Smart Machines and Service Work: Automation in an Age of Stagnation
(Reaktion Books, 2020)
Jason E. Smith’s Smart Machines and Service Work is being published this month as the fourth Field Notes book to be brought out by Reaktion Books in association with the Brooklyn Rail. Tony Smith, Professor of Philosophy at Iowa State University and author of Technology and Capital in the Age of Lean Production: A Marxian Critique of the “New Economy,” among many other books, took the occasion for a meeting of the Smiths.
Tony Smith (Rail): First of all, congratulations on the publication of Smart Machines and Service Work. It is one of the very best books on the social consequences of technological change I have read, far more insightful than the technology books that get so much attention in the mainstream press. Many of those works defend techno-utopianism, arguing that if we just wait a bit longer and just come up with the right policies advanced technologies will set off a new age of growth and prosperity. Others take a techno-dystopian stance, predicting unprecedented levels of technological unemployment and social chaos. How would you define your position in relation to those alternatives?
Jason E. Smith: They are both wrong. Both share the assumption that advanced capitalist economies are currently experiencing, or poised on the cusp of, a thoroughgoing, machine-driven transformation whose primary effect will be a sudden surge of labor productivity and economic growth. They differ only in their consideration of the knock-on effects of this renewed dynamism, with “techno-dystopians” emphasizing the presumably catastrophic social consequences for class stratification and job markets: an exacerbation of income inequality and, above all, “mass” unemployment. In my book, I focus on the latter. Episodes of mass unemployment result not from technological change, but from economic collapse. Were a robust, automation-driven reanimation of the high-income economies to take place, the historical evidence suggests an altogether different trajectory. We could expect temporary dislocations of the job market, as labor processes are revised, jobs redefined, labor reallocated away from high-productivity sectors to more labor-intensive ones, entirely new industries created, and new divisions of labor (both social and technical) imposed. A new class composition would emerge, with new stratifications of skill, gender, race, or locale. In the US, one need only look back to the period just after World War II, extending to around 1965 or 1970—what I call “Automation 1.0”—to trace such a pattern. To be sure, over the long-term, such a transformation would most likely result in rising unemployment, with most new employment in low-wage service sector jobs, not least personal services. Misery and upheaval for many, to be sure. But a sweeping technological make-over of the advanced economies is neither currently underway nor imminent.
Claims that these economies are on the verge of a technological rupture originate primarily in schools of business or “management,” as well as in Silicon Valley, and are then funneled, remasticated, and echoed by journalists and commentators. They come complete with buzzy names: “second machine age,” “third industrial revolution,” “industry 4.0,” take your pick. This hype trickles down to the Left and becomes linked to speculative UBI schemes or even proposals to “nationalize” social media platforms. Such projections are being made against a backdrop of unremitting crisis. (In 2018 the Bank of England could report that the British economy had “endured the worst decade for productivity growth since the 18th century.”)
The rhetoric that has emerged and consolidated around automation can be construed as part of a broader initiative to fuel a historically unprecedented equities bubble, driven primarily by a handful of so-called tech or internet stocks (the well-named “FAANG” stocks). The innovation story of the past decade is primarily confined to the financial sector and monetary policy: stock buybacks ($800 billion in 2018), near-zero borrowing rates, massive leveraging of private companies, round after round of quantitative easing. Tsunamis of cheap money have washed over the world’s richest economies, a sizable chunk of it lavished on urban real estate. With the onset of the pandemic, we got a King Kong-scaled dose of it, forcing stock markets to all-time highs as entire economic sectors shut down and tens of millions of workers in the US lost their jobs.
These fictions of technological change are vitally important for a capitalist class that imagines itself a progressive historical force, yet which presides over an economy that is profoundly stagnant, lurching from one deep crisis to the next. This class presents itself as a disruptive, even anarchic, historical force, whose extraordinary innovations pose problems (explosive productivity growth rendering half the workforce redundant, and so on) it alone can understand and solve (with UBI, a jobs guarantee, perhaps a “Green New Deal”). It is no wonder the buzzword of the decade was “smart” (smart phones, homes, factories, cars, and cities), a term reflecting the self-regard of those who coined it. Yet these people got rich off real estate and stock market bubbles.
Make no mistake, ours is absolutely an age of “social chaos,” to use your term: of social polarization and pulverization, of deepening debt and growthlessness, of broken labor markets, and sharpened yet fragmented and incoherent class conflict. Smart Machines and Service Work tries to size up this mounting disarray and offer a different account of why we’re caught up in it.
Rail: Most people believe we are living in an age of unprecedented technological change. Yet in your book you refer to “sustained technological inertia.” What do you mean by this striking phrase?
Smith: For the most part, the kinds of technological advances that have occurred over the past decade or more are irrelevant from the macroeconomic perspective, whether it is a matter of labor productivity growth, employment, rates of investment, GDP growth, or what have you. It is no accident that the consolidation of this rhetoric of imminent automation (machine-learning, algorithmic governance, platform revolution, the “sharing” economy) coincided with the sudden rise of companies like Facebook, Apple, Alphabet, Amazon, Alibaba, and Tencent. By mid-decade, these businesses had cemented their status as stock market leaders—their outsized valuations far outstripping older multinationals in banking, oil, pharmaceuticals, and cars—while also worming their way into the weave of everyday life of working and middle-class consumers. Social media companies like Facebook and internet monopolies like Alphabet/Google spent the decade promising a revolution in artificial intelligence or self-driving cars, while more than 90 percent of their revenue was made from selling advertising space to other companies (like banks and car makers). These platforms piled up massive profits over the past decade by creating and enforcing monopoly-like conditions in which to operate. Despite selling themselves as tech companies, they invest relatively little on R&D but spend lavishly to crush potential competitors, primarily by buying them out early on.
The “smartphone” stands in as the signal innovation, or contrivance, of the age, its “star commodity.” Its sheer pervasiveness, its presence on sidewalks, in boardrooms, classrooms, or at the dinner table, confirms its status as an epochal emblem. For the most part, it simply brings together older devices (the mobile phone, the personal computer). Providing access to a panoply of diversions—shopping, streaming music and video, interpersonal communication—by means of a single, interactive screen, these apparatuses complete a confluence underway for decades now: the fusion of commerce and news, entertainment and sociality, self-stylization and civic life on a one-size-fits-all, touch-sensitive LCD (or OLED) screen. Its user is torn between these registers, while performing them all at once; at a loss for bearings, their mood flickers between harmless diversion and inarticulate rage. Yet the heavy hand the largest technology companies have had in equities markets, combined with the concussive force they have unleashed on leisure, consumption, personal identity, and public discourse—all already in the throes of erosion and decomposition for decades—gave rise to claims for this core technology that far exceed its impact on how we shop, consume media, or interact with friends, family, and strangers. In the workplace these innovations promised to lead to what Paul Mason heralded as an “exponential takeoff in productivity.” That’s precisely what has not happened. What we got instead are increasingly tight webs of surveillance and tracing, on the streets and in workplaces.
It is telling that smartphones and social media platforms took off in the midst of a deep recession that never quite “broke.” The iPhone was first marketed on the eve of the 2008 financial crisis. The way people correspond, get their news, watch movies, shop, or share photographs will never be the same. But Robert Solow’s “productivity paradox,” first formulated in 1987—“You can see the computer age everywhere but in the productivity statistics”—has stood the test of time. The past decade saw the weakest growth in labor productivity gains in decades, even in the manufacturing sector. Yet the slowdown in labor productivity growth set in as early as 1970 or so, i.e. at the very moment the world’s first microprocessor, Intel’s 4004, made its debut.
Rail: This brings us to one of the abiding mysteries of the contemporary economy, captured in the phrase “secular stagnation.” How does your account of the disconnect between the apparent innovative dynamism of the last decades and the relative lack of economic dynamism differ from others who have drawn attention to it?
Smith: In late 2013, just as the rhetoric anticipating an automation-driven explosion in productivity was heating up, another fraction of the US ruling class weighed in with a very different perspective. Larry Summers, one-time Treasury Secretary under Bill Clinton, opined that the US and other mature capitalist economies were facing the prospect of a deep-rooted stagnation in which high unemployment, slow GDP growth, and wage stagnation could persist far longer than the short-lived downturns of typical business cycles.
The performance of the US economy seemed to prove Summers right. The promised lift-off never arrived. The decade during which books with titles like Rise of the Robots (2015) were given prominent places in public discussion was also a decade defined by an unrelenting global economic crisis the likes of which had not been seen since the 1930s. The opening round of this debacle was marked by a string of spectacular failures in the financial sector, as overleveraged investment banks faltered, either collapsing outright or being bought for pennies by less exposed firms. What unfolded next was as predictable as it was devastating: lost years scarred by unemployment rates not seen in decades, combined with plummeting labor participation rates, as laid-off workers dropped out of the job market (or, in some cases, were reclassified as “disabled”). As demand for labor dropped, wages for many workers fell. As workers were idled, so was capital. Throughout the crisis decade, capacity utilization rates, measuring the discrepancy between what an economy can produce and its actual output, reached the lowest levels in postwar history, far below those of the crisis years of the 1970s. GDP growth sputtered, even as corporate borrowing throughout this period skyrocketed. In the US and Europe alike, a phenomenon first observed during Japan’s “lost decade” of the 1990s surfaced: the ghostly presence of “zombie” companies able to stave off ruin by constantly refinancing their debt, even as their operations contracted. Most significantly, at the very moment when so many commentators heralded the prospect of a new machine age, private business investment in fixed capital cratered, bottoming out at rates unprecedented for the post-war era. US labor productivity figures unsurprisingly exhibited abysmal growth rates, increasing at less than one percent annually, even in the historically dynamic manufacturing sector.
The drawdown in capital spending has been especially acute, but in no way an aberration from what preceded. A few years into the crisis, one study showed that, measured as “share of GDP, business investment has declined by more than three percentage points since 1980.” Since the 1970s, only the decade of the 1990s stands out as an anomaly, during which a group of economic indicators (GDP, labor productivity, business investment) nudged modestly upward. But in the period between 2000 and 2011, business investment rates barely budged, growing at a mere tenth of the level that prevailed in the 1990s.
Insofar as they highlight the drying up of business investment, the stagnationists are not wrong. But their account of why high-income economies across the world are mired in a seemingly horizonless crisis—the Keynesian go-to response, insufficient demand—is a meager one. It should be remembered that Alvin Hansen, Keynes’s primary American advocate, first sketched the theory of secular stagnation in response to the abrupt downturn of 1937, after the failure of Roosevelt’s countercyclical fiscal strategy to prop up collapsing demand and boost private investment. This failure forced Hansen to consider the possibility of a chronic and intractable listlessness, and to speculate why mature capitalist economies tend toward stasis and drift (population decline? closing the frontier?). Yet, today, the policy prescriptions of those in this camp are still predicated on new rounds of large-scale deficit spending. They remain dazzled by the apparent successes of Keynesian demand management for a few decades after WWII, the better to repress this school’s rout in the 1970s, when these same policies helped birth a macroeconomic monster—“stagflation”—they could neither theoretically account nor devise antidotes for.
In 1981, the share of US government debt relative to GDP was only 31 percent; even before the massive spending bill passed in March of this year, that number was over 100 percent, very near the number registered in 1945–46, when financing defense spending for a global war was in place. It is certainly much higher now. By the same token, US government spending as a share of GDP has grown steadily since 1970, peaking at 43 percent in 2010, a year into the “recovery” from the 2008 crisis. The size of the private capitalist economy continues to contract, relative to total economic activity. What mainstream economists, Keynesian and neoclassical alike, do not acknowledge is the fundamental distinction between private capitalist activity and public expenditures, paid for out of funds appropriated (in the form of taxes or debt) from the private sector. When governments buy goods and services from private companies to stimulate demand, the result might be short-term upticks in employment. But, as Paul Mattick demonstrated with great clarity some time ago in Marx & Keynes (1969), this kind of spending is simply a form of large-scale, government-driven consumption, paid for out of the pool of profits (or “surplus value”) generated by the private economy. Government spending of this sort simply redistributes this share of total profit to specific capitalists, like Raytheon, Pfizer, or Purdue Pharma. In the same way, when governments directly produce services, like public education, these services are not sold on the market, and do not produce profits to invest in expanding production. Though education or health care spending by the state often addresses real needs, from the point of the capitalist system itself these are unproductive expenditures. They do not produce value or surplus value directly, but are paid out of the surplus value extracted by the private sector.
Rail: The categories of “productive” and “unproductive” labor are not found in mainstream economics. Could you talk about the distinction a bit more, since it plays such a crucial role in your book?
Smith: This distinction was a decisive one for classical political economy, for Smith, Ricardo, and Malthus, as well as for the great critic of that school of thought, Marx. I think it’s acutely felt in people’s everyday experience as well, which is why David Graeber’s misleading slogan, “bullshit jobs,” had the resonance it did. Similarly, Adair Turner has recently spoken of “zero-sum activities” to characterize the growing fraction of economic activity devoted not to the production of wealth but to the struggle over its distribution. Yet this fundamental conceptual distinction is completely lost on mainstream economists.
Economists do not differentiate between value-producing activities and those that circulate or distribute value. Nor do they see any need to account for the way the profits accruing to certain types of capital—banking capital, commercial enterprises—represent shares of what Marx calls “surplus value” originating in properly productive employments of labor. Rather than distinguishing between activities that produce value and those that capture surplus value redistributed by means of inter-capitalist competition, economists more or less adopt the commonsense notion of “productivity” used by business owners and the business press. Any economic activity that generates income is said to be productive, and the productivity of labor is gauged by dividing output, expressed in money terms, by labor units. Naturally, the existence of an expansive public sector not subject to the rigors of inter-capitalist competition while supplying goods and services not sold on the market poses some problems for this simplistic notion. But there are subtle accounting tricks to plaster over the cracks.
Let’s come back to the so-called “productivity paradox” mentioned above. The solution to this riddle was proposed, it seems, in a famous paper by William Baumol. He argues that as certain economic sectors introduce labor-saving innovations whose net effect is a reduction in the demand for labor, the newly redundant labor will be more or less seamlessly reallocated to more labor-intensive and less productive sectors. Many of these workers will be shunted into what economists call “service” sector employment. Baumol’s model predicts that as productivity gains are unevenly distributed across what he calls technological “progressive” and “stagnant” sectors, more and more labor will be concentrated in less productive employment, resulting in dwindling gains in the labor productivity of the workforce as a whole. Extrapolated over the very long-term, the growing disparity in productivity gains between sectors will result in an economy in which productivity growth approaches zero.
This story is conceptually flawed. It relies on a notion of productivity that is confused or contradictory, even on its own terms. In my book I explore some of the contradictions that arise when one tries to compare labor productivity across sectors, measured sometimes in physical units, sometimes in money units. How do you measure the productivity of the financial sector, whose output is hard-pressed to characterize in physical terms? Does it even make sense to measure the productivity of an activity that merely intermediates between other economic activities, without producing “use values” consumed by businesses or households? Economists do it all the time. How do you measure the productivity of public school teachers, who perform services administered primarily by local governments, and are not exchanged for money on the market? Despite the radically different labor processes and social functions of these examples, they are both lumped into a single, incoherent category of “services.”
More significantly, Baumol does not discriminate between activities that produce value and those that do not. He makes no distinction between goods and services provided by the public sector and those produced by the private capitalist economy and, within the latter, between activities that directly produce value and those that only circulate or distribute it. The exploration of these conceptual distinctions makes up a central preoccupation of Smart Machines and Service Work. If we employ these categories we arrive at a very different notion of productivity than that relied upon by economists and business owners. Many labor-employing activities generate revenue but do not augment the total wealth of society; many activities that create “use-values”—provided by the state or private households—do not produce value or exchange-value. A significant number of so-called service sector jobs are value-producing, however labor-intensive and resistant to technological change they may be; others produce no value at all, and entail labor processes that are susceptible to labor-saving reformatting. The distinction between productive and unproductive labor cuts across this category, and renders it analytically irrelevant.
This distinction is essential because, as I noted earlier, unproductive activities must be paid for out of the total pool of surplus value generated by the private economy: they are a cost incurred in the accumulation process. National income accounting conventions count these costs as income. One of the long-term tendencies of a mature capitalist economy is a rise in the number of unproductive activities, relative to productive ones, necessary for accumulation: carrying out portions of the exchange process, facilitating capitalist activities through financial operations, renting land and buildings to productive firms. This growing overhang of labor-employing activities that circulate or distribute value rather than create it is both a condition for capital accumulation and, the more this ratio of unproductive to productive activities rises, an obstacle to it. This is a thorny issue, and my thinking on it owes a lot to Mattick and to economist Fred Moseley’s work. The upshot is that there is a differential rate of productivity growth between the productive and unproductive dimensions of the economy; the labor-productivity gains of the value-producing activities, with important exceptions, tend to grow more quickly than those that circulate or distribute value. The resulting relative expansion of the unproductive sector exerts a crippling, downward pressure on the overall profit rate. The only hope of relieving this pressure is for a surge in labor productivity in the unproductive “sector” (a misleading term, since the distinction between productive and unproductive activities cuts across sectors and even individual businesses). But for reasons I’ve already spelled out, such a scenario is highly unlikely: not least because the constriction on the rate of profit drives down rates of investment.
Even among firms that do directly extract surplus value in the labor process, there is no correspondence between the quantity of surplus-value they exploit and the surplus-value they take in, in the form of profits; these profits reflect the maximum share of the total surplus-value produced by the economy as a whole that firms are able to appropriate in the distribution process. As accumulation slows down and capitalist firms intensify competition to appropriate a dwindling mass of surplus-value, they will devote more and more of their resources to Adair Turner’s “zero-sum distributive activities.” Often, these activities are supervisory in nature, as heightened workplace discipline requires additional personnel to enforce workplace speed-up in the absence of refinements in production techniques. But they just as often take the form of so-called “business” services, as more and more resources are devoted to accounting, advertising, and financial operations, or to efficiencies in marketing and commercial processes. The net effect of this war over distribution across the economy is a still further slow-down in accumulation, precisely because these activities represent additional overhead costs paid by capitalists out of the total pool of surplus-value created through exploitation in properly productive activities. As the profit rate is constricted, the shrinking pool of surplus value requires businesses to allocate even more resources toward appropriating, rather than producing, this surplus value, driving the profit rate down still further. This is the maelstrom-like dynamic of an unrelentingly stagnant economy.
Rail: At the conclusion of your book, you seem pretty pessimistic about both labor unions and community-based forms of struggle, calling for new forms of organization. What is the basis for this pessimism? Do you have any thoughts about the shape these new forms might take?
Smith: I am pessimistic only about a revival of the old labor movement, a prospect so many in the US left hold fast to. I find the way the conflict between classes is currently playing itself out to be promising and invigorating, even as the process remains fragmented, disoriented, and full of surprises.
Since the turn of the century, almost all job growth in the US has been in low-productivity “services,” and recent projections by the Bureau of Labor Statistics anticipate the fastest expanding segment of the labor market over the next decade to be in low-wage positions requiring no formal education. This pattern is a dire one, exacerbating a dynamic that has been in place for decades. In a certain sense, we are still caught in the undertow created by the great wave of capitalist innovation that unfolded between 1920 and 1960 or so. I dub this Automation 1.0, but this wave includes the development and widespread diffusion of the internal combustion engine, the building up of infrastructure on a properly capitalist scale, and the “promises” and perils of nuclear power, in addition to developments more narrowly associated with the automation of factories. It is no secret that real wages for American workers have barely budged since the mid-1970s. Many attribute this long-term wage stagnation to the defeat of organized labor dating from the early 1980s. To be sure, unionization rates have been halved in the meantime. But the defeat was not simply political. The material situation that made possible the consolidation of union power in the post-war decades began to erode as early as the mid-1960s, as the composition of the working class, and the nature of work itself, mutated. Wage stagnation was closely tied to the onset of dramatic declines in the rate of labor productivity growth. The US Bureau of Labor Statistics show that, for the period from 1973 to 1990, the productivity of US workers inched upward at an annual rate of just 1.3 percent, a fraction of the gains recorded in the two decades following World War II. Growth in workers’ real wages requires rising output per labor hour. This is why post-war arrangements between capital and labor in the US and Europe explicitly tied increases in pay to bolstered productivity: workers and owners would “share” the benefits of rising hourly yields. When such gains are hard to come by, long the case across Europe, North America, and Japan, any potential wage increase for workers would entail a corresponding drop in the profits of business owners. This prospect the capitalist class has fought, and will fight, tooth and nail.
The changing nature of the labor market, of class composition, and of work itself has had other crippling effects on the labor movement. As more and more workers are shunted into jobs in the distribution process rather than in production, or are concentrated in the low-wage end of the so-called service sector—in stores, call centers, hospitals, or day care centers—they are spread out over myriad industries and, unlike their parents and grandparents, who were often concentrated in large worksites bringing together thousands of laborers, they tend to be dispersed in space, in smaller workplaces, often working with very little fixed capital. If there is a defining feature of the broad service sector, where much “unproductive” labor is concentrated, it is a negative one: it lumps together widely divergent concrete labor processes whose sole shared feature is their labor-intensity. A decisive material condition for the post-war growth, in size and power, of trade unions was the homogenizing effects of the capitalist rationalization of the manufacturing core. In past periods of rapid industrialization, technological breakthroughs in one industry quickly generalized across lines of production, making labor processes converge; workers formerly divided by skill, class, region, gender, and wages found themselves performing increasingly similar laboring activities, their skills and wage levels converging. As older, craft-based skill differentiations eroded and became externalized in large-scale machinery, and as this convergence of labor processes spurred leaps in labor productivity, workers found it much easier to define themselves as workers tout court, defined over and against the capitalist class, rather than as employees of a specific firm, whose grievances were lodged against this-or-that boss.
As workers are expelled from the core, capital-intensive industries, crucial material conditions for class coherence give way. Despite the automation enthusiasts’ speculation, most service sector jobs remain impervious—by their very nature—to mechanization. And when they are susceptible to it, prevailing low wages dissuade business owners from undertaking top-to-bottom overhauls of these activities (delivery services, cashiering, security guards, hotel cleaning, taxi rides). Low productivity gains, persistent low pay, the nature of the labor itself (which for many takes the form of personal services) and above all a lack of solidarity are demoralizing for workers. There is very little sense that they form a class in any positive sense, that they represent some prefiguration of a society to come, to be built in their own image. In such conditions a heightened sense of conflict can prevail among them, feeding on longstanding identity formations (race, ethnicity, gender) that divide them. During the pandemic, these divisions have grown to include the distinction among those deemed “essential,” and therefore forced to risk their lives to continue working, those who lost work altogether, and those, often middle-class employees, who migrated easily to online platforms.
Despite the crumbling of the conditions that gave rise to the old workers movement, the past few years have nevertheless seen extraordinary initiatives taken by workers, both in workplaces and the streets. Let’s not forget that it was the real threat of an illegal strike by TSA workers in 2019, with airline workers poised to join them, that put an end to the government shutdown. For some years now, public school teachers have also been willing to undertake large-scale actions; these have often been in supposedly conservative states, yet were met with overwhelming popular support. Public school teachers have remained largely insulated from labor-saving mechanization of the sort that has transformed some industries, and their place within the broad social division of labor gives them extraordinary social leverage. In France, we recently got a glimpse of what a revolt in what Phil Neel calls “the hinterland” can look like, as the gilets jaunes movement—with all its contradictions—took aim at city centers and traffic roundabouts for months. God help the capitalist class if workers in distribution centers and the logistics networks decide to attack the flows of goods through the ports and along the arteries of the just-in-time networks. Just months ago, National Guard troops were patrolling American streets under curfew, as anti-police riots and demonstrations spread across the country, in the midst of a deadly pandemic.
True pessimism, to end on a personal note, was to watch hundreds of thousands demonstrate against the coming attack on Iraq, in 2002 and 2003, knowing how impotent these numbers were. Despite the ambient misery and even trauma inflicted by the crisis years, it feels now like we might be on the verge of a real break, a rupture. But whatever figures of struggle the next few years throw up, they are unlikely to revert back to models of labor movement in its mid-20th century heyday. Despite all that colludes against them, in both material and political terms, workers will have to grope their way toward something new.