by Robert Stavins, via An Economic View Of The Environment
The recent demise of serious political consideration of an economy-wide U.S. CO2 cap-and-trade system and the even more recent resurgence in interest among policy wonks in a U.S. carbon tax should prompt reflection on where we’ve been, where we are, and where we may be going.
Almost fifteen years ago, in an article that appeared in 1998 in the Journal of Economic Perspectives, “What Can We Learn from the Grand Policy Experiment? Lessons from SO2 Allowance Trading,” I examined the implications of what was then the very new emissions trading program set up by the Clean Air Act Amendments of 1990 to cut acid rain by half over the succeeding decade. In that article, I attempted to offer some guidance regarding the conditions under which cap-and-trade (then known as “tradable permits”) was likely to work well, or not so well. Here’s a brief summary of what I wrote at the time:
(1) SO2 trading was a case where the cost of abating pollution differed widely among sources, and where a market-based system was therefore likely to have greater gains, relative to conventional, command-and-control regulations (Newell and Stavins 2003). It was clear early on that SO2 abatement cost heterogeneity was great, because of differences in ages of plants and their proximity to sources of low-sulfur coal. But where abatement costs are more uniform across sources, the political costs of enacting an allowance trading approach are less likely to be justifiable.
(2) The greater the degree to which pollutants mix in the receiving airshed or watershed, the more attractive a cap-and-trade (or emission tax) system will be, relative to a conventional uniform standard. This is because taxes or cap-and-trade can – in principle – lead to localized “hot spots” with relatively high levels of ambient pollution. Some states (in particular, New York) tried unsuccessfully to erect barriers to trades they thought might increase deposition within their borders. This is a significant distributional issue. It can also be an efficiency issue if damages are nonlinearly related to pollutant concentrations.
(3) The efficiency of a cap-and-trade system will depend on the pattern of costs and benefits. If uncertainty about marginal abatement costs is significant, and if marginal abatement costs are quite flat and marginal benefits of abatement fall relatively quickly, then a quantity instrument, such as cap-and-trade, will be more efficient than a price instrument, such as an emission tax (Weitzman 1974). With a stock pollutant (such as CO2), this argument favors a price instrument (Newell and Pizer 2003). However, when there is also uncertainty about marginal benefits, and marginal benefits are positively correlated with marginal costs (which, it turns out, is a relatively common occurrence for a variety of pollution problems), then there is an additional argument in favor of the relative efficiency of quantity instruments (Stavins 1996).
(4) Cap-and-trade will work best when transaction costs are low (Stavins 1995), and the S02 experiment showed that if properly designed, private markets will tend to render transaction costs minimal.
5) Considerations of political feasibility point to the wisdom of proposing trading instruments when they can be used to facilitate emissions reductions, as was done with SO2 allowances and lead rights trading, less so for the purpose of reallocating existing emissions abatement responsibility (Revesz and Stavins 2007).
(6) National policy instruments that appear impeccable from the vantage point of Cambridge, Massachusetts, Berkeley, California, or Madison, Wisconsin, but consistently prove infeasible in Washington, D.C., can hardly be considered “optimal.”
Implications for CO2 Policy
In the same article, I noted that many of these issues could be illuminated by considering a concrete example: the “current interest” in applying cap-and-trade to the task of cutting CO2 emissions to reduce the risk of global climate change. Some of the points I made in this regard in my 1998 article were: