(I manage the FP Climate Fund, so my incentives run counter to my take).
I hadn’t had the time to examine this paper in detail, but I am quite skeptical of this Rethink Priorities paper. I think it is quite easy to string together assumptions that yield a high social cost of carbon, but I wouldn’t treat this as an unbiased estimate.
For example, if I understand this correctly based on your description, they use the Rennert et al (2022) paper to derive the SCC from which they make adjustments.
The assumptions of that paper are clearly extremely pessimistic, probably by 2022 standards, but definitely by what would now be the consensus view.
For example, they assume close to 20/Gt annual emissions in 2100 as their median scenario and high emissions continue well into the 23rd century. In other words, we are more than a 100 years late in achieving net-zero in their median scenario despite all technological trendlines rendering this quite implausible.
Combining this with a low discount rate will give a high SCC, but I don’t think this is close to a reasonable baseline for what a median expectation should be. (Obviously good to have a low discount rate from an EA perspective, but this requires that the modeling of the future is a bit more careful). Essentially, this means that most marginal carbon reduction modeled for the SCC will happen in worlds where this is implausibly valuable thereby inflating the SCC value.
This alone probably leads to an overestimate of the SCC of a factor of 5x or more and this came up from looking at the paper for 5min.
Regardless of what any individual thinks about the underpinning assumptions and methodology, the Rennert et al (2022) SCC is widely used and “mainstream”.
It is certainly true that there is high uncertainty around the SCC, and a wide range of estimates. But the Rennert et al (2022) paper is just about as mainstream, “blue-chips” of an estimate as you are going to get, and therefore I think is a reasonable anchor for the Rethink analysis.
A few data points showing how popular the Rennert et al (2022) analysis is (and the underlying GIVE model plus the probabilistic assumptions on emissions pathways that underpin it):
This work was used heavily by the Biden administration’s EPA as part of a comprehensive update of their internal social cost of carbon, and their estimate was ultimately adopted. (detailed EPA report here.)
The German Environment Agency uses a very similar approach built on Rennert, but applies some different assumptions (lower discount rate, equity weighting), and comes to an estimate of 345 EUR (assuming 1% pure rate of time preference)
From Opus 5.0: “There is no single consensus number, but there is a fairly clear structure to where the field has landed. Short version: ~$190/tCO₂ (2020 USD, 2% near-term discount rate) is the de facto reference value”
Of course this doesn’t mean the paper or its estimated SCC is “right”, but I do think it’s a highly defensible reference SCC for Rethink’s analysis. As correctly mentioned by Vasco, there are numerous critiques of the Rennert paper, arguing that a lower SCC is more appropriate. But there are also influential critiques in the other direction. In the post I discussed how incorporating effects on economic growth can lead to much higher SCCs. For instance, a recent paper by Bilal and Kanzig (2026) argue for a much higher SCC of $1200 (even assuming a 2% discount rate).
Despite being “mainstream”, is Rennert et al (2022) unreasonably pessimistic about future outcomes?
If I understand correctly, Rennert et al (2022) use distributions of future states of emissions based on Rafferty et at (2017), which use historical data to assign probability distributions to the different IPCC emissions scenarios. If historical trends are not good predictors of future ones (for instance, due to rapid technological advancement), these estimates could be off.
Personally I’d agree with Johannes that the Rennert future emissions distributions feel pessimistic, but I would not agree with the modifiers “clearly extremely”. I think their mean estimate would fall within one standard deviation from the mean in my personal distribution.
But in any case, one could re-do the analysis with more optimistic distributions. Resources for the Future has a handy calculator that allows one to re-calculate the SCC with different models and different parameters. If we switch from the emissions distribution in Rennert (RFF-SP) and instead move to the more optimistic SSP-2 (which is the closest option to my beliefs, though I’d still be a bit more optimistic), SCC falls by 15%, from 185 to 158. Unfortunately the calculator does not allow us to move the discount rate to zero [edit: actually, we want to move the pure rate of time preference to zero, not the discount rate], which is really what we need to test the sensitivity of the Rethink estimate to different emissions futures. One would need actually re-run the models to calculate exactly, but a 5x wedge feels quite off to me.
Thanks, Dan, for your reply. I think I still disagree pretty strongly and I’ll briefly outline how and why: (1) I don’t think we should give credence on studies based on government usage, (2) 22 century emissions are clearly very implausible, (3) there are reasons for this to really matter and (4) being able to find this in 5min should make us deeply skeptical of the wider paper. (5) As you, I deeply care about effective climate action but I don’t think we should propagate bad evidence to inflate the importance vis-a-vis other causes.
Should we give special credence to something because the Biden admin and the German EPA used it?
I think the answer is pretty clearly “no”.
As you also mention, SCCs are inherently very uncertain estimates, but they also are clearly political projects; they are often produced or at least amplified to serve political purposes such as justifying more or less ambitious policies.
Given the biases in the climate science community, I think it is upon us to evaluate studies on their merits and to not assume that studies that are well-cited or used by climate hawkish administrations are therefore inherently more credible.
In case that sounds conspiratorial, it’s worth remembering that RCP 8.5 (the most extreme emissions scenario by the IPCC) was essentially abused prominently for 5-10 years as a baseline after it was obvious to all serious observers it was not a baseline but worse than a conceivable worst case scenario.
Given this record of the emissions scenario modeling community, we should always check what they assume.
Why it’s obvious that 22nd century emissions shouldn’t have that high a median
This is obvious to Dan, but for the wider group. The reason the 22nd emissions are so implausible is because they essentially assume that nothing much changes by the 22nd century, 3-5 investment cycles from now. Yes, decarbonizing the world by 2050 might be hard, but decarbonizing the world by 2100 or 2150 is not hard. It’s simply not plausible to believe in a 22nd century that is both richer than today, a hundred years in the future, exposed to a lot of climate damage, and not able to decarbonize.
This is much more than a 15% difference!
This is not about a 15% difference and a 5x over-estimate from this alone is quite plausible.
The calculation you run is – as you note – not informative because it ignores the discounting and the implausibility of the Rennert et al paper lies primarily in the 22nd century+ and those will get outsized importance if changing the discounting assumptions.
Yes, by 2100 Rennert looks mildly pessimistic but not widely so, but that’s not the point – once discounting is very low the 22nd century matters and drives most of the result.
The 22nd century is discounted by more than 75% on the RFF explorers lowest discount rate (2%) so this is not a useful tool to explore whether implausible 22nd century assumptions massively inflate the SCC estimate.
And, indeed, we should expect the 22nd century to drive a massive difference when discounting is low. By Rennert et al’s assumption temperature keeps increasing well into the 22nd century, climate damage is non-linear, so on low discounting this will drive a massive amount of the damage and lead to strongly changed estimates.
I am not an economist, but when I Claude the emissions assumptions of the Rennert et al paper and stipulate to make conventional assumptions about climate damage at 0 discount rate, 96% of the damage is after 2100.
And we can see here why this is. In the Rennert world temperature keeps rising till after Captain Kirk with 25% probability mass above 4.5C. This is very different from what I think we both believe about the future.
This goes beyond this one example
I picked this example because it was quick for me to find, having seen this pattern (making deeply implausible assumptions in papers that get a lot of policy traction) so often. But I am sure one could find other instances. The point that it is so easy to find a highly biased assumption cited as mainstream evidence is that one should distrust the broader paper more.
Why I care
As you and many readers here, I deeply care about effective climate action and we are together on this project of making the climate response more effective.
But I do think we should not propagate bad evidence to inflate the importance of climate vis-a-vis other causes in the same way that we would also – and correctly – criticize a motivation of AI risk that only relied on Yudkowsky’s p(doom) estimate.
Implausible assumptions rarely come alone and when a paper makes such implausible assumptions (that are not simpler than more reasonable assumptions, i.e. this is not driven by parsimony) then I think we should be quite skeptical of the conclusions of that paper.
Thanks so much for interacting with our work. For broader context around this report, it’s important to note that our approach from the outset was to build a series of models to show what you would need to believe for climate to be competitive with GHD. We therefore aren’t wedded to any particular SCC, and certainly can imagine different ways of approaching the problem.
To your specific points:
It is our view that Rennert is the best middle ground available from current peer-reviewed work. It sits between DICE-2023 (~$66) at the lower end, and estimates that incorporate growth effects like Bilal & Känzig (2024) (~$1,000+ at the same 2% discount rate) at the higher end. So, whilst we agree that some parameters in the GIVE model can be questioned in isolation (your discount rate and impact horizon critique seems reasonable), we feel it is still a sensible middle ground when taken as a whole, in comparison to other peer reviewed estimates currently available.
On adaptation, we feel less optimistic that the principle of ‘richer future = strong adaptation’ will hold for all people (e.g. those most exposed and least able to adapt in LMICs) or all sectors (e.g. agriculture), with the intersection between the two being particularly noteworthy. However, we accept this is genuinely uncertain and that reasonable people will weigh it differently.
Overall, we think our analysis at least supports the re-visiting of climate as a cost-effective cause area within the EA space and we appreciate the various thoughtful comments so far. Thank you for engaging!
Hi Johannes, thank for all the detailed thoughts. I think we’re probably reaching diminishing marginal returns in debating point by point. I will say that we agree that the Rennert et al projections are too pessimisitic, and difficult far future projections can matter a lot with low discounting. Though I think we will need to agree to disagree on the magnitudes of these changes.
There is one technical point that I’d like to bring to your attention, though, that I think does critically matter. The Rethink application of a new discount rate is NOT a 0% discount rate. It is a 0% “rate of pure time preference”, which is one component of the discount rate. (I’m going to modify my original post to clarify.) Rethink’s calculation is built on calculations from the German Environmental Agency, so I will directly reproduce their explanations for how they did discounting:
“we use the social discount rate developed by Frank Ramsey (Ramsey 1928), which combines the two aspects above: expected consumption growth, weighted by its effect on the marginal utility of consumers, and the pure rate of time preference (PRTP).
In the GIVE model the consumption growth rate is a dependent variable. Therefore, it is not possible to specify the exact discount rate used for the climate costs”
I am not familiar enough with these models to get a good intuition for what discount rate would be, and therefore how much weight will be places on the far future. But as a reference, the Stern Review also used Ramsey discounting and a low (.1% PRTP), and resulted in an overall discount rate of 1.4%.
All of this to say, I think a simple mental model of “no discounting”, and therefore the vast majority of damages being in the far future, is not a good description of the Rethink analysis.
For example, if I understand this correctly based on your description, they use the Rennert et al (2022) paper to derive the SCC from which they make adjustments.
The assumptions of that paper are clearly extremely pessimistic, probably by 2022 standards, but definitely by what would now be the consensus view.
I agree. Here is a critique of Rennert et al. 2022 by David Friedman.
In summary, I found four problems with the mortality calculation from Cromar as used in Rennert[2]: neglecting the effect of income on temperature-related mortality, ignoring the pattern of temperature change implied by greenhouse warming, neglecting the effect of technological change, ignoring adaptation by migration. The first two can be corrected by substituting the result in Carleton for that in Cromar, reducing the SCC due to temperature-related mortality from $90/ton to something between $17.1/ton and $36.6/ton. Correcting the third and fourth should further reduce it by a large but unknown amount.
The values of 17.1 and 36.6 $/t above respect representative concentration pathways (RCPs) 4.5 and 8.5. I think RCP 4.5 is much more plausible trusting the temperature changes below from Wikipedia.
I am not sure the comparison stays entirely symmetric once endogenous growth is added on the climate side. If climate damage can permanently change the growth path, then at least some GHD interventions should also have effects on growth through health, schooling, productivity, fertility and so on.
I assume some of that is already captured in the GiveWell/Coefficient models, but probably not the same kind of macro second-order effects. Since climate seems to become competitive mainly when these more uncertain effects are included, I would be interested in how much this matters.
Fair point Alex. We mostly interrogated the climate side of things, and in that sense the work itself is asymmetric. GW does incorporate second order effects into some of its CEAs, but I guess the larger question is whether or not there is any such large multiplier to apply to some GHD interventions for some amount of time. My quick reading of the literature there is simply that there isn’t strong evidence on the direction of growth effects for GHD, whereas the estimates we found on the climate side were at least consistent in direction (though I may be wrong here).
Hi Tom. Which discount rate did you use for the benefits from endogenous growth? GiveWell (GW) uses a discount rate on future consumption of 4 %/year.
Good q. We did not directly adjust the endogenous growth term through any discount, it enters the model only as a multiplier. That said, the multipliers we use have underlying discount rates that they use (generally larger than 0). We probably could’ve harmonized in a smarter way to be able to answer your question, but the general point you are trying to make, I think, is that higher discount rates here might be appropriate because of long stream of future harms/benefits (or perhaps just uncertainty). This would mean that our SCC is too large. Fair; and maybe!
Recall that our main SCC estimates are log-linear in income. In a linear model with 0% time preference, yes, the estimate would balloon because of the long tail. But in a log-linear model where people are getting richer, a 0% time preference essentially sets the effective discount to the growth rate (because future people are wealthier, and so additional dollar damages mean less to them). So this would mean that our initial SCC is probably smaller than you’d think *and* it is less sensitive than linear models to discount rates (by a nontrivial but not super large amount I would say).
Now the multipliers themselves, as I said above, use some discount rate and are not log-linear in income the way our base estimate is. So the actual discount rates are important here and might mean our multipliers modify our baseline SCC too muuch (our log-framework already downweighs far-off damages, but our current multiplier does not, making it large) or not enough (if the pure time preference is really closer to 0% than the rates they were estimated at, given how back-loaded growth damages are, making it too small).
All to say, the multiplier bit was a significant source of uncertainty to us when we wrote it (we labeled the endogenous growth multiplier as an adjustment with high uncertainty in our report) and happy to have had a bit more time to rethink some of these things and share our views on it now.
My understanding is that GWs pure time preference is similar to ours and that any differnece in effective discounts is related to how they treat uncertainty over the future, but I have’t looked into this.
Thanks for the clarifications. They made sense to me.
My understanding is that GWs pure time preference is similar to ours and that any differnece in effective discounts is related to how they treat uncertainty over the future, but I have’t looked into this.
Here are the details (linked here) about GW’s discount rate of 4 %. Pure time preference does not contribute to this.
We considered five reasons to assign a discount rate.
Increases in consumption over time meaning marginal increases in consumption in the future are less valuable. We chose a rate of 1.7% based on an expectation that economic consumption would grow at 3% each year, and the function through which consumption translates to welfare is isoelastic with eta=1.59. (Note that this discount rate should be applied to increases in ln(consumption), rather than increases in absolute consumption; see calculations here)
Temporal uncertainty. Uncertainty increases with projections into the future, meaning the projected benefits may fail to materialize. James recommended a rate of 1.4% based on judgement on the annual likelihood of an unforeseen event or longer term change causing the expected benefits to not be realized. Examples of such events are major changes in economic structure, catastrophe, or political instability. This does not include the probability that a person will die before realizing the full benefits of the intervention, which is captured elsewhere in our cost-effectiveness analysis.
Pure time preference (beneficiaries). People act in such a way that implies they would prefer spending now to later. We did not apply an additional adjustment for this factor. We believe the common use of self-commitment mechanisms such as savings accounts indicate that a preference for short term benefits is often an involuntary action (independently of reasons 1 and 2).
Pure time preference (donors). Donors may prefer to achieve benefits now rather than later independently of the relative benefit to the beneficiaries. We do not have this preference.
Compounding non-monetary benefits. There are non-monetary returns not captured in our cost-effectiveness analysis which likely compound over time and are causally intertwined with consumption. These include reduced stress and improved nutrition. We chose a rate of 0.9% to account for this based on discussion.
For me this is the crux and the key line.”Climate’s SROI becomes competitive with GHD when accounting for high-risk, lower-certainty effects like tipping points and endogenous growth”
If you’re the kind of person who both has a high risk appetite, and believes we have some plausible control now over low certainty events in the far future then I would agree with the thesis that we should put more money into climate stuff.
Personally I think the medium-far climate future might be as difficult to predict as the trajectory of AI. We have political uncertainty (e.g. Trump scrapping wind), economic uncertainty (e.g. solar became unexpectadly cheap), potential AI tech progress, potential rogue actors (someone could release sulfur). We could be at net zero by 2040 or never.
Anyway I think the uncertainty with climate is so large that you need to be a certain kind of giver for that to make sense—especially with the “existential risk” component perhaps not being there.
I don’t find much solace and meaning in error bars that wide—but many others do.
I’m curating this post. It’s a great summary of some under-discussed research, and it has the potential to kick off a very valuable conversation (which we are seeing the start of—thanks jackva). I’d love to see more of an in depth comparison of investments in climate mitigation vs global health vs economic growth, especially accounting for more funding coming to global health. This post is a great start. Thanks Dan!
More EA Near/Medium-Termist Money Should be Flowing to Climate Change Mitigation
As Abbie says in the report, climate can’t compete with GHD on high-certainty interventions. It can compete if we: (a) interpret the economic effects of climate damage as mostly affecting growth rather than inflicting many one-off costs and (b) count effects well beyond the next century. Assumption (b) looks particularly shaky to anyone with a high credence that AI-powered and other growth will radically increase global resilience to climate change.
Thanks Stan. I think this is a concise and reasonable critique- I mostly agree. A few thoughts.
My understanding is the CG recently did an internal calculation of SCC along the lines of the Rethink analysis, but one key difference is that they cut off damages around 2100 (not sure the exact year). This was just to have a like-for-like comparison with other interventions where they project benefits out for a fixed amount of time. They came to a conclusion of an SCC around 700. So if you take $1/ton as a reasonable benchmark for climate you get 700x, which doesn’t reach the CG bar but is in the ballpark. If I understand correctly, this CG SCC doesn’t include growth effects. So my point is that you don’t necessarily need long term and growth effects- with just one I think GDH gets competitive under reasonable assumptions, though perhaps not “winning”. (I haven’t actually seen this CG SCC analysis, but talked to people there about it. Hopefully it gets made public some day!)
For (b) I don’t think you need damages past 2200 to be competitive with GHD (especially if you do have endogenous growth effects), but the point is well-taken that you do need long-term effects, up to 2100 and beyond. My personal central estimate is that by 2100 we’ll have reached near net-zero, at warming of ~2.5 degrees C. But even though emission stops, the warming will persist for centuries, and therefore has the ability to continue to cause damages (compared to a no or lower-warming counterfactual). Of course adaptation is the wildcard. Will AI-powered growth radically increase resilience such that climate damages will drop to zero even in the face of persistent increased temperatures? I am not as optimistic as you are, but I see your perspective.
My sense is that the social cost of greenhouse gases carries fairly wide uncertainty bounds, driven by discounting assumptions and long-run projections. I’m underqualified to judge how robust these are, though I’d guess the biases could run either way.
What I’m more interested in is the second-order effect on cause prioritisation. More incoming philanthropic funds are likely to saturate grantmakers in global health and development, pushing the cost-effectiveness bar down, and we’re already seeing that with GiveWell and Coefficient Giving.
But I wonder if part of what we’re seeing there is really an absorptive capacity constraint (a limited set of organisations able to scale up delivery), rather than a genuine shortage of cost-effective interventions. Scaling delivery takes time, people, and systems, and that constraint doesn’t come through clearly in your piece. Whether climate faces the same constraint isn’t clear to me either.
You mention Giving Green’s large 2025 grant will take roughly a year to fully disburse. How should we think about absorptive capacity in climate versus GHD?
If climate absorption is genuinely constrained, then GHD interventions, with likely smaller uncertainty bounds, could see absorptive capacity grow over two to four years as delivery organisations scale. That would push the cost-effectiveness bar back up and soften the piece’s headline conclusion. Keen to hear how you’d think through this dynamic.
Thanks a lot for these thoughts, Tony. I’m not sure I have all the answers, but a few ideas to throw in the mix:
A lot of what we think of as cost-effective interventions in GHD are on-the-ground, direct-delivery type stuff (bednets, vaccines) that require a lot of logistics and therefore large, sophisticated orgs to manage them. This isn’t really true for climate (at least in my opinion) where we are focused more on charging policy, technology, and markets. There are still constraints in org capacity, but they look pretty different.
The climate philanthropy ecosystem actually has a pretty impressive build-out of various funding vehicles that allow lots of money to move very quickly. For instance, there are big cross-sectoral regrantors like Climateworks, tons of sector and region specific regrantors (Industry Hub, Tara Climate Foundation), and focused research organizations working on specific topics (Carbon2Sea, Reflective). These organizations frequently see their missions as building up capacity in a topic as well as funding it. From someone with an EA-aligned perspective, if you believed that the mission of any of these orgs was super cost-effective, they would provide a vehicle to move lots of money quickly. For instance, Founder’s Pledge made a ~23M grant to Deploy/US based on this logic. We at Giving Green have so far shied away from giving to regrantors, since we think that keeping tighter control of our grants allows money to move faster (no middleman), and allows us to really focus on what we think is most impactful. But one could definitely debate that choice, and if lots more money came our way we would certainly revisit this assumption.
As a grantmaker in climate, I haven’t felt the absorptive capacity of the field to be a huge constraint. Sometimes we have said “we really want to fund orgs doing XYZ, but can’t find people doing it”, but I’d say that’s the exception rather than the rule. Maybe if I was moving much larger amounts of money (like GW), I would feel these constraints more strongly.
I really hope that organizations delivering really cost-effective GHD work can scale to meet the moment, that the OP/GW bar goes back up, and that the headline conclusions of this research are indeed softened. That would be great for the world!
Has anyone tried to estimate the social cost of carbon taking into account AI? I’ve seen several well-calibrated people estimating that there is a 20% probability that AI fizzles in the next few decades. If you believe that that would be permanent and AGI/ASI would solve the climate problem or result in catastrophe, to a first approximation, that would mean the social cost of carbon would be about 1⁄5 as much as people typically calculate. In reality, AGI/ASI may come further in the future, and there could be some benefits of reduced carbon even with AGI/ASI.
More EA Near/Medium-Termist Money Should be Flowing to Climate Change Mitigation
My understanding of EA funding is that almost none goes to climate. I (loosely) tend think that it would be good that a bit more resources go to Giving Green and other environmental prioritization projects.
However, I also gather that such projects attract funding far more easily than most other EA-consensual-ish causes.
Thanks for this contribution. I hadn’t come across the Rethink Priorities analysis, so this was a great nudge. Responding to both your post and the RP research, I think it’s important that the EA community recognize that certain climate change mitigation interventions double as global health and development interventions. And I’m not just referring to countering the social cost of carbon.
Climate change mitigation interventions, particularly in low- and middle-income countries, can be designed to not just address carbon emissions but also enable better healthcare and education delivery.
To give an example, there are an estimated 6.5 million diesel generators across Sub-Saharan Africa. These generators are polluting, prone to breakdowns and fuel for them is expensive. Yet households, businesses, health clinics or schools often have no other power option because they aren’t connected to the electricity grid, or if they are, the grid is prone to outages. So, diesel generator it is! Until there’s a fuel shortage. Or the price of diesel spikes. Or the generator, which was probably bought second-hand, fails. Then surgeries get cancelled and vaccines spoil from no refrigeration.
This is why a technology like a solar mini-grid, which can deliver electricity at a levelized cost up to 60 per cent lower than diesel, should be viewed as more than just a climate change mitigation solution. It provides a critical input to healthcare and education delivery, including some of the interventions that the EA community funds (think vaccination rollout).
Obviously, there’s a lot more to unpack in terms of what it takes to replace something like 6.5 million diesel generators with clean energy alternatives, including the cost of doing so. I just think that it’s important to recognize that the human impacts of clean energy deployment go beyond alleviating the social cost of carbon.
Has the EA community already looked at this nexus issue? Would love to connect with anyone thinking about or researching the full impact of clean energy, particularly its role as an enabler of interventions in GHD.
Hey Stephen, we at Giving Green actually started a project looking exactly at this- trying to identify cost-effective interventions and funding opportunities at the nexus of health/livelihoods and climate mitigation. We put the research on pause because of other near-term priorities and some uncertainty over whether there was hunger for this kind of thing among donors. But it’s something we hope to pick up again next year.
Great to hear that you and the team have been exploring this! I’d be keen to learn more about how far you got in your thinking/analysis, but realize it’s on the backburner for the time being. Let me know if I can be of any help when/if you resurface the topic.
More EA Near/Medium-Termist Money Should be Flowing to Climate Change Mitigation
Basically all EA money at this point should be going to AI safety (where AI safety is broadly defined to include not just loss of control / human extinction, but also things like concentration of power, value lock-in, and AI welfare). Anything else seems like a distraction.
14% of all coral reefs died between 2009 and 2018. At 1.5°C temperature rise, 99% of coral reefs are expected to be dead by 2050. 44% (48 million km2) of the world’s habitable land is used for agriculture. 33% of all soil is already degraded and 90% is estimated to be degraded by 2050. Vertebrate populations have declined by 73 % between 1970 and 2020 and insect populations have seen a 75% decline in biomass. The United Kingdom says ecological collapse is the biggest threat for food security. All these things and many other climate related things might affect billions of people.
Thank you for your post. I agree that climate change is underfunded in EA, and I’m happy to see Rethink Priorities updating its models. I also wonder whether GHD is somewhat overfunded by EA, especially since some interventions can accelerate economic development in ways that increase environmental pressures if they are not accompanied by systemic change. Climate change will itself have major consequences for global health and development. I’m particularly skeptical of approaches that rely on producing mosquito nets, medicines, etc. in Europe and shipping them elsewhere rather than strengthening local capacity and resilience.
On another note, I came to the Forum because I want to raise money for climate action during a demonstration in Geneva. Having been part of the EA community for a few years, I naturally looked at Effective Altruism Switzerland’s climate fund, but something didn’t quite click for me. I had a similar feeling when looking at Giving Green: it seems quite focused on technological solutions. I’m therefore looking for another organization to support, although I’m open to changing my mind. I’ve been a fan of GFI for years, for example, but I’m not convinced that plant-based meat and dairy analogues are a sufficient answer to climate adaptation in agriculture—they are only one part of the solution. What about the campaign beans is how? I also looked briefly at Deploy/US (though obviously nowhere near as deeply as Giving Green), and I have the impression that its European counterpart, The Shift Project, takes a somewhat more grounded approach. More broadly, I sometimes worry that focusing on decarbonizing aviation, cement, heating, or meat production can lead us to optimize the existing system rather than question what kind of system we actually want. I would like to see more resilient communities, sufficient food in both quantity and quality—fruits, vegetables, cereals, legumes, and perhaps some meat, including from agricultural residues or as part of agroecosystem management—and accessible healthcare. I’m not opposed to technological solutions when they are genuinely the most cost-effective way to achieve these goals, but I would value more diversification in climate philanthropy. Otherwise, what happens if we successfully decarbonize cement, aviation, and energy in 20 years, but then discover that transforming the underlying food, health, democratic, and social systems still takes another 30–50 years? I worry that we could end up having optimized the technologies around the existing system without having invested enough in making the system itself more resilient.
Existing climate giving is quite large, and it makes more sense to make that giving more effective than diverting EA resources to climate.
There is weak evidence that that the Giving Green Fund or Founder’s Pledge Climate Fund achieve $1-2 dollar/tCO2e mitigation. The interventions in these funds are based on poorly validated hypothetical expected value models—many closed/not shared, if based on models at all—and some bets have been particularly bad, as I discuss in this lengthy post. I cover issues with models from both organizations models in Part 8. There are wildly different (and weaker) evidentiary standards that GG and FP use compared to GiveWell in their effectiveness claims. In my estimation GG is improving, FP is not.
Other commenters have pointed out issues with the SCC calculation. I’ll further add that if you are going to include long term economic effects in the social cost of carbon, you need to do likewise for global health interventions when comparing. That was not done here as far as I can tell.
(I manage the FP Climate Fund, so my incentives run counter to my take). I hadn’t had the time to examine this paper in detail, but I am quite skeptical of this Rethink Priorities paper. I think it is quite easy to string together assumptions that yield a high social cost of carbon, but I wouldn’t treat this as an unbiased estimate.
For example, if I understand this correctly based on your description, they use the Rennert et al (2022) paper to derive the SCC from which they make adjustments.
https://www.nature.com/articles/s41586-022-05224-9/figures/1
The assumptions of that paper are clearly extremely pessimistic, probably by 2022 standards, but definitely by what would now be the consensus view.
For example, they assume close to 20/Gt annual emissions in 2100 as their median scenario and high emissions continue well into the 23rd century. In other words, we are more than a 100 years late in achieving net-zero in their median scenario despite all technological trendlines rendering this quite implausible.
Combining this with a low discount rate will give a high SCC, but I don’t think this is close to a reasonable baseline for what a median expectation should be. (Obviously good to have a low discount rate from an EA perspective, but this requires that the modeling of the future is a bit more careful). Essentially, this means that most marginal carbon reduction modeled for the SCC will happen in worlds where this is implausibly valuable thereby inflating the SCC value.
This alone probably leads to an overestimate of the SCC of a factor of 5x or more and this came up from looking at the paper for 5min.
Thanks for the comments.
A couple of quick responses.
Regardless of what any individual thinks about the underpinning assumptions and methodology, the Rennert et al (2022) SCC is widely used and “mainstream”.
It is certainly true that there is high uncertainty around the SCC, and a wide range of estimates. But the Rennert et al (2022) paper is just about as mainstream, “blue-chips” of an estimate as you are going to get, and therefore I think is a reasonable anchor for the Rethink analysis.
A few data points showing how popular the Rennert et al (2022) analysis is (and the underlying GIVE model plus the probabilistic assumptions on emissions pathways that underpin it):
This work was used heavily by the Biden administration’s EPA as part of a comprehensive update of their internal social cost of carbon, and their estimate was ultimately adopted. (detailed EPA report here.)
The German Environment Agency uses a very similar approach built on Rennert, but applies some different assumptions (lower discount rate, equity weighting), and comes to an estimate of 345 EUR (assuming 1% pure rate of time preference)
From Opus 5.0: “There is no single consensus number, but there is a fairly clear structure to where the field has landed. Short version: ~$190/tCO₂ (2020 USD, 2% near-term discount rate) is the de facto reference value”
Of course this doesn’t mean the paper or its estimated SCC is “right”, but I do think it’s a highly defensible reference SCC for Rethink’s analysis. As correctly mentioned by Vasco, there are numerous critiques of the Rennert paper, arguing that a lower SCC is more appropriate. But there are also influential critiques in the other direction. In the post I discussed how incorporating effects on economic growth can lead to much higher SCCs. For instance, a recent paper by Bilal and Kanzig (2026) argue for a much higher SCC of $1200 (even assuming a 2% discount rate).
Despite being “mainstream”, is Rennert et al (2022) unreasonably pessimistic about future outcomes?
If I understand correctly, Rennert et al (2022) use distributions of future states of emissions based on Rafferty et at (2017), which use historical data to assign probability distributions to the different IPCC emissions scenarios. If historical trends are not good predictors of future ones (for instance, due to rapid technological advancement), these estimates could be off.
Personally I’d agree with Johannes that the Rennert future emissions distributions feel pessimistic, but I would not agree with the modifiers “clearly extremely”. I think their mean estimate would fall within one standard deviation from the mean in my personal distribution.
But in any case, one could re-do the analysis with more optimistic distributions. Resources for the Future has a handy calculator that allows one to re-calculate the SCC with different models and different parameters. If we switch from the emissions distribution in Rennert (RFF-SP) and instead move to the more optimistic SSP-2 (which is the closest option to my beliefs, though I’d still be a bit more optimistic), SCC falls by 15%, from 185 to 158. Unfortunately the calculator does not allow us to move the discount rate to zero [edit: actually, we want to move the pure rate of time preference to zero, not the discount rate], which is really what we need to test the sensitivity of the Rethink estimate to different emissions futures. One would need actually re-run the models to calculate exactly, but a 5x wedge feels quite off to me.
Thanks, Dan, for your reply. I think I still disagree pretty strongly and I’ll briefly outline how and why: (1) I don’t think we should give credence on studies based on government usage, (2) 22 century emissions are clearly very implausible, (3) there are reasons for this to really matter and (4) being able to find this in 5min should make us deeply skeptical of the wider paper. (5) As you, I deeply care about effective climate action but I don’t think we should propagate bad evidence to inflate the importance vis-a-vis other causes.
Should we give special credence to something because the Biden admin and the German EPA used it?
I think the answer is pretty clearly “no”.
As you also mention, SCCs are inherently very uncertain estimates, but they also are clearly political projects; they are often produced or at least amplified to serve political purposes such as justifying more or less ambitious policies.
Given the biases in the climate science community, I think it is upon us to evaluate studies on their merits and to not assume that studies that are well-cited or used by climate hawkish administrations are therefore inherently more credible.
In case that sounds conspiratorial, it’s worth remembering that RCP 8.5 (the most extreme emissions scenario by the IPCC) was essentially abused prominently for 5-10 years as a baseline after it was obvious to all serious observers it was not a baseline but worse than a conceivable worst case scenario.
Given this record of the emissions scenario modeling community, we should always check what they assume.
Why it’s obvious that 22nd century emissions shouldn’t have that high a median
This is obvious to Dan, but for the wider group. The reason the 22nd emissions are so implausible is because they essentially assume that nothing much changes by the 22nd century, 3-5 investment cycles from now. Yes, decarbonizing the world by 2050 might be hard, but decarbonizing the world by 2100 or 2150 is not hard. It’s simply not plausible to believe in a 22nd century that is both richer than today, a hundred years in the future, exposed to a lot of climate damage, and not able to decarbonize.
This is much more than a 15% difference!
This is not about a 15% difference and a 5x over-estimate from this alone is quite plausible. The calculation you run is – as you note – not informative because it ignores the discounting and the implausibility of the Rennert et al paper lies primarily in the 22nd century+ and those will get outsized importance if changing the discounting assumptions.
Yes, by 2100 Rennert looks mildly pessimistic but not widely so, but that’s not the point – once discounting is very low the 22nd century matters and drives most of the result.
The 22nd century is discounted by more than 75% on the RFF explorers lowest discount rate (2%) so this is not a useful tool to explore whether implausible 22nd century assumptions massively inflate the SCC estimate.
And, indeed, we should expect the 22nd century to drive a massive difference when discounting is low. By Rennert et al’s assumption temperature keeps increasing well into the 22nd century, climate damage is non-linear, so on low discounting this will drive a massive amount of the damage and lead to strongly changed estimates.
I am not an economist, but when I Claude the emissions assumptions of the Rennert et al paper and stipulate to make conventional assumptions about climate damage at 0 discount rate, 96% of the damage is after 2100.
https://www.nature.com/articles/s41586-022-05224-9/figures/1
And we can see here why this is. In the Rennert world temperature keeps rising till after Captain Kirk with 25% probability mass above 4.5C. This is very different from what I think we both believe about the future.
This goes beyond this one example I picked this example because it was quick for me to find, having seen this pattern (making deeply implausible assumptions in papers that get a lot of policy traction) so often. But I am sure one could find other instances. The point that it is so easy to find a highly biased assumption cited as mainstream evidence is that one should distrust the broader paper more.
Why I care As you and many readers here, I deeply care about effective climate action and we are together on this project of making the climate response more effective.
But I do think we should not propagate bad evidence to inflate the importance of climate vis-a-vis other causes in the same way that we would also – and correctly – criticize a motivation of AI risk that only relied on Yudkowsky’s p(doom) estimate.
Implausible assumptions rarely come alone and when a paper makes such implausible assumptions (that are not simpler than more reasonable assumptions, i.e. this is not driven by parsimony) then I think we should be quite skeptical of the conclusions of that paper.
Thanks so much for interacting with our work. For broader context around this report, it’s important to note that our approach from the outset was to build a series of models to show what you would need to believe for climate to be competitive with GHD. We therefore aren’t wedded to any particular SCC, and certainly can imagine different ways of approaching the problem.
To your specific points:
It is our view that Rennert is the best middle ground available from current peer-reviewed work. It sits between DICE-2023 (~$66) at the lower end, and estimates that incorporate growth effects like Bilal & Känzig (2024) (~$1,000+ at the same 2% discount rate) at the higher end. So, whilst we agree that some parameters in the GIVE model can be questioned in isolation (your discount rate and impact horizon critique seems reasonable), we feel it is still a sensible middle ground when taken as a whole, in comparison to other peer reviewed estimates currently available.
On adaptation, we feel less optimistic that the principle of ‘richer future = strong adaptation’ will hold for all people (e.g. those most exposed and least able to adapt in LMICs) or all sectors (e.g. agriculture), with the intersection between the two being particularly noteworthy. However, we accept this is genuinely uncertain and that reasonable people will weigh it differently.
Overall, we think our analysis at least supports the re-visiting of climate as a cost-effective cause area within the EA space and we appreciate the various thoughtful comments so far. Thank you for engaging!
Hi Johannes, thank for all the detailed thoughts. I think we’re probably reaching diminishing marginal returns in debating point by point. I will say that we agree that the Rennert et al projections are too pessimisitic, and difficult far future projections can matter a lot with low discounting. Though I think we will need to agree to disagree on the magnitudes of these changes.
There is one technical point that I’d like to bring to your attention, though, that I think does critically matter. The Rethink application of a new discount rate is NOT a 0% discount rate. It is a 0% “rate of pure time preference”, which is one component of the discount rate. (I’m going to modify my original post to clarify.) Rethink’s calculation is built on calculations from the German Environmental Agency, so I will directly reproduce their explanations for how they did discounting:
“we use the social discount rate developed by Frank Ramsey (Ramsey 1928), which combines the two aspects above: expected consumption growth, weighted by its effect on the marginal utility of consumers, and the pure rate of time preference (PRTP).
In the GIVE model the consumption growth rate is a dependent variable. Therefore, it is not possible to specify the exact discount rate used for the climate costs”
I am not familiar enough with these models to get a good intuition for what discount rate would be, and therefore how much weight will be places on the far future. But as a reference, the Stern Review also used Ramsey discounting and a low (.1% PRTP), and resulted in an overall discount rate of 1.4%.
All of this to say, I think a simple mental model of “no discounting”, and therefore the vast majority of damages being in the far future, is not a good description of the Rethink analysis.
Hi Johannes.
I agree. Here is a critique of Rennert et al. 2022 by David Friedman.
The values of 17.1 and 36.6 $/t above respect representative concentration pathways (RCPs) 4.5 and 8.5. I think RCP 4.5 is much more plausible trusting the temperature changes below from Wikipedia.
I am not sure the comparison stays entirely symmetric once endogenous growth is added on the climate side. If climate damage can permanently change the growth path, then at least some GHD interventions should also have effects on growth through health, schooling, productivity, fertility and so on.
I assume some of that is already captured in the GiveWell/Coefficient models, but probably not the same kind of macro second-order effects. Since climate seems to become competitive mainly when these more uncertain effects are included, I would be interested in how much this matters.
Fair point Alex. We mostly interrogated the climate side of things, and in that sense the work itself is asymmetric. GW does incorporate second order effects into some of its CEAs, but I guess the larger question is whether or not there is any such large multiplier to apply to some GHD interventions for some amount of time. My quick reading of the literature there is simply that there isn’t strong evidence on the direction of growth effects for GHD, whereas the estimates we found on the climate side were at least consistent in direction (though I may be wrong here).
Hi Tom. Which discount rate did you use for the benefits from endogenous growth? GiveWell (GW) uses a discount rate on future consumption of 4 %/year.
Good q. We did not directly adjust the endogenous growth term through any discount, it enters the model only as a multiplier. That said, the multipliers we use have underlying discount rates that they use (generally larger than 0). We probably could’ve harmonized in a smarter way to be able to answer your question, but the general point you are trying to make, I think, is that higher discount rates here might be appropriate because of long stream of future harms/benefits (or perhaps just uncertainty). This would mean that our SCC is too large. Fair; and maybe!
Recall that our main SCC estimates are log-linear in income. In a linear model with 0% time preference, yes, the estimate would balloon because of the long tail. But in a log-linear model where people are getting richer, a 0% time preference essentially sets the effective discount to the growth rate (because future people are wealthier, and so additional dollar damages mean less to them). So this would mean that our initial SCC is probably smaller than you’d think *and* it is less sensitive than linear models to discount rates (by a nontrivial but not super large amount I would say).
Now the multipliers themselves, as I said above, use some discount rate and are not log-linear in income the way our base estimate is. So the actual discount rates are important here and might mean our multipliers modify our baseline SCC too muuch (our log-framework already downweighs far-off damages, but our current multiplier does not, making it large) or not enough (if the pure time preference is really closer to 0% than the rates they were estimated at, given how back-loaded growth damages are, making it too small).
All to say, the multiplier bit was a significant source of uncertainty to us when we wrote it (we labeled the endogenous growth multiplier as an adjustment with high uncertainty in our report) and happy to have had a bit more time to rethink some of these things and share our views on it now.
My understanding is that GWs pure time preference is similar to ours and that any differnece in effective discounts is related to how they treat uncertainty over the future, but I have’t looked into this.
Thanks for the clarifications. They made sense to me.
Here are the details (linked here) about GW’s discount rate of 4 %. Pure time preference does not contribute to this.
For me this is the crux and the key line.”Climate’s SROI becomes competitive with GHD when accounting for high-risk, lower-certainty effects like tipping points and endogenous growth”
If you’re the kind of person who both has a high risk appetite, and believes we have some plausible control now over low certainty events in the far future then I would agree with the thesis that we should put more money into climate stuff.
Personally I think the medium-far climate future might be as difficult to predict as the trajectory of AI. We have political uncertainty (e.g. Trump scrapping wind), economic uncertainty (e.g. solar became unexpectadly cheap), potential AI tech progress, potential rogue actors (someone could release sulfur). We could be at net zero by 2040 or never.
Anyway I think the uncertainty with climate is so large that you need to be a certain kind of giver for that to make sense—especially with the “existential risk” component perhaps not being there.
I don’t find much solace and meaning in error bars that wide—but many others do.
I’m curating this post. It’s a great summary of some under-discussed research, and it has the potential to kick off a very valuable conversation (which we are seeing the start of—thanks jackva). I’d love to see more of an in depth comparison of investments in climate mitigation vs global health vs economic growth, especially accounting for more funding coming to global health. This post is a great start. Thanks Dan!
As Abbie says in the report, climate can’t compete with GHD on high-certainty interventions. It can compete if we: (a) interpret the economic effects of climate damage as mostly affecting growth rather than inflicting many one-off costs and (b) count effects well beyond the next century. Assumption (b) looks particularly shaky to anyone with a high credence that AI-powered and other growth will radically increase global resilience to climate change.
Thanks Stan. I think this is a concise and reasonable critique- I mostly agree. A few thoughts.
My understanding is the CG recently did an internal calculation of SCC along the lines of the Rethink analysis, but one key difference is that they cut off damages around 2100 (not sure the exact year). This was just to have a like-for-like comparison with other interventions where they project benefits out for a fixed amount of time. They came to a conclusion of an SCC around 700. So if you take $1/ton as a reasonable benchmark for climate you get 700x, which doesn’t reach the CG bar but is in the ballpark. If I understand correctly, this CG SCC doesn’t include growth effects. So my point is that you don’t necessarily need long term and growth effects- with just one I think GDH gets competitive under reasonable assumptions, though perhaps not “winning”. (I haven’t actually seen this CG SCC analysis, but talked to people there about it. Hopefully it gets made public some day!)
For (b) I don’t think you need damages past 2200 to be competitive with GHD (especially if you do have endogenous growth effects), but the point is well-taken that you do need long-term effects, up to 2100 and beyond. My personal central estimate is that by 2100 we’ll have reached near net-zero, at warming of ~2.5 degrees C. But even though emission stops, the warming will persist for centuries, and therefore has the ability to continue to cause damages (compared to a no or lower-warming counterfactual). Of course adaptation is the wildcard. Will AI-powered growth radically increase resilience such that climate damages will drop to zero even in the face of persistent increased temperatures? I am not as optimistic as you are, but I see your perspective.
Thanks for synthesizing and sharing this, Dan.
My sense is that the social cost of greenhouse gases carries fairly wide uncertainty bounds, driven by discounting assumptions and long-run projections. I’m underqualified to judge how robust these are, though I’d guess the biases could run either way.
What I’m more interested in is the second-order effect on cause prioritisation. More incoming philanthropic funds are likely to saturate grantmakers in global health and development, pushing the cost-effectiveness bar down, and we’re already seeing that with GiveWell and Coefficient Giving.
But I wonder if part of what we’re seeing there is really an absorptive capacity constraint (a limited set of organisations able to scale up delivery), rather than a genuine shortage of cost-effective interventions. Scaling delivery takes time, people, and systems, and that constraint doesn’t come through clearly in your piece. Whether climate faces the same constraint isn’t clear to me either.
You mention Giving Green’s large 2025 grant will take roughly a year to fully disburse. How should we think about absorptive capacity in climate versus GHD?
If climate absorption is genuinely constrained, then GHD interventions, with likely smaller uncertainty bounds, could see absorptive capacity grow over two to four years as delivery organisations scale. That would push the cost-effectiveness bar back up and soften the piece’s headline conclusion. Keen to hear how you’d think through this dynamic.
Thanks a lot for these thoughts, Tony. I’m not sure I have all the answers, but a few ideas to throw in the mix:
A lot of what we think of as cost-effective interventions in GHD are on-the-ground, direct-delivery type stuff (bednets, vaccines) that require a lot of logistics and therefore large, sophisticated orgs to manage them. This isn’t really true for climate (at least in my opinion) where we are focused more on charging policy, technology, and markets. There are still constraints in org capacity, but they look pretty different.
The climate philanthropy ecosystem actually has a pretty impressive build-out of various funding vehicles that allow lots of money to move very quickly. For instance, there are big cross-sectoral regrantors like Climateworks, tons of sector and region specific regrantors (Industry Hub, Tara Climate Foundation), and focused research organizations working on specific topics (Carbon2Sea, Reflective). These organizations frequently see their missions as building up capacity in a topic as well as funding it. From someone with an EA-aligned perspective, if you believed that the mission of any of these orgs was super cost-effective, they would provide a vehicle to move lots of money quickly. For instance, Founder’s Pledge made a ~23M grant to Deploy/US based on this logic. We at Giving Green have so far shied away from giving to regrantors, since we think that keeping tighter control of our grants allows money to move faster (no middleman), and allows us to really focus on what we think is most impactful. But one could definitely debate that choice, and if lots more money came our way we would certainly revisit this assumption.
As a grantmaker in climate, I haven’t felt the absorptive capacity of the field to be a huge constraint. Sometimes we have said “we really want to fund orgs doing XYZ, but can’t find people doing it”, but I’d say that’s the exception rather than the rule. Maybe if I was moving much larger amounts of money (like GW), I would feel these constraints more strongly.
I really hope that organizations delivering really cost-effective GHD work can scale to meet the moment, that the OP/GW bar goes back up, and that the headline conclusions of this research are indeed softened. That would be great for the world!
Has anyone tried to estimate the social cost of carbon taking into account AI? I’ve seen several well-calibrated people estimating that there is a 20% probability that AI fizzles in the next few decades. If you believe that that would be permanent and AGI/ASI would solve the climate problem or result in catastrophe, to a first approximation, that would mean the social cost of carbon would be about 1⁄5 as much as people typically calculate. In reality, AGI/ASI may come further in the future, and there could be some benefits of reduced carbon even with AGI/ASI.
My understanding of EA funding is that almost none goes to climate. I (loosely) tend think that it would be good that a bit more resources go to Giving Green and other environmental prioritization projects.
However, I also gather that such projects attract funding far more easily than most other EA-consensual-ish causes.
Thanks for this contribution. I hadn’t come across the Rethink Priorities analysis, so this was a great nudge. Responding to both your post and the RP research, I think it’s important that the EA community recognize that certain climate change mitigation interventions double as global health and development interventions. And I’m not just referring to countering the social cost of carbon.
Climate change mitigation interventions, particularly in low- and middle-income countries, can be designed to not just address carbon emissions but also enable better healthcare and education delivery.
To give an example, there are an estimated 6.5 million diesel generators across Sub-Saharan Africa. These generators are polluting, prone to breakdowns and fuel for them is expensive. Yet households, businesses, health clinics or schools often have no other power option because they aren’t connected to the electricity grid, or if they are, the grid is prone to outages. So, diesel generator it is! Until there’s a fuel shortage. Or the price of diesel spikes. Or the generator, which was probably bought second-hand, fails. Then surgeries get cancelled and vaccines spoil from no refrigeration.
This is why a technology like a solar mini-grid, which can deliver electricity at a levelized cost up to 60 per cent lower than diesel, should be viewed as more than just a climate change mitigation solution. It provides a critical input to healthcare and education delivery, including some of the interventions that the EA community funds (think vaccination rollout).
Obviously, there’s a lot more to unpack in terms of what it takes to replace something like 6.5 million diesel generators with clean energy alternatives, including the cost of doing so. I just think that it’s important to recognize that the human impacts of clean energy deployment go beyond alleviating the social cost of carbon.
Has the EA community already looked at this nexus issue? Would love to connect with anyone thinking about or researching the full impact of clean energy, particularly its role as an enabler of interventions in GHD.
Hey Stephen, we at Giving Green actually started a project looking exactly at this- trying to identify cost-effective interventions and funding opportunities at the nexus of health/livelihoods and climate mitigation. We put the research on pause because of other near-term priorities and some uncertainty over whether there was hunger for this kind of thing among donors. But it’s something we hope to pick up again next year.
Great to hear that you and the team have been exploring this! I’d be keen to learn more about how far you got in your thinking/analysis, but realize it’s on the backburner for the time being. Let me know if I can be of any help when/if you resurface the topic.
Basically all EA money at this point should be going to AI safety (where AI safety is broadly defined to include not just loss of control / human extinction, but also things like concentration of power, value lock-in, and AI welfare). Anything else seems like a distraction.
Thank you for your post. I agree that climate change is underfunded in EA, and I’m happy to see Rethink Priorities updating its models. I also wonder whether GHD is somewhat overfunded by EA, especially since some interventions can accelerate economic development in ways that increase environmental pressures if they are not accompanied by systemic change. Climate change will itself have major consequences for global health and development. I’m particularly skeptical of approaches that rely on producing mosquito nets, medicines, etc. in Europe and shipping them elsewhere rather than strengthening local capacity and resilience.
On another note, I came to the Forum because I want to raise money for climate action during a demonstration in Geneva. Having been part of the EA community for a few years, I naturally looked at Effective Altruism Switzerland’s climate fund, but something didn’t quite click for me. I had a similar feeling when looking at Giving Green: it seems quite focused on technological solutions. I’m therefore looking for another organization to support, although I’m open to changing my mind. I’ve been a fan of GFI for years, for example, but I’m not convinced that plant-based meat and dairy analogues are a sufficient answer to climate adaptation in agriculture—they are only one part of the solution. What about the campaign beans is how? I also looked briefly at Deploy/US (though obviously nowhere near as deeply as Giving Green), and I have the impression that its European counterpart, The Shift Project, takes a somewhat more grounded approach. More broadly, I sometimes worry that focusing on decarbonizing aviation, cement, heating, or meat production can lead us to optimize the existing system rather than question what kind of system we actually want. I would like to see more resilient communities, sufficient food in both quantity and quality—fruits, vegetables, cereals, legumes, and perhaps some meat, including from agricultural residues or as part of agroecosystem management—and accessible healthcare. I’m not opposed to technological solutions when they are genuinely the most cost-effective way to achieve these goals, but I would value more diversification in climate philanthropy. Otherwise, what happens if we successfully decarbonize cement, aviation, and energy in 20 years, but then discover that transforming the underlying food, health, democratic, and social systems still takes another 30–50 years? I worry that we could end up having optimized the technologies around the existing system without having invested enough in making the system itself more resilient.
Thank you for your time,
Alexis Schoenlaub
aschoenlaub@hotmail.fr
Two main critiques:
Existing climate giving is quite large, and it makes more sense to make that giving more effective than diverting EA resources to climate.
There is weak evidence that that the Giving Green Fund or Founder’s Pledge Climate Fund achieve $1-2 dollar/tCO2e mitigation. The interventions in these funds are based on poorly validated hypothetical expected value models—many closed/not shared, if based on models at all—and some bets have been particularly bad, as I discuss in this lengthy post. I cover issues with models from both organizations models in Part 8. There are wildly different (and weaker) evidentiary standards that GG and FP use compared to GiveWell in their effectiveness claims. In my estimation GG is improving, FP is not.
Other commenters have pointed out issues with the SCC calculation. I’ll further add that if you are going to include long term economic effects in the social cost of carbon, you need to do likewise for global health interventions when comparing. That was not done here as far as I can tell.