Mankiw, one of the leading proponents of a carbon tax, claims to be agnostic
about the carbon tax vs cap-and-trade debate,
at least if carbon credits under a cap-and-trade system were auctioned rather
than freely allocated:
Of course, selling emission allowances under a cap-and-trade system makes
the system equivalent to a comparably-sized Pigovian tax.
Mankiw is surely wrong here: the "of course" guarantees that, even
if he’s right on substance. It’s conceivable that an auction-based cap-and-trade
system might end up being equivalent to a comparably-sized carbon tax. But it’s
hard to know that ex ante, and it’s certainly not obvious.
The biggest difference, of course, is that even if government revenues are
the same under both systems, total carbon emissions are not. In a cap-and-trade
system, emissions are capped. That’s the whole point. (And it’s why "safety
valves" are a very bad idea.) There’s no cap on emissions under a carbon-tax
system.
Also worth reading, in Mankiw’s comments, a case for why a cap-and-trade system
actually involves less bureaucracy and overhead than a carbon tax:
You’ve yet to establish that a system that requires bureaurcracy, invoicing,
billing, and payment processing every year (and all run by the federal government)
is as cheap as a system that only requires costs when caps are transacted
or people are looking for caps to purchase (whether you choose the initial
purchase or not), all run by the private sector.
I also happen to suspect that you could make a case for lower enforcement
costs under caps too (if random checks are feasible) versus a system where
you must measure *every single participant* to bill them correctly.