Private firms, public industries

If socialism is public ownership of the means of production and capitalism is private ownership of the means of production, then what is "social democracy", a "mixed economy"? The factory is either owned by the state, or it is not. Where is the middle ground?

There are lots of answers to this question. Many are reasonable and they are not mutually exclusive. Some production can be performed by state-owned firms while other production is performed by private firms. Firms can be jointly owned, with some shares held privately and some held on behalf of the state. Firms may be notionally private while the state synthesizes a partial ownership position, deriving cash flows by taxing the firm and exercising control through regulation. In most contemporary economies, we observe all three of these forms of mixing, in varying degrees.

I want to propose a different way of thinking about the question. In a mixed economy, we should consider firms to be substantially private, but industries to be a province of the public.

Firms are concrete. We transact with them, we work for them, we buy shares in them. But from a public perspective, what we care about is industries as a whole. If a firm is badly run, it may go bankrupt. That may be difficult for its private owners, but it is not the state's concern. What matters to the state is that, if this one goes under, that one takes over, or new ones appear. Through all the churn and competition, the public must remain continuously served. That is a characteristic of the industry, rather than of the firms that happen constitute it at a given moment.

Of course, as individuals, we have preferences and relationships with particular businesses. But as a public, we are basically indifferent to firms when industries are well arranged. It is the industries that we rely upon.

This sounds dismissive of firms, even mean. But it is this characteristic that renders it practical to think of firms as private at all! If the welfare of the broad public hangs directly on particular firms, then private control of those firms is not private in its consequences. In a democracy, management of such firms could not be left to some narrow, self-interested group of owners. Only the publicness of industries reconciles private ownership of firms with democratic control.

What does it mean, in practice, to say that industries are public? It means first and foremost that defining market structure is a province of the state, and the structure so defined must preserve the publicness of the industry. The state must guard the frontier between firm and industry.

Suppose a great and wonderful firm emerges, a fountain of efficiency that spews consumer welfare, whose shareholders content themselves with little. Absent state intervention, the firm, on its genuine merits, drives all of its competitors out of business. The public is delighted.

The state must not tolerate that, any more than it would tolerate the emergence of a benevolent dictatorship with hereditary succession. The new king may be public spirited! He may ensure that his ministers consult excessively with citizens, to ensure that his rule balances and takes into account the preferences of all! He may teach his son to follow in his footsteps! But his son may prove less wise or virtuous or capable. That is why we insist upon political democracy.

The very best firms begin like our benevolent king. Eventually, inevitably, they age into craven princelings. We structure the institutions of government to take the impermanence of virtue into account. We distinguish permanent roles from the transient tenures of humans who occupy them. We allocate roles by a mix of meritocratic competition and public consent, and then insist upon periodically reallocating them. We work to attract new talent and groom successors.

Industries rule us as much as any other organ of the state. A demand the food industry makes as a condition of your child's nutrition is as irresistible as anything a police officer might require at the point of a gun. It is the state's duty to manage police forces to ensure they exercise only the minimal coercion necessary to enforce the law. It is the state's duty to structure each industry, so that it demands as little as is practicable to supply a full range of goods and services, while consistently innovating in both process and product. Private firms seek to maximize profit. Public industries must be architected to enable firms to enjoy ordinary “accounting” profit, while keeping the contest for "excess” profit challenging. Excess profit should be transient and aligned with improvements that trickle up permanently to the level of the industry.

Like Federal lands, every industry is a public trust that requires active stewardship. Read this wonderful piece by Phillip Longman about the railroad industry, describing a history in which the state stepped up to its obligation, and then stepped down, with catastrophic results. (ht David Roberts)

It is not enough for the state merely to counter monopoly — although the state should, must, aggressively counter monopoly. As Sanjukta Paul has emphasized, "economic coordination of one kind or another is inevitable", because it is necessary. Absent an affirmative alternative structure, the emergence of one or a few dominant firms is simultaneously what antitrust law mans to prevent, and yet the only path to orderly efficiency antitrust law leaves available, because collusion between firms is unlawful but coordination within firms is normal. "Break 'em up" is only the beginning of the state's role. The state must provide rules of a game under which firms can, and must, stay broken up, with each firm succeeding or failing on its merits while the industry as a whole thrives and delivers for the public.

Industrial policy is all the rage these days and it should be. Forty years we wandered in a desert of delusion. We pretended we could have no policy and "the market" would choose for us the best and most optimal path. We are learning the hard way that what we do not manage we in fact concede to others who do.

But before we get into the business of choosing which firms to subsidize, we must ensure they are embedded in well arranged industries. We must make affirmative choices about the market structure we intend. We must define what constitutes the commodities bought, sold, and priced. We must design a tax and regulatory environment under which competing, profit-maximizing private firms will compose to the structure we desire.

Of course policy’s twin is unintended consequence. We will err. But no invisible hand can do this work for us. We have to do our best.

It is perpetually the work of the state — our work as a democracy — to define for each of our industries the market structure we desire, to design and then continually revisit policies in order to ensure the structure we target is roughly what prevails.

And it is perpetually the work of the state to police the frontier between private firm and public industry. We must punish firms who trespass merely for dispossessing us, however much, for the moment, we might admire the goods and services they produce.


Midsize is the right size

Last week there was a brawl on Twitter between people all of whom I mostly admire, including Eric Levitz, Matt Stoller,Basel Musharbach, Matt Bruenig, and David Dayen, among others. It centered around Levitz's claim

Small biz is not inherently more desirable than large-scale enterprises. Economies of scale exist. Large firms pay better than small firms, and are also easier to regulate (with large chains, a spot check at one location can incentivize compliance at hundreds of locations while policing the practices of every individual small business is not logistically feasible).

A lot of different things are true. It is true that

  • Bigger businesses often (not always) treat workers better than "Mom & Pop" firms can manage, especially in terms of benefits and flexibility.

  • The state is capable of enacting useful social policy — workplace safety standards, accommodations of disability, antidiscrimination law, etc — through bigger firms in ways it cannot through very small firms.

  • There are often technical economies of scale in production that very small firms cannot exploit. Societies that rely too heavily on very small firms can end up impoverished by low productivity.

These factors tilt in favor of "bigger business" and against "very small" or "Mom & Pop" firms.

But there are some benefits and advantages of very small firms:

  • In some industries, economies of scale are modest, and much of the value produced at an industry level lies in diversity of output, which favors very small firms. Restaurants and cafés and microbreweries, spas and nail salons, retail boutiques, artists and musicians — all at least sometimes have these characteristics.

  • Very small firms create opportunities for agency in production, which some (though far from all) people really value. Many people are perfectly happy to trade obedience for a paycheck. Others bristle at toiling in the belly of a hierarchy, being told what to do. An economy too heavily weighted towards larger firms may leave too few opportunities for those who value this kind of agency.

  • Very small firms create opportunities for agency in consumption. A small customer's interactions with a large firm are typically characterized by contracts of adhesion.

    « These are the terms on which our product is offered, this is the price, click the checkbox to accept or else fuck off. »

    A small customer has a greater likelihood of negotiating bespoke terms with the proprietors of a very small firm than with agents of a bigger firm.

  • "Relationships" can be valuable in commerce. A relationship exists when parties to repeated transactions sometimes accept a worse deal than they might find elsewhere, not out of laziness or inconvenience, but in well-grounded expectation that each party's occasional indulgences will be reciprocated. This serves as a form of mutual insurance. It may ipso facto enhance the perceived value of goods and services. Smaller customers may be capable of forming more reliable and resilient relationships with very small firms than with larger firms.

  • Many of the best bigger firms start as small firms. This is not always true: Investors with large pools of capital can spin-up big firms de novo. Big firms sometimes spin out other big firms. Nevertheless, small firms that grow big constitute an important source of new entrants in most industries. These firms may be especially important for innovation. Very small firms can experiment at modest risk, and then scale up practices that eventually improve whole industries. If an industry is inhospitable to very small firms, this source of innovation may be blocked.

The net effect of all this is that it's silly to argue over which is better. We should expect and want both kinds of firms. We should want a restaurant industry with a much larger fraction of "Mom and Pops" than an automobile industry. But even in a restaurant industry, we'll want some Chipotles and Panda Expresses. And even in an automobile industry we'll want some startups producing novel products by novel means at initially inefficient scale.

The fact that entrepreneurs and customers value the kind of agency and relationships that very small firms enable means we should expect and want more "Mom and Pops" than a narrow focus on efficiency or regulability would predict.

But overall, we should expect and want most industries to be dominated by bigger firms. Even in industries where economies of scale are only modest, the value of diversity tops usually out at a pretty low level as well. Those modest technical economies of scale, the insurance economies of scale that allow larger firms to treat employees better, plus susceptibility to regulation all tilt the balance toward larger firms.

But how large? I keep using terms like "bigger" and "very small". How big?

This is where the anti-trusters have the better of the argument. Almost always, all of the advantages of "bigger" firms are fully exploited once enterprises reach medium size. You don't need behemoths that dominate whole industries.

The state has little trouble perceiving and taxing and regulating midsize firms. Midsize firms are fully subject to labor and safety and other regulations that "Mom & Pops" would find unaffordable, and so are left exempt from. Midsize firms can tolerate worker flexibility (say time off for emergencies, or parenthood), because they are big enough to have alternate workers who can pick up the slack.

Most importantly, for an economy the size of the United States, midsize firms are large enough to exploit technical economies of scale almost completely. Your friendly neighborhood monopolist will deny this — they will finance studies to snow you — but neverthless it is true. Yes, any firm fixed cost can be made smaller if amortized over a larger number of customers. But even in theory, the incremental benefit of new scale is ever diminishing, and becomes increasingly negligible. In practice, most "fixed costs" hit capacity limits by midsize scale, after which they become variable again. You can only sell so many cars before you have to build a new factory.

What do I mean by midsize? It is best to talk in industry terms. Policy seeks to shape market structure, not churn out identical firms of the size the commissar deems optimal. Firm sizes will vary within industries, and reasonable firm sizes will vary by industry.

But in rough numbers, for the US national market, I'd describe an industry as dominated by midsize firms if the top ten firms in the industry together account for no more than 75% of the market, each have at least 3% share, and no one firm accounts for more than 20%. It's an industry in which most customers have meaningful access to tens of viable competitors — not hundreds or thousands of tiny "Mom & Pops", not the handful (less than ten) that has become the norm in most US industries. Midsize firms exploit technical and insurance economies of scale, are legible vehicles for socially valuable regulation, and compete with one another vigorously.

When industries are composed of firms larger than midsize, when the top few firms account for most of the share, the result is oligopoly. Rather than being susceptible to regulation as Levitz suggests, the state finds itself dependent on dominant firms. Boeing, Intel, all the "systematically important financial institutions" are too big to regulate or prosecute, and too critical to be permitted to fail. Very large firms become specialists in exploiting market and political power rather than technical processes of production. They develop "efficiencies" in a financial sense, and come to see the preconditions of hard-to-measure "quality" or "innovation" as fat to be trimmed. They get away with it, because they (and perhaps a few other giants, who adopt the same "best practices") are the only game in town.

It is true that while "Mom & Pops" have a role, they are not a great way to organize the bulk of a modern economy. It is also true that firms durably larger than midsize — defined by the scale at which they bestride their industries — ought to have no role whatsoever in a well run economy. Any "natural monopolies" should be public, or put under effective public control by regulation.

Some industries — in which economies of scale are modest, diversity is highly valued, and durable relationships between small customers and businesses are important — should be regulated in ways particularly friendly to "Mom & Pops". But for most industries, the goal of policy should be to construct a regulatory environment that favors vigorously competing midsize firms, and smaller upstarts that seek to join or displace them.


Rule-of-law is incompatible with a sharply polarized two-party system

A while back I wrote that two parties make us stupid.

Effective public deliberation requires a neutral-ish population who can help adjudicate. For any matter of public concern, there should be a sizable group whose interest in the question flows only through a broad public interest, rather than from any parochial stake in the outcome.

In a courtroom, we choose juries randomly, rather than letting the prosecution and the defense stand their own jurors. We understand that if we let each team hire six jurors, nearly every jury would be hung. There'd be little point in holding a trial. Partisan jurors would not base their verdicts primarily on good-faith evaluation of the evidence presented. They'd deliver the verdicts they were hired to deliver. If we want a defendant to be judged without prejudice on the basis of evidence presented at trial, we need neutral-ish jurors with little skin in the game.

Similarly, if nearly the whole of the politically engaged public is segmented into two camps, each of whose members share an overriding interest in winning contests that enhance the power of their side, then there can be no arbiter on behalf of a broader public interest. In every matter of public controversy, one of the two camps will have adopted a position which the other will oppose. Partisans on each side will tend to support their own party's position, regardless of cases earnestly made by advocates and activists on the broader merits of the issue. Substantive public deliberation will be short-circuited.

If members of the public are segmented into five or six different camps, a matter of public controversy may have entrenched advocates on one side or the other in some of the parties. But the issue may be orthogonal to the remaining parties' core allegiances. A multiparty system leaves room for neutral-ish adjudicators who serve as a good faith audience to public deliberation, who will constitute the "swing" vote when the question is decided in a legislature or referendum. A two-party system can pull this off only when it sustains a population keenly interested in public affairs, but with only weak factional identities — people who are politically engaged but routinely cross party lines. That is, the parties cannot be sharply polarized.

Criminal justice is largely exempt from this dynamic. While the guilt or innocence of a shoplifter is, in a certain sense, a matter of public controversy, the direct interests at stake are so small as to render the matter effectively private. The accused has a direct interest. The victim may perceive a direct interest. But for most of the rest of us, our stake in the outcome is only that justice be done. Courts can call jurors, filter away those whom there is some reason to think prejudiced, and conduct something like a fair trial.

This becomes more difficult, however, when the defendant is herself a public figure, particularly when the defendant is aligned or associated with a major political faction. In a sharply polarized two-party system, whenever a prominent figure of either party faces criminal charges brought by members of the opposing party, people strongly identified with the defendant's party will call the prosecution illegitimate and politically motivated. People associated with the prosecuting party will claim merely to be applying the law, without fear or favor.

Who is right and who is wrong? Without a credibly neutral public to adjudicate, the question is not decidable. You may be absolutely sure that your view (and, coincidentally, your side's view) is correct on the merits. The other side will have alternative facts.

In a sharply polarized, two-party system, there is simply no fixed point, no neutral vantage from which anyone persuasively judge. Except in cases so egregious that even copartisans concede a necessity, the question of whether a prosecution is invidiously political can find no authoritative answer. It becomes the macro analogue of a he-said, she-said controversy. You may have a strong view, and have your own deep understanding of the merits of the case. But then you would have that view, wouldn't you.

Again, in a multiparty system, or in a system with only weak factional identification, a less interested public can serve as arbiter and decide these questions. They might be complicated and hard and fraught, but when the bulk of a public not strongly identified with either the prosecutor's nor the defendant's faction comes to a decision, that decision can stand. No decision will engender a consensus or become authoritative while the public remains sharply polarized into two camps. Convictions will not be durable, unless transfers of power are prevented and one faction rules in perpetuity.

A system that, to a reasonable approximation, administers rule-of-law in cases involving ordinary people, but which cannot credibly impose criminal consequences on people who are famous and political and wealthy, is not rule-of-law at all. The fissures between the factions become the basis of, and a veil over, a deeper divide, between those whom the law protects but does not bind, and those who, the law binds but does not protect.

We can argue all we want over whether Trump or Elon Musk deserve imprisonment. I think there are strong cases against them both. I would, wouldn't I? People on the other side are sure Joe is the pater familias of a "Biden Crime Family".

I know I am right and they are wrong. They know the same. We're collectively so polarized and sorted there is no one who can serve as jury, no way to credibly adjudicate the question.

If we want rule-of-law, we'll need electoral reform. Not so much because we'll elect different people, but because of the way elections, by structuring citizens into constituencies, organize society as a whole. We need an electoral system that segments us into multiple camps, not just two, so that on many public controversies, most of us will have no factional interest, and can adjudicate rather than advocate.


A Westphalian order is project enough

What makes a sovereign state?

First, there is a territory. The territory has inhabitants. The territory is governed by a state. The state is legitimate.

What makes a state "legitimate"? I think there are two distinct but related criteria, which I'll call internal and external legitimacy. About a year ago, I described what I now think of as internal legitimacy:

Legitimacy sounds very subjective, almost populist. Is a state “legitimate” because the masses agree with it, because people like it? No. It’s nice when they do, maybe even helpful, but nice isn’t what we’re after.

We measure the legitimacy of a state by whether those on its territory both conform to it and resort to it. When the state commands, in the civilized tone of some legal notice (the threat of violence echoing faintly from the pages), do citizens obey? Or do various factions succeed at resisting, ignoring, and defying its edicts? When residents enter into dispute, do they draw their own weapons, or make use of the state’s courts and legislatures and police? Legitimacy is revealed by how people behave, not by what they say they think.

External legitimacy, I think, is even simpler. A state is externally legitimate if it has effective control over violence that projects outward from its borders. If there are marauders or militias that ravage neighboring territories without the state's direction and consent, then the state lacks external legitimacy.

When a state lacks external legitimacy, neighboring states cannot rely upon negotiation to address threats to their interests or to their own legitimacy. Sovereignty is a status that derives from reciprocality among peers. A state that cannot direct or suppress violence from its borders — violations of the sovereignty of neighbors — is not a peer at all. If a state is incapable of respecting neighbors' sovereignty, the state commands no sovereignty for its neighbors to respect.

Finally there is formal recognition. States, and interstate bodies like the United Nations, proclaim by various procedures their recognition that a sovereign state exists, and include those states in ceremonies and institutions reserved for peer sovereigns. The United Nations offers a seat in the General Assembly. Diplomats are dispatched to, and accepted from, the formally recognized state. Among most states, a consensus emerges to describe a community of widely recognized states, which includes most states.

A fully sovereign state demands that all of these criteria are met. The state genuinely governs its territory, suppresses — or in war directs — any violence that might project from its borders, recognizes and is recognized by other, peer, sovereign states.

Of course these criteria are never perfectly met. Every state includes some parties who prosecute disputes via freelance violence ("crime"), even somewhat institutionalized freelance violence ("organized crime"). No state can entirely prevent odd cases of residents traveling from its borders and pulling a gun or attempting some adventure or coup. Any given state may not be formally recognized by some one or few other states. None of these blemishes undo state sovereignty, as long as they are small, idiosyncratic, marginal. It's a judgment call, but it's not usually in practice a hard judgment call. We are usually able to distinguish weirdoes who commit random crimes from organized militias that consistently prepare for and perform violence beyond the capacity of the putative state to control.

The excellent Samantha Hancox-Li writes:

Human history is a river of blood. There is no justice for the numberless dead. Justice is for the living and the yet to be born. It is our responsibility to give not vengeance but peace to those who will come after us. The lesson of history is that it does not matter where you draw the lines on the map. What matters is what kind of society lies on each side of that line. Liberal democracy is the only thing yet discovered that offers a chance for climbing out of the bloody river. Until Palestinians and Israelis both choose liberal democracy, there will be no peace. I do not know how to get there. I only know it is where we must go.

Although I too prefer liberal democracy, I think Hancox-Li sets her ambitions too high, her requirements too stringently. We have had more peace in the world, remarkably, than can be explained by liberal democracy. Saudi Arabia and Egypt are not invading their neighbors, at least not when their neighbors are sovereign states by the three criteria above. (Yemen, in civil war or partial occupation by an Iranian proxy, has not met those criteria for some time.) Iraq in 1990 and Russia in 2014 are exceptions that prove the rule. Invasion of fully sovereign states by fully sovereign states are rare, big-fucking-deals that other sovereign states can be organized to oppose and reverse. Failure to organize and reverse, or at least severely punish, these norm violations invites further catastrophe.

For any sovereign state, a war whose battlefield will include its own territory is terrible. For autocrats as much as for democrats, it is the ultimate negative-sum game. Sovereign states can negotiate and be negotiated with. They can join alliances. Understandings between states can be meaningful and effective. Sovereign states act and can be deterred "rationally", whatever their form of government. After the paroxysms of the mid-20th Century, war between sovereign states became infrequent. We did in fact learn something.

In the vast majority of cases, war comes when the criteria set out above are not met. Violence emerges when a government claims sovereignty over territories beyond its internal legitimacy, beyond its capacity to control external violence, or where the boundaries of the territory are not widely agreed.

Neither Israel/Palestine nor Lebanon are sovereign states. It matters not a whit that they have seats at the UN or internationally recognized borders. It matters even less that Israel refers to itself, absurdly, as a liberal democracy. Israel/Palestine, Lebanon, Syria, post-Saddam Iraq, none of these meet the basic preconditions of sovereign states. So these places are riven by violence, internally among groups that do not recognize the same sovereigns, externally as no sovereign can credibly negotiate terms to suppress the projection of violence into neighboring territories.

The United States, but also China, all the powers and places in the world that enjoy the fruits of peaceful modernity, share a core foreign policy interest in a Westphalian order, in a world be organized into sovereign states that are legitimate internally and externally and widely recognized with clearly demarcated borders. Sovereign states are frequently tempted to undermine the sovereignty of neighbors and rivals, but it's like burning carbon. In the long term, it can only make you better off if everybody else sustains globally the system that you are undermining locally.

Supporting and reinforcing a Westphalian order is a relatively modest ask, compared to universalizing liberal democracy. It's a project the world's great powers and most states might agree upon. It doesn't foreclose geopolitical competition. The world remains a chessboard. But a chessboard has squares.


Industrial policy without national champions

Last Weekend Adam Ozimek and the Economic Innovation Group hosted an "Econtwitter IRL" meetup at Decades in Lancaster, PA, a wonderful venue Ozimek cofounded. I'm isolated these days. I am very grateful to Ozimek for bringing together an online community I really value but that usually feels distant.

During that event, Cardiff Garcia did a fascinating interview with Paul Krugman, at the end of which Joe Wiesenthal asked a critical question. The United States, under Joe Biden, is embarking on an aggressive program of industrial policy even as it pursues increasingly vigorous antitrust enforcement. Aren't there tensions between these goals? Responding to an antitrust investigation of NVIDIA, Dylan Matthews similarly asked, "Here you have a tremendously successful national champion in a strategically critical industry. Is that exactly what you want[?]"

It's a good question!

South Korea, an industrial policy success story, famously elevated a small number of "chaebols", who dominate that economy to this day. Japan's successful MITI gave the world automobile champions like Toyota and industrial conglomerates like Yamaha. Smaller countries may have no other choice than to endow national champions, boulders among pebbles in their domestic economies, in order to achieve technical economies of scale necessary to compete in world markets.

But industrial policy often fails, and usually when it fails, it does so for simple reasons of political economy. All industrial policy requires that the state subsidize firms, one way or another. But successful industrial policy demands that states also force firms to compete, so the subsidy becomes invested in efficient production, rather than simply captured by shareholders. Here's Vivek Chibber:

In country after country after country, you found [firms] hanging on to the subsidies and the general import-substitution long after everyone realized, ‘Look, we've reached about as far as we can go with the subsidy side, now we need to push these guys into competition.’ Long after they understood that, they stuck with a self-defeating policy.

The question is, why? And the answer is political. Basically, it's this: what all the economists assumed was a government and a state that's essentially free...and powerful enough...to tell firms to do whatever it wants them to do, and they're going to step into line. But the fact is, in any modern industrial capitalist economy, these firms who you're trying to push into exports are also the people with the most political power, the most political influence. They have the lobbyists, they have all the money, they fund elections, and they run the economy.

So here's the trap you were in. You gave them a bunch of free money and now you are asking to give up that free money and go into essentially shark infested waters. And country after country, what most of them said was, ‘Nah, we're not going to do it. Actually, we like it just the way it is.’ And, since we fund all the politicians, whoever gets into power, we're going to make sure they keep plowing this money towards us.

A national champion is at best a double-edged sword. It has the resources to become a world-class competitor, if those resources are allocated adroitly rather than captured as cash flow to shareholders. But those same resources enable it to specialize in exploiting market and political power, rather than in any technical facet of production.

Without competition to inform and discipline a "champion" while it is an infant, competing mostly in domestic markets, often under tariff protection, it may have no way even to learn how to excel at world-class production. The skills required to buy the loyalty of politicians, on the other hand, are accessible, reliable, and widely distributed.

The largest economy to succeed at industrial policy is China. What the Chinese example teaches is that, at least for large economies, there is no need to risk capture by national champions. Instead of targeting a few, particular firms for subsidy, one targets a competitive domestic industry, and funds a whole menagerie.

In 2022, China's leading manufacturer of photovoltaic modules produced less than 16% of industry output. The top four manufacturers accounted for less than 60% of production. Brad Setser points out that in China's "capex heavy steel sector…every province has its own local champion". China bestrides the planet like a colossus in both of these industries, but there are no national champions to be found.

China's experience suggests one can use the conditions of funding to affirmatively structure the market in ways that ensure continuing competition, rather than rely solely on after the fact policing by antitrust authorities. China's subsidies come via provincial governments, whose officials require their champions remain local. In the US, firms might simply be forbidden contractually from selling equity or assets to rivals or roll-ups.

However protected the domestic players might be from more advanced foreign competitors, competition in the domestic economy generates innovation and know-how. Despite subsidy, no domestic firm can rest on its laurels.

As competition brings the quality and efficiency of production towards world-market standards, it becomes politically plausible to reduce trade barriers, placing the now adolescent industry in global competition. Reducing tariffs won't destroy jobs at firms that produce competitively. Firms confident they can compete often lobby for open borders, because they gain access to a much larger global market by ceding some access to their own smaller pond.

So, for larger economies, there is no tension between industrial policy and antitrust. On the contrary, they are essential complements. Successful industrial policy requires that subsidized domestic firms vigorously compete.

National champions are a bad idea. Like athletic champions, they inevitably grow old. A vibrant capitalism (like a vibrant political system) depends on older incumbents ceding dominance to upstarts. Productivity growth is a process of small increments as stalwart firms iterate, but also great leaps as dissidents jump ship from staid incumbents and recombine industry know-how in novel, creative, threatening ways. National champions interfere with this process. They prove capable at discouraging rivals. They defend their turf, until their inevitable collapse — this too must pass — means collapse of a whole nation's industry, rather than a passing of the baton.

We require industrial policy. But we should eschew, even dismantle, "national champions".