One can mark the dawn of the Industrial Revolution with James Watt’s patent for the steam engine in 1769. The Industrial Revolution led to the first sustained increase in average living standards for millennia, but also centuries of almost unchecked fossil fuel burning.
Greenhouses gases like CO2 and methane trap heat heading out from the Earth and so have a warming effect. Since the Industrial Revolution, average global temperatures have increased by around 1°C, with most of the warming coming after 1980.
By emitting CO2, we all contribute to a problem with implications for the planet that will last millennia. Each time you drive your car, a sizeable fraction of the CO2 you release will stay in the atmosphere for millennia. Atmospheric CO2 concentrations and temperatures will only return to pre-industrial levels after hundreds of thousands of years.
In spite of increased awareness of climate change, more than half of the CO2 that has ever been emitted has been emitted in the last 30 years. As a consequence, atmospheric CO2 concentrations have risen to 415ppm — which is a 40% increase from what they were in 1769, and higher than they have been for at least hundreds of thousands of years. If current trends continue, by the end of the century, CO2 concentrations and global temperatures will be higher than they have been for tens of millions of years.
Making the transition to clean energy is also important from a near-term point of view. Burning fossil fuels not only causes climate change, but also kills millions of people every year via air pollution. Moreover, nearly a billion people lack access to reliable electricity and almost 3 billion rely on biomass (wood, animal dung, and crop waste) for cooking, with terrible consequences for health. Low-carbon energy sources — such as solar, wind, and nuclear — tend to have much lower health costs.
Thus, transitioning to low-carbon energy would be a win from both near-term and long-term perspectives.
Our aim is to analyse how much good people can do by donating to climate change. Not only this, we also want to compare climate change to other problem areas. This makes our focus here quite different to most discussions of climate change. For people aiming to do the most good with their donations, it is crucial to understand not only that donating to effective climate charities is a good thing to do, but also that it is better than donating to all other charities. This is an extremely high bar. Along with climate change, the world faces many other huge challenges, such as pandemics, nuclear war, steering the trajectory of AI development, global poverty and disease, factory farming, mental illness, and so on. When deciding where to donate, we need to prioritise between these problems. This requires difficult tradeoffs.
The discussion that follows should be understood in that context. We are not comparing climate change to trivial things, like the consumption of luxury goods, but rather to the best possible ways to do good.
This report is also not focused on what society as a whole should do. We think it is clear that society should be spending far more on climate change. But our audience is individuals deciding how to have the most impact on the margin with their donations. Thus, when we argue that other problem areas may be more pressing, we are not saying that social spending on climate change should be reduced. Rather, we are saying that, given the current allocation of spending, individuals may be able to have an even greater impact by donating to other problem areas.
It is widely agreed that climate change will be bad for the world, but just how bad it will be depends on three things:
All three of these factors are uncertain, so we cannot say with confidence just how bad climate change will be. A useful starting point is to consider the impacts of the most likely ‘business as usual’ trajectory, as well as an extreme ‘worst-case’ scenario in which we burn through all of the fossil fuels.
The following chart shows estimates of how likely these scenarios are, and the levels of warming they imply.
Due to extraordinary progress on renewable energy sources and batteries, combined with increasingly stringent global climate policy, it now looks as though we can avoid some of the most extreme emissions scenarios. Recent studies now suggest that, on current policy, we are most likely to follow what the Intergovernmental Panel on Climate Change (IPCC) calls a “medium-low” emissions pathway. This implies a most likely level of warming of 2.5–3°C in 2100, relative to pre-industrial levels. There is even some evidence that these studies might be too pessimistic because they underestimate likely progress on renewables, batteries, and other technologies, and do not account for existing targets and stronger climate policy in the future.What about a worst-case scenario? While it seems unlikely that we would burn through all of the , it is not impossible, and it is useful to consider just how bad climate change could be. Although we now have clear ways to decarbonise many sectors — such as increasing the share of renewables in electricity, and replacing petrol cars with electric cars — decarbonising the entire energy system is a much harder challenge.
The chart below shows the proportion of carbon emissions different industries are responsible for. The protruding parts of the pie chart highlight the industries that will most struggle to eliminate their emissions.
All in all, difficult-to-decarbonise sectors account for around a quarter of total fossil fuel emissions. If we fail to completely decarbonise the economy, emissions could continue for a very long time indeed. If emissions fall to a quarter of their current levels, then we would burn through all of the recoverable fossil fuels in about 750 years. If we did, there would most likely be around 7°C of warming relative to the pre-industrial period, and a 1-in-6 chance of warming of more than .
So, on business as usual, we look set for around 3°C of warming, but in the worst case, there would be more than 7°C of warming.
How bad would these different scenarios be for the world? The answer to this question depends in part on difficult ethical questions. In this post, we will broadly consider two different ethical positions: neartermism and longtermism.
Longtermism is the idea that ensuring that the long-term future goes well is a key priority of our time. The motivating idea behind longtermism is that future people and current people count equally. Since there could be so many people in the future, and there seem to be ways to make their lives go better, ensuring that the very long term goes well is highly morally valuable. Longtermists take a variety of approaches to protecting the long term, including working to prevent extinction or major global catastrophes that could derail civilisation; preventing prolonged economic stagnation; and preventing the lock-in of suboptimal values.
In contrast, neartermism is especially focused on improving the lives of people alive today, for example by reducing poverty or improving global health. One argument for neartermism appeals to particular views in population ethics that place more weight on the interests of people alive today than possible future people. Another argument is that we don’t really know how to influence the long term, so we should focus on things with more concrete and predictable outcomes.
Climate change is important from both neartermist and longtermist points of view, but taking one perspective over the other might lead you to focus on one particular set of risks.
In what follows, we focus on what we see as the most important humanitarian consequences of climate change:
In this context, a ‘tipping point’ refers to some threshold that might — when exceeded — cause a sudden, substantial, and potentially irreversible change to the climate, which does significant damage to human society. If there are such tipping points, it’s vital we know what the consequences of reaching them are.
A good place to start is to look at the past — have there been tipping points before?
Historical tipping points
Earth’s climate is not naturally stable. The chart below shows global temperatures over our current geological era, the Cenozoic Era (which started 66 million years ago, following the extinction of the dinosaurs):
Agriculture and written history only started in the Holocene Epoch, around 12,000 years ago. But this is only 4% of the total lifetime of our species. The stable Holocene climate is the exception rather than the rule.
At the global level, these climate changes happened slowly compared to future man-made warming. However, there have been more dramatic regional climate changes — regional droughts that set in over the course of a few years (lasting for decades or longer), regional warming of more than 4°C over the course of a century, and so on. These changes were triggered by relatively small but poorly understood triggers. So, we need to take seriously the possibility that greenhouse gas emissions could trigger global and regional tipping points.
What tipping points could we hit, and how bad would it be?
There is a good discussion of some potential tipping points in this article by Carbon Brief. Some of the most concerning are rapid warming and regional droughts, a collapse of the Atlantic Meridional Overturning Circulation (a major current system in the world’s oceans which plays a crucial role in regulating the climate), and changes in monsoons. All of these could have large negative effects on agricultural production in certain regions, though they would probably fall well short of reducing food production by more than 10%.
The risk of passing these tipping points is difficult to quantify, but seems non-negligible, and increases with warming. For example, the risk of collapse of the Atlantic Meridional Overturning Circulation is minimal today, but at 4°C increase, it would rise to around 1 in 20.
It is clear that there is scope for climate change to do unexpected damage at the regional level. So, from a neartermist point of view, climate tipping points are clearly important. How it compares to other neartermist priorities, like improving global health and development, is a very difficult question to answer because the effects of climate change are so broad, diffuse, and difficult to quantify.
From a longtermist point of view, the most important tipping points are those that could cause global societal collapse or even human extinction.
Here, we will cover what we see as the most worrying potential tipping point: loss of clouds leading to rapid warming. We’ll then consider what the effects of significantly increased warming would be in general.
The reason we are especially concerned with the loss of clouds is highlighted by a recent study which found that once CO2 concentrations reach around 1,300ppm, stratocumulus clouds will burn off, and there will be 8°C of warming over the course of several years — on top of the 6–7°C we would already have lived . This threshold is within reach on the worst-case scenario, but not on business as usual.
Scientists disagree on how plausible this finding is, but it is difficult to know how serious the risk is because CO2 concentrations have not been greater than 1,300ppm for at least tens of millions of years. But even a low probability that this level of warming could happen should greatly concern us.
Whether or not it occurs through this tipping point or another, what would this level of warming mean? Unfortunately, there has been little to no scientific research into this question. Agriculture would probably still be possible, but it is difficult to know exactly what would happen. Lethal limits for major food crops are between 40 and 50°C. With 15°C of warming, some places in the tropics would pass these limits for part of the year. However, North America, Europe, and China would not.
Although the risk of tipping points is real, there seem to be limits to the risks to life.
Firstly, although regional and global climate systems are not naturally stable, they also cannot be extremely unstable: it is not easy to trigger tipping points that threaten to kill all life on Earth. Two possible extinction scenarios are ‘moist greenhouse’ and ‘runaway greenhouse’ effects. In either scenario, temperatures become so hot that the oceans would be lost to space (as has occurred on Venus). The existing models suggest that it is not possible to trigger a runaway greenhouse effect on Earth with fossil fuel It also seems unlikely that we could trigger a , but if we did, CO2 concentrations would naturally decline over hundreds of thousands of years, well before the Earth’s water would be lost to space. Also, 56 million years ago, the Earth was 16°C warmer than today without triggering runaway greenhouse effects.
Secondly, life on Earth seems to be remarkably resilient to climate change: global biodiversity has increased steadily since the last mass extinction and ecosystems generally seem to persist in the face of even dramatic climate change.
Indeed, humans and our hominid ancestors have survived huge natural climate change. Humans lived through an ice age that was 4°C colder than today. Hominids survived when temperatures were 3°C warmer than pre-industrial levels. At this time, the hominids would all have been in Africa, so would have had to survive temperatures that were probably 13°C warmer than modern-day Britain.
In many ways, our hominid ancestors were in a much more fragile situation than we are: they were much less numerous, technologically advanced, and capable of a rational response to problems than people are today. Thus, their survival should provide some reassurance that we will make it through 3°C of warming.
However, one key difference between modern humans and our hominid ancestors is that they were hunter-gatherers, whereas we rely on agriculture. Agriculture has only flourished in the relatively warm, high CO2, and climatically stable Holocene. It is therefore crucial to examine how the changing climate might affect agriculture.
Global warming affects crop yields, soil moisture, and the productivity of agricultural workers. Overall, both business as usual and worst-case scenarios look set to have damaging effects on agriculture, especially in poor agrarian countries in the tropics, though the most likely effects fall short of global catastrophe. However, low-probability tipping points potentially pose a much greater threat.
The chart below summarises the impact of climate change on crop yields for the major food crops in different regions. Note that to fully understand this chart, you may need to read more of the paper it was part of.
The chart anticipates mixed, but largely negative, consequences from increased temperature on crop yields. The effect is most severe in tropical regions, and less bad (but still negative) in temperate regions. The chart only shows agricultural impacts of up to 5°C of warming; studies on even more extreme warming are scarce, but generally suggest that impacts increase as warming increases.
Bear in mind that these impacts on crop yields will occur in the context of substantial and consistent improvements in agricultural productivity. In the last 60 years, yields for most foods have increased by 200% or more in all regions.
Modelling studies suggest that climate change will damage crop yields, but that overall yields will improve due to technological progress. The Food and Agriculture Organization and others have found that technological change up to 2050 will outpace the effect of climate change for almost all scenarios: almost all crops, almost all agricultural systems, and all regions.
The effect of climate change on drought is determined by changes in precipitation and higher temperatures, which increase the rate of evaporation and soil moisture loss.
Models suggest that climate change will have mixed effects on precipitation, but will generally dry out soils due to faster evaporation. The chart below shows the effect of global warming of 4–5°C relative to today.
One way to adapt to droughts is irrigation. In India, 38% of agricultural land is irrigated and 60% in Bangladesh, so even relatively low-income countries can use this technology.
Productivity of agricultural workers
People are less productive in warm and humid conditions, so global warming would start to create problems for people working outdoors.
The chart below shows how rising heat stress would affect current labour capacity in different regions, as the global population is currently distributed, and without adaptation.
As you can see, tropical regions are currently at 75% of their labour capacity. However, 3°C of warming would decrease this to 50% of its potential, while 8°C of warming would decrease it to 20%.
Air conditioning, migration, and mechanisation of farm labour could mitigate these effects in some situations, but there are limits to these adaptations in poor agrarian economies.
One counterargument to this is that countries will no longer be poor and agrarian by the time that global warming passes 4°C. However, this seems unlikely given past trends in economic growth. The table below shows how well-off different regions will be in 2100 if recent trends continue. (Of course, trends may not continue, and we should expect growth to slow down, especially in richer regions.)
|Region||Per capita income 2019 (USD)||Median growth 1960-2019||Income per capita given median growth 2100 (USD)|
|East Asia & Pacific||$11,527||3.5%||$186,387|
|Europe & Central Asia||$24,696||1.6%||$86,422|
|Latin America & Caribbean||$8,847||1.7%||$33,480|
|Middle East & North Africa||$8,105||0.4%||$11,081|
As this shows, average growth has been very slow in sub-Saharan Africa. This average masks substantial variation: some African countries (such as Rwanda and Botswana) have recently enjoyed extraordinary growth, while many other economies in Africa (such as Niger and Democratic Republic of Congo) have essentially never experienced sustained improvements in living standards. There is a good chance that these countries will still be poor even in 2100, and so will have limited ability to adapt to climate change.
The bottom line: How much could climate change impact agriculture?
All in all, it looks like low-income countries in the tropics could be hit hard by the agricultural effects of climate change, but that the most likely effects will fall well short of global food catastrophe (i.e., something that could kill upwards of 10% of the global population).
However, lower-probability tipping points could do much more damage. Most importantly, if there were an abrupt collapse in the Atlantic Meridional Overturning Circulation, temperatures would decline by 3–5°C in the Northern Hemisphere (on top of a warmer starting point due to climate change). Cooling is generally much worse for crops than warming, because one day of frost destroys the entire growing season. There would also be large changes in rainfall patterns.
The Northern Hemisphere today accounts for almost all agricultural production. The effect that collapse of the Atlantic Overturning Circulation might have on global agriculture has not been studied extensively. Britain would experience some of the most dramatic climatic effects, but would be able to offset a lot of the impact by using irrigation. Whether poorer countries would be able to adapt to the changes in rainfall patterns is less clear, but the humanitarian consequences could well be very bad.
Climate change could also indirectly increase risks, such as conflict.
The effect of climate change on conflict to date is controversial. Some studies suggest that climate change has caused increased levels of civil conflict in Africa, but there is consensus in the literature that it has so far been a small driver relative to other factors such as low per capita income and states’ low capacity to .
It is hard to say what the effect on conflict would be on the worst-case scenario, as this is far outside of lived human experience. If poor agrarian countries do start to experience serious agricultural disruption, droughts, or flooding, that might lead to political instability and damage international coordination. Indirect risks such as this are difficult to quantify, but plausibly account for a substantial fraction of the global catastrophic risk from climate change.
The most important potential future conflict is war between the great powers, including the US, China, India, and Russia. Few scholars have raised climate change as an important driver of this conflict; other factors (such as the shifting global balance of power) seem much more important.
Overall, the evidence suggests that business as usual scenario warming will impose major costs on poor agrarian countries in the tropics, and that worst-case scenarios pose serious but difficult-to-quantify direct and indirect global catastrophic risks.
From a neartermist point of view, climate change is clearly a very important problem, though it is hard to compare to other areas, such as global health and development. Quantifying all of the possible costs of climate change is extremely difficult, and economic models that attempt to do this are often criticised for being unmoored from reality. However, without quantifying, it is simply hard to know how climate change compares to global health and development in terms of scale. This is an active area of research for effective altruism research organisations, and we are likely to know more in the next few years.
From a longtermist point of view, it is important to consider not just how bad climate change might be, but also how it compares to other important global problems. Researchers on global catastrophic risk generally think that, although climate change is one of our greatest challenges, other emerging threats pose even greater risks. For example, Toby Ord argues in The Precipice that the direct extinction risk of unaligned artificial intelligence and engineered pandemics are each 100 times greater than that of climate change.
Indeed, both the business as usual and worst-case scenarios for climate change look less severe than the risks in other top priority problem areas, such as nuclear security, engineered viruses, and advanced artificial intelligence.
Take the risk of engineered viruses as an illustration. Estimates from forecasters and experts put the chance of an engineered virus causing human extinction this century at . The risk of smaller biological catastrophes killing upwards of 10% of the global population must be much larger because it is so hard to kill everyone on Earth. In contrast, it is difficult to see how climate change could cause extinction this century.
However, these risks are difficult to work out — and we (the two authors of this piece) have differing views. For example, John thinks that the risk of climate change killing more than 10% of the global population this century seems lower than 1%, whereas Johannes is more uncertain on this. John’s view is informed by the fact that it seems the most damaging tipping points can only be triggered after very high emissions, which would require a reversal of current climate policy to occur this century.
The indirect risks of climate change are even harder to quantify, but climate change seems to be quite a weak lever on potential great power conflicts this century. Moreover, other priority risks are also likely to have important indirect effects and, in John’s view, it is reasonable to think that these would be larger because greater direct risks have greater indirect risks. It would be surprising if the indirect effects of climate change were large enough to outweigh both the direct and indirect effects of these other risks. Johannes is more uncertain about this.
Overall, climate change is indeed one of humanity’s greatest challenges, but seems to pose lower risks than other global risks, such as engineered viruses, advanced artificial intelligence, and nuclear war.
When we are prioritising problems, the scale of the problem is not the only thing that matters. We also need to consider how easy the problems are to solve.
There are two main reasons to think that climate change is more tractable than other problems we need to address to safeguard the long-term future.
Firstly, there is a clear success metric for climate change: we know we are winning if we reduce carbon emissions. Compared to other problems like AI safety, nuclear security, and biosecurity, it is much clearer whether we are making progress on climate change.
Secondly, because success is relatively easy to measure, it is easier to identify the most promising ways forward. There are now several climate success stories which suggest that progress on climate change is possible if efforts are carefully designed. For example:
Because climate change has such a clear success metric and different solutions are now so well-tested, it is one of the more tractable major global risks.
From a neartermist point of view, it is less clear how climate change compares to other interventions, such as fighting malaria or global poverty.
When we are deciding how to make the biggest impact on the margin, it is important to consider how neglected the problem is: how many resources are already going towards the problem? Problems that are more neglected still have low-hanging fruit, and diminishing returns have not set in.
Climate change is clearly much more neglected than it should be. However, other causes and global catastrophic risks discussed in the effective altruism community are even more . Each year, hundreds of billions of dollars are spent on climate change by governments and the private sector. In recent years, philanthropic spending on climate change has also increased significantly: in 2021, around $5 to $10 billion US was spent on climate philanthropy worldwide. Spending on other global catastrophic risks is orders of magnitude lower than this.
Still, given that climate change mitigation is a vast cause area that covers the entirety of global economic activity, it would be a mistake to assume that increased climate funding means there are no neglected spaces (see our discussion on this below).
Taking into account scale, neglectedness, and tractability, climate change overall is one of the most pressing problems in the world — it is clear that society should put much more effort into fixing it.
First, let’s consider a longtermist point of view. While other problems (such as risks from advanced artificial intelligence or biosecurity) are more severe and neglected, climate change’s higher tractability makes it unclear how cost effective it is at the margin. Also, because so many resources are going into climate change, we can have significant leverage by affecting how these resources are allocated — indeed, we see this as the primary pathway to high cost effectiveness in climate philanthropy (see below).
Nonetheless, both authors of this report currently believe that AI safety and biosecurity are more promising for donors who are neutral across causes and seeking to have the greatest impact on the margin.
It is much harder to say how climate change compares to other cause areas from a neartermist point of view. It is very hard to compare the scale and tractability of climate change to global health and development. This is an active area of research for effective altruism research organisations, and we are likely to know more in the next few years.
There are numerous ways to try to tackle the problem of climate change. In this section, we discuss the best ways to make a difference by establishing a framework for how to think about climate impact, and then highlight solutions that we think look particularly promising using that framework.
When evaluating which approach is most promising for donations, we need to consider four key facts:
Let’s first look at trends in emissions in different regions. Emissions in Europe and North America have probably already peaked, while they are rising rapidly across Asia.
The US and the EU are expected to contribute at most about 15% of emissions in the 21st century. So, if we are going to avoid the worst-case outcomes, we need to find ways to reduce emissions outside the West.
However, emerging economies still consume much less energy per person than Western countries. Since energy use is strongly correlated with living standards, there are strong humanitarian reasons to ensure that people can meet their growing energy needs in the future. We therefore need to find solutions that reduce emissions in emerging economies without damaging living standards.
Emissions reduction is a global and intergenerational public good: each country captures only a fraction of the benefits of cutting their emissions. As a result, countries tend to underinvest in emissions reduction.
Efforts to reduce emissions in one country can also be undermined by ‘carbon leakage.’ For example, if the US imposes emissions restrictions on the steel industry, the industry may just move somewhere with more lax restrictions.
Therefore, we need to focus on solutions that do not rely on extensive global coordination.
Philanthropic spending on climate change is 100 to 200 times lower than public and private spending. This means that climate philanthropy can have significant leverage by focusing on the most effective interventions to improve the response of governments and the private sector.
By and large, climate philanthropists’ attention is not currently where future emissions are — some key sources of future emissions remain severely neglected.
Founders Pledge’s climate philanthropy report contains a chart highlighting the distribution of climate philanthropy by region and sector. We summarise its key findings below.
Climate philanthropy outside the EU and US is almost entirely focused on two sectors — clean electricity (overwhelmingly renewables), and forests and other natural climate solutions — paying only marginal attention to large buckets of emissions (such as industry and agriculture).
While there is relatively more funding across a variety of sectors in the US and EU, funding is still far from balanced — it’s heavily tilted towards electricity and electrified transport, and neglects the hardest challenges most in need of additional effort (such as industry, transport, carbon removal, and agriculture).
For a more detailed discussion of how climate philanthropy is spent and how this compares to emissions sources across sectors and geographies, see the Founders Pledge report.
These four facts inform which approaches to solving climate change are likely to be most promising. In particular, they speak in favour of approaches that:
The rest of this section discusses some leading approaches in addressing climate change:
A lot of climate philanthropy aims to accelerate the adoption of mature and popular technologies, such as solar panels, wind turbines, and electric cars. For example, renewables-targeted philanthropy receives at least $500 million US per year.
These were early-stage technologies 10 to 20 years ago, and were critically dependent on policy support. Philanthropy played an important role in building this support, and improving those technologies produced enormous global benefits. But as these technologies mature, the returns on this work declines.
Consider this example of a typical learning curve cumulative deployment to cost reductions:
This model, roughly fitted to the learning curve observed in solar photovoltaic (PV), shows that increasing solar PV from 1 to 2 units has more than 400 times the value of the shift from 99 to 100 units. Of course, the unit from 99 to 100 is also considerably cheaper, costing about one seventh of the first unit. However, on this , the cost-reduction returns to buying the first unit are still about 30 times greater when assuming that deployment-related factors (learning by doing, economies of scale) drive about 60% of cost reductions.
Applied to real numbers from solar deployment and cost, adding 40 GW in 2010 (from ~42GW in 2010, i.e., a doubling) would have reduced cost by around $1,301 US per kW, whereas the same addition of capacity in 2017 would have only reduced cost by around .
Importantly, early cost reductions can create a virtuous cycle: increased consumer demand drives further cost reductions, which drives further demand, and so on. Moreover, once an industry grows, it develops political clout and special interest power that make the effects of additional philanthropy less effective. These dynamics provide a strong argument for supporting early-stage technologies rather than mature technologies.
There are several other reasons to think that supporting mature technologies like wind, solar, and batteries is not the highest-impact strategy:
One possible exception to this is if deploying mature technologies quickly could prevent locking in carbon-intensive infrastructure (discussed below).
One especially promising approach to climate change (as identified by Founders Pledge and Let’s Fund) is to focus on clean energy innovation. The great virtue of innovation in low-carbon energy is that it has potential to impact emissions in the sites of future energy demand, but does not require cooperation.
For example, Germany’s subsidies for solar power (mentioned above) drove down the cost of solar for Germans, but much more importantly, helped to drive down the cost of solar for the entire world. Crucially, Germany and other jurisdictions involved in this effort (such as the US, Australia, and China) played different roles in this effort, .
This theory of change can be summarised as follows:
Gigaton impact (massively improving the world!) is the end goal.
Even under quite conservative assumptions, this type of advocacy can be highly cost effective due to the leverage of advocacy and global technology spillovers. The benefits are especially large for neglected low-carbon technologies for the reasons discussed above.
While support for solar, wind, and electric cars has been a major success story, certain other key technologies — such as nuclear power, zero-carbon fuels, carbon capture and storage, enhanced geothermal systems, and energy storage are lagging behind.
There are four main objections to an innovation-focused approach to climate change:
While some of these objections somewhat weaken the case for innovation, on balance we still think it is one of the most promising solutions to the climate problem. This theory of change should play a much bigger role in OECD economies, where future emissions are low and partially constrained by policy, but innovation capacity is large.
Carbon lock-in occurs when decisions made today have long-lasting effects via carbon-intensive assets, such as building new coal and steel plants that will last for decades. For example, although the declining cost of renewables affects the prospects for new coal plants, it is much more costly to prematurely retire coal infrastructure that has already been built. This is one strong reason to try to avoid building new fossil fuel infrastructure in emerging economies.
The most promising work on avoiding carbon lock-in will target:
There are two primary reasons to expect this philanthropic strategy to be cost effective:
There are three primary reasons to be sceptical of this strategy:
As of 2022, Founders Pledge thinks that avoiding carbon lock-in is likely one of the highest-impact theories of change, and will continue to do more research and grantmaking in this area.
Some economists believe that carbon pricing is the ideal solution to climate change. However, despite decades of advocacy from economists, progress on carbon pricing has been poor. At the moment, the global average carbon price is around $2 US per tonne and the vast majority of emissions are not priced at all.
Indeed, it is worse than this. Once we take account of fossil fuel subsidies, the true global net carbon price is negative. The Information Technology and Innovation Foundation estimates that in 2019, the effective net global carbon price was minus $10.49 US per tonne, despite decades of advocacy. The playing field is not even level, let alone tilted towards clean energy.
Importantly, this is not for lack of trying, with climate leaders such as Sweden enacting carbon pricing in the mid-1990s; a major failed effort in the US; and advocacy by the EU, the World Bank, the International Monetary Fund, Nobel Prize–winning economists, and others.
The reason carbon pricing has failed so far is that it faces severe political barriers. In particular:
For carbon prices alone to drive deep decarbonisation, the global average carbon price would have to rise to above $100 US per tonne — about a 50 times increase from the current global average — a price that has only been achieved in a handful of small jurisdictions for a subset of sectors.
For example, to meet its 2030 climate targets, Canada would need to impose an economy-wide carbon tax that rises to $160 US per tonne by 2030. This would increase gasoline prices by 40 cents per litre. For context, the gillet jaunes protests in France started due to a plan to impose a carbon tax that would increase the price of gasoline by roughly . Although some countries, such as Sweden, have managed to impose high carbon taxes, this does suggest that the risk of political backlash is significant.
It is difficult to see how carbon pricing can drive deep decarbonisation, given the stark difference between current ambition levels and price levels needed to become an effective climate policy. To meet the Paris Agreement goal of limiting warming to 1.5°C, the typical rich country will have to reduce emissions by 11% per year. However, most current carbon pricing schemes only reduce emissions by 0–2% per year. Even the Swedish carbon tax — by far the most ambitious carbon pricing scheme in the world, with a carbon price of $130 US per tonne of CO2 for much of the economy — only reduces emissions by 6% per year.
From the perspective of global emissions, non-pricing policies have ironically been much more cost effective than carbon pricing. Policies that were highly locally cost ineffective — such as early subsidies for renewables and electric cars (with implicitcarbon prices in the hundreds of dollars per tonne) — have been globally more cost effective than real-world carbon pricing policies. This is because they have been stringent enough to drive down the cost of low-carbon technology for the whole world.
Of course, carbon prices could be much higher in the future. However, the evidence suggests that carbon pricing is a tool that would be near-ideal in hypothetical worlds, but can only make a small contribution to deep decarbonisation in the real world.
One promising philanthropic strategy is to advocate for effective policy leadership: if one country can effectively reduce emissions at reasonable cost and without political backlash, that could be an example that other countries could follow.
Because each country’s emissions are small compared to global emissions and countries constantly copy policies from each other, it may be highly impactful to push for policies in countries in the hope that they will spread globally — even if emissions reductions in the initial country would be low.
For such policy leadership to be effective, two conditions must be met:
Obviously, it is crucial for policies to be effective: some policies have diffused widely without being effective at reducing emissions. For example, there are now 65 carbon pricing schemes around the world, up from only three from Nordic countries in 2000. However, as discussed above, most of those policies are at such low price and ambition levels that they will have minimal effects on emissions.
There are already climate success stories that could serve as inspiration for the rest of the world. For example:
Overall, policy leadership is a promising approach, provided we can identify tractable solutions that are not already supported by other philanthropists.
In 2015, $330 billion US was spent on subsidising the fossil fuel industry. These subsidies are widely agreed to be economically inefficient and harmful to the environment, and removing them has been proposed as one of the most cost-effective ways to reduce emissions.
However, removing fossil fuel subsidies has limited tractability. The geographical split of fossil fuel subsidies breaks down as follows:
Removing all fossil fuel subsidies would require advocacy campaigns across many culturally and politically distinct jurisdictions — especially oil- and gas-exporting regions in the Middle East, Russia, and Latin America. Influencing fossil fuel subsidy policy in any one of these regions seems extremely difficult, and each campaign would require a highly distinct strategy.
Moreover, influencing policy in one of these regions would produce limited technology spillovers for the rest of the world, because it would be theoretically equivalent to imposing a modest carbon tax. Even under an expansive definition of fossil fuel subsidies, removing them entirely would only reduce emissions from fossil fuels and industry by 3–9%.
Some analysts argue that forest conservation is an especially low-cost way to reduce CO2, with costs below $10 US per tonne of CO2 avoided.
However, these estimates fail to account for several practical barriers to effective forest protection.
Defining reference levels for progress. It is very difficult to know how to judge the success of a deforestation effort, because it’s often unclear what the ‘reference level’ (the level of forestation there would have otherwise been) should be. Indeed, countries have incentives to pick reference levels that make their forestry progress seem more impressive than it really is. There is evidence that Brazil did this when choosing a reference level for its UN forestry submission. In fact, Brazil submitted separate reference levels for the Amazon Fund and the United Nations Framework Convention on Climate Change: each of which was very different, so yielded very different estimates of success at preventing deforestation.
Long-term sustainability. Optimistic estimates of the cost effectiveness of forestry protection often don’t include the long-term costs of preventing deforestation. If an area of forest is protected for only five years and then cut down, this makes no difference to cumulative emissions — which is what determines peak warming. A study of the costs to reduce deforestation in Guyana found that:
In general, ensuring that an area of forest is protected for decades is a costly and uncertain endeavour.
Inadequate protection measures. Monitoring and verification is inadequate to ensure true forestry protection. Even in countries with high capacity, monitoring, reporting, and verification remains imperfect. As of 2018, Brazil did not have a definition of forest degradation, and so degradation was excluded from reference levels for the Amazon and Cerrado biomes — despite being a major source of emissions in those biomes. Studiesacross many different countries show huge discrepancies between countries’ own estimates of forest loss against more reliable methods.
Leakage. One ongoing concern about measuring forestry protection is leakage — that efforts to prevent deforestation merely displace deforestation somewhere else (within or across national borders). Given the other three problems with forestry protection, there will always be a risk of leakage, due to imperfect compliance within or between countries.
Accounting for all of these factors, the true cost of CO2 abatement by protecting forests is much higher than commonly estimated, and it is doubtful that many claimed emissions reductions have any impact. In short:
Of course, the ineffectiveness of current forest conservation is not itself sufficient evidence that there could not be effective philanthropy targeted at improving the international forest conservation regime (REDD+), or bilateral or national efforts to protect rainforests. While this cannot be ruled out, we think the practical problems are sufficiently difficult that protecting forests is unlikely to ever be highly cost effective.
There is still significant uncertainty about several aspects of climate change, so the social returns to climate science research could be very high. Some scholars have proposed that it would be especially valuable to improve our knowledge of both:
Let’s first consider research into ‘climate sensitivity’ — the measure of how much average global temperatures change after the concentration of CO2 doubles (compared to pre-industrial concentrations). There are several reasons to think that the value of information from research into climate sensitivity is low.
Firstly, the latest IPCC report significantly narrowed uncertainty about climate sensitivity. In 2013 the IPCC thought, with 90% confidence, climate sensitivity was between 1.5 and 6°C. In the 2021 report, they are 90% confident that it is between 2.5 and 5°C. Carbon Brief visually represents this narrowing in its coverage of the IPCC’s sixth report.
Importantly, this reduction in uncertainty was produced not by expensive climate models, but by formally applying Bayesian statistical techniques to all the existing evidence, including from climate models and the paleoclimate (as discussed by Carbon Brief). It seems unlikely that this sort of statistical research is underfunded, given that it is a major input into the IPCC reports.
One potential way to reduce our uncertainty about climate change would be to improve climate models by funding international modelling efforts with supercomputers. But this would cost around .
It is clear that in a rational world, we would want to know a lot more about climate sensitivity. Does this mean that the value of information from researching climate sensitivity is high? Not necessarily. The value of information would only be high if additional information would make a difference to how people act: what governments, the private sector, and philanthropists would do with the information.
One possible outcome is that further research would produce good news and show that climate risk is lower than we previously thought. This would change the optimal level of mitigation. However, because mitigation is suboptimal today, this seems unlikely to mean that we should reduce the level of mitigation below its current level.
The other possible outcome of further research is bad news: that climate risk is higher than we previously thought. The benefits of this seem small. Political actors have thus far been largely insensitive to warnings from scientists about extreme warming. The 2013–14 IPCC reports plausibly suggested that on current policy, there was around a 1-in-10 chance of more than . This information has been in the public domain now for almost a decade, but seemingly had little effect on political willingness to mitigate emissions.
With the Paris Agreement of 2015 and the goal to limit warming to 2°C, almost all public discussions about ambition levels are focused on how compatible jurisdictional targets are with this global goal. It is difficult to see how increasing research on the probability of scenarios far outside of this goal would affect global policy responses.
This suggests that some estimates of the value of climate science research are too high. For example, Chris Hope estimates that the value of information from further research on the sensitivity of the climate to emissions is on the order of $10 trillion US. However, Hope’s model assumes that political actors will respond in a rational and benevolent way to learning more. The evidence we have so far suggests that this is very unlikely.
This does not mean that estimates of climate sensitivity have no impact on policy. For example, government agencies might use them to calculate updated estimates of the social cost of carbon. But it does mean that it is implausible that additional climate science research would have extremely high expected value.
For similar reasons, further research into the impacts of high levels of warming is also unlikely to be valuable. The available literature already focuses extensively on the impact of 4–5°C of warming. So, improvements could only feasibly be made by increasing research on more than 5°C of warming. But since the global policy community is focused on avoiding more than 2°C of warming, it is hard to see how this could have much effect on global political efforts.
Solar geoengineering (sometimes called solar radiation management) involves cooling the Earth by reflecting sunlight back to . The most researched proposed form of this is called stratospheric aerosol injection (SAI), which involves injecting particles such as sulphates into the stratosphere (the upper atmosphere), which would be distributed around the planet by stratospheric winds and would reflect sunlight back to space. Other forms of solar geoengineering, such as marine cloud brightening, have only regional effects and so have different costs and benefits.
The evidence suggests that if SAI were deployed in a certain way and could be governed properly, it would reduce the damages from climate change, with trivial technical implementation costs (less than $3 billion US per year).
The main problems with solar geoengineering centre on:
[Unilateral deployment scenarios] overstate the distribution of capabilities and thus the risk of unilateral action, because they focus too narrowly on financial cost as the determinant of capability and neglect other, non-financial, requirements and constraints.
Further, an SAI programme large enough to make a nontrivial sustained impact on the climate would be hard to conceal and vulnerable to military attack:
[U]nilaterally achieving a climate alteration that matters would require not just the money, technological capability, and delivery assets, but also the command of territory, global stature, and ability to deploy and project force necessary to protect a continuing operation against opposition from other states, including deterring their threats of stopping it through military action.
Indeed, Harvard researcher Joshua Horton has persuasively argued that SAI is actually characterised by a logic of multilateralism. Given the incentives created by conventional military threats, the decision to deploy SAI would not be taken lightly. SAI would affect the weather in all regions, so it is unlikely that, for example, Brazil would start a highly visible geoengineering scheme that would affect the weather in the United States, without the consent of the United States. It is difficult to see how SAI could ever be deployed unless it had the agreement of all major powers.
Moreover, to be useful in reducing the costs of climate change, SAI would have to be deployed for many decades. Thus, there would have to be multi-decadal agreement at least among all major powers on a global weather modification scheme. This would include democracies, in which political leaders change every five years or so.
One concern with such global coordination is that some countries will experience adverse weather events while SAI is deployed, and will attribute these events to it, even if it is not responsible. The climate system is highly unpredictable and chaotic, so at best we will be able to attain probabilistic causal attribution of adverse events to SAI. Climate models are poor at predicting regional warming and precipitation, but trust in the probabilistic models would have to be high for severe adverse weather events not to be blamed on solar geoengineering by the public or political leaders. If countries do blame severe adverse weather events on SAI, the response is likely to be angry and irrational, likely including suspicion about the motives of the controlling coalition.
In light of this, securing the multi-decadal agreement required looks extremely difficult, and possibly insurmountable. Moreover, states will foresee this, which would be a disincentive to deploy solar geoengineering in the first place.
This suggests that SAI will be deployed only once climate change starts to impose quite severe costs (at least on all major powers), such that there is significant within-country demand for solar geoengineering. This is not the case today, and it seems at the very least .
A common worry about SAI research is that it risks obstructing emissions mitigation efforts, which almost all researchers believe to be a top priority. This claim is very difficult to assess, and researchers are divided on how plausible it is. But since it seems unlikely that SAI will be deployed in the near future, there is little to be lost by delaying SAI research efforts until the prospects for deployment seem more realistic. In the meantime, we could focus on emissions reduction. Indeed, since momentum on emissions reduction is building, now seems the wrong time to explore a highly risky backup option.
The costs and benefits of research into regional solar geoengineering are different because the effects would be restricted to particular countries or regions, so the governance barriers are much lower. Indeed, many states already practice regional geoengineering. China’s weather modification scheme employs tens of thousands of people, and is mainly focused on influencing rain patterns through cloud seeding. Since the governance barriers are lower, the case for research into regional solar geoengineering may be stronger.
When climate activists and philanthropists examined why Obama’s landmark climate bill, the American Clean Energy and Security Act, had failed, it became part of the conventional wisdom that climate philanthropy and activism were too much of an inside game — that a popular climate movement had been lacking, and this was an important contributor to failure.
The years since have seen a massive uptick in philanthropic support for public engagement to build the case for climate action to be around $250 million US per year) and have also seen the rise of the progressive and youth movements, such as the Fridays For Future movement in Continental Europe, Extinction Rebellion in the UK, and Sunrise in the US. The pressure to act on climate change has reached unprecedented levels in most Western countries.
This prominence of climate attention, climate movements, and large environmental nongovernmental organisations — as well as the focus on this lever by major philanthropists — make it quite unlikely that there are high-impact unfunded opportunities by supporting mainstream organisations in this space. Once an organisation building public support for climate action has reached such prominence that it is covered in the national news, it is unlikely that it is funding-constrained.
This does not mean that nothing in this space is worth funding, but rather that one has to look deeply into the space and consider early-stage or behind-the-scenes organisations that could be funding-constrained. There is currently an effective altruism research effort called the Social Change Lab, which is studying whether early-stage social movements in climate and other causes could be worth funding.
While these movements’ work in increasing attention on climate change has been critical for bold climate action, some of the messaging and pressures applied may have some downsides. These include:
We think the best place to donate to mitigate the effects of climate change is the Founders Pledge Climate Fund. The strategy of the Fund is to find and fund the highest-impact opportunities based on the considerations laid out here, focusing on avoiding the maximum amount of climate damage. It does this by analysing neglectedness and theories of change, and making time-sensitive as well as multi-year grants. Individual donors contribute to the Fund, which is then distributed by experts in the field to the most effective projects. (Learn more here.)
So far the Fund supports four different theories of change:
The charities that they currently recommend include:
Note that it is likely significantly more effective to donate through the Fund than to the individual charities. This helps ensure each organisation receives the funding they need, and avoids situations where some organisations are unable to carry out projects due to a lack of funding, while others receive more funding than was needed.
Please help us improve our work — let us know what you thought of this page and suggest improvements using our content feedback form.