In David Roberts’ post on the carbon policy dilemma, David defines an “efficient” carbon policy as follows:

First, in a given sector, you set up a system that transfers capital directly from those over-emitting to those reducing emissions, in an agnostic fashion — that is, preferencing no particular set of technologies or practices. A ton of CO2 ought to be worth the same no matter how it is emitted or prevented, and there should be no net loss of capital in the sector (as there would be if the feds took the revenue and spent it on other things). Second, you remove existing regulatory barriers to that capital flow. As long as capital continues flowing from emitters to savers, you’ve got a perpetual economic motion machine.

My guess is the use of a perpetual motion machine as a metaphor was a message from David’s subconscious, because it is impossible to set up a mechanism where the transfer “to” is as efficient and automatic as the transfer “from.”

There are two (or maybe two and a half) very efficient ways of putting a price on emissions. You tax emissions and include an escalator clause that increases until they drop as far as you want. You can put a cap in place, auction the resulting permits, and keep dropping the cap until you get the reductions you want. Or, in a hybrid system, you put the same cap and auction in place, but with a minimum permit price to reduce volatility.

But feeding the revenues back into fighting emissions — a negative carbon tax or emissions fee — well, there is no really automated way to do that. Sean Casten made a very smart stab at a formula. Set a standard based on the ratio of emissions to useful BTUs delivered. If someone is more efficient than the standard, they have permits to sell. Someone who is less efficient than the standard has permits to buy.

The problem with this, like most attempts at measuring general industrial efficiency, is that excludes many types of efficiency. Increase your boiler efficiency when cleaning raw material as part of producing fabrics, and you gain permits. Increase the efficiency with which you cut fabrics so that you scrap less material and decrease emissions by exactly the same amount, and you gain no permits under this formula.

It’s not that Sean’s formula is a poor one. It’s just that no single formula will work. Let’s try a more sophisticated equation: ratio of emissions to value added. But then financial services get rated as very efficient even though you have to have some real goods and services in the world for financial services to have any value. (And it sure looks like the value of financial services have been heavily overrated in any case.)

What about David Roberts’ refinement: partition off sectors? Unfortunately, such partitioning is not an easy decision. Do we simply partition off industry from residential and transport? As a commentator pointed out, we might want to trade-off between them by building more housing in dense urban areas so people can transport themselves less, rather than merely more efficiently. For that matter, as the financial services example showed, we might want to partition off various sub-sectors of industry, which carries its own arbitrariness. I guarantee that if we ever traveled that path, it would be a highly politicized process.

So while you can set an automatic process for penalizing emissions — the “stick,” as Sean calls it — when it comes to positive incentives, “carrots,” there is no such automatic process. If you want positive incentives, there is no way to avoid establishing them via a political process. Yes, horror of horrors, if you want carrots, you have to “pick winners.” If you think “picking winners” is too unspeakable for words, then support a carbon tax or 100 percent auctioned permit system and nothing else.

If you want carrots, get over the prejudice that equates politics with evil and any economic decision-making by political processes with pork.

One last point: let’s look at the perpetual motion machine metaphor again. Note that David seems to be operating from a strong intuition that if we get the price of carbon right, it will generate exactly the revenue we need to invest to for carbon reduction. But a carbon price is ultimately an incentive. There is no reason revenue from such a carbon price should equal even approximately the investments we need, any more than revenue from liquor taxes automatically generates the right amount of revenue to fund the ideal amount of investment in alcohol treatment. The intuition that it should be is very similar to the intuition that led people to think you could keep balls rolling up and down an inclined plane forever.

There is no particular efficiency gain in dedicating revenue from a greenhouse tax to greenhouse purposes — rather than taking the money from, say, military spending. And there is the disadvantage that, if done right, greenhouse-gas revenues will be a declining source.