One thing we teach our students in environmental economics and other policy-related courses is that market-based pollution regulation is a great idea: it can be expected to deliver the same reductions in environmental damages as more rigid "command-and-control" regulations at substantially lower cost. Hence economists' strong advocacy of a carbon tax or cap-and-trade for fossil-fuel emissions. (Pope Francis, are you listening?)
Economic theory suggests why cap-and-trade is efficient: The government limits the total amount of pollution by limiting the number of permits to pollute that it issues. Then, firms that can reduce pollution at low cost have an incentive to do so, and sell their excess pollution permits to firms with higher abatement costs. Same overall pollution reduction, less total cost. And that's not all! We have always been able to point to evidence that it really works: the highly successful cap-and-trade program for SO2 reduction implemented by the EPA to cut acid rain.
Now along comes this paper by Chan et al, which if I read it correctly estimates that the SO2 trading scheme was on net actually quite a bit worse than a counterfactual uniform performance standard (command-and-control) that would have achieved the same aggregate emissions. While it's true that the power plants achieved the emissions reductions at lower total cost, these gains were more than offset by worse pollution damages in the form of adverse health consequences. (Side note: the major gains from cleaning up coal did not come from reducing acid rain, but from reducing health problems associated with particulates, etc.)
What went wrong? This finding points to a well-known potential problem with cap-and-trade systems. If it matters where the pollution goes, then administering a permit system that covers a wide area may sometimes lead to more pollution where it does more damage. In this case, the dirty eastern coal plants bought permits from their western counterparts, allowing the eastern plants to pollute more, as the western plants polluted less. Unfortunately, more people live downwind of those eastern plants. It would have saved more lives to force both the eastern and western plants to reduce their pollution by a uniform amount. (Now that we know there is a problem, there are well-known partial fixes to this problem—for example restricting permit trading to a "bubble" within a certain affected region. What is most interesting here is how important these effects turned out to be in practice.)
Now, back to climate change. Fortunately for carbon cap-and-trade (or a carbon tax), the damages associated with greenhouse gases are truly global in scope: The CO2 goes up into the atmosphere, spreads around, and warms the whole planet. So the localized damages are not a concern when it comes to climate change policy, and market-based solutions look pretty good again. Francis, are you still paying attention?
BUT... One caveat. One can never lose sight of the extraordinary, more localized impact of burning coal in places like China, where particulate air pollution causes hundreds of thousands of premature deaths annually. Pricing carbon globally is not the answer to that set of problems.