Myths of the Old Order: The automation illusion (part two)

In today's blog, I will continue last week's discussion of the automation myth.

Chapter XII of my book, “Politics for the New Dark Age” provides a lay introduction to productivity, and its role in generating long-term growth. The fact of the matter is when it comes to the technological miracles promised by Silicon Valley, we can see them everywhere except in the economic statistics (the Solow paradox). Since the 1980s, labour’s contribution to economic output per hour work has remained flat, while the return on capital has grown by leaps and bounds. If labour productivity had continued expanding over the last forty years, we’d already be working shorted hours for higher wages. Instead, we have seen the opposite: low productivity, flat or declining real wages, and chronic levels of employment stress.

What’s really going on?

Automation, then, does not made us more productive. So what is it?  My answer is that the drive for automation now, as sometimes in the past, is capital intensification. Think of the difference between productivity and intensity like this: productivity increases output if either your workers or equipment become more efficient; intensity increases output when you add either more workers or more equipment to the production process. It is intensity rather than productivity gains that we have seen under neoliberalism: the outsourcing of labour, de-unionisation and greater labour uncertainty leading to modes of production that rely more and more on expensive capital equipment (including IP) and finance and less on labour input.

Capital intensification is not necessarily a bad thing: in the 1700s, capital needed to be concentrated out of the landed aristocracy and into the bourgeois banks in order to facilitate loans to the first factories (in both the trading and manufacturing sense of that word). Sometimes, productivity leaps do require capital to be concentrated. In an era of low labour productivity, when eighty per cent of labour worked farmland, capital intensification created new industries that in turn soaked up workers. It paid them for their increased productivity and expanded output to match, creating the middle-class consumer out of nothing and increasing prosperity for all. But that’s not what’s happening today, especially post-GFC. Consumer demand remains weak everywhere, and rates of return on expanded output are at historic lows. 

Automation in the modern era, then, is being driven by capital intensification in existing workplaces (especially wholesale, healthcare and other essential services). High levels of inequality mean that capital-owners have historic financial reserves and no productive outlet for investment. Governments bailed out the banks, keeping them flush, and in some places began a historic era of quantitative easing, literally showering banks and the corporate sector with cash in an effort to get them to invest in the real economy. Governments across the world are trying and failing to get corporations to spur growth by investing these savings, rather than using them to big up their stock proces and private wealth. The savviest businesses discovered that they could use this cheap credit, not to expand output, but to produce the same output using more capital and less labour. By doing so, they maintain or increase their share of profit in an era in which economies as a whole are growing slowly.

Automation is, therefore, a capital bubble (Chapter XIII of my book) that will only increase inequality and lead to further capital accumulation. It is for this reason that pushing tax cuts as a spur for economic growth (as governments in Australia and the US are trying to do) is catastrophically ill-timed on both a micro- and macro- level. Since new investments are normally tax-deductible, lowering marginal corporate tax rates will decrease the incentive to invest; higher tax rates increase the incentive for employing additional workers and equipment. At the macro-level, increasing the rate of corporate profit while impoverishing the social programs propping up what remains of consumer spending will only increase capital concentration, slow the economy and increase inequality. Why would governments do this?

Automation is leading to permanent job losses but is doing so for the same social and policy reasons that labour productivity has remained stagnant for forty years. Power in the workplace has shifted decisively in favour of capital (with government assistance) and employment is less secure than at any other time in a century or more. While a minority of those laid off will find new employment as technicians or programmers, the majority are being forced into the retail or ‘gig’ economy as low-skilled personal service workers: Walmart ‘greeters’ and Uber drivers, for example. Such jobs are not only less secure, lower paid and offer fewer benefits than the jobs they replaced, but actively alienate workers from one another and de-skill them, further reducing their collective bargaining power. 

Speculating about the future of work

While the service sector may be the most important part of a highly developed economy, transitioning everyone to a service sector job will not be our saviour from automation because not all services are high-productivity and high-wage. A totally service-dominated economy with flat levels of production and consumption would look very much like a feudal aristocracy. Those who control capital (once upon a time: land; now: financial assets) will live very handsomely indeed on the rents they extract. Everyone else will eek out an existence predicated on hierarchical personal service to such individuals: a true ‘trickle down’ economy. Not only would such an economy be technologically and socially stagnant, I argue it would be prone to potentially revolutionary social disruption.

Truly forward-thinking prophets (or fantasists) of the automation apocalypse warn us that the true threat to the contemporary economic and social order is not automation but AI. AI algorithms that will not only replace high-skilled service sector employees such as doctors, lawyers and teachers, but also manage capital better and more effectively than humans and produce cultural goods on their own. Quite frankly, we’ll cross that bridge when we come to it. True AI would lead to such a profound re-structuring of human life on Earth that the question of economic organisation might seem like a second- or third-order problem. 

An optimistic response to the possibility of hyper-competent decision algorithms is that it would increase the value of non-analytical, emotional and social skills that humans are likely to continue to hold and advantage in, a point Ezra Klein puts emphasis on. While I’m sympathetic to the argument that humanity has a unique social and emotional toolkit that would be hard for us to [reliably] replicate in an artificial intelligence, we presently compensate such skills poorly. The modern cult which privileges rationality as the basis of competence would have to change pretty dramatically, and if we could somehow create a discourse that privileged humanity’s social nature so highly, there would much better uses we could put it.

Policy consequences and responses

From my perspective, the policy solution for the automation dilemma is obvious. Increasing the labour share of income (either through higher wages or greater redistribution) would soak up additional production and create new demand so that the capital intensification of the last decade doesn’t go to waste. Increasing labour power and giving workers a democratic say over the workplace decisions that affect their lives is the best way to manage the transition to new forms of production: if automation is truly unavoidable, then let’s make it work for the many and not the few.

Unfortunately, the automation debate is not intensifying calls for democratising the means of production. Quite the contrary: the right is using it as a cudgel to threaten workers and the centre-left appears paralysed with indecision and fear. For centrists, the policy consequences of automation are expressed in terms of anxiety about its consequences for social stability and the political consensus underpinning the continuation of status quo policies of de-regulation and unrestricted trade. The reason for the sudden surge of interest in harebrained schemes like universal basic income is that the centre is looking for ways to buy off revolution and discharge its social responsibilities to the economic losers without having to re-examine or change the fundamental ways in which production is organised.

For the centre-left in particular, there is a second aspect to their obsession with the automation narrative. For well-educated elites in the public sector and media, there is a fantasy that if the working class comprises enterprising, globally-open “knowledge-workers” (like themselves) then it will also come to share in their elite culture. Under this narrative, elites don’t have to close the gap that has opened between themselves and the proletariat: instead, the workers will come to them. But they mistake labour flexibility for labour uncertainty, and their enthusiasm for using Uber is not necessarily matched by the desire of former manufacturing worker forced to drive a taxi for uncertain wages. The bourgeoise thought this way about the disciplining effect of manufacturing too; as did the feudal landlords and slaveholders before them. Sadly, while culture is in fact a function of economic patterns of production, so long as those production patterns contain exploiters and exploited, the interests of elites and the working class will remain divergent.

Myths of the Old Order: The automation illusion (part one)

This series of posts will continue to examine myths or tropes that I hear repeated by people trying to make sense of the "New Dark Age" that is our fragmented social reality. If you have an idea for a trope to address, contact me on Twitter @Askews2000.

We begin with two narratives of automation. 

There’s a scene in “White People Renovating Houses”, the South Park Season 21 (!) premiere, in which a tiki-torch wielding mob of rednecks marches through the town, demanding the destruction of digital personal assistants such as Siri and Alexa. The episode was panned by critics, who thought the writers didn’t adequately skewer the presumed target of mockery (i.e. white nationalists). But what critics failed to understand is that white nationalists were not the intended target (that would have been too easy). They themselves were. Or rather, the narrative that fear of automation was driving the economic anxiety of Trump’s racist base. The incongruity between the recent facts (neo-Nazis in Charlottesville) and the explanation (fear of automation) is the point of the joke. 

As a historical counterpoint, there's a moment a third of the way into Sven Beckert’s magisterial history of capitalism, “Empire of Cotton: A Global History”, at the dawn of the Industrial Revolution in Britain when a few hundred ‘spinning jennys’ (machines for cotton weaving) begin to displace the work of hundreds of thousands of  artisans in the manufacture of cotton fabrics. Within a generation, automation had destroyed a global industry which had remained largely unchanged for a thousand years, and begun the “Great Divergence” by which the wealth of Western Europe eclipsed that of the rest of the world. Beckert describes the opposition and mob violence which was often visited upon the mills but which failed to ultimately prevent the radical transformation of land and labour relations they represented. 

Everybody’s talking about automation

Discussing automation, and its effect on social, political and economic relations, is a recurrent feature of 2017’s “New Political Dark Age”. There are new books, conferences, speeches, and essays on the topic almost daily; we must tackle the subject, we are told, to be taken seriously as a political and economic thinker. Automation, in this telling, is a miracle bestowed upon us by Silicon Valley, but which demands unemployment and political resentment as its price. It is both a moral good to be embraced and a governance challenge to be managed: the perfect talking point for a certain kind of forward-thinking technocrat. 

The automation narrative came [back] to the forefront of progressive politics during the Obama administration, by some accounts. And indeed, the topic reeks of Very Serious People  justifying their own failure to achieve lasting economic change while grappling for control of the narrative with radicals (i.e. the Sanders crowd) who see further compromises with the status quo as the problem. When you look into the issue, you find that almost every article, think piece or editorial  on the subject cites back to a famous 2013 Oxford paper, which found that 47 per cent of job categories defined the researchers were ‘at risk’ from automation. 

I don’t find that paper’s methodology particularly convincing, but smarter people than me have looked into it and come to two sorts of conclusions. Firstly, stating that automation will eliminate half the jobs in the economy is probably a wild over-estimate: the OECD have estimated that the true figure is closer to 9 per cent. Even so, those job losses won’t occur all at once, leaving plenty of time for workers to adjust, and are likely to be compensated for by new jobs created by automation. The automation problem is not, therefore, the total number of jobs in an economy but to whom those jobs are going to be distributed. Because if technological change is too quick, the prophets of automation say, then there’s no guarantee that particular people that lose their income will find a replacement.

Automation and inequality

When you boil it down, the automation narrative is similar to a moral panic: highly anecdote driven and largely divorced from any sense of historical perspective. It’s easy to talk about the installation of touchscreens at McDonalds, speculate about driverless trucks, and mourn the loss of analogue film to Snapchat. But we’ve been here before, and recently: people worried once upon a time that ATMs would replace banks, TVs would render cinema irrelevant, or that vacuum cleaners would lead to idle house-wives [sic]. McDonald's isn't even innovating: 'automats' were a thing as far as as 1895! Jason Furman, President Obama’s final Chief Economic Advisor, made this point quite well in a December 2016 presentation that still available online. The process of creative destruction is so inherent to material progress (see Chapter XII of my book) that, with the benefit of hindsight, we look back with bemusement at  luddites who feared steam-powered cotton spinning.

What really controls employment levels is the supply and demand for goods and services in the economy; the suppressed consumer demand caused by contemporary levels of inequality is what’s standing in the way of full employment, not technological churn. Increasing the productivity of labour and/or capital increases social output if and only if that output can be consumed. So technological progress doesn’t change the number of jobs in an economy, but it does change the skills those jobs require. Labour’s capacity to capture a share of the increased output and consume is dependent, in a wage-economy, on whether or not workers have the skills to work higher productivity jobs. If not, the share of profit capture by capital and high-knowledge workers increases while the labour share of income for already marginalised workers does the opposite, making inequality worse and dragging the economy. 

Furman was therefore correct to state that the problem with automation is not a risk of mass unemployment but of growing inequality. The Industrial Revolution made Britain’s merchant class fabulously wealthy, while impoverishing both peasants in India and smallhold farmers at home. When an industry goes extinct, there’s likely to be a mismatch between the demand for skills and those possessed by the existing labour force. Normally, the mismatch is resolved over time through retraining, the retirement of older workers and the entry into the market of new cohorts. This is the laissez-faire ‘attrition’ model of development and social equality; in the alternative, society can play an active role in controlling the pace of change through industrial policy, providing active support to retraining into new industries, or increasing redistribution in favour of those who can’t or won’t adapt to new patterns of work. This is how progressives deal with problems of inequality in all its forms. 

That concludes Part One, looking at the contours of the debate. Return next week for Part Two where we dive deeper into the economic and policy consequences.