By May 2026, Chinese shipyards controlled 70.9 percent of the global market for new ship orders. A decade ago, that share sat near 25 percent. No industry on earth has been captured this completely, this fast, by a single country. If you wanted a single image of state-directed industrial policy succeeding exactly as designed, this is the one Beijing would choose to show you.
Now look at what the state actually got for the money.
Between 2006 and 2013, China spent roughly $91 billion subsidizing its shipbuilding industry: cheap coastal land to encourage new shipyards, subsidized steel and inputs, low-interest loans to buyers and builders alike. Three economists, Panle Jia Barwick, Myrto Kalouptsidi, and Nahim Bin Zahur, spent years building a model rigorous enough to measure what that money actually returned. Their finding, published in the Journal of Economic Perspectives: a gross return rate of 18 percent, measured as the lifetime profit gains of domestic firms divided by the total subsidies paid. Not an 18 percent profit. An 18 percent recovery. Kalouptsidi put it plainly in an interview: you give a dollar, you get back twenty cents.
Sit with that. Ninety-one billion dollars, and the firms it was spent on gained back sixteen billion in lifetime profit. The other seventy-five billion did not vanish, exactly. It went somewhere. It built shipyards that should not have been built, kept afloat builders who could not compete, and expanded China’s share of an industry by taking customers from Japan and South Korea rather than creating new demand. Seventy percent of the entire market-share gain, the same paper finds, came from business stolen from rivals, not business that did not exist before.
Daewoo Shipbuilding and Marine Engineering sat on the losing end of that transfer. Years of eroding order books and mounting losses eventually forced a debt restructuring, then a 2023 buyout by Hanwha, which now runs the yard as Hanwha Ocean. Korea’s shipbuilders are not strangers to state support of their own, and DSME’s fall is not a clean story of an unsubsidized firm losing to a subsidized one. It is closer to a firm losing a subsidy race because its own government didn’t back it generously enough to win, which says less about Korea’s discipline than it does about how little of this market anywhere was ever actually tested by demand. Nobody subsidized DSME’s mistakes into permanence, whatever help it did receive stopped short of that. Its lenders and shareholders absorbed the losses, its leadership changed, and the company itself ceased to exist under its own name. That is what happens on the other side of the table from a firm that has to earn its price one contract at a time.
The subsidy did not stop at the shipyard gate. Part of the $91 billion went to cheap financing for ship buyers, the carriers themselves, lowering the cost of capital for fleets that would have been more expensive to build at true market rates. Shipping has repeated the same self-inflicted cycle for the better part of two centuries: demand spikes, new capacity takes years to arrive because ships take that long to build, prices swing wildly in both directions, and owners order the most new tonnage right when they should be ordering the least, extrapolating a boom that is already ending. Subsidized loans do not fix that problem. They remove the one thing that might discipline it, the true cost of the ship, from the buyer’s own calculation, and hand cheap capital to owners at precisely the moment in the cycle when cheap capital does the most damage. The subsidy did not just prop up bad shipyards. It magnified the amplitude of a cycle the industry has never learned to correct on its own.
A firm chasing profit in a market economy has one honest measure of whether its ships were worth building: did anyone want them enough to pay a price that covered the cost of building them, plus a profit? China’s shipbuilding program never had to answer that question, because the state was the one asking it, and the state was also the one grading the answer. Entry subsidies, mostly free or nearly free coastal land, made up 69 percent of the total spent. That is not money that built ships. That is money that built the right to try to build ships, regardless of whether trying was a good idea.
Ludwig von Mises named this problem a century ago, arguing that a planner without market prices for capital has no way to tell whether resources are creating value or destroying it. He was writing about central planning writ large, but the shipbuilding numbers are almost a laboratory demonstration of the specific mechanism. Profit and loss are not bureaucratic housekeeping. They are the only signal that tells an economy where to stop. Remove that signal, replace it with tonnage produced and market share captured, and you get exactly what China got: an industry that looks unstoppable by the metric the state chose, and an industry that burned three-quarters of the capital thrown at it by the metric that actually measures whether wealth was created or merely moved around.
This is not a new failure mode for the man who runs the system. This newsletter has already traced what a working correction mechanism looks like in practice, in “Big Defense Just Lost Its Alibi”: Ukrainian units price their own battlefield information in real time, rewarding the manufacturers whose drones actually survive contact and starving the ones that don’t, no ministry official required to know which is which. China’s shipbuilding program has no equivalent channel. What the shipbuilding numbers show is the mirror image of that same mechanism, running in the economic register instead of the military one. A market punishes a bad shipyard the way Ukraine’s front-line commanders punish a bad drone: automatically, continuously, without waiting for permission. Remove the correction mechanism in either domain and the failures do not announce themselves. They accumulate, get counted as tonnage or as inventory, and wait.
The pattern has not stayed in the past tense. Merics, the German institute tracking Chinese industrial policy in real time, reports that investment in priority manufacturing sectors remains elevated even as domestic demand weakens, a combination that pushes down prices, compresses margins, and increases the share of loss-making firms across the board. Chinese producers, unable to sell enough at home, lean harder on exports to keep output moving, the same broken circuit this newsletter traced this spring in “The Consumer Who Wasn’t There”: a production machine large enough to build anything, paired with a domestic demand base structurally prevented from absorbing what it builds. The ships get built. The orders get won. The margins that would tell a market-tested firm to stop building keep sliding, and nobody in a position to act on them would dare go against the directive.
None of this means the ships are imaginary. China builds enormous numbers of them, at scale no competitor can currently match, and that fact carries real strategic and commercial weight regardless of what it cost to produce. The seen and the unseen are not in competition for which one is real. They are in competition for which one gets counted.
Washington is currently being told, by people across its political spectrum, that the answer to China’s industrial strategy is an American industrial strategy: subsidize the shipyards, subsidize the chips, match the state with a state. The Barwick, Kalouptsidi, and Zahur paper is not an argument against ever using industrial policy. It is a very precisely measured argument that the tool you are being told to copy did not do what it looks like it did. It moved seventy percent of a global market from one country’s yards to another’s. It did not create seventy percent more shipbuilding worth having.
The standard defense of a program like this is that a state isn’t optimizing for producer profit, it’s buying sovereignty, capacity, and time, and early losses are the expected cost of a payoff that comes later. That defense has a real shelf life, and it is the same one that gets reached for after the fact to justify Apollo by way of Tang and memory foam rather than the moon landing itself, a reward assigned after the spending rather than priced before it. It does not explain why, two decades after the subsidies began, Chinese shipbuilding's margins are still compressing and its loss-making firms are still multiplying. An infant industry is supposed to grow up. Shipbuilding is still losing money at scale in 2026.
A ship is either something a buyer wanted badly enough to pay its true cost, built by a yard that could only survive by earning that price, or it is tonnage moved by capital that never had to answer for itself, on either side of the transaction. China’s tonnage is real. Whether it was worth ninety-one billion dollars is a question the tonnage cannot answer, and the only people positioned to answer it, the shipyards and the carriers alike, never had to.
Enduring economic lessons plucked from the headlines, seen through an Austrian lens. Ground truth from 40 years in the Colón Free Zone.