Do Carbon Offsets Really Work?

A friend recently pointed me to an article arguing that the Clean Development Mechanism may be counter-productive if it mis-estimates how additional its supported projects are, and ends up allowing more emissions elsewhere than it actually reduces. Does this mean offsets are useless? As an individual, does buying and retiring offsets still hold water as a climate charity? Let’s work through a contrived example and see what happens.


For context, the CDM is the world’s largest carbon-offset market. If I understand correctly, the idea is this: There are relatively easy-to-regulate carbon emitters, like fossil-fuel power plants. They can be subjected to a cap-and-trade scheme, where the regulatory authority demands that each one emit no more mega-tons of CO2-equivalent than its allowance. To take advantage of markets to maximize efficiency, the regulatory authority allocates a single total allowance (cap) for the whole sector, and auctions the individual allowances under the cap to whoever is willing to pay for them. Thus, power plants that can reduce emissions the most cheaply can do that, and those that can’t buy more of the allowances, achieving the required emissions limit at the lowest cost. As an individual, you can further push the world to reduce carbon emissions by buying some of the allowances and retiring them without emitting anything, effectively paying to lower the cap.1

So far so good. Where do offsets and the CDM come in? Here the idea is that reducing emissions from even the easiest-to-clean power plants may not be the most effective use of money. Other reduction projects are harder to measure and regulate, though, so the regulatory authority can’t do it reliably. Enter the CDM. They are supposed to do a bunch of due diligence evaluating various emissions reduction projects according to their criteria, notably including “additionality”. If a project meets the criteria, the CDM funds the project, and creates an offset, which is a legal certificate saying “because we paid them, this project reduced emissions by X mega-tons that otherwise wouldn’t have happened, and it’s all for real.” Then the regulatory authority in our cap-and-trade power market accepts this certificate as equivalent to a tradable X-mega-ton emissions allowance. If the CDM’s projects are cheap enough, the regulated emitters will buy these certificates, thus funding the CDM’s projects instead of performing their own (presumably more expensive) emissions mitigations, ultimately achieving the required emissions reductions even more cost-effectively.


Sounds great! But, as Calel et. al. point out, what if the CDM is wrong, and its projects aren’t actually as additional as it thought? Then it’s handing out money to projects that would have happened anyway, and letting regulated emitters off the hook by selling them offsets that don’t actually correspond to real reductions. How bad is this?

Let’s work through some contrived numbers. Suppose we are working with a regulated power market that, in the absence of the CDM, would clear at $26/T for 30 GT of emissions allowances. Suppose the CDM funds $10B worth of emissions reductions projects that the CDM thinks reduce emissions by 1 GT. Following its rules, the CDM proceeds to create 1 GT worth of offset certificates and (I assume) offer them on the regulated market at $10/T.

The referenced paper is a study of wind farms in India, comparing those the CDM funded with similar ones it did not. It comes to the conclusion that at least 52% of the farms would have been built anyway, so let’s go with that: we assume in our example that the CDM is wrong by 2x, and in fact its $10B only produced 0.5 GT of really additional emissions reductions. What happens now?

Scenario A

As a baseline, what if the CDM just doesn’t issue those offsets in the first place? Then the regulated market is irrelevant, and the CDM is charity that thinks it’s getting reductions at a rate of $10/T, but it’s actually paying $20/T. Given the severity of the climate disaster, that’s probably still worth it.

Scenario B

CDM makes its offsets and offers them on the market. Let’s assume that the increased supply means the market now closes at $25/T for 31 GT of allowances instead of $26/T for 30 GT. Then the CDM:

  1. made a profit of $15B;
  2. increased total emissions by 0.5 GT;
  3. reduced in-market costs faced by emitters by (26 * 30 - 25 * 31) = $5B;
  4. reduced out-of-market compliance costs for emitters by some amount between $25B and $26B (because at those market clearing prices, that must be what that 1 GT of reductions would have cost to implement by the regulated emitters); and
  5. reduced the revenue of the regulatory authority by $30B.

So in this scenario, the CDM is itself a polluter with revenue $50/T and profit $30/T; which cheated the regulatory authority by increasing the effective cap in the sector by 0.5 GT. The authority can fix it post-hoc by lowering the cap by 0.5 GT, but the CDM’s existence is still post-hoc “justified” if we buy the logic of cap-and-trade markets, as it’s generating “more value” from the limited pool of allowed emissions than the marginal other market participant.2

Even in this setting, buying offsets remains a reasonable climate charity. The existence of the offset mechanism has silently increased the real emissions cap, but as long as that cap remains binding, buying allowances still forces marginal reductions (in this case at a cost of $25/T).

Scenario C

CDM catches itself, and creates 0.5 GT of offsets instead of 1 GT, which it offers at $20/T instead of $10/T.3 Supply in the emissions market again increases, but not as much as in Scenario B, so maybe now the market clears at $25.5/T. Then the CDM:

  1. made a profit of $2.75B;
  2. was emissions-neutral;
  3. reduced in-market costs faced by emitters by (26 * 30 - 25.5 * 30.5) = $2.25B;
  4. reduced out-of-market compliance costs for emitters by some amount between $12.75B and $13B; and
  5. reduced the revenue of the regulatory authority by $15B.

So now the CDM is doing what it was designed to do: achieving the emissions targets set by the regulatory authority at lower overall cost (and therefore probably higher political feasibility).


Of course, all the above has lots of assumptions and moving parts, but a few themes seem durable to me (assuming that one trusts regulated cap-and-trade markets to work as advertised, and to be big enough to absorb the offsets):

  1. Carbon offset markets can indeed backfire, in the sense of accidentally increasing effective emissions caps (but not by more than the volume of the new offsets).

  2. If the realized additionality of funded projects can be measured after the fact, the backfire can be fixed by lowering caps to compensate.

  3. Charitable offset purchases still work, but they only work as advertised because they reduce the effective cap in the regulated market into which those offsets would otherwise be sold.

  4. If the offsets are not tied to a trustworthy regulated market, we’re in Scenario A. A buyer can still obtain 1 T of reductions by buying offsets, but it becomes necessary to pay $20 for 2 T worth of offset certificates to do so.

I like point 2 above. Everything is easier to measure after the fact than before, including how additional an emissions reduction project really was. I wonder whether some energetic economist can work out a good policy for cap-and-trade market authorities to follow when accepting offsets, along the lines of “We’ll honor your offset certificates today, but we’ll measure how good they were later and compensate with automatic additional cap reductions.” Maybe with some mechanism for pushing money around too, so that not-really-additional offset projects don’t become too good of a swindle.


  1. Assuming, of course, that you trust the caps to be set wisely and exogenously, based on scientific climate mitigation goals rather than on observed prices for the allowances. In the latter case, burning allowances just causes the authority to slow down the pace at which the cap is lowered, defeating the purpose. That problem is out of scope for this essay, though.

  2. What about the money? Who is really paying for what here? Well, that depends on what our modeled entities do with their money. One (charitable!) assumption is that the regulatory authority and the CDM, being public institutions, return all their profits to the public as a dividend or as reduced taxes; and the emitters are forced by competition to pass their savings along to the public as well, in the form of reduced electricity prices. Then the public gains $15-16B in aggregate, which came from giving $25-26B less in work to the power-plant-emissions-reduction industry and instead giving $10B more to the CDM-funded projects. It also cost 0.5 GT more emissions, coming out at $30-32/T. The gain of $15-16B is uneven, though: $25-26B in savings are distributed in proportion to consumption of electricity (including indirectly through the cost of energy-intensive but competitively priced goods), while $15B of costs are imposed in proportion to payment of taxes.

  3. Why do I bother with the CDM’s offer price, if the market clears above it anyway? Because the gap between offer and clearing is the CDM’s signal to fund more or fewer projects. If we assume the CDM has the organizational capacity to find and fund enough projects to equalize its cost to the clearing price in the regulated market, then pricing offsets higher to account for diligence errors will reduce the number of projects the CDM can fund. This is presumably good, because it lowers the magnitude of the mistake the CDM is making under these assumptions.