We All Live Downwind

Living downwind from a chemicals plant can be strange. The water tastes wrong, the air is worse, the kids in school (and their parents) fall ill with odd diseases. A neighbor's cancer, your own asthma, and the illnesses moving through a whole community become economic weaknesses. And, in turn, the store closes, jobs follow it out the door, some families go hungry. The chemicals firm, meanwhile, reports record earnings.

We all live downwind. Literally, from a factory or a freeway shedding microplastics with every mile driven. Metaphorically, in a world shaped by the consequences of others — a warming climate, an emptying ocean, harms we inherit rather than choose. The costs are real; the accounting is incomplete.

And this is not a new observation, but it is only in the last century that economists have tried to account for it at all.

The dark and sinister mills of the industrial revolution spewed coal dust and smoke, and caused injuries and miseries as they lifted England from rural poverty to a global powerhouse. These costs swiftly caught the attention and imagination of artists and writers. But it was only in the early twentieth century, through Marshall, Pigou, Ramsey, and others, that economists began to wrestle with the idea that such costs needed accounting for at all. A hundred years later, the job is not done, even as the scale of externalities has soared. Three figures shaped how we still argue about it.

Pigou (1920) framed externalities as a divergence between marginal private and marginal social cost. If a producer doesn't bear the full cost of what he does, he produces too much of it, and the otherwise efficient market misallocates. The Pigouvian remedy is a corrective tax equal to the marginal external damage, and a bounty for positive externalities, which restores the price signal without abolishing the market. This remains the theoretical basis of every carbon price. The weaknesses are real: it assumes a state that knows the costs of damage and is willing to act on them; the revenue goes to the general treasury rather than to mitigating the harm that justified it; and in inelastic markets the harm continues, tax or no tax.

Coase (1960) attacked exactly that assumption about government, and reframed the problem as reciprocal. The factory harms the laundry, but restraining the factory harms the factory. To Coase, asking "who is the cause" is the wrong question; the right one is which allocation of rights yields greater total value. In a world of well-defined property rights and costless bargaining, the parties reach an efficient outcome regardless of who holds the right. Coase understood perfectly well that transaction costs are never zero, so the theorem is a benchmark, not a policy. What follows from it is that you must compare real, imperfect institutional arrangements, courts, firms, regulation, doing nothing, on their actual costs, rather than comparing a flawed market against an idealised regulator.

Friedman (1970) had an incoherent response to the growing scale of externalities. He conceded neighbourhood effects as one of the few legitimate grounds for state action, citing airborne pollution by name, and parts of his writing back Pigou on taxes. He then spent considerable energy on the reasons not to use them: damages are hard to measure, intervention is hard to confine once granted, and the costs of government, he theorised, can exceed the market failure being corrected. His position on corporate responsibility pushes the problem back to the legal system, a firm should pursue profit within the rules of the game, so fixing externalities means fixing the rules, not exhorting managers. He also broadly accepted that companies would work to shift the rules of the game in their own favour, which rather undercuts the destination he was pointing at. He is the government-failure counterweight, and a major reason a Pigouvian tax has never been an easy sell.

Why so little has moved

The three positions are stable rather than resolved, because each is right about the others' weaknesses while each holds an unsustainable central position of its own. But the deeper reasons progress has been scant are practical.

The number doesn't exist. You cannot tax, litigate, insure, or bargain over a harm nobody has credibly quantified. Damage functions are contested, uncertainty bands are wide, and where the science is unsettled, the operative default is zero — the one value known to be wrong. That coal's external costs run to some multiple of its revenues is a finding beyond dispute, but no balance sheet anywhere reflects it.

Attribution crosses jurisdictions and generations. The emitter and the bearer are commonly in different legal systems, sometimes in different centuries. There is no forum in which the affected party has standing.

Coasean bargaining doesn't scale. It may work between a factory and a nearby laundry. It fails when the counterparties number in the millions, and are diffuse, poor, or unborn.

The political economy is asymmetric. The costs of internalising are concentrated, visible and immediate: a particular plant, jobs at risk, higher prices. The benefits are diffuse and statistical. Political pressure follows the asymmetry. The result is visible in carbon pricing, where observed prices remain well below many estimates of the damages emissions impose.

Even agreed physics doesn't produce an agreed number. Ramsey's 1928 discounting formula, written to answer how much a nation should save for its successors, turns out to determine what future damages are worth today, and the Stern–Nordhaus dispute over its parameters moves the present value of identical damages by a factor of four. A century on, the argument that made the accounting possible is also what keeps it permanently open.

Where Calx sits

Calx works on the problem that comes before the policy choice, which is establishing what the harm actually costs. A century of argument has been about which instrument, tax, legal right, or nothing. Very little of it has been about the input all three require: a defensible, decision-usable estimate of the harm.

For Pigou, we supply the missing numbers, with explicit uncertainty, rather than waiting for standardised reporting or a scientific consensus that may never arrive. Producing usable estimates while the science is still unsettled is our methodological bet. A well-sourced range with stated confidence is vastly more useful than silence.

For Coase, we lower the informational component of transaction costs. Rights can't be traded over a harm that hasn't been attributed or priced; making harms legible gives insurers, litigants and regulators something to contract over. The information asymmetry underpinning those assumed-prohibitive transaction costs is exactly what agentic search, AI analytics and the Calx taxonomy are built to reduce.

For Friedman, we route around the state as the sole determinant of outcomes. Information delivered to lenders, insurers, investors and procurement lets an externality be priced through cost of capital and insurability, driven by private actors acting in their own interest, no statute required, and no new agency to be captured.

Beyond these three, two additions were unavailable to earlier generations. We aim to model harms as interacting and cascading rather than as separable line items, because a coal facility's carbon, mercury and particulate harms compound through the same populations and systems. And we treat the operative question as which intervention changes outcomes most per unit spent, pushing toward innovation, invention and investment, and turning an unresolvable welfare-economics problem into a tractable question of engineering and return.