Should a carbon tax rise or fall over time?

A carbon tax has been proposed as one possible way of reducing the emissions of ''greenhouse gases''. Much of the recent work on carbon taxation has concentrated on analysing the effects of introducing such a tax at a level which would meet certain emission targets at some fixed date. Less attention has been devoted to the time profile of this tax. Yet in securing international agreement on the introduction of a carbon tax it could matter greatly whether what is being proposed is a tax which is initially low but rising, or initially high but falling. We show that in a wide class of cases the optimal time path of the tax will be that it initially rises and then falls as the exhaustion constraint starts to bite. However there are few general conclusions, and much depends on more detailed modelling assumptions on which the scientific evidence gives no firm guidance. (author)