The debates.

The case that markets can price them: Coase's 1960 answer was that externalities are not a market failure but a property-rights failure. Where rights are clearly assigned and bargaining is cheap, the parties will trade their way to the efficient outcome regardless of who holds the right, so the state's job is to define and enforce entitlements rather than to set prices. The practical descendants are tradable-permit systems, which use a market to find the abatement cost a regulator cannot know. Elinor Ostrom's fieldwork strengthened the case from an unexpected direction: studying irrigation systems, fisheries, and forests, she documented communities governing shared resources sustainably for centuries through self-organized rules, neither privatized nor state-run, which is evidence that the choice is not simply market or state.

The case that they cannot: Coase's own conditions are the objection. Where transaction costs are high and the affected parties number in the millions and are not yet born, as with climate, there is no bargaining table to sit at, and the theorem describes a world that does not exist. Weitzman's 1974 result cuts deeper into the tradable-permit answer: under uncertainty, prices and quantities are not equivalent instruments, and which one performs better depends on the relative slopes of marginal cost and marginal benefit, so the market-based remedy still requires a regulator to make a judgment the market cannot make for it. Ostrom's own design principles, on this reading, describe conditions (clear boundaries, monitoring, graduated sanctions) that a global atmosphere conspicuously lacks.

The case that it is separable: Stigler's 1971 analysis is usually read as an indictment of capitalism, but its logic points the other way: regulation is captured because regulation is worth capturing. Concentrated benefits and dispersed costs mean an industry will always lobby harder over its own rules than the public will lobby against them, which makes the size of the discretionary state the variable that matters. On this account cronyism scales with the prize available, so the remedy is fewer discretionary levers to buy, and the wide variation in measured corruption across rich capitalist democracies is the evidence that institutions make the difference.

The case that it is not: The reply is that this treats political power as exogenous to economic power when it is produced by it. Capitalism generates concentrated wealth by design through capital accumulation, and concentrated wealth buys the rules, so a system that produces billionaires produces the lobbying that follows. Olson's 1982 argument sharpens this into a prediction rather than a complaint: stable societies accumulate distributional coalitions over time, and those coalitions slow growth and resist removal, so sclerosis is the expected end state of a successful capitalist democracy rather than an avoidable accident. On this view the shrink-the-state remedy is self-defeating, because deregulation is itself something concentrated interests lobby for.

The case for crediting market liberalization: The fall is concentrated in the places and periods where markets were opened, which is not a coincidence. China after 1978 and India after 1991 account for a very large share of the total, and in both the turning point followed the abandonment of central planning and autarky rather than preceding it. The mechanism is not mysterious: price signals directed capital toward labor-intensive export manufacturing, which absorbed hundreds of millions of workers out of subsistence agriculture at wages they voluntarily accepted. Deaton's account treats growth as the engine and stresses that the escape from poverty and the escape from early death are the same story.

The case against crediting it: The strongest version does not deny growth mattered; it denies the label does the work. If the two decisive cases are China and India, then what is being credited is a one-party state that retained the commanding heights and a post-colonial developmental state, neither of which is the model this site's center band describes. Critics including Hickel add that the result is sensitive to the poverty line chosen, and that a threshold this low measures the elimination of destitution rather than the achievement of decent life. Others allocate more of the gain to public health, vaccination, sanitation, and literacy, which are largely state-provided and improved even in periods and places where growth did not.

The case that it misallocates: The claim is not that state direction never works but that it cannot correct itself. Where credit is allocated by policy rather than price, loss stops functioning as information, and the errors compound instead of clearing, which is what a property sector that ran to a quarter of output and a local-government debt overhang look like from inside the model. Lardy's work supplies the internal evidence: China's fastest growth came when private firms were gaining ground, and the post-2013 resurgence of state enterprise coincides with a measurable decline in trend growth, which suggests the market component was doing the work and the state component is now a drag.

The case that it is the counterexample: The reply is that the model's own record on development is worse than it admits. Studwell's comparative account of Japan, South Korea, and Taiwan argues that every late-developing economy that caught up did so through land reform, directed credit, and export discipline rather than through open markets, and that the countries which followed liberalization advice did not catch up. Rodrik's version is more general: the policies that produce growth are not the ones the textbook prescribes, and the binding constraint differs by country, so a state able to identify and relieve it will outperform a state forbidden from trying. On this reading China is not an anomaly requiring explanation but the largest instance of the actual pattern.