The Healthcare Market Cannot Heal Itself: Notes on Health, Markets, and Power

In 1963, Kenneth Arrow published what remains one of the most important papers ever written about healthcare markets. “Uncertainty and the Welfare Economics of Medical Care,” which appeared in the American Economic Review, did not argue that medicine should be nationalized or that doctors were greedy. It argued something more precise and more damning: that the market for medical care is structurally different from other markets in ways that make the standard case for competitive efficiency as an organizing principle simply inapplicable to healthcare. The demand for care is irregular and unpredictable. The product cannot be inspected before purchase. The seller knows vastly more than the buyer. And the consequences of a bad transaction, unlike, say, a bad haircut, can be catastrophic and irreversible. Arrow concluded that these factors combine to ensure that the normal supply-and-demand apparatus would fail in healthcare, and thus some form of social intervention was not merely politically called for but economically necessary.
‍

More than sixty years later, Arrow's diagnosis has held. What he could not fully anticipate was how the political economy of that social intervention would itself become a source of failure, arguably as consequential as the market failures it was meant to address. Political economy, in the classical sense, is the study of how power and economic interest shape the rules under which markets operate: not markets as they exist in models, but markets as they are actually constructed, governed, and contested by organized actors pursuing advantage. It is the discipline that asks not only what the optimal policy is, but who benefits from the status quo, who has the power to change it, and why they so rarely do.
‍

*****
‍

The information asymmetry Arrow identified cuts in multiple directions. Patients generally cannot evaluate the comparative quality of care they receive, which gives providers pricing power unconstrained by the usual competitive discipline. Insurers cannot perfectly observe patient behavior, generating moral hazard on the demand side. And employers, who in the United States serve as the de facto administrators of the insurance market, cannot easily evaluate the actuarial quality of healthcare plans they offer which further blunts competitive pressure on the supply side.

But information asymmetry is only the beginning. Healthcare also exhibits features of a public good (i.e., goods that are non-rival and non-excludable) in certain dimensions; for example,  epidemic controls and vaccinations generate positive externalities that private markets systematically under-provide. The healthcare system involves massive fixed costs in research and infrastructure, creating natural monopoly tendencies. And it is subject to severe supply constraints: the number of licensed physicians, hospital beds, and approved drugs is not determined by market equilibrium but by regulatory gatekeeping. The result is a sector that combines the worst features of monopoly, public goods failure, and asymmetric information simultaneously.

Orthodox welfare economics has a standard toolkit for these types of situations: regulate monopolies, subsidize public goods, and mandate disclosure. While the United States has deployed versions of all of these, what it has not reckoned with seriously enough is the political economy that governs how those interventions get captured and ultimately defanged.
‍

*****

The political logic was laid out with sharp clarity by Mancur Olson in The Logic of Collective Action only a couple of years after Arrow’s paper. Small, well-organized groups with concentrated interests will consistently outmaneuver large, diffuse groups even when the latter represent a majority. The American Medical Association spent decades successfully lobbying against government-run health insurance, not because patients lacked preferences but because patients lacked coordination. Pharmaceutical companies invest heavily in regulatory lobbying not because the investments are certain to succeed but because the expected returns, measured in patent extensions, favorable formulary placements, and weakened price negotiation authority, are extraordinary relative to the lobbying cost.

The data support this reading. A 2018 study in JAMA by Papanicolas, Woskie, and Jha found that the United States spends nearly twice as much on healthcare per capita as comparable high-income nations, without achieving commensurately better outcomes on most population health metrics. The gap is not primarily explained by greater utilization of services—Americans do not, for example, see doctors particularly more often than the Swiss or Germans. Rather it is explained overwhelmingly by prices: higher prices for drugs, for procedures, for administrative overhead, and for physician compensation. Those prices are not set in a competitive market. They are the product of negotiated arrangements between concentrated suppliers and imperfectly organized buyers, arrangements shaped at every level by political intervention.
‍

Consolidation and vertical integration across the system illustrate the mechanism with compounding force. On the insurer side, Dafny, Duggan, and Ramanarayanan (2012) exploited the natural experiment created by Aetna's 1999 acquisition of Prudential Healthcare to show that rising insurer concentration causally increased employer-sponsored premiums by roughly seven percentage points across affected markets while simultaneously depressing physician employment and earnings, a finding consistent with the exercise of monopolistic buying power. On the hospital side, Dafny (2009) found that hospital mergers produced price increases well in excess of 40 percent, with essentially no quality offset.

But the most striking contemporary example of consolidation-as-extraction may be the creation of pharmacy benefit managers (PBMs). Originally conceived as neutral intermediaries to help insurers process drug claims and negotiate manufacturer discounts, PBMs have transformed into something far more powerful and far more conflicted. In 2004, the top three PBMs managed 52 percent of prescription drug claims; today, CVS Caremark, Express Scripts, and OptumRx together manage 79 percent of claims for roughly 270 million Americans. More consequentially, each of these leading PBMs is now fused with a major insurer: CVS with Aetna, Cigna with Express Scripts, and UnitedHealth with OptumRx. The resulting structures control the insurer, the drug benefit administrator, and frequently the pharmacy itself. This vertical integration creates incentives that are almost perfectly misaligned with patient welfare: a PBM can charge insurers more for a drug than it reimburses the dispensing pharmacy (a practice known as spread pricing) while simultaneously steering patients toward the pharmacies it owns, and negotiating rebates from manufacturers that favor higher-priced drugs over cheaper therapeutic equivalents. An ongoing investigation by the FTC has concluded that this consolidation may have enabled PBMs to reduce competition and inflate drug costs across the supply chain.
‍

Private equity's rapid acquisition of physician practices represents a fourth and newer vector of the same dynamic. Between 2012 and 2021, PE-acquired physician practice sites grew from 816 locations across 119 metropolitan areas to 5,779 locations across 307 metropolitan areas for a sevenfold expansion in under a decade. In markets where a single PE firm achieved high penetration, prices were between 1.5 and 3 times higher than in comparable markets, across specialties including gastroenterology, dermatology, and obstetrics. The mechanism in this case is structurally different from hospital consolidation: PE firms acquire practices through rapid “add-on” roll-ups, extract returns over a 3-to-7-year window, and then sell, most typically to another PE firm, in a secondary buyout that restarts the cycle.

While each link in these chains of consolidation are legally permissible, the aggregate result is a sustained transfer of control and value from patients and payers to financial intermediaries who have little or no long-term stake in care quality and face limited regulatory constraint to calibrate their actual market behavior.

*****

None of this is uniquely American, though the American version is unusually severe. Every developed country, for example, faces some version of the pharmaceutical pricing dilemma: the fixed costs of drug development are enormous and largely sunk, while marginal production costs are low, creating a genuine tension between innovation incentives and access. Every country with a fee-for-service payment structure faces provider-induced demand, where the volume of services delivered reflects supplier revenue optimization as much as patient need. This phenomenon has been rigorously documented in the US in work like Gruber and Owings' (1996) analysis of cesarean section rates increasing following negative income shocks to obstetricians.

What varies across systems is not the underlying incentive structures per se, but rather the institutional capacity to resist regulatory capture. Single-payer or tightly regulated multi-payer systems in Canada, Germany, and the United Kingdom have not eliminated rent-seeking by suppliers, but they have concentrated the bargaining power of the buyer (i.e., the government) in ways that partially counteract it. The result is not optimal, but it is less distorted.

Germany's 2011 pharmaceutical pricing reform (the Arzneimittelmarkt-Neuordnungsgesetz, or AMNOG) offers a case study in what concentrated buyer power actually looks like in practice, and what it can and cannot accomplish. Before AMNOG, manufacturers launching new drugs in Germany were free to set their own prices for the first year on the market, subject only to ex-post rebate negotiations. The reform changed the game structurally: pharmaceutical companies are now required to submit a dossier to the Federal Joint Committee demonstrating added patient-relevant benefit in mortality, morbidity, and quality of life. That assessment then becomes the basis for price negotiations with the National Association of Statutory Health Insurance Funds which, because it negotiates reimbursement amounts for private insurance programs as well as government programs, effectively acts as a monopolist on the demand side. The empirical results vindicate the design: a triple-difference analysis of drug launches across 24 OECD countries found that the AMNOG negotiation process successfully reduced drug prices by 16.4%.

Yet the German case also illustrates the limits of non-market based systems designed to alter incentives and behaviors. Under the reformed German system, pharmaceutical companies retain the option to withdraw a drug from the German market rather than accept a negotiated price—and some have exercised it, a form of rent-seeking that simply relocates rather than eliminates the underlying tension. The creation of countervailing market power through a monopoly can lead to reduced and delayed introduction of drug innovations despite temporary patent protection which creates a genuine access cost that policymakers must weigh against the pricing discipline the centralized system imposes. The German experience does not, in other words, fully solve the political economy of pharmaceuticals. It demonstrates, rather, that an empowered government / monopoly buyer can partially counteract supplier pricing power, but also pushes rent-seeking into new channels that future regulation will need to address.

Australia's Pharmaceutical Benefits Scheme (PBS) offers a second, and in some ways more revealing, international case study. Where Germany's AMNOG is a reform grafted onto an existing multi-payer system, the PBS is a founding institution of the modern Australian welfare state, operating continuously since 1948. Its logic is similarly anchored in concentrated buyer power and evidence-based gatekeeping: a medicine is listed on the PBS only if an independent expert committee recommends it after drawing on evidence for clinical and cost-effectiveness. Following such a recommendation, the government negotiates its price with the supplier. The results are dramatic by any measure. Atorvastatin, a cholesterol medication among the ten most prescribed drugs in Australia, costs Australians the equivalent of roughly $21 per prescription; Americans pay $2,628 for the same drug—a price difference of 125-fold. Similarly, a GAO analysis of 41 high-expenditure, brand-name drugs confirmed the pattern systematically, finding that U.S. prices were higher on average than prices in Australia, Canada, and France across the sample.

What makes the Australian PBS particularly instructive as a case study in political economy is the reaction it provokes. The Pharmaceutical Research and Manufacturers of America has long remained critical of the PBS's reference pricing mechanism and in March 2025 prominent American pharmaceutical companies urged the U.S. President to consider Australia's PBS an unfair trade practice, pressing for retaliatory tariffs on the grounds that the scheme systematically devalues U.S. medicines. The industry's argument that Australia is essentially free-riding on American patients who pay high prices that cross-subsidize global R&D contains a kernel of truth that the PBS's defenders do not always acknowledge. But it also confirms the central claim here: a well-designed buyer-side institution can extract substantial price concessions from pharmaceutical suppliers, and the suppliers know it. Their lobbying effort is itself the most persuasive evidence that the PBS works.

These international examples suggest something counterintuitive to libertarian sensibilities: that countries which achieve more efficient healthcare outcomes are not necessarily those with the freest markets, but rather those with the strongest concentrated negotiating capacity. The market logic that applies to most goods (competition drives prices down and quality up) does not straightforwardly apply here, because the conditions for competitive equilibrium (price transparency, homogeneous goods, informed buyers, free entry) are systematically absent.

*****

Against this backdrop, the enthusiasm for digital health technologies as a potential structural corrective also deserves scrutiny. The argument runs as follows: machine learning can reduce diagnostic error, wearables can enable continuous monitoring, interoperable electronic records can reduce redundant testing, and telemedicine can increase access. Each of these claims is plausible in isolation. But technology does not operate in an institutional vacuum. Electronic health records, promoted in the US with enormous federal subsidy under the HITECH Act, were designed primarily to facilitate billing rather than care coordination, a consequence of the political economy that shaped their implementation. Interoperability has been strongly resisted by incumbent vendors who profit from data silos. And the platform dynamics of digital health threaten to reproduce, in a new register, the same consolidation dynamics that produced today's hospital oligopolies.

Artificial intelligence, the latest technology to be offered as a structural corrective, appears to be following the same institutional logic. The most telling evidence is where AI investment is actually concentrating. Under the current fee-for-service framework, developers and health systems face strong financial incentives to invest in administrative AI applications such as billing, documentation, and prior authorization where returns are clearer and independent of coverage decisions; by contrast, adoption of clinical AI has lagged due to limited assurance of reimbursement. In essence, the payment system is selecting for AI that optimizes extraction over AI that improves care. The prior authorization arms race is perhaps the purest illustration: both physicians and payers are investing in AI to speed up the same adversarial process, each side trying to gain an advantage, while the fundamental problem of misaligned incentives and fee-for-service payment models remains untouched. Meanwhile, at the market structure level, healthcare incumbents and private equity firms recognize that AI implementations simultaneously drive revenue growth and margin improvement. And they are acquiring AI companies accordingly, folding them into the same consolidating structures described above. Technology, in the absence of institutional reform, does not disrupt the political economy of healthcare. It inherits it.

*****

The political economy of healthcare is not, in the end, a technical problem awaiting a better model. It is a problem of power, who has it, how they preserve it, and what it would take to redistribute it. Arrow gave us the intellectual framework to understand why health markets fail. What we have been slower to reckon with is that the political response to those failures is itself shaped by the same forces that produced them. Insurers, hospitals, and pharmaceutical companies are not villains in this story. They are economically rational actors in a system designed largely by themselves through decades of legislative and regulatory engagement to reward their rationality at collective expense.

Healthcare reform will not arrive because the counter-arguments finally became compelling. The arguments have been compelling since 1963. It will arrive, if it does, because the political coalition sustaining the current arrangements finally fractures, because some concentrated interest calculates that a different settlement serves it better, or because the costs to diffuse interests grow acute enough to generate the organization that Olson's logic and years of experience demonstrate is so difficult to create and harness.

‍