Subsidy, Standards and Borders

Having established that a carbon price is the efficient instrument, an inconvenient fact: it is not what produced most of the emissions reductions achieved so far. Deployment subsidies, efficiency standards and outright bans did more of the measurable work.

The clearest case is solar. Germany's feed-in tariff, and later Chinese manufacturing support, bought an expensive technology in volume. Every doubling of cumulative installed capacity cut the price by roughly a fifth, and the relationship held for four decades across an order of magnitude of cost reduction. Judged as emissions abatement at the time, the subsidies were extravagantly expensive per tonne and were criticised on exactly those grounds. Judged as an investment in permanently cheaper energy for everyone afterwards, including every country that contributed nothing, they were among the better uses of public money in the period.

The mechanism is the learning curve, first documented in aircraft manufacturing in the 1930s and since found across a wide range of manufactured goods. Unit costs fall regularly with cumulative production, through process improvement, scale and accumulated engineering knowledge. Its policy significance is that early purchasing is an investment rather than pure expenditure, and standard cost-per-tonne accounting misses this entirely, because the benefit lands outside the project's boundary and mostly outside the purchasing country.

Then there is leakage, the standard objection to acting ahead of your trading partners: production relocates to where emissions are untaxed, and global emissions do not fall. The measured effect has generally been smaller than predicted, because energy is a modest share of total costs in most industries and firms weigh many other things when choosing where to produce. It is genuinely significant in a handful of energy-intensive sectors (cement, steel, aluminium, fertiliser) which is precisely where border measures have been aimed.

A border carbon adjustment charges imports for the carbon embedded in them, matching the domestic price. It closes the loophole by removing the advantage of producing somewhere unregulated and selling into a regulated market. It is also, viewed from outside, a charge on foreign producers that domestic ones escape, which is a serviceable description of a tariff. Whether it survives challenge at the trade bodies is one of the more consequential open questions in the field, and the answer will shape whether any country can price carbon seriously without waiting for everyone else.

Which leaves the uncomfortable conclusion. The second-best policies won. A subsidy is less efficient per tonne than a price and it creates a constituency (manufacturers, installers, the workers in a factory) that defends it at the next election. A carbon price is more efficient and creates a constituency that attacks it, visible at the fuel pump and attributable to a named government.

For a problem requiring several decades of consistent policy across changes of government, surviving is a property worth a great deal. That suggests political durability belongs inside the analysis rather than as an afterthought appended to it, which is a harder thing for the profession to absorb than any of the technical results, because it implies the efficiency ranking is not the whole ranking.