A Low Demand For Labor
As part of our end of year sale, we are posting excerpts from some of our new titles. In this excerpt from Automation and the Future of Work, Aaron Benanav analyzes the future of employment, what the true causes might be for the low demand for labor, and how this analysis plays into political strategizing going forward.
Working at Any Cost
Starting in the 1970s, unemployment rates in wealthy countries began to rise from historically low levels. Outside of the United States, they remained stubbornly high for decades.5 In that context, unemployment insurance programs went into crisis: they had been designed for short bouts of cyclical unemployment in fast-growing economies, not long-term unemployment in stagnating economies. To coax the unemployed back to work, governments began to reduce labor market protections and scale back unemployment benefits. Active labor market policies replaced passive income-support systems as the main response to job loss. In Denmark and Sweden, govern- ments tried to balance inducements to work by spending almost 1 percent of GDP in 2016 on placement services, training programs, and employer incentives, but they achieved mediocre results in slow- growing economies. In most wealthy countries, such programs were even less in evidence: spending on active labor market policies (not including direct job creation) averaged just 0.3 percent of GDP in OECD countries in that same year.
Under these conditions, few workers remain unemployed for long. No matter how bad labor market conditions become, they still have to try to find work, since they need to earn an income in order to live. As growing numbers of workers find themselves without reserves, the present-day world economy comes to look more like the one Marx analyzed in the mid nineteenth century, in Capital. In a stagnant economy, Marx explained, the stagnant portion of capitalism’s “indus- trial reserve army” or “relative surplus population” will tend to grow. “Recruited from workers in large-scale industry and agriculture who have become redundant,” this stagnant surplus population comes to form a “self-reproducing and self-perpetuating element of the working class,” which takes “a proportionally greater part in the general increase of that class than the other elements.” Since their work is “characterized by a maximum of working time and a minimum of wages,” their “conditions of life” tend to “sink below the average normal level.” The expansion of this population was, for Marx, an “absolute general law of capitalist accumulation.” Written over 150 years ago, Marx’s analysis has become contemporary once again. In the slow-growing economies of the past few decades, job losers have been obliged to join new labor market entrants in low-quality jobs—earning less-than-normal wages in worse-than-average working conditions. Unlike in Marx’s time, this phenomenon is mediated, today, by postwar welfare-state institutions, which have continued to shape labor-market outcomes even as those institutions have deteriorated over time. Cross-country institutional differences determine the degree to which experiences of precarious- ness diffuse through the workforce or remain concentrated within specific sections of the population.
Such shifts are easiest to document in the United States, where only unionized workers are afforded basic employment protections. Almost all other employees are hired at will and, barring outright discrimination, can be fired at any time. Between 1974 and 2019, unemployment rates were on average 30 percent higher than they were between 1948 and 1973, primarily due to lower rates of job crea- tion following recessions. Over the same period, private sector union- ization rates declined significantly: from nearly 30 percent in the early 1970s to 6 percent in 2019. Firms were therefore able to take advantage of higher average unemployment rates, which left many workers fearing for their jobs, to put the squeeze on employees. Given the difficulty many workers would face finding new work were they fired from their jobs, they have been compelled to accept relatively stagnant real wages as the condition of working at all.
Some economists have argued, on the contrary, that over the past few decades it is only US workers without college degrees who have truly faced deteriorating labor market conditions. In a less extreme version of the automation thesis, these economists claim that technological change has hollowed out the American job market, destroyed middle-wage jobs, and polarized employment opportunities between high- and low-wage work. The automation of routine tasks is said to have generated a rising college wage premium, setting off a race between education and the machines. It is certainly true that, in the United States, experiences of precariousness are strongly modulated for individual workers by education levels, as well as race. Unemployment levels in the United States are significantly higher for workers with low educational attainments and for people of color. It is also true that, in the 1980s and early ’90s, some Americans were able to insulate themselves from downward pressure on wages by getting a college degree. However, by the early 2000s—when the automation of economic activity was supposed to be accelerating—the college wage premium had stabilized, since the wages of most college educated workers had begun to stagnate. The median American college educated worker earned a lower real wage in 2018 than in 2000, even though the total value of outstanding student loans rose dramatically over those years. The reason is that from 2000 on, economic growth rates slowed significantly—and so too rates of job creation—while college degrees became more common: 40 percent of prime-age workers had at least a college degree in 2019. Those degrees offered less protection from deteriorating labor market conditions. Workers with college degrees crowded out workers with lower levels of educational attainment in jobs that did not previously require such degrees. Meanwhile, the share of young college-educated workers with employer-sponsored healthcare halved, from 61 percent in 1989 to 31 percent in 2012. Despite earning higher wages than their less educated counterparts, many of these workers were precariously employed.
What makes the United States unusual, from an international comparative perspective, is precisely that experiences of economic precariousness diffuse throughout the workforce. Even regularly employed US workers find that they are highly exposed to potential job loss in a persistently low-labor-demand economy, since they can be fired at any time. The consequence is that, unlike firms in other countries, US firms face no particular need to construct alternative working arrangements to take advantage of vulnerable sections of the labor force. Some firms do utilize alternative working arrangements to get around US labor law—witness the small but significant boom in gig-economy jobs, like Uber and Lyft, which offer work through online platforms as a way of disguising their employees as independent contractors.
Set against this American case, the employment landscape in European and wealthy East Asian countries is more complicated. In these regions, postwar labor market institutions were mostly designed not by left-wing governments but by right-wing politicians who emphasized the importance of national-imperial identities, the formation of male-breadwinner households, and the maintenance of relatively fixed workplace hierarchies. In return for accepting corporatist arrangements, male heads of households received substantial job protections: unlike in the United States, regularly employed workers were not hired and fired at will. For a crude measure of the difference that has made, we can look to the OECD index of employment-protection levels, which measures, on a scale of 0 to 6, the degree to which employees are protected from individual firings. Permanent workers in the US barely register at all on this index (at 0.5), while workers in the UK (at 1.2), Japan (1.6), Germany (2.5), Italy (2.5), and France (2.6) have been much more protected (Figure 4.2). In the latter countries, heads of households who obtained permanent jobs were largely insulated from market pressures associated with a declining demand for labor. They remained free to fight for collective wage increases, even as economy-wide unemployment rates rose to 10 percent or more. Meanwhile, as compared to the US, unemployed workers in these countries received more generous out-of-work benefits.
In most high-income countries, rising rates of unemployment from the mid 1970s onward therefore did not initially cause workers’ real wages to stagnate as in the United States. The workers who suffered most were the unemployed, as well as the children and spouses of still-employed workers. The jobs crisis took the form of a worsening exclusion; it was concentrated on specific sectors of the population rather than widely diffused. Older unemployed workers were pushed into early retirement. Married women were discouraged from looking for work, which is why women’s labor force participa- tion rates remained low in many European countries and Japan— with Sweden as a key exception—into the 2000s.
The term “precarity” entered a wider lexicon precisely amid protests against laws that reduced job security for many workers, particularly for women and youth. For instance, the 2003 Biagi law allowed Italian firms greater “flexibility” in firing part-time and temporary workers; the 2004 Hartz IV reforms substantially reduced out-of-work benefits in Germany. Similar efforts to strip young labor market entrants of employment security in France were rebuffed by workers in 2006 and again in 2016. Yet despite bouts of resistance, labor markets in western Europe and wealthy East Asia have become steadily more bifurcated between workers in standard employment with relative job security, and a growing mass of (mostly younger) workers in nonstandard jobs who lack it. Between 1985 and 2013, the share of nonstandard employment in total employment rose: from 21 percent to 34 percent in France; from 25 to 39 percent in Germany; from 29 to 40 percent in Italy; and from 30 to 34 percent in the UK. In Japan, the “non-regular employment” share (a category similar to the nonstandard employment share) rose from 17 percent in 1986 to 34 percent in 2008, with similar trends unfolding in South Korea. Changes in the composition of employment were much more dramatic for new job offerings: 60 percent of jobs created in OECD countries in the 1990s and 2000s were nonstandard.
More and more workers were exposed to employment insecurity at a time when, due to slowing rates of job creation in anemic econo- mies, they would have trouble finding new employment were they to lose their jobs. These workers were forced to moderate their demands for wage increases. Across the OECD, real median wages rose by 0.8 percent per year between 1995 and 2013, even though labor produc- tivity rose by 1.5 percent per year, leading to a significant upward redistribution of income (although one that was less intense than in the United States alone, where those rates were 0.5 and 1.8 percent respectively).
As underemployment rises, inequality must intensify. Masses of people can only work as long as the growth of their incomes is suppressed relative to the average rate of income growth. As econo- mists David Autor and Anna Salomons note, “Labour displacement need not imply a decline in employment, hours or wages,” but can hide itself in the relative immiseration of the working class, as “the wage bill—that is, the product of hours of work and wages per hour—rises less rapidly than does value added.” The consequence is to further expand the gap between the average growth rate of real wages and of productivity levels—contributing to the 9-percentage- point shift from labor to capital incomes in the G20 countries over the past fifty years. Worldwide, the labor share of income fell by 5 percentage points between 1980 and the mid 2000s, as a growing portion of income growth was captured by a tiny class of wealth holders.
As I discussed earlier, increases in inequality have been worse than even these statistics suggest, since the distribution of labor income has itself become more unequal, with the largest pay raises going to managers. Between the late 1980s and the early 2010s, labor produc- tivity grew faster than average wages, which in turn grew faster than median wages across the OECD. Over time, immiserating employ- ment growth becomes self-reinforcing. Sectors of the economy expand by taking advantage of pools of underemployed labor and then come to depend on their continued availability. As thoughtfully depicted in Bong Joon-ho’s award-winning 2019 film Parasite, it begins to make sense for high-net-worth and managerial households to hire working-class households to perform more of the tasks they would otherwise do for themselves—as tutors, domestic servants, drivers, childminders, and personal assistants—simply due to large differences in the prices of their respective labors.
These trends suggest that the apocalyptic crisis of labor market dysfunction anticipated by automation theorists will not take place. Instead, unemployment will continue to spike during downturns— as we are seeing happen once again, and on a truly massive scale, in the present COVID-19 recession. Then, in the course of the tepid boom periods that follow, this unemployment will slowly but surely resolve itself into higher levels of underemployment and rising inequality. In Rise of the Robots, futurist Martin Ford says that his worst nightmare would be if the “economic system eventually manages to adapt to the new reality” of labor displacement. But in truth, it has. As Mike Davis put it, the “late-capitalist triage of humanity” has “already taken place.” Unless halted by concerted political action, the coming decades are likely to see more of the same: overcapacity in international markets for agricultural and industrial products will continue to push workers out of those sectors and into services, which will see their share of global employ- ment climb from 50 percent today to 70 or 80 percent by mid century.