Shipbuilding Subsidies Bought China Market Share at a High Domestic Cost

Myrto Kalouptsidi describes global shipbuilding as an industry whose center of gravity has repeatedly shifted. Britain led from the 19th century through much of the period before World War II, alongside other European builders, including Germany and Scandinavian countries. The United States became the world’s largest producer during both world wars, but its lead did not last. Japan rebuilt its industrial base after World War II with government support for shipbuilding and became the leading producer within roughly a decade. South Korea took first place by the late 1970s. China entered around 2005–06 with an explicit ambition to become a global leader; by 2010 or 2011 it held more than half the market, and today it has the largest share in many ship categories.

That sequence is not simply a story of countries accumulating know-how until one permanently wins. Leadership has moved as governments, firms, labor forces, and markets changed. The American wartime surge is especially revealing: the United States built ships at extraordinary scale, yet did not retain a lasting commercial advantage afterward.

Panle Barwick puts the prewar American position in context. A 1901 report by the U.S. Commissioner of Navigation described British shipyards as operating on a wholesale scale and American yards on a retail one. British capacity and output were about ten times the U.S. level. American iron and steel were more expensive, and wages were higher—by as much as 50 percent, Barwick said—making U.S. builders costly relative to British competitors.

The scale changed dramatically during World War II. The 1936 Merchant Marine Act had aimed to replace an aging American fleet. Wartime federal procurement then expanded both naval and commercial shipbuilding capacity on a vast scale. Barwick estimated government procurement spending at roughly $750 billion to $850 billion in today’s dollars. The Liberty ships, produced by the thousands, became the most visible result.

The production model changed as well. Henry Kaiser, a construction entrepreneur rather than an established shipbuilder, brought prefabricated blocks into shipbuilding. With large, reliable government orders, American production moved toward the wholesale model the earlier report had said it lacked. Wartime output demonstrated that the country could rapidly build capacity when demand and public funding were concentrated.

But wartime scale did not translate into a durable commercial lead. Barwick noted that Liberty ships were not what the postwar market wanted: buyers wanted larger tankers and bulk carriers. When government orders stopped, Kaiser left the industry. The glut of wartime ships also depressed U.S. demand for years; some surplus vessels were converted for commercial use and sold cheaply.

Knowledge and production methods crossed borders, too. Barwick described U.S. shipowner Daniel Ludwig as studying labor-cost differences and deciding that Japan was a better place to build. He brought welding techniques developed during the war and the prefabrication model to Japan. Japan absorbed those methods, combined them with lower wages and government support, and built its own capacity. For Barwick, that transfer helps explain why American wartime learning did not remain an American advantage.

Davis points to another part of the evidence: he says that even at the end of the war, U.S. yards used more labor hours to build a Liberty ship than British yards used to build similar merchant vessels. A production boom can lower costs through learning, but the comparison Davis draws suggests that the American surge did not erase the underlying productivity gap.

The historical lesson is not that wartime industrial policy failed at its wartime objective. It is that extraordinary procurement, capacity expansion, and learning did not guarantee a lasting commercial position once the conditions that supported production changed. Kalouptsidi calls the speed of both expansion and contraction striking: a sector can be established quickly and dismantled just as quickly. Shipbuilding’s volatility matters to that account. The postwar surplus weakened demand, while Japan’s support for its industry was beginning to redirect production and market share.

The discussion distinguishes several ideas often bundled together under “learning by doing.” Barwick separates learning within a firm or shipyard from spillovers that help other firms. Learning inside a single firm may lower its costs, but that alone does not establish a reason for government support: the firm can capture the benefit itself. A subsidy becomes more plausible if learning spills beyond the firm and creates gains that ordinary commercial incentives do not reward.

Barwick also questions how much of the measured wartime cost decline should be attributed to learning. Earlier studies, she says, did not adequately account for dock size and capital investment. Once those factors are included, the estimated learning effect shrinks substantially. The decline in costs may have reflected more than accumulated experience; larger facilities and investment mattered. She also says the evidence does not clearly establish industry-wide learning spillovers.

Davis agrees with the distinction, but presses a further implication. Whether the mechanism is scale economies or learning, he argues, the United States had an unusually strong test during World War II: a huge industry, extensive supporting capacity, and substantial production experience. Yet that was not enough to preserve a comparative advantage.

Barwick’s response is that scale requires more than productive capacity. It also requires enough market demand to keep that capacity in use. After the war, the surplus of ships left the United States with little demand for new commercial vessels for nearly a decade. Davis counters that ship markets are global. Barwick’s answer is that American yards still were not commercially competitive in that global market. The exchange leaves an important qualification: a country can build capacity quickly, but that does not mean it can keep orders flowing to that capacity at competitive cost.

Steven Davis frames the issue as a test of industrial policy’s economic rationale. If learning stays inside a firm, the case for subsidy is weak. If it spills over to other firms, the government still has to ask where the gains go. Spillovers may extend beyond one shipyard but also cross national borders. In that case, a country paying for the subsidy may not capture the resulting benefits.

Kalouptsidi says this question cannot be separated from how countries obtain and retain industrial knowledge. Japan and South Korea used joint ventures and other incentives to encourage knowledge transfer from abroad. Those efforts raise a policy question for the United States: whether it should use similar arrangements to bring expertise into domestic production.

The American experience, then, does not show that scale or learning never matter. It shows that neither mechanism alone guarantees durable national advantage. Market demand, labor costs, capital investment, technology transfer, and foreign competition all shape whether production remains viable after a government-backed expansion ends. Kalouptsidi adds “forgetting” to the account: when military contracts stop and production falls, capabilities may not remain in place. She also cautions against treating the U.S. case in isolation. Japan was supporting shipbuilding and other industries, including steel and petrochemicals, as it pursued industrialization and export financing.

Davis summarizes the Chinese shipbuilding program as a success in expanding market share and a failure on the measure of domestic economic welfare. The economists broadly accept that characterization, while adding that the results vary by policy instrument and that other potential benefits are difficult to quantify.

The researchers’ estimates distinguish sharply between the initial entry subsidies and the later support for production and investment. Davis summarizes their estimate for the early program as a gross return of about $18 for every $100 spent, with the remainder lost through inefficiency or accruing outside China. Barwick says the bulk of the losses came from subsidies that encouraged new entry. The government made valuable coastal land available to shipyards, many of which had little production experience. After the boom, high-cost producers struggled and some went bankrupt.

The government later imposed an entry ban and shifted support toward established firms, including through a “white list” of companies eligible for subsidies. Barwick says the later production and investment subsidies were much more efficient by comparison. Their estimates put the net surplus generated at about $70 to $80 per $100 in subsidy. Davis restates that as a remaining loss of roughly $20 to $30 per $100. Barwick agrees that the later losses remained real, while emphasizing how sharply they narrowed.

Kalouptsidi cautions that these calculations cover a particular time horizon. The researchers also considered whether cheaper ships lowered freight rates and thereby supported China’s expanding exports. If trade generated welfare gains for China, that channel could partly offset the costs of shipbuilding subsidies. But she says they do not have an agreed-upon welfare estimate for the effect. It remains a possible justification, not a measured offset.

Davis is skeptical that the export channel resolves the broader question. He argues that China’s export expansion was itself supported by subsidies in various forms, and says this does not establish an economic justification for shipbuilding subsidies or for Chinese industrial policy more generally.

A second question is whether findings from shipbuilding apply to other Chinese industrial programs. Kalouptsidi identifies sectors such as steel and solar panels as having some features in common with shipbuilding: fragmented markets and many firms, including regional clusters where multiple producers compete. She contrasts that structure with Japan and South Korea, where a small number of large conglomerates became prominent industrial players. In her view, market structure can affect how efficiently subsidies work.

She also says the researchers could not identify a clear, first-order market failure in shipbuilding that would justify large-scale intervention. The possible grounds for government action include market power, spillovers, or financing constraints. Their work did not uncover compelling evidence of those problems in this industry. Davis adds a further test: identifying a market failure is not enough by itself. Policymakers also have to make a case that government intervention can improve on the market outcome rather than make it worse.

Barwick stresses that this conclusion should not be turned into a rule against industrial policy generally. She cites electric vehicles as a contrasting case where, in her view, market power distortions, environmental benefits, and spillovers provide real grounds for intervention. She says her estimates of subsidies in that sector find positive welfare gains. The relevant question is not whether subsidies are good or bad in the abstract, but whether they address a specific market failure and do so effectively.

The policy’s effects also changed over time without eliminating political allocation choices. Barwick says the government moved from encouraging new entry toward supporting established producers, and that the later production and investment subsidies were more efficient in their estimates. But Davis asks whether inefficient state-owned enterprises still received favorable treatment. Barwick agrees that state-owned firms were favored and more likely to receive support. The shift toward established producers therefore did not mean that subsidies were allocated solely according to economic efficiency.

Davis presses the distinction between a policy meeting an output target and improving welfare. China reached its shipbuilding goals ahead of schedule and gained global market share. Those outcomes do not show that the policy was economically beneficial to China. Barwick agrees: if the objective is to grow output, the program succeeded; if the objective is domestic welfare, their results make the costs difficult to justify. Military objectives, resilience, or access to a critical input may change the calculation, but policymakers need to say which objective they are pursuing.

The Jones Act and related cabotage rules are a different kind of American shipbuilding policy. Davis describes the requirement as covering transport between U.S. ports using U.S.-built and U.S.-flagged ships. He says the policy has been in place in some form for more than a century, with the modern Jones Act dating from the 1920s, and asks why it has not sustained a larger U.S. shipbuilding industry.

Davis characterizes the result as a failure even on the policy’s stated industrial purpose. He argues that the United States now produces very few large ocean-going ships and has not maintained a large merchant fleet. He also says the restrictions make transportation more expensive for places including California, Puerto Rico, Alaska, and Hawaii.

Kalouptsidi agrees that the law likely raises prices for places that rely on maritime transport, though she notes that domestic maritime trade is relatively small in the U.S. economy overall. Barwick is careful to distinguish her assessment from a study of the law’s effects: she has not analyzed the Jones Act as closely as the Chinese shipbuilding program. Based on the literature she has read, she says U.S. protection sheltered domestic builders from foreign competition but did not make them competitive in global markets.

In Barwick’s account, American commercial shipbuilding was not competitive across the historical periods the researchers studied. High wages and reliance on federal orders—including during both world wars and the Cold War—were enduring features. That makes it hard to attribute the industry’s decline to a single law. The United Kingdom, too, lost its former dominance as Japan rose; Japan now competes with South Korea, and South Korea with China. The identity of the leading builder has changed repeatedly. Protection may shelter U.S. builders from foreign competition, but it did not create the cost reductions and capabilities needed to compete internationally.

Davis raises a separate national-security argument sometimes offered for the law: maintaining a merchant marine might provide capacity for wartime use. He says the policy has not delivered a large U.S.-flagged fleet on that measure either. The speakers do not assess the fleet’s security value in detail, but distinguish that objective from commercial competitiveness.

A larger shipbuilding industry might be worth maintaining even if it were not the lowest-cost commercial producer, if national security required it. Davis says security considerations can outweigh economic efficiency, as they did during World War II. He points to U.S. submarine production as an example of a concern that belongs in this separate category: he recalls that the United States produces fewer than two submarines a year, perhaps fewer than one, but says he cannot remember the exact number. His point is that the rate seems troubling to him from a national-security perspective.

The submarine example clarifies the limit of the commercial-welfare analysis. A finding that subsidies or protection do not pay for themselves commercially would not, by itself, answer whether the United States needs shipbuilding capacity for security. Conversely, a security rationale would not demonstrate that the Jones Act has produced the capacity or fleet it is meant to support.

Barwick identifies a shortage of skilled workers as a central constraint on any U.S. effort to rebuild shipbuilding. The industry needs welders, technicians, and people trained to build large ships. She contrasts China, which she says has 10 to 14 universities dedicated to marine technology, with the United States, which has none. In her view, a limited pool of skilled labor makes it difficult to produce competitively and risks making subsidies permanent rather than transitional.

Her account connects the labor constraint to the broader history of comparative advantage. Barwick says Britain once had a large pool of skilled shipbuilding workers and technicians, supporting shipyards able to operate at wholesale scale. The United States did not have the same workforce base in the earlier period. Today, she argues, a shortage of skilled people remains a problem even if the country adds protection or financial support. Without workers who can use the relevant techniques, subsidies may have to continue to offset the cost disadvantage.

She describes a U.S.–South Korea investment and training effort as a move in the right direction: South Korean firms could bring accumulated know-how and train American technicians. Barwick refers to a $150 billion investment agreement and says the initiative is intended to transfer expertise and develop the U.S. workforce. But she says the effort has been held back by immigration restrictions that prevent technicians from coming to the United States for short-term training work. Davis cites the arrest of South Korean nationals at a Hyundai-related facility while they were helping train local workers as an example of how policy can work against the stated industrial objective.

The history and the current effort point to the same constraint: subsidies can expand output, but durable capability depends on more than money. It also depends on skilled workers, transferred know-how, and enough demand to sustain production. Those conditions shape whether support creates an industry that can compete, or mainly shifts production and costs while the support lasts.