Varieties of overcapacity
Chinese overcapacity, MITI-era Japan, and notes on a capital theory of industrial policy
A year ago, I panned a piece by Varun Sivaram inaugurating the Council on Foreign Relations’ new “Climate Realism Initiative.” His essay was full of ham-fisted and slightly horrifyingly belligerent suggestions for how the U.S. startup ecosystem could innovate its way out of dependence on Chinese clean technology supply chains. I stand by everything I said. Thankfully, nobody talks much about the essay anymore.1 But its obsession with competing against China only reflects the sentiments that have long been percolating through the DC policymaking ecosystem.2
Quite a few of my peers here are now focused on gaming out what kind of industrial policy program the United States needs to parlay inbound Chinese investment (if it arrives) into sectors like EVs and batteries into a broader American industrial revival—one that, ideally, delivers for the American consumer in ways that neither Trump nor Biden did. Many of us in DC working on industrial policy, me included, are overawed by the scale of Chinese production. The city’s event circuit continuously stresses the differences between U.S. and Chinese capitalism.
I think what really exercises this loose “industrial policy community” is the promise and specter of Chinese “overcapacity”: at once an economic dislocation when viewed against orthodox theory and at the same time an object of extreme envy when viewed against the sclerotic American industrial landscape. We can say, truthfully, that we want to see the technological upgrading of America’s industries, particularly in clean technologies. But I think that we also just want a politics of “more.” In the spirit of Klein and Thompson’s Abundance and Dan Wang’s Breakneck (another qualified “China envy” kind of book), many of us have made the judgment that letting the “supply-side” of the economy rip is a necessary path toward affordability, abundance, decarbonization, reindustrialization—choose your catch-all vision.
So I think we’d all do well to read Jeremy Wallace’s latest essay in the BREAK—DOWN, “Capacity Returns.”3 If we’re going to have these collective psychic neuroses about overcapacity, then we might as well learn some more about its political economy. Jeremy has already written a great piece earlier this year in WIRED about how Chinese overcapacity is not a centrally planned affair. In his own words, it’s a mess. But “Capacity Returns” takes a step back to discuss the macroeconomic purposes of overcapacity itself—and what it reveals about capitalism.
I’ve been spending lots of time these past few months going down various rabbit holes concerning the political economy of innovation. Like Chinese polysilicon manufacturing, I am not sure if these research junkets will be particularly profitable—certainly not in the short term. But, reading Jeremy’s piece, I have a few observations about what Chinese industrial policy could teach U.S. policymakers about the macroeconomics of large and long-duration capital expenditure.
I also recently finished Chalmers Johnson’s MITI and the Japanese Miracle (1982), which, beyond being immensely interesting, stands at odds with the Chinese predilection toward “messy” overcapacity; wherever possible, MITI bureaucrats in the 1950s and 1960s kept trying to consolidate Japan’s industrial sectors to maximize their capital efficiency and arm them to compete in the global economy—even if the end result felt like “overcapacity” to the rest of the world, and the United States in particular. The apparent contrasts between these two systems hint at a broader capital theory of industrial policy centered on the role of finance and demand.
“Capacity Returns” states upfront that excess capacity across the Chinese industrial system is a quality that broadly affords resilience, albeit at the cost of profit maximization. The reasons why are fairly obvious, and only more so in light of the U.S. war on Iran. The option value of possessing spare industrial capacity is immensely “in the money” during a shortage of primary goods that will inevitably throttle the construction of new industrial capacity—as many observers of the U.S. energy market might tell you today in the context of breakneck artificial intelligence-buoyed demand growth.
Jeremy compares the Chinese frenzy for infrastructure capex, fueled considerably by loans from local governments, to its American technology counterpart:
An interesting counterexample [of the American failure to “adequately invest in capital-intensive manufacturing and infrastructure”] is the current AI investment cycle, where hyperprofitable tech firms have been placing bets by the billion on data centres and, in doing so, shedding their prior capital-light status and making huge decisions about the economy’s future. But this is an exception, however large, for American capitalism.
This exception is so large as to suggest its own rule: Hyperscale tech companies are spending so much that they not only can engage in indicative planning—the promise of their demand has prompted all kinds of firms, from General Motors to AllBirds, to pivot into providing power and GPU solutions to tech companies—but, in doing so, can fry any sense of prudent risk management among their counterparties. The way that American stock market froth essentially provides free equity capital to firms pivoting toward the American five-year plan seems similar, from a capital structure perspective, to the way Chinese local governments exercise an “extend and pretend” mentality with respect to their many loans to private entrepreneurial manufacturing firms.4
The difference, of course, is that China’s economy lets supply rip in ways that the U.S. economy doesn’t. There’s a vicious cycle here where prospective demand generated via an indicative plan actually does generate supply:
But while we credit these entrepreneurs for their shark-like brilliance at sensing opportunities, many are the fish in the sea: in almost every case, multiple firms see the dynamics at play and sense the same opportunities; supply then turns out to be greater than demand, and the result is a decline in prices. The story then shifts: the decline in prices leads to an upsurge in demand as firms and individuals sense new opportunities to take advantage of the low-cost offering. The demand has thus been created to match the new supply, albeit often at a lower price.
As evidenced by severe downward price pressures, this supply response seems disproportionate to the demand pull. This is where the Chinese overcapacity story gets “messy.” Local governments are all pushing capital into similar industries. Private Chinese firms in industries like polysilicon have tried to cartelize and consolidate so that they don’t fall over their own momentum. But the Chinese central government did not tolerate this attempt at capital discipline:
What had changed was the realization on the part of the Chinese government that, while it doesn't want to lose polysilicon, it has many more firms, jobs and bets tied up with the industry's consumers than in its manufacturers. Monopoly prices may mean profits for the base of the supply chain, but they would further squeeze the huge and powerful firms closer to the consumer end.
Monopoly is one way to interpret the polysilicon sector’s attempt at cartelization; Veblen’s notion of “sabotage” is another. But I’m also reminded of the story of offshore wind in Europe and North America post-pandemic, when input margins rose and “de-ranged” price expectations across two continents’ supply chains, forcing offshore wind majors to cancel projects and write down billions of dollars worth of investments. The reasons there are immensely different (I recently learned from an expert in the sector—I cannot remember ever hearing this reason specifically before—that immense demand for steel to supply Ukraine’s war effort cut directly into offshore wind majors’ steel procurement plans), but I suspect that the effects, applied to the Chinese case, would be similar: Capital discipline is both an opportunity for polysilicon firms to collect oligopoly rents and a threat to the coherence of all downstream supply chains dependent on them.
This notion of supply chain coherence is of utmost importance to an industrial economy pushing the technological frontier, in the very Galbraithian sense that de-ranged primary goods prices destroy investment planning and (a la Kalecki) clogs up the channels through which demand calls forth investment in supply. (Inayama Yoshihiro, the former President of Nippon Steel—whose bureaucratic counterparts we’ll discuss subsequently—did not call steel “the rice of industry” for nothing.) From the perspective of capital, local governments’ cut-rate loans might seem unprofitable (a financier might argue they have a negative net interest margin), but putting downward pressure on generalized price inflation actually helps preserve the present value of those loans, too, so the accounting here could be worse.
Jeremy also argues that Chinese firms’ inability to compete on prices—which, for the most part, ratchet downward—forces them to compete for quality and for future demand. “Process innovation” thus becomes an existential requirement for Chinese industry, insofar as the only way they can amortize their overcapacity “overhang” is to seize new or augmented markets; new revenue streams thus help roll over old debts. This is where the reliance benefit of overcapacity intersects with the innovation benefits:
while the end product manufacturers suffer from overcapacity, so do their suppliers and machine tool makers. Having interlocked industrial processes with similar dynamics can lead to a fractal-like intensification.
It’s easier to keep capital assets at work longer. It’s also easy to retool that capacity to test new ideas and build new product lines. But that only works with immense public and consumer demand churning through the economy.
As colleagues and I like to remind our fellow American policy wonks often, overcapacity is really only relative to demand. The Keynesian solution to this problem is to manage demand rather than to throttle supply.
The Japanese industrial policy bureaucrats of the 50s and 60s, in particular, had fundamentally different incentives structuring their work. First and foremost, they worked chiefly at the national level, rather than the provincial level. They were not interested in interregional competition; what mattered was national export performance and, as such, international competition. Second, international competition was, at least initially, one of catch-up growth and the maximization of productive capacity in globally mature industries such as steel and automobiles rather than innovation in a new sector—although innovation did follow. Third, bureaucrats at times exercised what amounted to authoritarian controls over industrial investment due to persistent postwar shortages of foreign exchange. MITI, at one point, controlled the foreign exchange budgets of large industrial conglomerates and conditioned its disbursement on industrial investment planning, therefore managing the import of capital goods into Japan with precision and, often, prejudice.
The bureaucrats at MITI, Japan’s old Ministry of International Trade and Industry, therefore had very different ideas about industrial management. They did not tolerate “messy” overcapacity whatsoever, and in fact were often wringing their hands over how to attenuate price wars.
The stylized facts I’m presenting all come from Chalmers Johnson’s incredible MITI and the Japanese Miracle: The Rise of Industrial Policy 1925-1975 (1982). The book ends with the 1970s oil shocks; as such, it has nothing to say about the Japanese financial crisis and sharp economic slowdown. But Johnson offers an exhaustive bureaucratic history of how MITI was built and how it amassed an armamentarium of draconian planning tools. The book goes perhaps too far to claim that the history itself proves that MITI is prima facie responsible for the Japanese growth miracle—and was clearly written with the U.S.-Japan economic rivalry in mind5—but the history is enough of a foundation from which I can offer some more thoughts on a capital theory of industrial policy.
Japan’s antitrust policy was oriented “to developmental and international competitive goals rather than strictly to the maintenance of domestic competition.” In other words, Japan did not really have the kind of antitrust policy that Americans would recognize:
[MITI] has always been hostile to American-style price competition and antitrust legislation. [Shigeru Sahashi, a notable former MITI leader,] likes to quote Schumpeter to the effect that the competition that really counts in capitalist systems is not measured by profit margins but by the development of new commodities, new technologies, new sources of supply, and new types of organizations.
MITI regularly worked around the Antimonopoly Law (a postwar gift from the Americans) to cartelize the Japanese economy—including its cotton, rubber, and steel sectors—up to a point where it could compete most effectively against foreign firms. The Important Industries Control Law, from 1931, legalized “treaty-like cartel arrangements” among firms to fix production levels, set prices, limit entrants, and share control over product marketing. In 1937, some bureaucrats argued that small enterprises did not contribute much to Japan’s export potential. They soaked up labor but did not earn foreign exchange for export industries. Better, in their view, to fold them up into larger productive units or for them to subcontract for large exporting enterprises. (The existing industrial cartels liked this idea a lot.) So MITI regularly engaged in “industrial rationalization” and “enterprise readjustment” to forcibly merge small producers. By 1980, there was “extensive subcontracting between large, well-financed final assemblers and innumerable small, poorly financed machine shops.”
As early as 1953, MITI’s Industrial Rationalization Council started forcibly assigning firms to trading companies (exporters) if those firms were not already affiliated with one through a keiretsu (Japan’s oligopoly of conglomerate holding companies). “Through its licensing powers and ability to supply preferential financing, MITI ultimately winnowed about 2,800 trading companies that existed after the occupation down to around 20 big ones, each serving a bank keiretsu or a cartel of smaller producers.”
Japanese politicians and bureaucrats were sometimes shockingly cold to the fate of small businesses in their drive toward heavy industrialization.6 But they always offered cheap financing for mergers, acquisitions, and rollups, for small companies and large ones alike:
During fiscal 1963 the Japan Development Bank set aside some ¥3 billion (enlarged to ¥6 billion during 1964) for “structural credit” loans to large firms that merged. The government had long used easy financing to encourage mergers among medium and smaller enterprises; now it extended such funds to the automobile, petrochemical, and alloy steel industries. In the merger of the Nissan and Prince automobile companies, which was finally consummated on August 1, 1966, Nissan is alleged to have received a reward in the form of an $11.1 million loan from the [Japanese Development Bank]. The government justified this kind of largesse as part of its export promotion policies, since economies of scale lowered the prices of export products.
In 1971, to head off a dispute with the Nixon Administration concerning Japanese textile overcapacity, MITI set up a ¥200 billion “relief program” for the sector, which included “governmental purchase of surplus machines, compensation for losses in exports, and long-term low-interest loans for ‘production adjustment’ and occupational change.” MITI was clearly empowered to use its financial firepower to clean up “messy” overcapacity when it needed to.
Very obviously, this isn’t like the China story at all. While the Chinese government seems reluctant to embrace cartelization, MITI bureaucrats thought rollups were all the rage. There are, of course, an abundance of historically contingent reasons why the two strategies diverge. But my first-cut capital theory of the case is that Japan was competing internationally to seize existing markets for textiles, petrochemicals, steel, automobiles, and the like—whereas, in the case of Chinese clean energy technology producers, particularly of solar and EVs, Chinese companies do not seem to have significant international competition whatsoever in these new markets. They are competing for leadership within China just as much, if not moreso, as for global leadership.
Japan’s national champions, the keiretsu, competed against one another, too, but Johnson gives no indication that any of them ever felt like the others could wipe them out at any point, save for a forced merger. So it’s clear that Chinese policymakers are taking a very different route toward the achievement of production economies of scale that are still conducive not just to downward pressure on input costs and consumer prices but to the ability to disseminate innovative technologies quickly. This route is not entirely intentional: There’s lots of literature about how local government officials’ careers are built on supporting new industries,7 and the political economy of “involution” seems to constrain the ability of local officials and planners to do anything about zombie firms, even if they’d want to.
Jeremy notes in a post earlier this year that, while Chinese planners may continue to resist cartelization, it’s still possible for leading clean technology firms to eat each other up piecemeal:
Will we see consolidation not via cartel but instead by sheer dominance of the main players? Will Tongwei and GCL try to buy up their peers until we have just 3 or 4 players who can sit around the table and agree that working together, silently perhaps this time, is a better strategy? Chinese hypercapitalism ends not via cartel but by oligopoly?
So the jury’s still out with respect to the role that competition policy will play in Chinese clean technology development.
Either way, what MITI bureaucrats and Chinese local governments share is a commitment to trigger-happy financing for the firms and sectors they care about. MITI had various forms of incredibly concessional capital tools that nearly eliminated default risks for borrowers—so long as those borrowers responded to indicative planning and administrative guidance.8
It’s easy to suggest here that the American wonk crowd, in its supply-side neuroses for “more,” ought to embrace the financing cannon. This is correct, and, of course, I’m a big fan, but it does not go far enough. My reading of Johnson and my understanding of Chinese clean technology innovation suggest that there is no way to commercialize technological innovation and suppress domestic price competition at the same time without dredging a demand pathway which forces firms to compete on quality rather than on price.
Japan had its export bias, of course, which forced it to learn how to market its goods abroad—but, in the postwar decades, it also had a rapidly growing domestic market. MITI ministers put the two together to drive economies of scale, which only got better as domestic demand boosted capital efficiency, reduced marginal costs, and provided for quicker dissemination of new technology.9 And China is not just the world’s preeminent supplier of the rest of the world’s manufactured goods, but has one of the world’s biggest consumer markets (which attracted numerous American companies in the 90s and aughts). So it’s in a similar position with respect to clean technology, even if consumer demand is apparently suppressed and industrial investment has to be stoked through indicative planning.
American policy wonks trying to make good on their overcapacity envy here at home would do well to remember that the kind of technologically advanced industrial production demands a domestic demand channel. More recent literature on globalization and production networks (Argyrous 1993 and Argyrous 2000 come to mind) suggests that industrial production networks of capital goods do not emerge without major sources of domestic demand—but, once domestic demand is established, those networks end up posting better export performance, too. There’s nothing stopping American policymakers from taking this approach—save for a lopsided economy, rising costs for basic services like healthcare, and price inflation of key primary goods, all of which clog up the United States’ demand channels.
And stalking behind this strategic approach is a counterintuitive conclusion. If demand for Chinese clean technology goods continues to increase, in China and abroad, then China’s capacity to push out the frontiers of clean technology innovation and commercialization might only accelerate, too.
Even Sivaram has moved on. He now works on Emerald AI and has not, to my knowledge, published with the Climate Realism Initiative since.
The ecosystem has sorted itself into, as best as I can see it, three general camps, each representing a broader strategic orientation toward U.S.-China investment diplomacy.
The first camp is full of Trump Republicans, who, at least publicly, seem more open to Chinese investment into the U.S. (Even if their bluster comes to nothing, many of them are in China right now.) The second camp seems to be full of Democrats embracing protectionism, whether out of jingoism or out of concern for the American auto sector and its labor unions (and their electoral college-advantaged votes). The third camp is a bunch of centrists and center-left types, mostly wonks, trying to split the difference and find viable pathways for ensuring that inbound Chinese clean technology investments help upgrade the U.S. industrial base and support American industrial development with superior technology. Many of these wonks were connected to the Biden Administration’s industrial policy wings. (I should say: This third camp—which, currently being disconnected from supportive political leadership, is maaaybe more of a loose grouping—is the camp I find myself in by virtue of my work.)
Highly recommend the whole publication, of course. I’m not biased at all…
Indicative planning relies on risk-less capital access—but, to be clear, that doesn’t mean that production itself is risk-less.
Johnson published MITI and the Japanese Miracle in 1982, right when the U.S.-Japan economic rivalry was at its peak. It’s clear the book was written with this in mind: Much of Johnson’s conclusion is devoted to telling English-language readers what the United States could learn and adapt (or not) from the Japanese experience. In no uncertain terms, these recommendations suck. Some of them are so anodyne as to remind me of contemporary commentary on what the United States can borrow from China: “Lacking a comparable consensus on goals, the United States might be better advised to build on its own strengths and to unleash the private, competitive impulses of its citizens rather than add still another layer to its already burdensome regulatory bureaucracy.” You could write this today. Johnson is clearly phoning it in at this point. Another contemporary echo is the fact that Johnson more or less refers to the United States as a nation of lawyers whereas Japan is a nation of bureaucrats and engineers. Dan Wang’s recently published Breakneck has a long pedigree, it seems… (I’m pretty sure people said the same thing about the Soviets too.)
Johnson describes a “slip of the tongue” from Finance Minister and MITI Minister Hayato Ikeda in 1952:
Ikeda was forced to resign as MITI Minister and accept a temporary setback in his political career because he had said too candidly in the Diet, “It makes no difference to me if five or ten small businessmen are forced to commit suicide” as a result of the heavy industrialization efforts. Ikeda must be recorded as the single most important individual architect of the Japanese economic miracle.
Here’s a great reading guide on Chinese local government political economy.
I have to footnote this part or else the main post will get too long. Johnson provides an exhaustive account of MITI’s financial capabilities.
American military truck procurement during the Korean War was a huge boon to Japan’s postwar economy—but many firms could not access capital fast enough to retool in time to meet American orders, and working capital access was not liquid enough to keep them solvent if American payments were delayed. So Japanese bureaucrats created the “city-bank overloan system,” where a city-bank loans at high leverage to priority industries, where the Japanese Development Bank loans to the city-bank, and where the Bank of Japan guarantees the JDB’s loans. This is how Japan created its “policy loan” system. In doing so, however, MITI killed Japanese capital markets, by eliminating equity and bond finance. Enterprises became dependent on banks well beyond any individual firm’s capacity to repay—and those banks overborrow from the Japanese Development Bank. The BOJ is exposed through its guarantees, but, in the process it gets detailed control over the policies and lending decisions of the various city-banks. Because the BOJ guaranteed everything, it eliminated risk for bank managers. “Bankers had only one thing to concentrate on―competition for expansion of bank’s share of loans and depositors. City banks did everything they could to discover and aid growth enterprises. Each bank had to have its entry in each new industry fostered by MITI or face being frozen out of riskless, profit-making new sectors.”
Later, the Economic Planning Agency, by issuing official statements of what industries the government is prepared to finance/guarantee in the near term, created excessive competition around zero-risk sectors. The combination of the overloan system and forced competition in foreign markets ended up forcing competition rather than throttling it.
MITI never had much personnel or budget, and relied instead on direct credit provisioning—which, counterintuitively, freed it from the influence of the Finance Ministry and allowed MITI to spray credit where it wanted to.
But other tricks helped. The postwar Capital Assets Revaluation Law “literally created capital where none had existed before by a (downward) reassessment of industrial assets for tax purposes, a process that was conducted some three different times between 1950 and 1955.” And, through the Rationalization Promotion Law of 1952, firms received direct subsidies for experimental installation and trial operation of new machines and equipment, plus rapid amortization and local tax exemption of all investments in R&D. It authorized certain industries to access 50 percent depreciation within the first year. And it committed the central government to build ports highways power and industrial parks at the public expense.
MITI also gave tax writeoffs to keiretsu spending on opening foreign branches, and provided contingency funds against bad debt trade contracts.
Correcting assumptions that Japanese industrial policy was entirely export-oriented, Johnson describes how MITI Minister Tanzan Ishibashi argued in 1954 that the domestic consumer market was essential for ensuring that Japanese industries could most quickly achieve production economies of scale:
Ishibashi pointed out that the key to exports was, of course, the lowering of costs, and the key to that was enlarging production to effect economies of scale. But to enlarge production, Japanese manufacturers needed more customers. And where were they to be found? In the huge potential market of Japan itself. The Japanese people had suffered from economic stringency for at least two decades; they were ready to buy anything offered to them at prices they could afford. Ishibashi’s idea was that MITI should promote both exports and domestic sales. When problems in the international balance of payments arose, the government could curtail domestic demand and promote exports; when the problems of paying for imported raw materials ceased, the focus should be on enlarging sales at home. If this could be achieved, Japan’s factories could keep operating throughout all phases of the business cycle. In Hira’s words, Ishibashi “combined export promotion and high-speed growth into a coherent theory.”