[updated] Global development interventions are generally more effective than Climate change interventions

by HaukeHillebrandt 2mo2nd Oct 201915 min read31 comments

97


Previously titled “Climate change interventions are generally more effective than global development interventions”.  Because of an error the conclusions have significantly changed. [old version]. I have extended the analysis and now provide a more detailed spreadsheet model below.

Word count: ~1800

Reading time: ~9 mins

Keywords: Climate change, climate policy, global development, global health, cause prioritization, prioritization research, comparing diverse benefits

Epistemic status: Uncertain and speculative. I don’t excessively hedge my claims throughout for clarity’s sake (#‘better wrong than vague’, #“say wrong things”, #"correct me if I'm wrong", #"All models are wrong, but some are useful").

Acknowledgments: Thanks to John Halstead, Danny Bressler, Sahil Shah, and members on the Effective Altruism forum, especially AGB, for helpful comments. Any errors are mine.

Comparative cost-effectiveness of climate change and global development

Summary

Does climate change deserve more attention within the effective altruism community?[1]

What is more effective: climate change interventions to avert emissions per tonne or single recipient global development interventions such as cash transfers?

Are targeted interventions to more fundamentally transform the lives of the poorest more effective than supplying broad global public goods such as a stable climate with comparatively small benefits to everyone on the planet?

To answer these questions, the following question is crucial:

“What value should we use for the social cost of carbon to adequately reflect the greater marginal utility of consumption for low-income people?”[2]

Here, I tried to answer this question. Surprisingly, I find that global development interventions are generally more effective than climate change interventions.

My spreadsheet model below shows that climate change interventions are only more effective than global development interventions, if and only if:

  • Money is worth only 100 times as much to the global poor than people in high-income countries (i.e. if utility to consumption is logarithmic) and not more
  • AND climate change interventions are very effective (less than $1 per tonne of carbon averted) AND/OR
  • under quite pessimistic assumptions about climate change (if the social cost of carbon is higher than $1000 per tonne of carbon).

Key claims

I base the above conclusion on the following three empirical claims:

1. New research on the income-adjusted country-level social cost of carbon allows us to compare global development interventions to climate change interventions.

The new research is the first to use climate model projections, empirical climate-driven economic damage estimations, and also socio-economic projections which take into account greater marginal utility of consumption for every country individually.[3]

In other words, this takes into account “your dollar does (>)100x or more good if you give to the poorest rather than people in high-income countries”). More on income weighting in Appendix 2.

Other more canonical Integrated Assessment Models (IAM) such as DICE have only have one value for the whole world, and, while the RICE IAM has 12 regions,[4] this still understates the heterogeneous geography of climate damage.

The new research first estimated the social cost of carbon for every country in the world. Then, the authors summed up all the country-level costs of carbon to arrive at the global cost of carbon: US$417 per tonne of CO2 (66% CI: US$177–805, data explorer).

Dividing this value by how much more worth a dollar is for the poorest vs. people in high-income countries, e.g. 100x, allows us to directly compare climate change to cash-transfers to the poorest and other global development interventions. For instance, dividin by 100 results in ~$4.17 per tonne of CO2 (66% CI: $1.77–8.05)).[5]

While this global estimate also includes costs to more privileged people…

  • ...in advanced economies (e.g. US, EU),
  • …in the future who are richer due to economic growth,
  • …in countries that are not as affected by climate change due to geography (e.g. some rich cold countries might actually benefit a little)

… these do not weigh as heavily in these calculations because the modelling adjusts for decreasing marginal utility of income.

The key point here is that the new model accounts more thoroughly for geographical heterogeneity and diminishing returns to consumption. We now need no longer worry as much that the social cost of carbon estimates obscure that climate change will be much worse for the poorest people in geographies more affected by climate change (who we could send unconditional cash transfers to).

The new paper’s social cost of carbon figure is controversial and has been criticized for being too high for various methodological reasons.[6] For instance, one very critical new paper also now estimates the social cost of carbon on a country-level, suggesting that the global social cost of carbon is only $24 (and, using various sensitivity analyses, values ranging from $3.38/tCO2e to $21,889/tCO2e).[7]

To account for the new paper overestimating or underestimating the social cost of carbon, below, we use sensitivity analysis to show how our model responds to over- or underestimating the true social cost of carbon by 10x.

2. Only some climate change interventions avert a tonne of CO2 for less than the global income-adjusted cost of carbon.

Only if a climate change intervention has a cost-effectiveness of social cost of carbon / income adjustment / X per tonne of CO2 averted, then it is X times as effective as cash-transfers.

So generally, climate change interventions create x more utility than cash-transfers, where


For instance, a climate change intervention with an effectiveness of $1 per tonne of CO2 averted would be ~4.17x more effective than cash-transfers if the social cost of carbon / 100 is $4.17. This is just the basic model. We complicate this analysis in the spreadsheet below with more parameters. For instance, some global development interventions are 17.5 more effective than cash-transfers.

3. Scalable climate change interventions are not generally as cost-effective compared to global development interventions.

Cash-transfers can absorb more funding with consistently high cost-effectiveness than any other intervention.

So even if many other high-risk, high-reward projects have a higher benefit-cost ratio than cash-transfers in expectation, they usually have smaller funding gaps and one needs to do more research to find them.

But climate change can also absorb large amounts of funding at scale with consistently high cost-effectiveness and only slowly diminishing returns.

Examples of interventions with cost-effective scalable interventions are:

  1. Ocean alkalinity might be a way to absorb large amounts of carbon up to 100 billion tonnes / year) for as little as $10 per tonne of CO₂ averted.[8],[9]
  2. Tropical reforestation between 2020–2050 could be increased by 5.7 billion tonnes (5.6%) at a carbon price of $20 per tonne of CO2 averted or by 15.1 billion tonnes (14.8%) at $50 per tonne of CO2 averted.[10]
  3. Avoided deforestation can prevent 55.1 billion tonnes at $20 per tonne of CO2 averted or 108.3 billion tonnes at $50 per tonne of CO2 averted.[11]
  4. The levelized costs of capturing CO2 from the atmosphere are projected to be $94-232 per tonne CO2[12] and could decrease to $35 by 2050.[13]

Multiplying the number of tonnes avertable by the (average) cost per tonne averted equals the funding gap. For some of these interventions this is in the trillions.[14] To get a rough estimate of the overall funding gap of climate change, we can multiply global emissions—roughly 37 billion tonnes per year—and assume an average cost-effectiveness of $50 per tonne. This suggests a funding gap of $1.85 trillion/year.

Yet, none of these scalable interventions can avert a tonne of carbon at scale for less than $4.17.

The figure below shows the cost-effectiveness of some more interventions and their overall abatement potential (though we should be skeptical of “free lunch” interventions with negative costs).[15]

Implications for cause prioritization

If none of these three claims above can be falsified, then it follows that, as a rule, we should consider prioritizing global development over climate change interventions. In other words, averting carbon should not replace unconditional cash-transfers to the poorest as the new ‘benchmark’ or replace global development as the new ‘mainstream EA flagship cause’.

However, under some pessimistic modelling assumptions, some select climate change interventions might be more effective than global development interventions and should be prioritized. Read on for our updated model on this.

When are climate change interventions more effective than global development interventions?

Consider the following spreadsheet (Google sheet). A more detailed description of the parameters, assumptions and scenarios are in Appendix 1.

Optimistic scenario

Assumptions:

  • Social cost of carbon is only $17.7
  • an η of 2 (i.e. cash-transfers to the poorest are worth 13,610x as much as to people in rich countries)
  • a high cost of $232 per tonne of carbon averted at scale,
  • and global development interventions being 17.5x as effective as cash.

This results in climate change interventions being only 0.00003% as effective as cash-transfers.

Realistic scenario

Assumptions:

  • Social cost of carbon is $417
  • η of 1.5 (i.e. cash-transfers to the poorest are worth 1,260x as much as to people in rich countries)
  • $10 per tonne of carbon averted at scale (e.g. deforestation prevention)
  • and global development interventions being  7.95x as effective as cash (median Givewell charity effectiveness)

This results in climate change interventions being only 0.42% as effective as cash-transfers.

Pessimistic scenario

Assumptions:

  • Social cost of carbon is high $8,050
  • η of 1 (i.e. cash-transfers to the poorest are worth 120x as much as to people in rich countries)
  • $0.02 per tonne of carbon averted at scale (e.g. lobbying for deforestation prevention)
  • and global development interventions are cash-transfers, which are give 83 cents of every dollar to the poorest

This results in climate change interventions being 4,041x more effective than cash-transfers.

Conclusion

Global development interventions seem generally more effective than climate change interventions. However, under pessimistic modelling assumptions, select climate change interventions might be more effective than global development interventions.

Only if utility to consumption is logarithmic (i.e. only if a dollar going to the poorest is not more than 100x as much as going to people in rich countries) AND a given climate change intervention is very effective (less than $1 per tonne of carbon averted) OR under quite pessimistic assumptions about climate change (if the social cost of carbon is higher than $1000 per tonne of carbon), then climate change interventions are more effective than global development interventions.

So for those wanting to maximize expected utility, one should support climate change interventions only if they are very cost-effective (i.e. lower than $1/tCo2e averted).

The results of the model are also very sensitive to the income adjustment parameter η – if it’s just 1, i.e. returns to consumption are logarithmic, and money to the poorest is only 100x as good at going to the poorest, and the social cost of carbon is just in the hundreds of dollars, then some effective, but not super scalable interventions such as deforestation prevention on the order of $1 per tonne of carbon averted can beat some global development interventions.

Because the confidence intervals between climate and development are wide and overlapping, the value of information of reducing uncertainty is high. For instance, the value of better information on the transient climate response has been estimated to be $10 trillion.[16] In other words, if new research would show that the social cost of carbon is actually much higher, then this might lead to more optimal allocation of resources.

Appendix 1: Additional info on model parameters

Social cost of carbon

General note on social cost of carbon: Generally, climate modelling is much more uncertain than global development interventions (which can be studied with RCTs) and the effects of climate change are in the future (see Appendix 3). Altruists with high risk / uncertainty aversion and/or high discount rates might want to not support climate change interventions for that reason.

Yet, the estimate from above social cost of carbon modelling uses sensitivity analyses to account for uncertainty and uses discounting and so the estimates are at least somewhat robust to different specifications.

Also, some commenters note that the quantification of climate modelling is essentially useless (also see Appendix 3). However, one study estimated a lower bound of the global social cost of carbon at US$125 and argues that:

“Quantifying the true SCC value is complicated because of various difficult-to-quantify damage cost categories and the interaction of discounting, uncertainty, large damages and risk aversion [...] The best that can be offered is a lower bound based comes from a conservative meta-estimate that aggregates studies using high and low discount rates, it does not account for various climate change damages owing to a lack of reliable information, and it does not consider a minimax regret argument addressing damages associated with extreme climate change.”

Also, as an aside, outside of prioritization, for optimal policy the social cost of carbon should be:

  1. Set to the marginal abatement cost, which can be optimal and easier to estimate.[17] or
  2. Set to err on the side of overestimating externalities[18] (while reducing other non-Pigovian taxes).

Optimistic assumption: The new study's estimate is 10x higher than more canonical estimates such as the EPA’s $42 per tonne, which is based on IAMs. This because it contentiously assumes impacts on GDP growth permanently alter a country’s GDP [19] using different damage functions, not because they are accounting for greater marginal utility of consumption to individuals with lower consumption levels.[20]

For that reason, in our optimistic scenario, we downward adjust by 10x on the paper’s lower 66% CI, so that the social cost of carbon is $17.8. Note that this is conservative in the sense of being much lower even than the estimated lower bound of the social cost of carbon.

Realistic assumption: In the realistic assumption, we use the study’s central estimate ($417), and assume that the model is correct.

Pessimistic assumption: This assumes that the study’s estimate is actually 10x too low and adjusts for this.

This is plausible because of contributors to social cost of carbon not fully captured by empirical, macroeconomic damage functions, and their likely impacts on the social cost of carbon (see Table S5 in the paper’s supplementary material and Table 1 in[21]). For instance:

  • Adjustment costs (short-term costs of adaptation)
  • Non-market damages (biodiversity loss, cultural losses, etc.)
  • Tipping points in the climate system (catastrophic climate events, hysteresis etc.)
  • High inertia effects of CO2 (ocean acidification, sea level rise)
  • General equilibrium effects (spillover, trade, etc.)
  • Macro-scale adaptation (long-term restructuring of economy)
  • Political instability and violent conflicts
  • Large migration flows
  • More extreme weather and natural disasters
  • Bresler finds that explicitly accounting for climate mortality costs triples the welfare costs of climate change.[22]
  • The highest social cost of carbon estimate in the literature is on the same order of magnitude ($1687[23]), and the highest figure amongst many in a recently published paper find that for 6 degrees of warming the cost will be (which has a substantial probability) is $21889 / per tonne) [24]

Income adjustment

The income adjustment takes into account that “your dollar does (>)100x or more good if you give to the poorest rather than people in high-income countries”). More on income weighting in Appendix 2. The optimistic case has an η value of 2, the realistic of 1.5, and the pessimistic of 1 (Source). This corresponds to 1 dollar being worth 120-13,610 more when it goes to the poorest people on the planet (e.g. via unconditional cash-transfers) than someone on Median US income (120 might be an underestimate according some of my calculations and it might be 250). For higher values going towards 2, this can dominate the analysis.[25]

Cost per tonne of CO2 averted

Optimistic case: The levelized costs of capturing CO2 from the atmosphere are projected to be $94-232 per tonne CO2[26] and could decrease to $35 by 2050.[27] We use the upper bound ($232) for the optimistic case.

Realistic case: For the middle case, we use ocean alkalinity which might be a way to absorb large amounts of carbon up to 100 billion tonnes / year) for as little as $10 per tonne of CO₂ averted.[28],[29] This is also roughly in line with the cost per tonne of CO2 averted through new clean energy generation over coal and other interventions, wind production tax credit in the United States, which have estimated carbon abatement costs ranging from $2-260 (Carbon capture and storage is on a similar order of magnitude, also see figure below for other interventions).[30]

Optimistic case: Preventing deforestation can have a cost-effectiveness of $0.57 per averted ton of CO2 at scale.[31] Donations to “Coalition for Rainforest Nations” for advocacy on deforestation prevention has been estimated to avert a tonne of CO2e for $0.12, with a plausible range of $0.02 - $0.72.[32] We use the lower bound, $0.02, as the optimistic case. Note that this is perhaps comparing apples to oranges, by comparing advocacy to direct interventions and that a fairer comparison would high-risk, high-reward global development science or policy interventions  which have been suggested to be perhaps 100x more effective than cash.[33]

Global development interventions effectiveness

Optimistic case: Some global development interventions have been estimated to be 17.5x more effective than cash-transfers (e.g. deworming).[34] We use this as the optimistic case.

Realistic case: The median Givewell charity effectiveness vs. cash is 7.95.[35]

Pessimistic case: The proportion of total expenses that GiveDirectly has delivered directly to recipients is approximately 83%.[36]

Appendix 2: Income-weighting

A recent paper estimates the country-level social cost of carbon, using not only climate, but also socio-economic projections.[37] For the marginal utility substitution, they use a μ-value of 1.5 as a central value.

What concretely does this mean?

All else being equal, money going to poorer countries or people is better than money going to richer countries or people. Weyl suggests that assuming logarithmic utility giving 1 dollar to an extremely poor person is like giving 66 dollars to an American[38]. (“That is, if marginal utility is declining in levels of income, say utility is the natural log of consumption, then the marginal utility is 1/consumption. This implies a dollar’s worth of consumption in utility terms of a person at the global poverty line is worth 64 times as much as a dollar to person in the highest decile of consumption in the USA (63.6=(1/(1.9*365))/(1/44,152) so transferring income from a rich person in the USA to a globally poor person produces, in and of itself, massively higher total global utility (even if not Pareto improving).”[39]).

Weyl suggests that logarithmic utility is canonical in economics and supported by a wide range of data, “including recent happiness studies (Stevenson and Wolfers, 2008) and labour supply decisions (Chetty, 2006)”.  This is also in line with work that finds a correlation between log income and happiness [40]:

The law of logarithmic utility can be found in other areas such as research funding as well [41].

The general form of modelling utility consumption relationships using isoelastic utility function is: [42]:

Ord [43] explains this function as follows:

“This equation has one free parameter, known as η (‘eta’, which sounds ‘e’ for ‘elasticity’), which represents how steeply returns to consumption diminish.  η must be between 0 and ∞, and can be estimated empirically.

The equation, for utility (u) at a given consumption level (c), and elasticity (η) is:

From this it follows that for η = 0 utility is linear in consumption, for η = ½ utility is the square root of consumption, and for η = 1 utility is logarithmic in consumption. Values of η above 1 correspond to utility having a finite upper bound, which is approached hyperbolically as consumption increases.

However, the main use of the equation is to just compare the slope of the curve at one consumption level to the slope at another consumption level. For example the ratio of the slope at $1,000 per annum to the slope at $10,000 per annum shows us the relative value of giving an extra dollar to someone with annual consumption $1,000 versus to someone with $10,000. When performing this calculation, the equation is very simple:

Giving a dollar to someone with k times as much consumption is worth only:

(1/k)^2

times as much.

There have been many attempts to measure η, and it is typically found to be between about 1 and 2. If η equals 1, then we have logarithmic utility of consumption and we have the very simple rule that a dollar is worth 1/k times as much if you are k times richer (and that doubling someone’s income is worth the same amount no matter where they start). If η equals 2, then we have to raise this to the power of 2, so being 10 times richer would mean a dollar is worth just 1/100th as much (and doubling your income is worth much less the higher your starting income). The truth is probably in between these limits.”

Appendix 3: Uncertainty around climate change modelling

Integrated assessment models have been heavily criticised. Consider the following quote by MIT Economics Professor Robert S. Pindyck from his paper “The Use and Misuse of Models for Climate Policy”:[44]

"In a recent article, I argued that integrated assessment models (IAMs) “have crucial flaws that make them close to useless as tools for policy analysis.” In fact, I would argue that calling these models “close to useless” is generous: IAM-based analyses of climate policy create a perception of knowledge and precision that is illusory, and can fool policy-makers into thinking that the forecasts the models generate have some kind of scientific legitimacy. IAMs can be misleading – and are inappropriate – as guides for policy, and yet they have been used by the government to estimate the social cost of carbon (SCC) and evaluate tax and abatement policies. What are the crucial flaws that make IAMs so unsuitable for policy analysis? They are discussed in detail in Pindyck (2013b), but the most important ones can be briefly summarized as follows: 1. Certain inputs – functional forms and parameter values – are arbitrary, but have huge effects on the results the models produce. An example is the discount rate. There is no consensus among economists as to the “correct” discount rate, but different rates will yield wildly different estimates of the SCC and the optimal amount of abatement that any IAM generates. For example, these differences in inputs largely explain why the IAMbased analyses of Nordhaus (2008) and Stern (2007) come to such strikingly different conclusions regarding optimal abatement. Because the modeler has so much freedom in choosing functional forms, parameter values, and other inputs, the model can be used to obtain almost any result one desires, and thereby legitimize what is essentially a subjective opinion about climate policy. 2. We know very little about climate sensitivity, i.e., the temperature increase that would eventually result from a doubling of the atmospheric CO2 concentration, but this is a key input to any IAM. The problem is that the physical mechanisms that determine climate sensitivity involve crucial feedback loops, and the parameter values that determine the strength (and even the sign) of those feedback loops are largely unknown, and are likely to remain unknown for the foreseeable future. 3. One of the most important parts of an IAM is the damage function, i.e., the relationship between an increase in temperature and GDP (or the growth rate of GDP). When assessing climate sensitivity, we can at least draw on the underlying physical science and argue coherently about the relevant probability distributions. But when it comes to the damage function, we know virtually nothing – there is no theory and no data that we can draw from. 4. IAMs can tell us nothing about the likelihood or possible impact of a catastrophic climate outcome, e.g., a temperature increase above 5°C that has a very large impact on GDP. And yet it is the possibility of a climate catastrophe that is (or should be) the main driving force behind a stringent abatement policy."

Yet, in a later paper Pindyck estimates the social costs of carbon through expert surveys to be in the hundreds of dollars range. This is in line with IAMs and leads me to believe that

The Social Cost of Carbon Revisited:[45]

“An estimate of the social cost of carbon (SCC) is crucial to climate policy. But how should we estimate the SCC? A common approach uses an integrated assessment model (IAM) to simulate time paths for the atmospheric CO2 concentration, its impact on temperature, and resulting reductions in GDP. I have argued that IAMs have deficiencies that make them poorly suited for this job, but what is the alternative? I present an approach to estimating an average SCC, which I argue can be a useful guide for policy. I rely on a survey of experts to elicit opinions regarding (1) probabilities of alternative economic outcomes of climate change, but not the causes of those outcomes; and (2) the reduction in emissions required to avert an extreme outcome, i.e., a large climate-induced reduction in GDP. The average SCC is the ratio of the present value of lost GDP from an extreme outcome to the total emission reduction needed to avert that outcome. I discuss the survey instrument, explain how experts were identified, and present results. I obtain SCC estimates of $200/mt or higher, but the variation across experts is large. Trimming outliers and focusing on experts who expressed a high degree of confidence in their answers yields lower SCCs, $80 to $100/mt, but still well above the IAM-based estimates used by the U.S. government.”[46]

References

[1] "Does climate change deserve more attention within EA? - Effective ...." 16 Apr. 2019, https://forum.effectivealtruism.org/posts/BwDAN9pGbmCYZGbgf/does-climate-change-deserve-more-attention-within-ea. Accessed 1 Oct. 2019.

[2] "Research Fellow - Open Philanthropy Project - Google Hire." http://web.archive.org/web/20190828072238/https://hire.withgoogle.com/public/jobs/openphilanthropyorg/view/P_AAAAAAFAACuGJXzLj5l5Sp?trackingTag=googleSearch.  

[3] ”Country-level social cost of carbon” - Nature Climate Change

[4] "Revisiting the social cost of carbon | PNAS." https://www.pnas.org/content/114/7/1518. Accessed 10 Sep. 2019.

[5] "The value of money going to different groups - Centre for ...." 2 May. 2017, https://www.centreforeffectivealtruism.org/blog/the-value-of-money-going-to-different-groups/. Accessed 13 Sep. 2019.

[6] "Valuing climate damages at the country level | Nature Climate ...." 24 Sep. 2018, https://www.nature.com/articles/s41558-018-0285-8. Accessed 13 Sep. 2019.

[7] "A social cost of carbon for (almost) every country - ScienceDirect." https://www.sciencedirect.com/science/article/pii/S014098831930218X. Accessed 26 Sep. 2019.

[8] "The Emissions Gap Report 2017 - UNEP Document Repository Home." https://wedocs.unep.org/bitstream/handle/20.500.11822/22070/EGR_2017.pdf. Accessed 9 Apr. 2018.

[9] "Negative emissions„Part 2: Costs, potentials and side ... - IOPscience." 22 May. 2018, https://iopscience.iop.org/article/10.1088/1748-9326/aabf9f/pdf. Accessed 29 Jul. 2019.

[10] "Potential for low-cost carbon dioxide removal through tropical ... - Nature." 27 May. 2019, https://www.nature.com/articles/s41558-019-0485-x. Accessed 29 Jul. 2019.

[11] "Potential for low-cost carbon dioxide removal through tropical ... - Nature." 27 May. 2019, https://www.nature.com/articles/s41558-019-0485-x. Accessed 29 Jul. 2019.

[12] "A Process for Capturing CO2 from the Atmosphere - ScienceDirect." 15 Aug. 2018, https://www.sciencedirect.com/science/article/pii/S2542435118302253. Accessed 31 Jul. 2019.

[13] "Techno-economic assessment of CO2 direct air ... - ScienceDirect.com." 1 Jul. 2019, https://www.sciencedirect.com/science/article/pii/S0959652619307772. Accessed 1 Aug. 2019.

[14] "Potential for low-cost carbon dioxide removal through tropical ... - Nature." 27 May. 2019, https://www.nature.com/articles/s41558-019-0485-x. Accessed 29 Jul. 2019.

[15] "The Cost of Reducing Greenhouse Gas Emissions - Harvard ...." 2 Aug. 2018, https://scholar.harvard.edu/files/stock/files/gillingham_stock_cost_080218_posted.pdf. Accessed 24 Sep. 2019.

[16] "Value of better information about TCR | Philosophical Transactions of ...." 5 Oct. 2015, http://rsta.royalsocietypublishing.org/content/373/2054/20140429. Accessed 9 Apr. 2018.

[17] Yew-Kwang Ng - “Optimal Environmental Charges/Taxes: Easy to Estimate and Surplus-yielding”

[18] Blogpost on policy by Paul Christiano

[19] "Valuing climate damages at the country level | Nature Climate ...." 24 Sep. 2018, https://www.nature.com/articles/s41558-018-0285-8. Accessed 13 Sep. 2019.

[20] https://forum.effectivealtruism.org/posts/GEM7iJnLeMkTMRAaf/climate-change-interventions-are-generally-more-effective?commentId=oE4jiWzi63HZxZvvw

[21] "A lower bound to the social cost of CO2 emissions | Nature ...." 26 Mar. 2014, https://www.nature.com/articles/nclimate2135. Accessed 13 Sep. 2019.

[22] "2nd Oxford Workshop on Global Priorities Research - 1-2 July ...." https://globalprioritiesinstitute.org/2nd-oxford-workshop-on-global-priorities-research/. Accessed 13 Sep. 2019.

[23] "The Economic Effects of Climate Change - American ...." https://pubs.aeaweb.org/doi/pdf/10.1257/jep.23.2.29. Accessed 13 Sep. 2019.

[24] "A social cost of carbon for (almost) every country - ScienceDirect." https://www.sciencedirect.com/science/article/pii/S014098831930218X. Accessed 25 Sep. 2019.

[25] "The value of money going to different groups - Centre for ...." 2 May. 2017, https://www.centreforeffectivealtruism.org/blog/the-value-of-money-going-to-different-groups/. Accessed 25 Sep. 2019.

[26] "A Process for Capturing CO2 from the Atmosphere - ScienceDirect." 15 Aug. 2018, https://www.sciencedirect.com/science/article/pii/S2542435118302253. Accessed 31 Jul. 2019.

[27] "Techno-economic assessment of CO2 direct air ... - ScienceDirect.com." 1 Jul. 2019, https://www.sciencedirect.com/science/article/pii/S0959652619307772. Accessed 1 Aug. 2019.

[28] "The Emissions Gap Report 2017 - UNEP Document Repository Home." https://wedocs.unep.org/bitstream/handle/20.500.11822/22070/EGR_2017.pdf. Accessed 9 Apr. 2018.

[29] "Negative emissions„Part 2: Costs, potentials and side ... - IOPscience." 22 May. 2018, https://iopscience.iop.org/article/10.1088/1748-9326/aabf9f/pdf. Accessed 29 Jul. 2019.

[30] "The Cost of Reducing Greenhouse Gas Emissions - Harvard ...." 2 Aug. 2018, https://scholar.harvard.edu/files/stock/files/gillingham_stock_cost_080218_posted.pdf. Accessed 13 Sep. 2019.

[31] "Cash for Carbon: A Randomized Controlled Trial of ... - NBER." http://www.nber.org/papers/w22378.pdf. Accessed 23 Sep. 2019.

[32] https://docs.google.com/spreadsheets/d/12lwvxlWLjwuSuXiciFvnBF2bkfcCkrusdqqT37_QWac/edit#gid=1267972809&range=E35

[33] "GiveWell's Top Charities Are (Increasingly) Hard to Beat ...." 9 Jul. 2019, https://www.openphilanthropy.org/blog/givewells-top-charities-are-increasingly-hard-beat. Accessed 2 Oct. 2019.

[34] Givewell CBA

[35] Givewell CBA

[36] "GiveDirectly | GiveWell." https://www.givewell.org/charities/give-directly. Accessed 23 Sep. 2019.

[37] ”Country-level social cost of carbon” - Nature Climate Change

[38] "The Openness‐equality Trade‐off in Global ... - Wiley Online Library." https://onlinelibrary.wiley.com/doi/abs/10.1111/ecoj.12469. Accessed 30 Jul. 2018.

[39] "Alleviating Global Poverty: Labor Mobility, Direct Assistance, and ...." 20 Mar. 2018, https://www.cgdev.org/publication/alleviating-global-poverty-labor-mobility-direct-assistance-and-economic-growth. Accessed 1 Nov. 2018.

[40] "Subjective Well-Being, Income, Economic Development and Growth." http://www.nber.org/papers/w16441. Accessed 30 Jul. 2018.

[41] "The law of logarithmic returns - Future of Humanity Institute." 20 Oct. 2014, https://www.fhi.ox.ac.uk/law-of-logarithmic-returns/. Accessed 30 Jul. 2018.

[42] "Isoelastic utility - Wikipedia." https://en.wikipedia.org/wiki/Isoelastic_utility. Accessed 30 Jul. 2018.

[43] "The value of money going to different groups - Centre for Effective ...." 2 May. 2017, https://www.centreforeffectivealtruism.org/blog/the-value-of-money-going-to-different-groups/. Accessed 30 Jul. 2018.

[44] "The Use and Misuse of Models for Climate Policy." http://www.nber.org/papers/w21097. Accessed 24 Jul. 2019.

[45] "The Social Cost of Carbon Revisited." http://www.nber.org/papers/w22807. Accessed 24 Jul. 2019.

[46] "The Social Cost of Carbon Revisited." http://www.nber.org/papers/w22807. Accessed 24 Jul. 2019.

97