← Writing archive

What the State Cannot Provide (first draft) — part 3 of 4

Source: public/assets/writing/humanstructures/What the State Cannot Provide (first draft).docx · humanstructures

The American healthcare system, then, is something like the following: a sector in which employer-sponsored third-party payment was created by wartime wage controls and sustained by federal tax preferences; in which hospital entry is restricted by state-issued permissions; in which pharmaceutical entry is restricted by billion-dollar federal compliance requirements; in which provider supply is restricted by federally funded residency caps administered through medical licensure monopolies; and in which forty percent of total spending is directly disbursed by federal payers using administered prices. This is not a market. It is a regulatory cartel operating under the nominal form of private provision, and its dysfunctions are the dysfunctions the framework would predict any such arrangement to produce.

The framework’s prediction can be tested against the parts of the sector that have escaped most of the regulatory apparatus. These are not academic examples. They are sizable sub-sectors of American healthcare operating under something close to actual market conditions, and they exhibit the predicted pattern with enough regularity to constitute evidence rather than anecdote.

Hearing aids are the cleanest recent natural experiment. For decades they were prescription medical devices sold through audiologists at prices of $2,000 to $8,000 per pair, requiring multiple fitting visits and bundled service contracts. In 2022 the FDA reclassified them to permit over-the-counter sale. Within three years, direct-to-consumer devices from Sony, Bose, and Apple appeared at prices from $200 to $1,000. The AirPods Pro 2 now functions as a clinical-grade hearing aid at roughly $250. The technical quality of 2025 OTC devices exceeds the prescription devices of 2015. This is not an elective cosmetic market; hearing loss skews older and lower-income, and the population served is not self-selected for affluence. The state released a single node it had captured, and the discovery mechanism began operating within months.

Telemedicine during and after COVID was a larger experiment at national scale. State scope-of-practice restrictions were relaxed in most US states during the pandemic, allowing physicians to practice across state lines and permitting remote consultations to substitute for in-person ones. Prices for routine primary care fell sharply: a Teladoc visit in 2024 costs roughly $75 without insurance, against $150 to $300 for an equivalent office visit. Wait times collapsed; same-day access became standard. Access in rural areas improved measurably. When states reimposed the restrictions after the emergency, the patterns reversed in those states. Treatment and control conditions inside the same country over the same two-year period, and the within-country variation tracks the regulatory variation.

Dental care has operated outside the health insurance apparatus for as long as both have existed. Most dental care is paid directly or through dental plans that function as discount cards rather than insurance. The sector exhibits the full predicted pattern: published prices, price competition among providers, cash-pay discounts, substantial innovation in service delivery, falling real prices for routine procedures. Dental implants have fallen from roughly $5,000 per tooth in 2005 to $1,500–2,500 in 2025. Orthodontics has been partly disrupted by direct-to-consumer clear aligners. Dental care is not elective in the way LASIK is; people need their teeth repaired. The pattern persists because the discovery layer has never been captured in dentistry the way it has in medicine.

Retail clinics and urgent care show the same pattern for non-elective routine medicine. CVS MinuteClinic, Walmart Health, and Walgreens Healthcare Clinics have scaled to several thousand locations nationally, delivering routine primary care at published prices: $89 for a sick visit, $125 for a physical, $25 for a strep test, with patient satisfaction scores consistently high. Urgent care chains like MedExpress and Concentra serve acute but non-emergency conditions at $100–300 per visit, typically without appointments, with wait times under thirty minutes. They compete with hospital emergency departments for the same services at roughly a tenth of the price.

Generic drugs show the mechanism on the pharmaceutical side. When a drug’s patent expires and generic manufacturers can enter, prices typically fall by 80 to 95 percent within two years. Statins fell from four dollars per pill under patent to ten cents per pill generic. Metformin, now the world’s most prescribed diabetes medication, costs pennies per dose. Insulin is the exception that confirms the framework: its price has not fallen to generic-equivalent levels despite being nearly a century old, because the FDA’s biologic-approval rules prevent straightforward generic entry in a way that the small-molecule pathway does not. The sub-sector where the state has captured the approval mechanism produces the prices the captured-sector framework predicts; the sub-sector where capture expires by design produces the universalization pattern.

Veterinary medicine provides the cleanest direct comparison because it uses largely the same technology as human medicine performed by specialists with comparable training. A veterinary CT scan costs $1,000 to $2,500; the human equivalent costs $3,000 to $10,000 at the point of care and bills out at considerably more. The differential is not explained by quality or technology. It is explained by payment structure: veterinary medicine has to price in ways that customers can actually afford, because there is no third-party payer insulating providers from that constraint.

These sub-sectors are not edge cases. They constitute a substantial fraction of how Americans actually obtain healthcare, and they vary across every dimension a critic might invoke to dismiss them: elective and non-elective, urgent and routine, affluent and mass-market, cosmetic and medical, technologically simple and complex. The common feature is that in each, the state has not captured the layer at which the sector’s decisions are made. The common output is the pattern markets produce when allowed to operate: falling real prices, rising quality, access that expands over time without anyone in particular being in charge of expanding it.

The reviewer in the tradition of modern health economics will respond with a specific objection: that these natural experiments operate in simple sub-sectors, and that insurance markets covering catastrophic and unpredictable medical events have structural features that prevent the discovery mechanism from operating even in principle. Adverse selection causes insurance markets to unravel; pre-existing conditions cannot be insured against after they arise; the entire category requires state intervention to function. There is a substantial literature making this case.

The argument treats as structural what are in fact features of a specific organizational form American insurance has taken. The adverse-selection literature describes a sector locked into mid-century architecture by the 1943 IRS ruling and the regulatory superstructure built on top of it. Within that architecture, the failures it documents are real. What the literature does not and cannot know is what the sector would look like if the architecture were permitted to reorganize. Long-term portable individual health policies on the life-insurance model, transparent catastrophic-only coverage with direct payment for routine care, direct-to-consumer distribution channels: obvious reforms that would address most of what the literature treats as inherent problems with health insurance. Whatever frictions remained would be the kind of thing entrepreneurs solve, in healthcare as in every other sector where they have been permitted to operate. The ridesharing entrepreneurs did not have a better theory of urban transportation than the taxi commission; they had a different organizational form. The insurance entrepreneur who would reorganize health coverage has not arrived because the regulatory architecture does not permit their arrival.

What this means for the argument is that the American case, properly understood, does not refute the thesis. It reinforces it. The United States and the United Kingdom have chosen two different architectures for the same basic intervention: the subordination of healthcare provision to political direction. Britain does it openly, through state employment and state ownership. America does it covertly, through tax preferences, regulatory moats, licensure cartels, and direct federal payment. Both produce the pathologies the framework predicts for any sector in which prices do not operate and calculation cannot occur. Both produce a class of rent-seekers whose income depends on the political arrangements sustaining the dysfunction. Both produce a population whose experience of healthcare is dominated by shortage, delay, and inability to discover or influence what any given treatment costs. The difference between the two is aesthetic; the underlying mechanism is the same.

The honest comparison is not NHS versus American market. It is NHS versus American regulatory cartel, with the observation that state direction of healthcare produces state-direction pathologies under whatever institutional form it takes. Neither country has tested what markets in healthcare would actually produce, because neither permits healthcare markets to operate. The evidence we have about what markets produce comes from the sub-sectors that escape the apparatus, and that evidence is consistent.

VI. The Word That Should Be Taken Back

I have been arguing negatively. State direction of a sector produces certain pathologies; the political system protects the institutions producing them; both effects are downstream of a single asymmetry in what failure modes the system can see. What I have not yet done is offer a positive account of what the alternative looks like. The usual move at this point is to gesture vaguely at free markets and trust the reader to fill in the content. I want to decline that move, because the positive claim is specific and inverts a framing that has dominated political discourse for a century.

The claim is that markets, not states, are the only force in modern history that has ever made goods genuinely universal. The state’s version of universalism is in most cases a legal promise of access to things it has prevented from existing in sufficient quantity for the promise to be honored. The word universal, as commonly used in political argument, has come to name the declaration rather than the delivery.

Consider what universal means in ordinary language. A good is universal if essentially everyone has it, everywhere, regardless of wealth or station. Universality is a feature of the world, not of a legal text. The question of whether a good is universal is answered by looking at the world and counting. By this definition, a small and specific class of goods have become genuinely universal in modern history. They are worth naming.

The mobile phone is universal. As of 2021, there were approximately 8.6 billion mobile subscriptions worldwide, more subscriptions than there are people on the planet. Roughly 85 percent of the human population lives within range of a wireless signal. Low-income countries now have approximately 60 mobile subscriptions per 100 people, meaning the average inhabitant of a low-income country is more likely to have a phone than not; the growth rates over the past two decades have matched the global average. Six of the world’s seven billion people have mobile phones, though only four and a half billion have access to a toilet. No agency declared the mobile phone a human right. No international treaty bound governments to provide one. The good became universal because the price mechanism, operating over forty years, drove the real cost from luxury levels to levels affordable by people whose daily income is measured in a few dollars.

Refrigeration is universal in the developed world and rapidly becoming so elsewhere. In 1920, fewer than one percent of American households had a mechanical refrigerator. By 1930, eight percent. By the end of the 1930s, despite the Depression, forty-four percent. By 1980, effectively one hundred percent. Washing machines and vacuum cleaners followed similar trajectories. The microwave oven went from seven percent of American households in 1980 to ninety-two percent by 2017. Air conditioning has multiplied more than fifteenfold in American adoption since 1960. None of these appliances was universalized by state provision. Each followed the same pattern: early adopters paid prices that would today be considered luxury, those luxury prices financed the production scaling and refinement that reduced manufacturing costs, and within a generation or two the good had reached saturation across the income distribution.

Calories are universal, which is the most striking and least appreciated universalization of the past two centuries. For essentially all of human history, the governing problem of food was insufficient quantity. The background fear was famine. In 2024, for the first time in any period for which we have data, more human beings in the world are obese than underweight. More than a billion people are now classified as obese; fewer than 550 million are underweight. Approximately 2.8 million people die each year from the consequences of being overweight, substantially more than die from undernutrition. Benjamin Franklin wrote in Poor Richard’s Almanack, I saw few die of hunger, of eating, a hundred thousand. In 1735 this was wishful irony. In 2026 it is a descriptive statistic. The transition from a world in which hunger was the dominant nutritional risk to a world in which excess calories are the dominant nutritional risk is a change so large that it barely registers as an observation, because the people for whom it is true cannot remember a time when it was not.

Extreme poverty has collapsed. In 1820, according to the most careful historical reconstruction available, approximately 84 percent of the world’s population lived in extreme poverty: on the equivalent of less than two dollars a day in modern purchasing power. By 1910, after ninety years of the Industrial Revolution, 66 percent. By 1950, 55 percent. By 1981, 42 percent. By 2018, 8.6 percent. As of 2025, roughly 10 percent, or around 800 million people. In absolute numbers, approximately 1.5 billion fewer people live in extreme poverty today than did in 1990, on a planet whose population has grown by roughly 2.5 billion over the same period. McCloskey calls this the Great Enrichment, and in sober historical perspective it is the single most important fact about the last two centuries. It was not planned. It was not declared. It was not provided. It emerged from the spread of markets, property rights, and the price mechanism into populations that had previously been outside them, and the timing of the change in each region corresponds with remarkable precision to the timing of that spread.

The Chinese case, often treated as a qualification, is the framework’s largest confirming instance. Between 1978 and 2025, roughly 800 million people were lifted out of extreme poverty in China alone, more than half of the global total. This happened not through the state’s capture of new sectors but through its selective release of sectors it had previously captured. Beginning with Deng’s reforms, the Chinese government progressively uncaptured specific discovery layers: agricultural prices in the late 1970s, coastal manufacturing in the 1980s, retail and services in the 1990s, housing markets in the 2000s. In each released sector, the universalization mechanism ran exactly as the framework predicts. The 800 million were not lifted by a state-run universalization program. They were lifted by the state’s partial withdrawal from the mechanisms it had previously captured, and by the operation of the universalization process in the space that withdrawal opened.

To these examples one could add antibiotics, which after patent expiration fell to prices measurable in dollars per course and are now distributed globally at scales the original discoverers could not have imagined; eyeglasses, which have moved from artisan-crafted luxury to mass-produced commodity; commercial flight, whose real cost per passenger-mile has fallen by an order of magnitude since the 1970s; internet access, which now reaches roughly two-thirds of the human population. The list is finite, but it is large, and it has a common structure.

The structure is the mechanism by which market universalization actually happens. It does not happen through declaration. It does not happen through redistribution. It happens through a process that, at every moment of its operation, looks like inequality. An early version of the good is produced at high cost and purchased by people who can afford luxury prices. Those luxury prices finance further production, further investment, further refinement. As production scales, unit costs fall. As unit costs fall, the good becomes affordable to a wider range of consumers, who buy it at prices still higher than the eventual equilibrium but lower than the initial ones. Their purchases finance additional scaling, additional cost reduction, additional price decline. The process continues until the good has reached essentially everyone who wants it, at a price essentially everyone can pay. Across the examples, the time from first introduction to near-universal adoption clusters around forty to seventy years. The inequality visible at any point in the process is the mechanism by which the eventual universality is produced. Remove the inequality, by, for instance, preventing the early high prices that finance the scaling, and you remove the universalization.

It is worth acknowledging what the state actually contributes to this process, because critics sometimes point out that the mobile phone depended on spectrum allocation, that the Green Revolution received public funding, that antibiotics emerged partly from state-funded basic research, and that international standards treaties coordinate the systems any of these goods operate across. The observation is correct but requires a careful distinction.

Markets cannot function without a substrate of institutional conditions only the state can provide: property rights that are legally secure, contracts enforceable in courts, protection against fraud and theft, a currency whose unit of account is stable, the basic legal infrastructure that allows strangers to transact at arm’s length. These functions are not contested within the market-oriented tradition. Without them the discovery mechanism cannot operate at all. The state as provider of these conditions is the precondition for everything I have argued markets do well.

Beyond this substrate, the state’s contributions become contested. Patent systems, public research funding, spectrum allocation, the specific architecture of international standards: these are particular institutional arrangements among several possible ones, with real costs as well as benefits, and the question of whether they are on balance helpful or harmful to the universalization process is genuinely open. The pharmaceutical patent system arguably accelerates some innovations while delaying others through the regulatory architecture built around it. Public research funding has produced important results, but it has also redirected research talent toward what grant committees reward rather than toward what markets would have directed it toward, and the counterfactual is unknowable. These are not examples of the state enabling market discovery the way rule-of-law functions do. They are specific interventions whose net effects are debated, and the framework gives us reason to think their costs are systematically underweighted because those costs are invisible in the way the framework predicts. The distinction that matters is therefore narrower than the critic’s objection implies. The state that enforces contracts and protects property is enabling the mechanism. The state that captures the layer at which sectoral decisions are made is suppressing it. Between these two clear cases lie a set of intermediate interventions whose effects are contested, and the framework does not require resolving every one of them to establish its central claim.

What makes this mechanism visible as a mechanism, rather than as a description of what happened to happen in certain places, is that its operation varies in intensity with the conditions that sustain it. The process is not all-or-nothing. It runs faster and more completely in jurisdictions whose institutions permit it, slower and less completely in those that constrain it, in a pattern that tracks the degree of constraint with some precision.

The cross-country evidence is the easiest case. A mobile phone, a refrigerator, an antibiotic, a pair of eyeglasses, a safe car, a varied diet: each of these is more universally available in countries with stronger market institutions than in countries with weaker ones, even controlling for raw income. Hong Kong under Cowperthwaite from 1961 to 1971 produced the fastest sustained universalization of consumer goods in any economy of comparable size, and did so under an administrator who deliberately refused to collect economic statistics on the grounds that they would invite intervention. The Baltic states post-1991 reached Western European consumption patterns within a generation through institutional reforms whose principal feature was rapid release of previously captured discovery layers. East and West Germany ran the cleanest natural experiment the twentieth century conducted: same population, same language, same inherited infrastructure, divided by an institutional boundary for forty years. By 1989, per capita consumption of essentially every market-provided good was several multiples higher on the Western side. The variation is not a matter of resources. It is a matter of whether the mechanism that produces universalization was allowed to run.

The within-country evidence is sharper, because it controls for everything that cross-country comparison cannot. A British citizen in 2026 lives in a country with one institutional environment: one set of property rights, one currency, one tax system, one regulatory culture. Yet the universality of goods available to that citizen varies dramatically by sector, in a pattern that corresponds exactly to how much of the sector the state has captured. Food is universally available, cheap, varied, of a quality a prosperous Edwardian could not have imagined. Clothing is universally available, cheap, varied, technically superior to the clothing of any previous generation. Consumer electronics are universally available at prices that have fallen continuously for forty years, with capabilities that expand every cycle. Household appliances, furniture, books, music, entertainment are all universally available and continuously improving. In none of these does a British citizen wait, ration, or treat the item as a scarce good whose allocation requires political attention.

In the sectors the British state has captured, the pattern inverts. Healthcare is rationed by queue. Housing is rationed by price and geography, with prices that have outrun wages for three decades. Childcare is rationed by availability, with prices that have outrun wages for two. Higher education is rationed by credential inflation, with costs that have tripled in real terms within a generation. Primary and secondary education is rationed by catchment area, with quality variation so severe that parents reorganize their lives around it. The same citizens, the same government, the same overall prosperity. The only variable distinguishing the sectors where everything works from the sectors where nothing works is whether the state has captured the allocation mechanism.

The Nordic case, sometimes raised as a counterexample, turns out on inspection to be a confirming one. Denmark, Sweden, Norway, and Finland run tax-funded healthcare systems that deliver outcomes at or above the European frontier. But these systems are not NHS-like state monopolies. They are tax-funded but institutionally pluralistic, with substantial private provision, patient choice, and competition among providers in ways the NHS explicitly lacks. Denmark’s dramatic improvement in cancer outcomes since 2000 followed reforms that introduced competitive elements into cancer pathways. Sweden has had the waiting-list pathologies the NHS has, and its recent reforms have moved toward more pluralism rather than less. The Nordic lesson is that tax-funded systems with preserved discovery-layer mechanisms outperform state-captured ones. The Nordic countries did not displace the Hayekian, Kirznerian, and competitive-discipline mechanisms; they preserved them inside a framework of universal tax-based financing. The outcomes track the preservation.

The stylized fact that captures the pattern most clearly is the divergence in real prices across sectors over the past fifty years. In every developed economy for which good data exist, the sectors with the steepest price declines are those where market competition has been most intense and state involvement least: televisions, computers, software, phones, appliances, clothing, toys, air travel. The sectors with the steepest price increases are those where state involvement has been heaviest: healthcare, housing, childcare, education. The divergence is not small; it is orders of magnitude. A television in 2026 costs, in real terms, perhaps five percent of what an equivalent-quality television cost in 1975, while a year of American college costs roughly three times what it did in 1975 for a product that has arguably not improved. The same citizens buy both. The same economy produces both. The only thing that differs is the mechanism operating in each sector. Where the discovery mechanisms operate, productivity grows and prices fall. Where they are suppressed, productivity stalls and prices rise, and other distortions like the federal student-loan system inflating tuition far beyond instructional cost layer additional rent extraction on top of the underlying capture.

The within-country divergence reveals a second property of the market mechanism that is less frequently noticed but equally important: the good being universalized is continuously improving, and the good the state provides on a universal basis is, in addition to being rationed, failing to improve at the rate it would under competitive provision. Market universalization does not deliver the same good to more people over time. It delivers a better good to more people over time. The mobile phone that reaches a farmer in rural Kenya in 2026 is not the mobile phone that reached a New York executive in 1985. It is a supercomputer with global information access, a camera, a translator, a banking terminal, a navigation system, sold at a fraction of the 1985 phone’s real price. The refrigerator in a modest American household in 2026 is quieter, more reliable, more energy-efficient, and colder than the one a prosperous household owned in 1970, at a lower real price. The eyeglasses a schoolchild receives from a charity program in Bangladesh are made with technologies the wealthiest aristocrats of the nineteenth century could not have bought at any price.

State-provided goods exhibit the opposite pattern. The same mechanism that restricts supply under state direction also arrests the quality improvement that markets generate as a byproduct of competition for customers. The NHS bed the patient waits eleven months for is not only late; it is, compared to the bed that same patient would occupy in a system where providers competed for them, worse. The consulting room is older, the equipment more dated, the scheduling more rigid, the follow-up less personalized, because none of the institutional actors involved face the competitive pressure that would drive these things to improve. The Cuban taxi still running in Havana in 2026 is there because it cannot be replaced; it is also a 1957 Chevrolet, while the same Cuban driver in a market economy would be sitting in a car with anti-lock brakes, airbags, electronic stability control, and fuel economy that did not exist in 1957 at any price. Both things are true, and they are the same thing. The state system delivered fewer cars and delivered worse cars, because the mechanism that would have produced more cars is the same mechanism that would have produced better cars, and displacing it displaced both.

The pattern generalizes. Soviet housing was both scarcer and worse than contemporary Western housing; it was shoddily built, poorly maintained, and architecturally uniform precisely because no producer faced any incentive to make it otherwise. British housing is in short supply; it is also smaller and of older average vintage than housing in European countries with more responsive supply. American healthcare in the sub-sectors the state has dominated is both expensive and technically behind: drug approval takes over a decade because the approval process is the constraint, and the state of the art in any given therapeutic area advances more slowly than it would if the approval process were shorter. State direction produces shortage and stagnation simultaneously, because both are downstream of the same missing mechanism.

This double failure is what makes the comparison with market universalization most stark. The person who receives market-provided calories in 2026 receives better calories than the wealthiest person of 1826 received, in greater variety, at lower real cost. The person who receives state-provided healthcare in Britain in 2026 waits longer for a worse consultation than the patient of 1986 did, despite four decades of intervening medical progress that the institution has partially failed to incorporate. The direction of travel is opposite. Markets move goods toward more people and better versions simultaneously. State systems, in the domains where they have displaced markets, produce both rationing and stagnation: the queue lengthening while the thing being queued for fails to improve.

The distinction this gradient is showing us is between two kinds of universalism that the standard political vocabulary conflates. State universalism is universalism-by-decree: a legal guarantee of access to a good, enforced by political authority, which the state then attempts to deliver by direct provision or administrative allocation. The NHS promising healthcare to every British resident. The Soviet state promising housing to every Soviet citizen. The various constitutional provisions that guarantee food, water, and shelter as human rights. Universalism-by-decree is easy to declare and almost impossible to honor, because the mechanism by which goods are produced in quantities sufficient to reach everyone, at qualities that track the technical frontier, is precisely the mechanism the declaration displaces. The NHS promise was made in 1948; the waiting list has been with Britain ever since, and the service itself has fallen behind peers on outcomes the medical frontier has made possible. The Soviet housing promise was made in 1917; the queue outlasted the state that issued it, and the quality of the housing delivered was so low it is now demolished as a matter of routine maintenance across the former Eastern Bloc.

Market universalism is universalism-by-production. No one declares it; it emerges. No minister announces that calories shall be universal; they become universal, and the calories themselves become cheaper, safer, more various, more nutritionally legible over time. No founding speech inaugurates the era of the universal mobile phone; the phone becomes universal, and the phone itself becomes a radically better device every few years. The good reaches everyone because the process of producing it at falling costs, over decades, eventually crosses the threshold of affordability for the last income decile. There is no queue, because there is no promise, because there is no institution empowered to make the promise. There is only the good itself, cheapening, spreading, improving, showing up in the hands of people who in previous generations would not have had it, in versions their ancestors could not have imagined.

The reason this observation has been absent from political discourse for most of the last century is the visibility asymmetry I started with, now applied to universalism itself. State universalism is legible. It has a founding date, a ministerial sponsor, a founding speech, a legislative text, a name, a logo, a Danny Boyle ceremony. It can be pointed at. It can be celebrated. It can be made a component of national identity. Market universalism has none of these. Nobody in particular made phones universal, and so phones’ universality cannot be attributed to anyone in particular. There is no Department of Mobile Phones, no founding minister of refrigeration, no civic religion around indoor plumbing. The process that did the work is distributed across millions of decisions by billions of people over decades, and its output is taken for granted because no one can point at the person responsible for it. When a citizen surveys the world for instances of universalism, the eye is drawn to what has a name. The institution with the logo wins the category it has not in fact delivered, against the mechanism that delivered the category but has no logo to display.