I don't think you are discounting here for the cap not being effectively binding (in the RGGI example) and also, last time I looked into this in depth (I used to work on RGGI at ICAP, which you cite), it was not true that additionality would always be 1.
Thank you for commenting! I believe you're saying that the cap could be large enough that there's excess allowances, so retiring one doesn't affect polluters' emissions.
The market price of allowances reflects whether the cap is binding or not. An allowance that polluters don't plan to use is worthless. If polluters did not plan to use them all, their market price would have to be near zero, as was the case in the EU ETS in 2007 (~$0.10/allowance) or in the Acid Rain Program today ($0.01/allowance in the 2025 auction).
For RGGI, a question is: do power plants plan to use all allowances given that they were willing to buy every last one for $20 in the last auction? I would think so.
Inside baseball comment here but what's the relationship between Ultra Civic and Climate Vault? Assuming that as a recent UChicago econ grad you're familiar with them. (I used to work for Levitt, my team briefly explored working with Greenstone on Climate Vault when the model was purely about retiring allowances, before they pivoted to combining this with removal tons)
Hi ethai, thanks for reaching out! I learned about Climate Vault through a friend and longtime Greenstone RA, who suggested the allowance retirement idea to kick off a crypto project that crowdfunds public goods. I loved the idea. I wouldn't say that Climate Vault and Ultra Civic are super different. As far as I've seen, they differ in that:
Gotcha. Since you're already in the UChicago network I think it's definitely worth getting in touch with someone about Climate Vault, as in the early days they were aiming at permanent retirement - I don't know the specifics of why they pivoted away from that model but I imagine you'll learn something useful.
Giving Green briefly looked into some brokers with similar models here as well; ultimately we're pretty skeptical. Even if your marginal additionality is 1 for small purchases, it's inherently not a scalable model. If the core of what you're doing is funding public goods and not selling carbon allowances, you should pitch that more! There are more cost-effective ways to just reduce emissions :)
That report is super valuable, thank you for sharing it. I thought there were no better climate actions hehe. Please let me know what's on your mind.
The point of my post was to clarify what retiring allowances accomplishes under what circumstances, and it is my fiduciary duty to address the criticisms ;) I'll keep Ultra Civic on the sidelines, as I'd prefer to discuss the mechanics of allowance retirements. I'll bring up inflationary taxation via crypto issuance as an alternative to finance public goods in another post.
I'm arguing that skepticism is not warranted in certain programs (like RGGI).
Thanks again for bringing all this up (and sorry for taking so long to respond---I've been travelling a lot)
Hmm. I think once we get into the territory of "plan to use", though, you end up with the same types of criticisms that apply to carbon credits and RECs, no? I really think that a nonzero price is not a good enough signal here! Polluters don't have perfect information about future climate regulations, future technologies, future market demand for their products, or even necessarily the future of the cap per Ville's point below. I read Making Climate Policy Work recently, which is a really thorough critique of the compliance markets; could be useful to see which of their criticisms apply to retiring allowances.
Giving Green hat on: if you want relatively certain and near-term emissions reductions at this price point, we usually recommend Tradewater. However, generally we would argue that unconstrained donors (i.e. not meeting some kind of net-zero reporting requirement) should donate to nonprofits working on policy or other systems change; GG's cost-effectiveness analyses land about an order of magnitude lower in $/ton than most allowances or credits, and I think also Founders Pledge's climate fund's analyses do as well.
Hi ethai! Thanks again. I had the chance to read Making Climate Policy Work and found it super insightful.
Regarding perfect foresight
You're right: I glossed over the extent of imperfect foresight when discussing uncertainties. For RGGI specifically, here are reasons why polluters likely have accurate expectations about their future allowance usage:
There are also valid reasons for uncertainty:
Whatever the foresight of polluters is right now, it should improve as third-parties retire allowances. Right now, a power plant's cost of purchasing a CO2 allowance it won't use is $20. The higher the price of allowances, the greater the cost of making mistaken purchases, so you'd expect their foresight to improve as more allowances are retired and prices increase.
Regarding carbon credits and Tradewater
I think it's useful to distinguish between carbon credits and the climate actions they incentivize. Any climate action has to answer: "How much does it reduce atmospheric CO2?" But carbon credits must also answer: "How much future climate action does purchasing this credit incentivize?" That's another hard question, and the impact of carbon credits is often unclear.
For example, Tradewater burns refrigerants and issues carbon credits, which they later sell. Even if burning refrigerants were perfectly additional, it's unclear how much more refrigerant they will burn because you paid them $21 for a carbon credit. Similarly, someone might retire CO2 allowances and issue carbon credits to sell later. Even if retiring allowances were perfectly additional, you can't know how many more allowances that person will retire because you bought a carbon credit from them.
Regarding policy, systems change, and Making Climate Policy Work
Cullenward and Victor emphasize the critical role of political feasibility. Carbon pricing programs (CO2 taxes + cap-and-trade) are politically costly and consequently haven't been very useful, whereas targeted regulations (e.g. car emissions standards) have quietly done the most to decarbonize society.
They do acknowledge that a global cap-and-trade program with allowance prices matching the social cost of carbon would efficiently eliminate harmful carbon emissions. It's just that it's infeasible.
There is a role for cap-and-trade programs, only smaller than previously thought. To be effective in driving decarbonization, they need higher allowance prices. Their policy recommendation is to stop overallocating allowances.
You can view retiring allowances as implementing this policy recommendation, only without the politics.
Thanks again for engaging. I will really appreciate hearing your further thoughts.
Paco
Equal Right is testing a similar model in Palau and Tuvalu: https://www.equalright.org/
I guess you have already heard of it, but I put the link here just in case, because I really like their idea. :)
Hi Ulf, thank you for bringing up Equal Right: I was actually not familiar with it!
From their cap-and-share proposal, I gather that they advocate for a cap on emissions with allowances that aren't tradeable. An argument in favor of trading allowances is that polluters can freely redistribute allowances towards those who value them more (i.e. emit to produce more valuable things), resulting in lower pollution abatement costs. Cap-and-share involves direct government control over individual polluters, which makes climate mitigation costlier: Greenstone et al. (2025) found that cap-and-trade reduced pollution abatement costs by 11% relative to the traditional command-and-control approach in India.
Newcomer here. I'm happy to edit for clarity or accuracy!
This post argues that retiring pollution allowances (the legal rights for polluters to emit CO2) is a cost-effective way to reduce global emissions. In government ”cap-and-trade” programs, polluters must buy a serial-numbered allowance to emit each ton of CO2. These allowances are sold in public auctions and secondary markets. Crucially, anyone can buy and retire allowances, reducing their availability and forcing polluters to emit one less ton. In well-designed cap-and-trade programs, allowance prices range from $20 to $85, significantly below the social cost of carbon (~ $230 according to the EPA) and most carbon-removal technologies ($300+ per ton). I discuss two key risks: leakage (the shifting of pollution outside the cap-and-trade jurisdiction) and political rollback.
Epistemic status: confident that retiring an allowance in well-designed cap-and-trade programs prevents one ton of regulated CO2 emissions at a cost equal to the market price of that allowance. Such “well-designed” programs meet the four conditions outlined in the additionality section. Less confident about the effect of retiring allowances on global fuel markets and about political risk. Numbers are for mid-2025 unless noted.
A regulator measures the CO2 of large emitters (power plants in the U.S. Northeast, heavy industry in the E.U.). It issues a fixed number of allowances each year, the “cap”. Polluters must surrender one allowance for each ton of CO2 emitted. Failure to comply results in substantial fines. Since allowances are both scarce and tradable, polluters buy and sell them—hence, “cap-and-trade”.
The key insight is that private individuals can also buy these allowances. By permanently removing them from circulation (retiring them) they effectively lower the cap. Fewer allowances mean polluters must either reduce emissions or buy increasingly expensive remaining allowances. This mechanism allows direct emissions reduction without relying on political interventions.
Retiring an allowance reliably forces polluters to reduce emissions by exactly one ton of CO2, provided the cap-and-trade program satisfies these four conditions:
Relative to retiring allowances, I argue that carbon credits and renewable energy certificates have a low certainty of CO2 impact, and that carbon removal is expensive.
These are tradeable certificates that registries award to those who develop climate projects. The goal is for third-parties to buy credits from developers and incentivize further climate action. They face three major challenges:
These are tradeable certificates granted to renewable energy producers per megawatt-hour generated. Similar to carbon credits, it's unclear how much more renewable energy generation RECs actually incentivize or how effectively renewables displace fossil fuel-based electricity.[2]
These include things like direct air capture or enhanced rock weathering. They’re necessary but currently expensive and uncertain. For example, Stripe’s carbon removal offerings cost between $300–$1,000, above the EPA’s $230 social cost of carbon. Additionally, some direct air capture methods (e.g., Climeworks) reportedly emit more CO2 than they capture.[3]
Criticism: Retiring allowances means local producers reduce production, but producers elsewhere increase production, merely shifting emissions geographically rather than reducing them.
Response: this criticism is partially correct; the magnitude of leakage depends on industry and region. Consider RGGI, the U.S. cap-and-trade program that regulates fossil fuel power plant emissions in the Northeast. Retiring an allowance forces Northeastern power plants to either switch to cleaner power or reduce output. While this may increase power imports from other regions, these imports aren't necessarily equally polluting. Power imports into RGGI states come partially from Canadian hydro and are cleaner than local RGGI power (0.27 tCO2/MWh vs. 0.45 tCO2/MWh). Even without increased local renewable production or reduced overall demand, retiring one allowance in RGGI results in a net reduction of approximately 0.4 tCO2 (calculation: 1 - 0.27/0.45).
Criticism: retiring allowances today is meaningless if governments shut down the cap-and-trade program tomorrow.
Response: shutdowns do not necessarily invalidate prior retirements. If polluters use all available allowances before the program ends, earlier retirements still prevent emissions during the program’s remaining active period. Conversely, if polluters have more allowances than they need to reach the end of the program, previously retired allowances lose their impact. Interestingly, this scenario would drastically lower allowance prices, creating an opportunity to cheaply retire allowances and reintroduce scarcity.
Criticism: Reducing allowances restricts production of valuable goods and services, potentially harming society overall.
Response: this criticism is incorrect. The net social impact of retiring an allowance equals the social cost of carbon minus the value of goods produced by the marginal ton of emissions prevented. The marginal ton of emissions is the least valuable ton emitted by polluters; its value is approximately equal to the allowance price (otherwise polluters would have continued emitting it). Given allowance prices ($20–$80) are substantially lower than the social cost of carbon ($230, EPA), retiring allowances yields a net social benefit.
The idea of retiring allowances originated in 1992 with David Webster’s Clean Air Conservancy under the Acid Rain Program (a cap-and-trade program targeting sulfur dioxide and nitrogen oxides). The organization dissolved after Webster's death in 2009. Recently, economist Michael Greenstone advertised the concept (2021 podcast).
Here's the Clean Air Conservancy describing allowance retirements in 2001:
I’ll use numbers for RGGI, since it offers the cheapest allowances, ticks all additionality conditions, and has a leakage that's easy to understand (see the discussion of leakage in 4.1.).
Cost: $20 to retire one RGGI allowance (latest auction prices here).
Benefit: Avoided social damage of $230 (EPA SCC, 2025).
Benefit-cost ratio: ~12:1 before leakage; ~5:1 under worst-case 60% leakage.
Comment, suggest edits, demand missing references, and criticize :)
See, for example, the definition of an Output Auditor under the puro.earth registry rules.
See the discussion and links herein: https://www.bccas.business-school.ed.ac.uk/impact-and-collaboration/renewable-energy-purchasing
See a recent article claiming that Climeworks’ capture fails to cover its own emissions.
One thing that speaks against this argument is that the climate targets and thus the cap is a political decision, and can be adjusted periodically (In EU, The EU-ETS cap was made stricter as part of Fit-For-55 package, for example. Also, market stability reserve and its cap reduction was due to low allowance prices). Also, EU-ETS cap so far is set until 2030, and following years will depend on upcoming decisions by politicians.
Due to the political nature, the cap is at least somewhat endogenous to the allowance price. The higher the price of allowances, the greater the cost for industries/households and the greater the threat of carbon leakage. Thus higher price decreases the political willingness to set a stricter cap. Because buying and retiring allowances increases the price (due to increased demand with constant supply in short run), it can thus lead to greater allowance cap in the long run.
Therefore, I would wager that retiring allowances will decrease emissions, but the effect is not 100% of retired allowances, but somewhere between 0-100% of the, depending on the elasticity of cap setting to the allowance price.
Hi Ville, thank you. I agree this is a central concern. If retiring 100 allowances leads the government to issue an extra 20, the climate effect is just 80%.
I also agree that politicians benefit from raising the cap in response to high allowance prices (ultimately driven by large-scale retirements). Raising the cap
My point is that politicians also face significant costs of raising the cap:
Historically, the political problem in cap-and-trade has been lowering, not raising, the cap. For instance:
Your point that the cap is endogenous to allowance prices is well-taken. Historically, what politicians want is lower caps and higher prices---exactly what retiring allowances delivers. For current and moderately higher allowance prices, my sense is that the risk of politically motivated cap increases remains modest.
Lastly, you can retire allowances without causing cap increases in cap-and-trade programs where the relationship between the cap is an increasing and publicly known step function of the price of allowances. For example, RGGI employs a uniform-price auction and releases additional allowances only if the clearing price exceeds a publicly-known trigger (it's called the cost containment reserve). To make sure that your bidding does not inadvertently raise the clearing price and raise the cap, you just bid below the trigger price. If the auction clearing price surpasses the trigger, your bid is below the lowest winning bid, so you don't affect the outcome.