If I'm reading the arguments made here correctly, the lay of the land is as follows: We agree that in theory a cap-and-trade system and a tax can be designed to be similar with respect to their coverage, their effects, and, to some extent, their administrative costs. For example, both can be equally imposed upstream or downstream, equally grandfather existing emissions, raise the same revenue, and, at least roughly, provide the same level of price certainty and emissions certainty. But this all depends on how well the systems are designed. The disagreement seems to be largely about which one will be susceptible to bad design decisions, or more crudely, which one will Congress screw up more. Congress is good at writing bad tax laws; would it also bungle a cap-and-trade regime? And will the administrative costs of a cap-and-trade regime as actually implemented exceed those of a tax as actually implemented?
It's unclear how to determine the answer to these sorts of questions. We cannot perform an experiment, nor do we have natural experiments or historical data series to analyze. Existing carbon tax regimes aren't very good, but neither are the existing carbon cap-and-trade regimes--most significantly the EU Emissions Trading System (ETS). However, the existing carbon taxes and the EU ETS were enacted at different times with different information, so it's not clear that they can be directly compared. At best, we have intuitions about likely effects. My intuition tends to lie with Janet Milne's--a tax would be simpler because the institutions already exist. Perhaps one thing we might get out of the current financial crisis is how difficult it is to design well-functioning markets. On the other hand, I'm a tax lawyer by training and have plenty of experience with bafflingly complex tax laws.
Let me suggest two different lines of argument that might help resolve the issue in the absence of information about likely administrative costs. The first is that a cap-and-trade system must have a price ceiling (often called a safety valve) and a price floor. The goal of putting a price on emissions is to make polluters face the full costs of their actions. A price is necessary because the market price doesn't reflect the external harms that come from pollution. In a cap-and-trade regime, if the price spikes or tanks, polluters will very likely no longer face the true external costs of their actions. For example, if the price drops precipitously as happened in the EU ETS, industry would be able to pollute even if the external costs remain high. Similarly, if the price spiked because of short-term demand as happened in the Californian RECLAIM system, the regulatory cost would exceed the true external costs.
This lack of tracking between the trading price and the external harm is the major deficiency of a cap-and-trade system. A price floor and ceiling remedy the problem by ensuring that permits trade within a reasonable band. My guess is that many cap-and-trade advocates wouldn't accept a system with a ceiling and floor. Those who want absolute certainty on total emissions wouldn't accept a ceiling. Industry, which argues for a ceiling, wouldn't want a floor. To me, this means that we should start with a tax rather than risk getting a flawed cap-and-trade system.
Second, I believe that the central issue for choosing a pricing instrument, an issue that has unfortunately received scant attention in our discussions so far, is how each of the systems would be implemented internationally. A climate regime must involve all major emitters and the choice of instrument can affect whether we achieve broad participation and whether a treaty is enforceable. The claimed advantage of a cap-and-trade regime is that it increases participation because countries can be bought off through extra emission allowances, as was done in Kyoto. In theory, exactly the same thing could be done with taxes by transferring a portion of the tax receipts, but the transparency of such transfers may make this infeasible. On the other hand, buying off governments would likely involve giving billions of dollars of permits to badly behaved governments, something that most of the world would not agree to. And if our current foreign aid budgets reflect how much we're currently willing to give to foreign governments, there's no reason to think that we would spend orders of magnitude more in a climate treaty. Moreover, governments have little incentive to enforce caps on their own industries: allowing domestic cheating helps local industries and imposes externalities on the world. Taxes, I think, dominate in the international context.