The Deindustrialization Myth

In Automation and the Future of Work, Benanav argues that the artificial intelligence (AI) and robotics revolution cannot deliver on its promise of a post-work future because capitalism has been in terminal decline, with deindustrialization holding back growth. For Benanav, the causes of the crisis lie in production, not distribution, because deindustrialization means that the growth engine has stopped firing on all cylinders.

The key to his approach is the question of industrial capacity. He argues that overcapacity reflects a competition-led crisis of profitability and has led to a global lowering of labor demand, largely manifesting in underemployment rather than unemployment. Manufacturing overcapacity, he argues, was a key reason why US firms built global supply chains. This downplays the importance of new technologies and the breakdown of the postwar Bretton Woods system in facilitating this trend.

If we look at the period since the 1970s, taking into account ups and downs due to recessions, capacity utilization rates in the United States return to or exceed the normal engineering rate as defined by the Organisation for Economic Co-operation and Development (OECD) until the year 2000, when there is a longer slump. During the same period, Brazil and China rise to above-average levels, which is unsurprising given the development of those countries. This doesn’t fit with the “long downturn” or “deindustrialization” narratives since the United States largely had already shifted to an economic model predominantly based on services rather than manufacturing by the 2000s.

Benanav’s adjustment to Brenner’s long downturn thesis hinges on the implications of deindustrialization for global capitalism. Yet even in the United States, deindustrialization has not been a negative development in the way that Benanav claims. Manufacturing employment has declined relative to the total working-age population. This is due largely to automation, the creation of new service jobs, and the offshoring of low-value-added processes though global supply chains to countries like China and Mexico. In absolute numbers, however, US industry (including construction) employed 26,644,300 people in 1965 and 27,622,400 in 2022.

The world economy has not deindustrialized in absolute terms, either. World manufacturing value added more than doubled from $6.129 trillion in 1997 to $16.05 trillion in 2021. Outside the United States, Europe, and Japan, most countries had a higher proportion of the workforce employed in manufacturing in 2011 than in 1970.

Has increased global competition in manufacturing caused a slowdown, as Brenner argues? Perhaps this argument would make sense if we were witnessing a great rebalancing between traditional economic powers in Europe and North America and the rising economies of Asia. But rich countries have consolidated control of high-value service activities and intellectual property rights. Apple, for example, has most of its physical hardware manufactured in China through Foxconn, while the service labor of design and development is still based in the United States.

The Nature of Services

Benanav argues that the growth of the service sector is inherently a drag on growth, referring to an old economic theory called the Baumol effect, named after its originator, William Baumol. Simply put, Baumol divides an economy into two sectors: manufacturing, which is technologically progressive, and services, which are technologically stagnant.

For Baumol, this imbalance in technological dynamism drives a shift in resources to less dynamic sectors, since higher productivity activities require progressively less labor to meet demand. The stagnant sector absorbs more labor because the barriers to productivity mean that more workers are required to meet demand in relation to the falling number of workers in the technologically progressive sectors.

However, this theory cannot account for the reality of service-based capitalist accumulation today. First of all, the category “services” lumps together a widely divergent range of sectors such as retail, hospitality, and transport, not to mention professional and personal services from education and health care to finance and real estate. Many of these sectors are productive in the Marxist sense of the term, generating surplus value, and indeed are extremely profitable.

Second, the assumption that manufacturing possesses unique growth-inducing properties and capacity to scale doesn’t really make sense in the face of services based on information and communication technology (ICT), digital automation, and platform scalability. Third, Baumol’s model is outdated. Service workers represent a far greater proportion of the labor force today than they did in the mid-twentieth century when he developed that model. Baumol also uses a narrow set of service occupations as the basis for his theory, like beauticians and elementary school teachers.

Taking the United States as an example, from the 1990s to the 2000s, jobs and productivity growth was mostly driven by ICT, logistics and transport, the retail trade, insurance, finance, and other services. Labor productivity in services grew at a rate of 2.6 percent a year between 1995 and 2001 — the rate for manufacturing was 2.3 percent — accounting for 73 percent of US labor productivity growth. The share of services in US GDP as a whole rose to 80 percent in 2007 from 60 percent in 1947. Service sector employment accounted for more than 83 percent of total employment in 2013, compared to 60 percent in 1947.

Looking beyond rich countries, we can also see shifts to ICT-intensive service exports. In 2014, India was the largest global exporter of ICT services with a value of $74 billion due to both scale and productivity. Total factor productivity in the Indian service sector grew by 2.4 percent from 1980 to 2006 — twice the rate of industry and agriculture.

We should not dismiss service industries as “stagnant sectors” of low-skill, low-wage jobs that are a drag on growth. Many countries around the world are seeing services contribute an increasing share of growth over time. Services may have historically played a small part in capitalist growth, often resisting mechanization as they remained outside the circuit of capital in domestic reproduction. But this is no longer the case today.

Brenner and Benanav ignore Joseph Schumpeter’s crucial insight that the destruction of capitals through crisis tends to open up new areas of innovation and investment, resolving the contradictions that gave rise to a particular crisis and enabling new growth. This is exactly what we have seen in ICT services and now global platforms.

Indeed, service exports have grown at a faster rate than exports of manufactured goods, expanding 5.4 percent per year on average between 2005 and 2017. Services account for nearly half the value added of the international goods and services trade. ICT and platform technologies reduce transaction costs, make exponential scalability possible, and facilitate networked effects, thereby boosting manufacturing productivity. The World Trade Organization predicts that the share of the service sector in global trade could increase by 50 percent by 2040.

Unemployment and Class Power

Benanav’s thesis has clear implications for the plight of labor. He argues that deindustrialization has run down the world economy’s growth engine and that the rise of insecure and low-paid service work indicates a low demand for labor. However, while there has been a downward trend in labor’s share of income since the 1980s for forty-two out of fifty-nine countries, that does not necessarily indicate a lower demand for labor overall. It could simply mean that workers have weak bargaining power.

For Benanav, low labor demand chiefly manifests itself through “underemployment” rather than outright joblessness. Underemployment is an extremely broad term that can refer to anyone who is underutilized, underpaid, or overqualified for their current post. Yet Benanav extends the term even further to include “nonstandard employment.” This refers to anything other than a Monday to Friday, nine-to-five job of the kind that has only ever existed for a small portion of the world’s population.

However, as Kim Moody has pointed out, the actual growth of nonstandard employment has been moderate, with most of that growth occurring before 1985. Part-time employment, whether permanent or temporary, is not a novel phenomenon, and involuntary part-time work accounts for only a third of it. The International Labour Organization actually recorded a reduction in underemployment before the pandemic.

Benanav holds that deindustrialization and underemployment are two aspects of the same phenomenon. This ignores the fact that garment manufacturing, meat processing, construction, and mining continue to be some of the most underpaid, “nonstandard” occupations in the world. This includes high rates of misclassification through bogus “self-employment,” especially in the United States. From 2010 to 2018, temporary worker contracts registered the biggest increases in sectors like food processing, software development, and industrial truck and tractor operators.

It is my contention that the crisis of work has more to do with political choices at the level of institutions than with deindustrialization. A series of neoliberal assaults undermined the collective-bargaining capacity of workers and the threat of strike action, shifting the balance of power from labor to capital.

Benanav would presumably argue that these choices stem from the crisis of profitability in 1970s. But the assault on labor has combined with the deregulation of capital to generate an increasingly volatile world economy, reducing viable prospects for investment. While the problem of “secular stagnation” has become part of the mainstream economic consensus, it may be questions of distribution and bargaining power rather than competition and production that explain the crisis.

As many commentators have pointed out, capitalists are guided by the marginal rate of profit they anticipate in the future, rather than the averages realized in the past. Problems may arise if there is an unexpected acceleration in technological innovation and diffusion, leading to a fall in prices before capitalists have been able to pay off their fixed costs. However, this does not necessarily lead to a terminal decline in the average rate of profit as growth will be faster in more technologically dynamic sectors, which will also attract more investment.

Reversing Labor’s Downturn

What are the political implications of Benanav’s argument? Perhaps it is easier to believe in a terminal crisis of capitalism than endure the system’s persistent revolutions. Benanav views the rise of service work as an inherent drag on productivity growth and seems to extend this limitation to the capacity of workers to collectively organize and make demands for a larger share of the product. The “long downturn” thesis thus appears to hold out the prospect of terminal decline for the workers’ movement.

Yet as Nate Holdren has shown, levels of labor militancy do not correspond to the rate of profit or GDP growth. The 1920s, for example, saw huge GDP growth in the United States, but unions were in decline. The 1930s, by contrast, was a period of economic crisis and recession, yet there was exceptional union growth and militancy. The British economy has seen sluggish growth for decades, yet there have been more strikes in the past two years than at any time since the 1980s.

Capitalists will continue to lower (and steal) the wages of workers to the greatest possible extent — this is what the system incentivizes, whether in times of boom or bust. The only time in history that workers got a fairer share of the product in Europe and North America was during the postwar decades, when workers’ power was at its strongest. It is only by rebuilding that power, building on the progress of recent years, that the long downturn in labor’s fortunes can be reversed.