Putting a Price on It

The economist's answer to emissions is a century old and fits in a sentence: charge for the damage, and let everyone find their own cheapest way to emit less. Arthur Pigou's argument was that a decision-maker weighs only the costs they bear, so a charge equal to the costs they impose on others makes the private calculation match the social one. The mechanism has been essentially uncontroversial ever since.

The virtue of a price is informational, and it is the same argument Hayek made about prices generally. No regulator knows where the cheap reductions are. The firm knows its own boiler, its own process, its own suppliers. Set a price and every firm compares it against its own abatement costs, and reductions happen wherever they are cheapest without anybody central having to find out where that is. A regulation specifying technologies cannot do this, because it requires the regulator to possess knowledge that is distributed across thousands of firms.

There are two ways to implement it. A tax fixes the price per tonne and lets the resulting quantity fall wherever it falls. A cap fixes the quantity of permits and lets trading discover the price. Martin Weitzman showed the choice depends on which uncertainty you fear more: if the damage from an extra tonne rises sharply, guaranteeing the quantity is worth more; if abatement might prove ruinously expensive, certainty about the price protects the economy. Hybrid schemes with price floors and ceilings exist because neither pure form is obviously right.

The European scheme spent its first decade as the standard example of failure. Initial permit allocations were based on industry self-reporting and were issued into a period that then included a recession, leaving a surplus that took years to clear and a price far too low to change any decision. The fix, when it came, was structural rather than political: a market stability reserve that automatically withdraws permits when the surplus grows. Since it took effect the price has behaved as theory always said it would, which is a useful data point about whether the instrument works or merely the implementation failed.

Distribution is where the instrument meets the electorate. A carbon price falls harder on poorer households, who spend a larger share of income on energy and transport. What happens to the revenue decides the net effect entirely: a flat per-person dividend leaves most low-income households better off, because they use less energy than average in absolute terms. Regressivity is therefore a design choice, not a property of the instrument: a point that survives being made repeatedly without much changing how the policy is received.

Around a quarter of global emissions now sit under some carbon price, and most of those prices are far below any serious estimate of the damage per tonne. That gap is where the real argument lives. The instrument is largely settled among economists; the level is set by people who face an electorate afterwards, and it comes out systematically low.

Which raises the question the next chapter has to answer: if the efficient instrument keeps being set too low to work, what has actually been reducing emissions?