomic risk than demographic change. They have produced a generation of centre-right political leaders content to tell voters that the source of their hardship is not their lack of control and fair remuneration at work, but rather public spending, immigration, the European Union, environmental protection, wokeism, and a university education. The long-term economic consequences of this agenda will be devastating.
Where do we go from here? A focus on education investment retains its appeal for some progressive thinkers. We do of course need to spend more on education, but not because it represents an investment in ‘human capital’ that will help citizen-workers navigate the age of automation. We cannot engineer children and young people for the jobs of tomorrow — instead we can enable them to imagine the economy of tomorrow, while also providing better incentives to encourage them to train for the occupations needed today.
Universal Basic Income (UBI) may be part of the solution, if automation means there are fewer (good) jobs available in some industries and some places. Everybody should be insured against obsolescence, and UBI is becoming increasingly popular among progressives in the wake of the COVID-19. But there is a less wholesome side to the idea too, explaining its popularity with Silicon Valley tech bros. It could enable new forms of conditionality and disciplining. And the redistributive nature of UBI would have to be weighed against the prospect of — in combination with automation — surrendering the struggle for more control for workers within production processes.
For automation to be a development that actually improves the function of the economy from a progressive perspective, rather than simply something which — best case scenario — creates a bigger pie so that losers can be compensated, we will need to confront the disempowerment of workers that it threatens to accelerate.
For example, although automation is itself unlikely to destroy the total number of jobs, this may in fact be an outcome worth fighting for, with automation harnessed by workers to enable reduced working hours. Short of this, advancing automation must trigger the strengthening of employment protection, reform of corporate governance and ownership practices, and a reorientation of taxation towards rent-seeking activities.
Workers must of course be given the opportunity to themselves shape how automation affects their workplace. This was the organising premise of Hilary Cottam’s recent project for IIPP on work quality. It is a prospect the trade union movement has been assessing, resulting for instance in TUC guidance on ‘people-powered technology’ and collective agreements on the automation of management processes. And to bring this post back to where it started — the rail industry — a recent report by TUC Wales highlighted the Work 4.0 agreement negotiated by the EVG union in Germany (representing Deutsche-Bahn workers), which embeds a series of protections for workers when new technology is introduced in the rail network.