Brave New World

Forty years ago, in 1984, Michael Piore and Charles Sabel published a MacArthur award-winning book, The Second Industrial Divide: Possibilities for Prosperity, that shaped a decades-long debate about the future of markets and the social orders supporting them.¹ Piore and Sabel characterized the 1970s and early 1980s as a second critical inflection point in the history of capitalism. The first shift had been from craft to mass production in the early 1900s. Mass production matched large fixed investments in product-specific production equipment with semi-skilled workers struggling for long-term employment stability. P&S argued that those investments had created an overly rigid production structure vulnerable to almost any shock. And the 1970s and early 1980s presented a multitude of shocks—strikes, unstable oil prices, rising inflation, intensifying conflict between the US and Soviet Union in terms of nuclear threats and proxy wars, the beginning of deregulation and privatization, and new global public health challenges like the AIDS epidemic. Political struggles to reorganize society created an opening for a watershed in economic relations that gave their book its name: a second industrial divide. They argued that enterprises could and would increasingly shift towards what they termed “flexible specialization”—smaller, more equally sized firms using newer general-purpose equipment, run by a more egalitarian workforce, in a more competitive market—because this form was more resilient in the face of shocks. The rules of the game everywhere were about to be re-written, and their speculative sketch of a possible future won their place in intellectual history. 

The big questions about the re-written rules were, of course, by whom, how, and with what outcomes? Forty years later, we now know P&S were wrong about how the second industrial divide eventually played out. They were right, however, about the fact of a historical divide in the interaction of labor relations and the supply side of the economy, and about the stakes inherent in the massive shift in the organization of production, consumption, and innovation this entailed. Today we face a similar inflection point—a third industrial divide. 

These macro-sociological inflection points are moments of political and social openness, when struggles between workers and firms, among firms themselves, and among various social groups determine everything important: whether or not growth will resume, what the distribution of income will look like, and who will have status and power. Echoing the 1970s and 1980s, inter-imperial rivalry using proxy states, separatist wars, and heightened trade tensions raise fears of great power conflict today. Democratic governance is under siege, secular stagnation and its associated deflation threaten debtors, and global pandemics are no longer science fiction plot points.² Just as with the mass production era P&S analyzed, the normal processes of endogenous decay inherent to an economic growth wave have created a series of overlapping crises that are transforming the world economy and the societies it structures. The growth wave dating to the 1980s, propelled by investments in information and communications technology (ICT) and first-generation recombinant-DNA biotechnology,³ is now decaying. While partly a technical phenomenon, this endogenous decay is even more so an erosion of the social and political basis for that growth wave. Why decay? And what might happen in the next few decades?

The conflicts unleashed by this process of endogenous decay in both the mass production and ICT economies will determine how this third divide plays out. Going back to P&S’s analysis helps understand these processes, not simply because it was an immensely influential book describing the decay of an earlier growth wave and forecasting a new one, but because we can only really understand the full shape of the current growth wave retrospectively—as Hegel said, the Owl of Minerva flies at dusk.⁴ What history has exposed as the flaws in their forecast help us understand where we might ourselves go wrong in analyzing the potential pathways out of what I will argue is the waning ICT and biotech 1.0 era.

The exhaustion of the mass production growth wave in the 1970s

The Second Industrial Divide responded empirically to the evident exhaustion of the mass production era in the 1970s, and intellectually to the flood of books and analyses attempting to explain this exhaustion and advance ameliorative strategies. Those analyses had three main foci. The first, exemplified by two books edited by the British academic John Goldthorpe⁵ and the OECD sponsored McCracken Report,⁶ identified inflation as a manifestation of social disorder and recommended with varying degrees of emphasis tighter monetary policy and a weakening of union power—a kind of unilateral incomes policy. The second, exemplified by MIT Dean Lester Thurow⁷ in the United States and French economist Michel Albert⁸ in Europe, argued for more vigorous state intervention and corporatist bargaining—formally negotiated incomes policies—to bridge the valley between older declining industries and potential new growth industries through targeted investment. And the last, exemplified by labor historian Mike Davis,⁹ identified the profound weakness of the American labor movement and the impossibility of formal negotiations—echoed in the other Anglo-economies and industrializing Asia—as the source of the breakdown of the post-war social compact balancing supply and demand. Finally, though ignored by P&S, Keynesian economist Hyman Minsky¹⁰ was already warning about the tendency for financial systems to generate systemic financial crisis when an era of stability encouraged ever riskier innovation in credit creation.

An exogenous shock—an asteroid—may have caused the Cretaceous-Paleogene dinosaur extinction. But changing industrial eras and the waves of growth that define them, P&S showed, are more like dinosaur die-offs from internal, endogenous dynamics, like over-grazing of foliage or slow-moving mammals. P&S thought the arguments advanced by the “usual suspects” above viewed raw materials shocks and the 1960s expansion of the welfare state too much like asteroids. They had missed the degree to which the technological imperatives of mass production created the potential for an unstable economy. Mass production, in their view, was neither more efficient than other production systems nor an inevitable outcome of technological progress. Politics mattered in the choice of production systems and thus in the shape and stability of the economy: it was the rigidity of 1960s-style mass production in the face of shocks like rising oil prices or wildcat strikes, as much as the shocks themselves, that were the major cause for the crisis and the coming industrial divide. Just as politics had shaped the outcome of the first industrial divide, they argued, politics in the 1980s would condition the way industrial societies adapted to their new technological possibilities. 

Those possibilities were important to understand because of the exhaustion of the mass production growth wave—that is, the era of investment in corporate-organized semi-skilled workers running expensive, long-lived, and specialized machinery. Mass production was highly productive, and thus potentially highly profitable, if—and this is the critical “if”—that expensive, long-lived, and specialized machinery could be run at something close to full capacity. But specialized machinery and the need to run at full capacity created enormous rigidities. As Thorstein Veblen¹¹ pointed out eight decades before P&S, the need for constant and consistent inputs to feed those specialized machines required application of what he called the “machine process” to nearly all parts of the economy: not only uniformity of materials, standardization of parts, synchronization of delivery times, and precision in machinery, but also assured demand through marketing and advertising. As Aldous Huxley¹² satirized in Brave New World, the machine process could extend to the production of workers themselves: as differentiated but internally standardized groups of selectively bred humans, even if the post-World War 2 compromises trading wage increases and employment security for unilateral managerial control (what academics typically stylize as “Fordism”) made those workers well-paid. 

With everything so tightly linked, any internal or external shock could send the mass production economy into a deflationary (1930s) or inflationary (late 1960s-early 1970s) spiral that rapidly propagated through national economies and thence the global economy. P&S thus argued that the 1970s crisis opened a space, a potential second divide, for a return to flexible specialization, a more stable, less rigid, and more humane form of production rooted in the earlier craft tradition. Fordism’s huge vertically integrated firms had become too clumsy to compete successfully against smaller, nimbler, vertically disintegrated craft-based firms. The networks of smaller, less hierarchical firms typical of Italy’s Emilia-Romagna region were P&S’s ideal model for a better future: in those industrial districts, firms often changed places between prime and sub-contractor, workers circulated across firms as they needed different skill sets, and firms essentially self-financed. By using general purpose machinery, the specialized firms around Bologna, and those who adopted their organization and strategy, could adapt to whatever the market demanded. 

P&S thus sought to inject new ideas into those policy debates around inflation, unions, and the state’s role in the economy. They saw only two paths out of the problems animating the debate. One path was what they saw as an improbable expansion of mass production and thus Keynesian policy of demand management to the global level. The other was a transformation of the industrial base in the direction of flexible specialization, which they thought would help reestablish cooperation between capital and labor, allowing for a more communitarian politics capable of taming inflation. Though P&S heeded Yogi Berra’s famous advice about prediction,¹³ they clearly favored an economy centered on those flexible industrial networks. A flood of academic and business publications identifying and lauding Japanese and German industrial districts as similar congeries of flexible firms followed: industrial districts normatively and empirically seemed to be the way to go.¹⁴

The reality of the ICT and biotech growth wave and the inadequacy of growth models

Alas, the next forty years did not confirm their preferred outcome. Writing in the early 1980s, P&S were fairly pessimistic about a new growth wave and could barely imagine the rapidity of subsequent technological change. Computers feature only briefly in their account; biotechnology not at all. Industrial districts in their sense remained important—but constituted a narrow slice of industrial output. The Pearl River Delta, upstream from Hong Kong, became arguably the biggest and most important industrial district in the world. Elsewhere, most rich country economies increasingly revolved around service provision rather than manufacturing, much of which became the province of low-wage, highly exploited labor. Even the poster child region for flexible specialization, Emilia-Romagna, increasingly relied on cheap immigrant labor to remain competitive. Italy’s economy as a whole saw no increase in per capita income after 2000.¹⁵

While the unstable 1970s and 1980s eventually gave way to the great moderation of the 1990s, GDP growth in the largest economies slowed in each decade after 1992. The new growth wave, which started around the time P&S were writing, built upon ICT and biotechnology 1.0 and arguably lasted until the 2008–2010 financial crisis. As this ICT and biotech 1.0 wave reached its financial, physical, and social limits, rich country economies in the decade before Covid-19 experienced the growth slowdown contemporary analysts called “secular stagnation.”¹⁶ Global smart phone sales plateaued in 2017; privacy, geopolitical, and psychological concerns limited continued growth of social media; semiconductor chip manufacturing technologies approached physical limits in terms of shrinking transistor size but more importantly faced rapidly rising capital costs; biosimilar pharmaceuticals proved outrageously expensive for consumers and public health authorities; climate change called into question the entire fossil fuel- and petrochemical-based systems powering energy, agriculture, and packaging. 

Perhaps the most comprehensive attempt to examine this decline came in the growth model literature, which sought to move away from equilibrium-style analyses of national industrial structures and from excessively supply side-oriented analyses back to a neo-Keynesian focus on the sources of demand, albeit at the national level.¹⁷ Explicitly political, these analyses trace the origins of production structures back to fundamental compromises among powerful actors. The growth model literature, though, lost sight of two key features in P&S analysis: Where P&S were explicitly Schumpeterian in their focus on innovation and organizational change, the growth model literature has tended to focus on decomposing economic aggregates to see the sources of demand, neglecting innovation and corporate organizational change. Second, the growth model literature is typically methodologically nationalistic, focusing on domestic patterns in isolation from the broader global economy.¹⁸ Consequently the growth model literature struggles to explain change or the arrival of inflection points. 

What, then, can we learn from P&S about the current inflection point in the economy? As with the mass production wave, the decay of this wave emerged endogenously from the very processes that initially drove growth, from the very success of ICT technology as an economic and social phenomenon.

Five critical errors in forecasting the ICT and biotech growth wave

Five critical errors of omission in P&S’s analysis clouded their predictive ability. The first omission involves their understanding of distributional conflicts in capitalism. Veblen’s classic anthropology of business enterprise clarifies this. Capitalism has two core distributional conflicts: the first is between capital and labor—firms and their workers—over the respective share of profit and wages in total output; the second and equally important is between firms over shares of the subsequent pool of profits. P&S attended closely to the first conflict, matching their analysis of production structures—the supply side of the economy—with an analysis of “labor relations”—the income distribution and thus demand side of the economy. But they largely ignored the second. 

Firms strive for monopoly profits, as Veblen and Schumpeter argued. But as the evidence about profits and public equity returns shows, few actually attain them.¹⁹ The dominant mass production firms had pursued monopoly and oligopoly profit by controlling large volumes of physical capital and integrating vertically. This helped to deter market entry by potential rivals and enabled direct control over the entire value chain. After the divide, firms pursued monopoly profit through possession of intangible assets protected by intellectual property rights (IPRs—patent, copyright, brand, trademark), a strategy that enabled them to shed tangible assets and labor. IPRs create legally protected monopolies without the risks and obligations of an employer, and without the risk of under-utilized physical capital. P&S did not anticipate this change in how firms pursued monopoly profit.

The second omission ironically concerns how the distributional conflict among firms drove the vertical disintegration that P&S foresaw and desired—but created vertical inequalities in power and income rather than an egalitarianism of horizontal competition. As Alfred Chandler²⁰ argued, profit strategy determines organizational structure. P&S correctly foresaw that firms could and would shrink themselves via outsourcing and demerger to reduce the rigidity accompanying mass production. The IPR-based profit strategy motivated lead firms to shift rigidities and risks downward onto their subcontractors and to suppress the independence of satellite firms in their value chains and industrial districts. Even the iconic Fordist-era automobile firms spun component production out into independent firms and began subcontracting some assembly operations. From a legal point of view, this vertical disintegration created the kind of specialized producers P&S favored. But subordinate firms’ de jure independence concealed lead firms’ continuing de facto control. Rather than the fluid hierarchies P&S envisioned, with lead and subcontractor firms exchanging position as needed, both older Fordist firms and newly emerging technology firms constructed sets of dependent subcontractors who competed against each other. Meanwhile, lead firms’ possession of monopoly-generating IPRs enabled them to capture the lion’s share of profits from those hierarchies.²¹

The third omission concerns the sources of innovation. P&S’s networked Italian firms were highly innovative with respect to variation in their existing line of products. They reliably generated novel shoe, tile, and clothing designs. But as Schumpeter said, “Add successively as many mail coaches as you please, you will never get a railway thereby.”²² One might similarly say, link as many flexible firms together as you please, and you will never design and build an airplane or semiconductor fab or a telecommunication switching server. P&S focused overmuch on what Schumpeter scholars call “Mark 1”-type innovation: heroic (and mythical in almost all senses) entrepreneurs toiling away in their (rich parents’ four car) garages and developing something entirely novel. But they ignored the systematic corporate “Mark 2”-type innovation that necessarily occurred inside a hierarchical management structure capable of contributing to and preserving the vast pool of human and operational knowledge—basic and applied science— behind the generation and production of complex products.²³ Though some segments of the software sector were Mark 1-type firms, and many large Mark 2 firms generated only incremental advances, most innovations constituting the new technology sectors emerged from Mark 2 innovation done by large firms, or captured by them from smaller subcontractor firms. The very kind of relationships P&S saw withering away in the second industrial divide defines modern supply chains, which are typically networks of subordinate firms dominated—and exploited—by lead firms that own IPRs. 

Fourth, P&S overlooked the inherently financial aspect of capitalist economies. Firms exist to make profits, but they need credit to operate, and the creation of varied financial assets in which to park cash profits. As Carlota Perez notes, the huge infrastructural development behind Schumpeterian growth waves requires equally huge financial investment.²⁴ But P&S had no way to understand how financial “innovation”—read, the predation and short-termism labeled the “shareholder value model”—would reshape corporate organization and affect the two central distributional conflicts in capitalism. Financial pressure embodied in the shareholder value model magnified the shift to a de jure vertically disintegrated, but de facto still-integrated, organizational structure that concentrated profit into a handful of firms through captive supplier markets.

Finally, P&S overlooked geopolitical competition and conflict, which incentivized most of the radical innovation that Mark 2 firms undertook. Historically, security reasons drive states to promote radical innovation and, equally important, the rollout of new infrastructure. DARPA’s role in funding what became the internet in the United States is well known, but merely echoes the roll-out of nineteenth century railroad systems, early- to mid-twentieth century electrification, and product standardization,²⁵ and mid- to late-twentieth century telecommunication. Likewise, security or actual war-fighting concerns drove much research in pharmaceuticals, electronics, and new (but now old) materials.²⁶ Put simply, war is the mother of almost everything related to basic research and the early stages of radical innovation, even if civilians ultimately field that technology. P&S missed the enormous US government promotion of ICT technology in the late 1970s and 1980s; that technology enabled private firms to develop the new “franchise“ organizational structures characterizing the ICT-plus-biotech 1.0 growth wave.²⁷

Explaining decay

These five errors of omission also indicate the endogenous dynamics that eventually crippled that growth wave, yielding the period of decay in which we are now living. Rather than achieving a new stability of horizontal competition, the turn toward the market brought about a vertical domination of brand-name intellectual property claims and captive supply chains. In any given sector, a handful of firms captured the bulk of profit. Thus, in the broad computer sector (NACE code 26), the standard gini inequality index for cumulative pre-tax profits from 2010 to 2019 was 0.907, well above the gini index for household income in the United States (.48) let alone Sweden (.29), because the top 20 out of 6511 firms captured over half of all profit. Pharmaceuticals (NACE code 21, including 2677 firms) similarly was .895 and 63 percent; the broad automobile sector (NACE code 29, 1190 firms) was .921 and 67 percent.²⁸

Two new general-purpose technologies powered the ICT wave. Ever cheaper computing meant ever more opportunities to deploy software in production and communications, which reduced some of the rigidity of labor and inventory costs in mass production. More flexible automation and better information about consumer desires allowed firms to move away from producing potentially unsellable inventory ahead of demand to something approximating a “sell one, make one,” on-demand model, in which lead firms directed subcontractors to make something only after a customer committed cash. Digitalization cut transportation and transaction costs for many services (no need to go to the bank or a travel agent, for example). While biotech 1.0 had a relatively smaller economic impact, recombinant DNA technologies allowed pharmaceutical companies to move away from exhausted discovery processes and small molecule drugs towards more precise diagnostics and larger biological drugs, and, significantly, manufactured human insulin. Biotech 1.0 also allowed the engineering of a range of agricultural products with better nutritive features, pest resistance, and yield (though it also sparked social resistance against “frankenfoods”). 

These two new general-purpose technologies powered a wave of investment, and thus growth, in the 1990s and early 2000s. Yet they also set in motion dynamics of decay corresponding to the strategy, organizational structure, innovation, geopolitical, and financial issues raised above. Certainly technological or physical limits to the ICT and biotech 1.0 wave existed. By 2014, semiconductor production technologies hit an apparent limit as the elements (“gates” or “nodes”) on a semiconductor chip became so small that sub-atomic effects potentially disrupted their function. Efforts to overcome this limit by shifting from two-dimensional to a more three-dimensional chip architecture have stretched out the historic doubling of transistor density every two years (Moore’s law). Worse, the shift to three-dimensional gates reversed the historic decrease in gate cost. The cost of producing a transistor gate began rising after 2012, reflecting a jump in the cost of building a cutting-edge semiconductor fabrication facility from roughly $5 billion in the early 2000s to $20 billion by 2020.²⁹ As smartphones absorb about one-third of global semiconductor output, and within that a much larger share of cutting-edge output, the peak of smartphone sales six years ago poses a potential damper on demand for semiconductors. 

But the endogenously emerging organizational and social limits were more profound than the technological ones. While biotech 1.0 yielded a range of sophisticated and effective pharmaceuticals, these largely sold at prices threatening fiscal stability for public health systems (including that in the United States, where public budgets finance roughly half of health care directly, or two-thirds when tax-expenditures are included). Government budget politics today reflect this impasse. From the point of view of equity markets, the increase in profits and thus return on a much smaller pile of physical assets justified rising stock market capitalization for those factory-less lead firms. Meanwhile firms doing actual production were thrown into a Hobbesian competition with each other as lead firms threatened to end contracts unless those subordinate firms delivered annual price reductions, or, in the case of grocery stores, payment for advantageous product placement. The winners spun off labor-intensive and physical capital-intensive production to newly formed firms, reinventing themselves as factory-less owners of intellectual property. Offshoring and outsourcing production both accessed and created cheaper labor. Subordinate firms tried to depress wages as much as possible to remain profitable, aggravating income inequality and thus slowing economic growth, or pursuing horizontal mergers to regain some market power. These individually rational firm choices meant that firms on the whole won the vertical distributional conflict against labor, and undermined prosperity in general. 

Lead firms’ victory in the distributional conflict among firms for shares of the profit pool also set a decay process in motion around innovation and investment. Lead firms had a low marginal propensity to invest in actual productive processes, which meant their profits were available for reinvestment. Some went into the normal passive vehicles—if they were a country, Microsoft and Apple’s collective $112.2 billion in US government bonds and $92.6 billion in corporate bonds in 2023 would make them the thirteenth largest foreign bondholder on the US government and corporations, ahead of all Korean, German, or Norwegian holders of fixed income debt.  Similarly, the logic of maximizing profits through an asset light strategy and intangibles generated what Cory Doctorow has called the “enshittification” of the app-economy narrowly and the internet more generally.³⁰ Once useful products become increasingly dysfunctional via search engine optimization strategies and ceaseless efforts to maximize attention, click-through rates, and purchases.

Signs of exhaustion

As a social process, the transformation from vertically integrated firms to the new de jure disintegrated structure played out slowly. Unions resisted layoffs, management had to master contracting and the new business process software, and states had to build the physical infrastructure to support offshoring. But the wage share of GDP declined, and with it demand relative to supply. Profits came in as cash but had to be parked in some new asset: the rising profit share thus had rising household debt as its necessary counterpart.³¹ But only rising incomes could sustain the rising debt powering growth. Absent that, a crash was inevitable—and one came in 2008. With central banks and politicians largely wedded to fears of inflation, most rich country governments opted for tepid fiscal responses to that crash, creating a lost decade in terms of growth, and powering an intensifying populist politics everywhere today.

Meanwhile, the bulk of profits at the peak of the last growth wave flowed into venture capital and private equity (PE) firms that favored investments with a low labor headcount and asset light profile.³² Indeed, PE is apotheosis of the model: purely financial firms that own but do not generate intellectual property or physical innovation. PE’s share of the US economy is relatively small compared to publicly listed firms.³³ The 3640 public firms employed 42 million people in 2019, while PE firms controlled 7200 firms employing only 5.4 million workers. But those numbers reveal that PE typically controls the smaller firms that drive Mark 1 innovation and grow into the new large firms doing Mark 2 innovation. Finally, PE, like all financial strategies, has also suffered from endogenous decay as more and more investors engaged in and funded PE-type buyouts, harvesting the low hanging fruit. From 1984 to 2015, a weighted average of PE firms’ returns was only about 10 percent higher than that of the S&P 500, and VC funds only outperformed during the 1990s. By the 2010s, PE firms were returning less than simply investing in an S&P 500 index fund: 13.8 percent versus 14 percent.³⁴

Finally, the ICT wave also exhausted itself by changing the geography of production and thus the balance of geopolitical power. The organizational shift to a three-layer production chain—IP owners, capital-intensive firms in markets with high-barriers to entry, and labor-intensive firms in markets with low-barriers to entry—moved significant slices of physical production out of rich countries, especially the United States, towards developing economies, especially China. By 2023, firms in China generated 35 percent of global manufacturing output and 29 percent of value added, versus only 12 percent and 16 percent respectively in the United States. Put differently, China’s share of global manufacturing output was double its share of global GDP while the US share of global GDP was half of its share of global profits. Value added is related to profitability, so the inversion of output and value added here reflects the lingering dominance of US lead firms relative to Chinese subcontractors. But wars are won with actual production, not with profits—you can’t DoorDash munitions. By the early 2020s, rich country governments and particularly the United States responded with a wide range of industrial policies and subsidies intended to mitigate or reverse dependence on China, undoing the globalization trends of the ICT era.

So, what’s next?

In principle, the decay of the ICT and biotech 1.0 wave should incentivize a new growth wave. The physical basis for a new wave is already visible. Solar and wind generated electricity, plus geothermal heating, are already often cheaper than and thus beginning to replace fossil fuels, mitigating the global CO2 load. AI-enhanced bio-genomics combined with CRISPR-Cas and other second generation, precision biotechnologies, along with AI-assisted additive manufacturing processes and new, non-biological, non-petrochemical based materials provide new general-purpose technologies. (Generative AI, though, is mostly a sideshow.) Electric vehicles, electric heating, and new materials like bioengineered spider silk (for textiles) are new consumer goods.

But everything that is important remains hidden in the fog and mist of the future. Those are the social and political struggles shaping new forms of production, corporate organization, and demand creation. Here we can have a moment of sympathy for P&S’s misreading the direction of change. While the important drivers of change are more visible, the probable outcomes hinge on local and geopolitical struggles. Those geopolitical struggles are inducing a re-verticalization of some production, particularly in the crucial EV space. And at the margin, in combination with local struggle and protest, geopolitical concerns have motivated some state efforts at expanding social policy and social protection again, potentially reprising the “home front” policies of the first Cold War. Finally, the slow-moving demographic collapse, which was partly driven by weak wage growth and rising debt, has helped shift labor markets back in favor of workers and rekindled interest in private sector unionization in the United States. 

But if P&S were correct about only one thing, it was the utter indeterminacy of any given divide. Yogi Berra, the Prophet of Prudence, suggests waiting a few decades to see what the Owl of Minerva says. But in the meantime, the openness of the future suggests organizing to shape that future. That future will not look like P&S’s ideal and idealized flexible specialization. But it also need not be a continuation of giant monopolies harvesting personal data while providing services of dubious value and goods produced with exploited offshore labor.