An externality is a cost or benefit produced by a transaction that falls on someone who was not party to it. Pollution is the canonical negative externality: a factory makes steel, a downwind neighborhood breathes the smoke, the price of steel does not reflect the health cost. Vaccination is the canonical positive externality: one person's inoculation reduces every other person's risk of infection, but no one pays the vaccinated person for that benefit.
Pigouvian Taxes and the Standard Fix
Arthur Pigou's The Economics of Welfare (1920) proposed the mechanism that still bears his name: tax the externality at the level of its marginal social cost. A carbon tax is a Pigouvian tax. Tobacco excise is a Pigouvian tax. Subsidies for childhood vaccination are inverse-Pigouvian transfers. The principle is simple — line the private price up with the social cost, and the market will allocate correctly on its own.
The principle is simple and the implementation is fraught. Estimating the marginal social cost requires knowing the damage function, which for climate change runs from $51 per ton (the US Interagency Working Group's 2021 central estimate) to $190 (the EPA's 2023 revision) depending on discount-rate choice. The estimate is the policy, and the political economy of selecting an estimate is what most carbon-pricing fights are actually about.
The Coase Critique
Ronald Coase's The Problem of Social Cost (1960) is the most cited counterargument. Coase pointed out that in a world of zero transaction costs, externalities solve themselves through bargaining between affected parties, regardless of who initially holds the property right. If a factory is polluting a downstream fishery, either the factory pays the fishers to accept the pollution or the fishers pay the factory to abate, depending on which is cheaper — and the outcome is efficient either way.
Coase did not claim transaction costs are zero. The whole point of his paper was that they are usually large, which is why governments have to intervene. But the "Coase theorem" got reframed in libertarian hands as a claim that all externalities can be handled by private bargaining if property rights are clearly defined. The actual conclusion of Coase's paper was almost the opposite: that institutional design, including tort law, regulation, and taxation, is unavoidable because transaction costs are real.
Why the Climate Externality Is Special
Climate change has features that make it the hardest externality problem ever posed. The emitters are dispersed across every country and every household. The damages fall mostly on future generations and on poor countries that emitted little. The atmospheric stock is global, so unilateral action by one country produces benefit shared globally — the standard free-rider problem, but at the scale of every nation simultaneously. And the lag between emission and damage runs decades to centuries, which is far longer than any electoral cycle.
The conventional Pigouvian fix — tax carbon at its social cost globally and let markets adjust — runs aground on every one of these features. The political economy of getting 195 countries to agree on a single global price has so far been impossible. The patchwork that has emerged — EU ETS, the California program, Sweden's $130/ton tax, the US IRA's subsidies — is the second-best solution that history has produced because the first-best is politically inaccessible.
Positive Externalities Get Less Attention
The asymmetry in public discourse — endless attention to negative externalities, much less to positive ones — has real policy consequences. Basic research is systematically underfunded because its spillover benefits cannot be captured by the researcher. Vaccination programs are politically embattled even when the social return on investment runs in the thousands of percent. Public-domain knowledge production — open-source software, open-access journals, Wikipedia — operates on volunteer labor because the market price for the contribution does not reflect its social value.
Mariana Mazzucato's The Entrepreneurial State (2013) is the most-cited recent book on the positive-externality side. Her argument is that nearly every component of the modern technology stack — internet, GPS, touchscreen, voice recognition, basic pharma research — originated in publicly funded work whose social return dwarfed the private return any single firm could have captured. The implication is that the standard market-failure argument for public R&D funding is stronger, not weaker, than it appeared in the 1960s.
The Honest Reading
Externalities are not a marginal case. They are pervasive. Every act of consumption has spillover effects on the climate, on traffic, on public health, on neighbors. Every act of production has spillover effects on workers, on suppliers, on downstream ecosystems. The question is not whether markets handle externalities well — they do not, by definition — but which externalities are large enough to warrant correction and which corrections actually deliver. The Pigouvian fix works in domains where the damage is measurable and the political economy permits a tax to stand. It fails in domains where the damage is contested or the politics is uncongenial. Most of climate policy operates in the second category, which is why the first-best solution remains unimplemented and the second-best is what we have.
The Recurring Pattern
Externalities are not a peripheral category. They are the everyday case across environment, healthcare, education, transportation, urban form, financial stability, and labor markets. The Pigouvian framework provides the cleanest theoretical fix but runs into political-economy obstacles whenever the affected industry is large enough to mobilize against the tax. The Coasian alternative — let private bargaining solve externalities — works only when transaction costs are low, which for most large externalities they are not. The empirical record is that externalities require sustained political work to address, that the political work is hard, and that the absence of the work means the externalities continue to be borne by parties who did not consent to them.