nature climate change

Article [Link]

Philosophers and economists agree on

climate policy paths but for different reasons

Received: 5 October 2022 Frikk Nesje , Moritz A. Drupp

1,2
, Mark C. Freeman
2,3 4
& Ben Groom 5,6

Accepted: 24 April 2023

Published online: xx xx xxxx The estimated value to society from climate change mitigation is highly
sensitive to the long-term social discount rate. Governmental discounting
Check for updates
guidance has almost exclusively been influenced by economists, although
it is not clear that they possess any special expertise on intergenerational
ethics. Here, by contrast, we report the views of philosophers, who are the
most trained in ethical matters. We show that, as a group, these experts
offer strong support for a real social discount rate of 2%, a value that is also
predominantly backed by economists. We find multidisciplinary support
for climate policy paths in line with the United Nations climate targets when
views on discounting determinants are applied within a recent update of
the DICE integrated assessment model. However, this apparent agreement
hides important differences in views on how the ethics o­f i­nt­ergenerational
welfare can be better incorporated into climate policy evaluation.

Appraisals of climate mitigation policies and projects are notoriously desirable9,10. This has not gone unnoticed, and the Biden administration
sensitive to the weight placed on future costs and benefits1,2. In cost– has been recommended to “seek broad input”, including from philoso-
benefit analysis using integrated assessment models (IAMs), this weight phers4 who do have specific competencies in such complex ethical
is determined through the social discount rate (SDR) and its determi- matters. While a number of philosophers have engaged with social
nants, with small changes to the SDR having notable policy implications. discounting11–19, a systematic account of the views of these disciplinary
The Trump administration’s recalculation of the social cost of carbon experts has previously been missing. We therefore undertook a com-
(SCC) showed that increasing the real SDR from 3% to 7% reduces the prehensive survey of philosophers with expertise on social discounting,
SCC by a factor of seven within its chosen model3. By contrast, New York focused on the appropriate intergenerational SDR, and compared the
State has raised its SCC from US$51 to US$121 per ton of CO2 on the basis results to those from a previous identical survey of expert economists20.
of a real SDR of 2% rather than 3%4, while recent interim estimates by We found that philosophers and economists exhibit strong agreement
the US Environmental Protection Agency (EPA)5 calculate a central SCC on a long-run real SDR of 2%.
of US$190 on the basis of a real SDR that declines from an initial value Our survey also allowed us to apply philosophers’ views on dis-
of 2%. Recent judgements on the SCC and its underlying SDR in the counting to determine an optimal climate policy path within a recent
United States (Louisiana et al. versus Biden et al.; Case 2:21-CV-01074) update21 of a prominent IAM, DICE22. This model is underpinned by the
have underscored the importance of this issue. simple Ramsey rule (SRR) within the discounted utilitarian framework,
Despite enormous sensitivities of policy appraisals to discount- which is commonly understood by experts from both disciplines and is
ing choices and their ethical content, recommendations on the SDR found in government guidelines across the world23–25. The SRR equates
have been dominated by economists6 who do not necessarily have the return to capital through real interest rates (r) on the production
any special expertise in providing ethical advice7. Rather, economics side of the economy with the social rate of time preference (δ + ηg,
has been described as being “value-free”8 due to a reliance on revealed where δ is the pure time preference, η is the elasticity of marginal utility
preference and economic fundamentals to identify what is socially and g is the real growth rate) on the consumption side: r = SDR = δ + ηg,

1
Department of Economics, University of Copenhagen, Copenhagen, Denmark. 2CESifo Research Network, Munich, Germany. 3Department of Economics
and Center for Earth System Research and Sustainability (CEN), University of Hamburg, Hamburg, Germany. 4School for Business and Society, University
of York, York, UK. 5Dragon Capital Chair in Biodiversity Economics, LEEP Institute, Department of Economics, University of Exeter Business School,
Exeter, UK. 6Grantham Research Institute on Climate Change and the Environment, London School of Economics and Political Science, London, UK.
e-mail: [Link]@[Link]

Nature Climate Change

Article [Link]

a b 1.0

3.0
0.8
2.5

Proportion
0.6
2.0
SDR (%)

0.4
1.5

0.2
1.0

0.5 0
–5 0 5 10 15

0 SDR (%)
Philosophers Economists
ts
s
er

is
ph

m
no
so

o
ilo

Ec
Ph

c d e f g
3.0 3.0 100 3.0 3.0

90
2.5 2.5 2.5 2.5

Elasticity of marginal utility

80
Risk-free interest rate (%)

Pure time preference (%)

Normative weight (%)

70
Growth rate (%)

2.0 2.0 2.0 2.0

60

1.5 1.5 50 1.5 1.5

40
1.0 1.0 1.0 1.0
30

20
0.5 0.5 0.5 0.5
10

0 0 0 0 0

s
s

ts
s

ts
ts

ts

ts
er
er
er

er

er

is
is

is
is

is

ph
ph
ph

ph

ph

om
om

om
om

om

so
so
so

so

so

on
on

on
on

on

ilo
ilo
ilo

ilo

ilo

Ec
Ec

Ec
Ec

Ec

Ph
Ph
Ph

Ph

Ph

Fig. 1 | Comparison of philosopher and economist recommendations g, Elasticity of marginal utility, η. The numbers of observations for philosophers
on intergenerational discounting and key determinants. a, Real SDR. (economists) from top left to bottom right are n = 11 (181), n = 11 (182), n = 10
b, Proportion of experts whose SDR range they are comfortable with (181), n = 10 (176), n = 13 (182), n = 14 (180) and n = 10 (173). The boxes in a and c–g
recommending includes the SDR value given on the x axis. c, Real growth rate, represent interquartile ranges (25th to 75th percentiles). The thick horizontal
g. d, Real risk-free interest rate, r. e, Normative weight. f, Pure time preference, δ. black lines indicate the medians, and the crosses indicate the mean values.

which can be calibrated in different ways (Methods). Here δ reflects the complex matter at hand to simple parameter values within the
societal impatience as measured by the discount rate on utility, while SRR, reflecting their discomfort with the discounted utilitarianism
ηg is the wealth effect, which places a lower weight on future consump- approach. Economists, by contrast, often recommend technical exten-
tion if the marginal utility of consumption is diminishing (η > 0) and sions to the SRR to account for uncertainty, distribution and limited
when future societies are predicted to be wealthier than today (g > 0). substitutability of non-market goods.
In addition to a direct elicitation of the SDR in its general formula- In addition, philosophers give systematically lower responses
tion, we asked for values of each of the components of the SRR: r, δ, η for δ than economists, reflecting their stronger preference for an
and g. Our results show that experts broadly support the parameter val- impartial perspective of social justice. This impartiality perspective
ues currently applied by the EPA in its latest SCC estimates, although on is also reflected in a greater median value for inequality aversion, η,
the basis of very different intellectual arguments. We then applied the although its sample frequency distribution does not differ significantly
philosophers’ median recommended values of δ and η as inputs to an between the two disciplines. Yet, with positive expected growth, g, a
updated version of the DICE model21. This resulted in an optimal climate higher median η leads to a higher SDR and less weight placed on future
policy path that is almost indistinguishable from that recommended by well-being. These two effects cancel out on average within the SRR, and
the median view of economists as previously reported in ref. 21. Both the implied climate policy paths are therefore similar to those recom-
are highly supportive of the United Nations (UN) Paris climate targets. mended by economists.
The median views from both disciplines translate into climate policy Finally, philosophers place greater weight on normative considera-
paths that limit global temperature change to under 1.5 °C by 2100, tions and less on the potential inefficiencies that arise from ignoring
with estimates for the SCC of around US$220 per ton of CO2 in 2020. production-side opportunity costs (r) than economists do. Therefore,
However, while there is agreement between the two disciplines on while providing multidisciplinary support for the UN climate targets,
these and other matters, both quantitative and qualitative responses philosophers and economists provide different practical and ethical
reveal that each has systematically different recommendations for insights into future methodological improvements for appraising
how intergenerational welfare can be better incorporated into cli- climate change mitigation policies. As emphasized in ref. 26, many US
mate policy analysis. Philosophers are much more reluctant to reduce statutes require agencies to use the highest-quality data to allow them

Nature Climate Change

Article [Link]

a 2.5 b 1,200

SCC (US$ per ton CO2)

Elasticity of marginal
2.0 1,000

800
1.5
utility
600
1.0
400
0.5
200

0 0
0 2 4 6 8 2020 2040 2060 2080 2100

Pure time preference (%) Year

Median philosopher view Median philosopher view

View of median philosopher run Median philosopher run
Individual philosopher view Individual philosopher view

c 40 d 3.0

Temperature change (°C)

Industrial emissions

2.5
20
(GtCO2)

2.0

0
1.5

1.0
–20
2020 2040 2060 2080 2100 2020 2040 2060 2080 2100
Year Year
Median philosopher view Median philosopher view
Median philosopher run Median philosopher run
Individual philosopher view Individual philosopher view

Fig. 2 | Philosophers’ views on climate policy paths. a–d, Optimal climate median run. The plots show views on the normative policy parameters (a), SCC
policy paths in an updated DICE–IAM for each philosopher’s view on the two per ton of CO2 in 2020 US dollars (b), industrial emissions in GtCO2 (c) and global
normative policy parameters, δ and η, and the parameters of the median view and mean temperature change in °C compared with 1850–1900 levels (d).

to best take into account the welfare of future generations, includ- A key finding, depicted in Fig. 1a, is that philosophers and economists
ing “intergenerational equity”. Additionally, Circular A-4 states that have very similar views on the appropriate value for the long-term
“special ethical considerations arise when comparing benefits and SDR. Using a non-parametric Wilcoxon rank-sum test, we were unable
costs across generations.” This paper explicitly addresses such ethical to reject the null hypothesis that the two sets of responses come from
questions, providing data that are of direct relevance to the EPA and the same distribution (P = 0.818). While economists have a slightly
other policymakers for evaluating climate policy paths. lower median response (2.0% versus 2.4%), the mean responses are
almost identical at 2.27%. The multidisciplinary agreement on the SDR
Expert recommendations for the SDR is further strengthened by examining acceptable ranges for the SDR
The survey (outlined in the Methods) asked the respondents for their (Fig. 1b). A real SDR of 2% is most often contained within this range
preferred value of the real, long-term (>100 years) SDR and the values for experts in both disciplines (for all but one philosopher and 77% of
for each of the components, r, δ, η and g, of one specific formulation economists). Expert support falls quickly for higher values of the SDR.
of the SDR: the SRR. We also asked each expert the extent to which the There are also no significant differences between the two disciplines
SDR should be based on normative issues, involving justice towards in the descriptive components of the SRR—g and r, shown in Fig. 1c,d
future generations, compared with descriptive issues, such as observed (rank-sum, P = 0.681 and P = 0.617).
or forecast market rates of return. ‘Acceptable ranges’, within which There are important areas of disagreement, however. Philosophers
each expert thought the SDR could reasonably lie, were also elicited, are much less persuaded by opportunity-cost-of-capital arguments
and an opportunity was given for open-ended qualitative comments. for setting the SDR than economists (Fig. 1e). The mean (median)
We identified many fewer philosophers with expertise in social recommended weights put on normative considerations are 78.5%
discounting (n = 46), selected on the basis of relevant publications, than (80%) for philosophers and 61.5% (70%) for economists, and the distri-
expert economists (n = 627)20. However, we received a higher response butions are statistically significantly different (rank-sum, P = 0.040).
rate from philosophers (63%, n = 29) than in ref. 20 (31%, n = 197), thus Figure 1f shows that, for the “two central normative parameters”29 of the
capturing a more complete account of the relevant expert popula- SRR, philosophers recommend statistically significantly lower values
tion. Yet, philosophers were much less willing to provide responses (rank-sum, P = 0.043) for δ than economists despite a few outlier high
to the quantitative questions. Only 52% (n = 15) provided quantitative responses. While it might be expected that this preference for intertem-
responses, with some questions receiving as few as 10 answers (Sup- poral equity would lead philosophers to provide higher values for η,
plementary Table 1). This is comparable to other recent expert surveys Fig. 1g shows that this view is not supported by the data, as the rank-
on key intergenerational matters27,28. sum test fails to reject the null that the two distributions of responses
We first illustrate the distributions of philosophers’ quantitative are identical (P = 0.539). The median response of philosophers is nota-
responses and contrast these with those by economists (Fig. 1); see bly higher than that of economists, though, and this has an important
Supplementary Table 1 and Supplementary Fig. 1 for further details. part to play in the next section.

Nature Climate Change

Article [Link]

a 2.0 b 800

SCC (US$ per ton CO2)

Elasticity of marginal
1.5 600

utility
1.0 400

0.5 200

0 0
0 0.5 1.0 1.5 2020 2040 2060 2080 2100
Pure time preference (%) Year

Median philosopher view View of median philosopher run Median philosopher view Median philosopher run
Median economist view Nordhaus discounting Median economist view Median economist run
Stern discounting Nordhaus discounting Stern discounting

c 40
d
2.0

Temperature change (°C)

Industrial emissions

1.8
20
(GtCO2)

1.6

0
1.4

–20 1.2

2020 2040 2060 2080 2100 2020 2040 2060 2080 2100
Year Year
Median philosopher view Median philosopher run Median philosopher view Median philosopher run
Median economist view Median economist run Median economist view Median economist run
Nordhaus discounting Stern discounting Nordhaus discounting Stern discounting

Fig. 3 | Philosopher and economist agreement on climate policy paths. Nordhaus22 and Stern30. b, SCC per ton of CO2 in 2020 US dollars. c, Industrial
a, Views on the normative policy parameters, δ and η, including the median emissions in GtCO2. d, Global mean temperature change in °C compared with
economist and median philosopher views, parameter choices in line with the 1850–1900 levels. All panels show optimal climate policy paths for different
median policy run (applicable for only philosophers; Methods), and views by views on the two normative policy parameters in an updated DICE–IAM.

Optimal climate policy paths 1.4 °C by the end of the century (Fig. 3d). The median philosopher
To consider implications for optimal climate policy paths, we applied an view exhibits a lower value of δ than the median economist (0.075%
updated version of the IAM, DICE21,22 (outlined in Methods). Key changes versus 0.5%) but a higher η (1.25 versus 1). Both effectively attach more
include an update of the climate module and more recent climate dam- weight to distributional issues, but with counteracting effects on the
age estimates. Discounting within DICE requires two input variables SRR. Within DICE, these cancel out and result in very similar optimal
from the SRR framework: δ and η. For the two summary measures, we climate policy paths. Furthermore, the median runs of philosopher and
use the median view from philosophers for these two parameters and economist policy paths are also virtually indistinguishable (Fig. 3b),
the median run of the resulting climate policy paths. For each param- with SCCs in 2020 of around US$120 (US$112 and US$130, respectively)
eter pair, the optimal climate policy is computed. The growth path is and temperature trajectories that stay below 2 °C and reach 1.8 °C by
non-constant and endogenized, not a constant long-term rate from the end of the century.
experts’ forecasts.
We illustrate climate policy paths for each of the ten philoso- Expert recommendations for intergenerational
phers that provided complete pairs of δ and η (Fig. 2). The median view discounting
represents this summary measure on both parameters individually Philosophers’ qualitative comments were particularly rich. These are
(Fig. 2a), while the specific combination (δ = 0.075, η = 1.25) is not held compared to those of economists20, which have not previously under-
by any particular philosopher. The median run of philosophers’ policy gone detailed analysis. Comments from both philosophers (denoted by
paths, in contrast, is supported by three philosophers with identical P1, P2 and so on) and economists (E1, E2 and so on) largely fall into the
views (δ = 0, η = 2). We observe that 60% of the runs result in tempera- three broad categories that form the basis for the subsections below.
ture changes below 2 °C by the end of the century (Fig. 2d). We highlight selected comments that showcase areas of agreement and
We next compared policy paths under median philosophers’ views differences in each category (Tables 1 and 2) and provide a complete
with an identical analysis by ref. 21 that uses the median economists’ list of anonymized qualitative remarks, including a more complete
views as reported in ref. 20. We also calibrated the DICE model using analysis of comments (Supplementary Information).
well-known parameter choices by Nordhaus22 and Stern30 for com-
parison. The median philosophers’ and median economists’ views Discounted utilitarianism and the SRR
translate into almost indistinguishable climate policy paths in terms Discounted utilitarianism underpins most discounting guidance25 and
of emission reductions and SCC estimates of around US$220 (US$219 often manifests as the SRR, a specific formulation of the SDR23. A num-
and US$227, respectively) in the year 2020 (Fig. 3a). These values are ber of economists and philosophers expressed discomfort with using
similar to estimates in a recent comprehensive analysis of the SCC28. these frameworks as the basis for evaluating intergenerational deci-
We also observe almost identical temperature changes of around sions, but for different reasons. A critique from several philosophers

Nature Climate Change

Article [Link]

Table 1 | Selected qualitative responses on the SRR, its components and normative versus positive weightings

Philosophers Category Economists

The Ramsey formula requires some empirical input, specifically Ramsey rule (nE = 17, nP = 3) My discount rate is less than implied by the Ramsey rule
the rate of economic growth. (P13) because I use the extended rule, incorporating uncertainty.
(E24)
To employ the Ramsey rule [positively]…we need some
serious research into the empirics of the Ramsey rule.…
[yet] there is no empirical support for the Ramsey rule as a
positive description. (E8)
I think it’s morally acceptable—and in many contexts even Pure time preference I see no reason to treat generations not equally. (E5)
required—for us to give greater weight to the concerns of those (nE = 10, nP = 7) Pure time preference should probably reflect some
nearer and dearer to us than to those further away. (P5) catastrophic risk: 0.2% or so. (E4)
I do not believe in pure time discounting at all. The idea is that I…favor a preference-based view of well-being,…some
time itself has no moral significance in itself. (P6) deference to individuals’ own preference parameters. (E17)
I mostly buy the Parfitt-Cowen response…but shoehorning
existential risk into the discount rate [is reasonable]. (P5)
0 in the case of wrongful harm to future generations (e.g. climate Elasticity of marginal utility [η]…being based on representative agent or [CRRA]
change)…1.5 for all other effects. (P2) (nE = 12, nP = 1) formulations…is largely a distributional/equity
consideration… (E13)
As climate change accelerates, and if the projections are at the Growth rate (nE = 14, nP = 3) I foresee a very bright economic future with a continued 2
medium to high end, this threatens to turn negative. I’d guess percent growth rate for the coming century. (E8)
somewhere in the range of +2% to −2%. (P1) I have never understood why economists always assume
that consumption will rise at a constant rate. (E1)
The normative question of what social discount rate is required Normative weight (nE = 16, I don’t think normative issues outside of an objective should
by justice to future generations can only be answered once nP = 10) have much of a role.…Indeed, under a different social
the descriptive issues have been answered. Therefore it is not welfare function there is a different positive description of
possible to assign proportions to these two. (P10) the appropriate discount rate. (E23)
The components of the SDR are overwhelmingly normative.
(E13)
The variables nE and nP refer to the numbers of qualitative remarks from economists and philosophers, respectively.

concerned the difficulties of embodying all concepts of fairness consist- However, these circulars also acknowledge the virtue of alternative
ently in such a limited ethical framework, resulting in discarding mor- approaches such as using a shadow price of capital to value displaced
ally relevant information (P15 and P3). This includes the intratemporal investment. Alternatively, the SDR can be estimated from the consump-
distribution of income and the pathways of physical and economic tion side (as also recommended by the use of observed savings rates
outcomes over time that are lost in intertemporal aggregation (also in Circular A-4) or calibrated to the SRR (as in the HM Treasury’s Green
E77). While sustainability rules may offer one way to introduce rights Book in the United Kingdom). The latter is seen as a predominantly
or duties into the utilitarian framework, some philosophers propose normative approach, although g is a descriptive variable25. Yet, rather
stepping back completely and reassessing the essential moral issues at than experts being polarized, the quantitative and qualitative responses
stake. One example is “moral modelling”31, which argues that decision show that a large majority of both economists and philosophers find
theory can embody many ethical approaches despite having largely both normative and positive issues relevant (P10, P2, P14, E26 and E17).
consequentialist foundations (P5; Table 2). Rank-ordered utilitarianism, As revealed by their quantitative responses, philosophers place
emphasizing a generation’s well-being rank32, was suggested from both stronger emphasis on normative issues, and this is also reflected in their
disciplines as an alternative to a zero or non-zero pure time preference comments. Within discounted utilitarianism, a number of philosophers
(P3 and E76). Some economists also proposed alternative criteria, such take a fully normative approach to determining the SDR, believing that
as sustainability rules and ‘tolerable windows’ approaches to rule out “uncertainty and justice should be key determinants” (P12). This posi-
certain adverse outcomes for future generations (E28 and E52). tion is also reflected in the comments of some economists. By contrast,
By contrast, rather than mainly proposing alternatives to dis- some economists, but no philosophers, are 100% ‘pure’ positivists,
counted utilitarianism itself, economists predominantly recommended believing that interest rates should be used to calibrate the SDR (E32).
technical extensions to the SRR. Examples include substitutability and Some philosophers do, though, report that normative questions can
environmental scarcity (E5, E51 and E48)33,34, distributional issues be answered only when the positive facts are known (P10; Table 1).
(E20)35,36, uncertainty (E8)37 and, relatedly, declining discount rates The responses from both disciplines give little support for the purely
(E42)1,38. Resolving the perceived inadequacies of the SRR via math- positive position being taken to discounting that underpins the latest
ematical extensions within the discounted utilitarian framework, estimates of the SCC published by the EPA5, where δ and η on the con-
rather than departing from it altogether, appeared popular among sumption side of the SRR are calibrated on a purely descriptive basis39,40.
economists. This may help explain Supplementary Fig. 2, which plots
the frequency distribution of SDR − (δ + ηg) across experts in each Calibrating the components of the SRR
discipline. This distribution has a mean of −1.21% (two-sided t-test: Philosophers and economists alike provided comments on calibrating
P < 0.000) for economists but is not significantly different from zero the components of the SRR and the difficulties involved. One broad
for philosophers (P = 0.517). critique from economists was that there is no empirical support for
the SRR as a descriptive model (E8; Table 1). While modal responses
Normative versus positive weight on pure time preference in both disciplines suggest that remoteness
The Stern–Nordhaus debate was ostensibly polarized between posi- in time itself has no moral importance19, this view is contested within
tive and normative approaches, and this framing continues to this day. both groups. Among those arguing for a non-zero δ, existential haz-
Within the SRR approach, the SDR can be calibrated from the opportu- ard risk is the most frequent motivation (P5 and E4), and both groups
nity cost of capital, r, as recommended in, for example, Circulars A-4 also recognize tyranny arguments between generations that can arise
and A-94 of the Office of Management and Budget in the United States. from an inappropriate choice of δ (P5 and E5). Positive rates of time

Nature Climate Change

Article [Link]

Table 2 | Selected qualitative responses on extensions and alternatives to discounting

Philosophers Category Economists

Because the discount rate is dependent on assumptions Limited substitutability (nE = 20, nP = 4) If future costs/benefits accrue, e.g., to environmental
about future scarcity,…etc. there should be no single discount amenities, I would argue for a very low discount rate,
rate across the board. Natural assets, monetary investments, based on an expectation of increasing relative prices
infrastructure project etc. should be discounted at different for these goods. (E48)
rates. (P16)
If there are real sustainability issues and we fail to take the path
towards a sustainable development, then the growth rate of real
per-capita consumption will be negative. (P2)
If we’re concerned about risk, then we should just model risk Uncertainty (nE = 20, nP = 4) We need to admit that the current state of the world
explicitly, but given the complexity of climate policy evaluation, is full of uncertainties. [Yet] most uncertainties
I think shoehorning existential risk into the discount rate is a are neglected, and sometimes few remain when
reasonable thing to do. (P5) these are considered most important,…or easiest to
accommodate. (E8)
Discounting rates are useful but do not provide
(accurate) market values of risky assets…but what
does? (E56)
An intergenerational sort of “differentiated love,” to use the Declining discount rates (nE = 20, nP = 3) I am more comfortable with declining discount
Confucian term, justifies a strictly positive rate of pure time rates…due both to declining time preference rates
preference that declines to zero over the course of a few and to uncertainty about future consumption
generations. (P5) growth. (E42)
A discount is certainly justified [because] marginal utility of
income will decrease with increasing material well-being,
especially in countries counting at present as developing
countries. But this consideration does not justify a discount
merely for temporal distance. (P8)
In the case of a project that potentially harms people living Heterogeneity, distribution and The normative approach is based on weighing
abroad in health and property, these harms should be given aggregation (nE = 19, nP = 1) WTP for different generation, with less weight on
equal weight in project appraisal. (P2) the richer ones. It makes no sense to use this in
combination with valuation studies where each
individuals WTP are not weighted similarly by
income. (E20)
As you know, the correct way to handle the ’opportunity cost of Opportunity cost of funds (nE = 8, nP = 2) SDRs should reflect the social opportunity cost of
capital’ is not by adjusting the social discount rate. (P13) funds. (E61)
An opportunity cost of investment funds should be
the instrument of capital rationing. (E15)
Instead of “cost-benefit,” or “discount rates,” long-term future Alternatives to discounting Instead of imposing a [SWF] and calculate the
assessments should be made according to a “sustainability (nE = 15, nP = 9) corresponding optimum, it is “better” to depict a
index,” which gives primary consideration, not to “market set of feasible paths of consumption, production,
preferences,” but rather to sustaining environmental conditions temperature, income distribution, etc., and let the
and resource availability. (P15) policymaker make a choice. (E12)
I think of…policy evaluation as something closer to Katie
Steele’s “moral modeling” than to some kind of Benthamite
calculus. (P5)
The variables nE and nP refer to the numbers of qualitative remarks from economists and philosophers, respectively.

preference result in weights approaching zero on the utility of very Discussion

far-future generations, which may be morally unacceptable. In contrast, Expert advice plays a key role in complex areas of public policy, includ-
δ = 0 can lead to ‘hair shirt’ outcomes for the current generation, who ing discounting and its application to climate policy appraisal1,41–43.
would be required to save ‘damaging’ amounts for the benefit of their Yet, academic guidance in this area has been sought predominantly
distant descendants (P5). from one discipline, economics, where experts do not have any spe-
Yet philosophers rely less on positivist considerations and pre- cial insights into the underlying ethical issues at stake. Our survey of
fer a variety of alternatives to classical utilitarian consequentialism, philosophers meets the need for broader multidisciplinary input into
including agent-relative ethics, suggesting that it is morally acceptable this policy field4.
to place greater weight on those nearer and dearer to inform our pre- We find considerable agreement between the two disciplines for a
ferred utility discount rate (P5). Economists more often used positive long-term real SDR of 2%, a value that has underpinned recent estimates
arguments for δ > 0 and a deference to individual preferences (E17). of the SCC28. Furthermore, median discounting values and median
Although one philosopher proposed that “people’s actual behaviour in policy runs from each discipline result in almost identical optimal
discounting reflects their genuine normative concerns” (P4), they were climate policy paths that support the UN Paris climate targets. Yet each
ultimately introspective on the validity of that assumption. While both discipline brings distinct expertise to bear. While economists provide
groups engaged with the consequentialist approach, philosophers guidance on technical extensions to discounting rules25, philosophers
tended to invoke rights and duties (P15). Finally, only one philosopher point to broader extensions to discounted utilitarianism: the limita-
commented on the elasticity of marginal utility, recommending a tions of consequentialism, the rights of future generations and the
different value of η for “wrongful harm to future generations” than duties of the present generation.
other effects, again leaning on the idea of rights and duties to future When estimating the welfare cost of future climate damages, the
generations (P2). For economists, the qualitative comments reveal SDR is critical. Embedding this within an intergenerational welfare
that η has different meanings, with others also questioning whether setting requires many competencies and may require deliberative and
it is appropriate to use a single value for this parameter in all contexts multidisciplinary perspectives4. The different motives held by phi-
(E18 and E5). losophers are complementary to those of economists and recommend

Nature Climate Change

Article [Link]

alternative approaches to current climate policy analysis. Philosophical 18. Mogensen, A. L. The only ethical argument for positive δ?
perspectives offer ethical checks and balances within IAMs to narrow Partiality and pure time preference. Phil. Stud. 179,
the set of acceptable consequentialist climate policies, or they can 2731–2750 (2022).
offer altogether different procedural lenses through which to evalu- 19. Parfit, D. Reasons and Persons (Oxford Univ. Press, 1984).
ate climate policy. Examples here include rank-ordered utilitarianism, 20. Drupp, M. A., Freeman, M. C., Groom, B. & Nesje, F.
sustainability rules and ‘tolerable windows’ approaches. That econo- Discounting disentangled. Am. Econ. J. Econ. Policy 10,
mists and philosophers can agree on policy outcomes builds academic 109–134 (2018).
expert consensus, a key input to any legal challenge (as recently seen 21. Hänsel, M. C. et al. Climate economics support for the UN climate
in the United States). These insights will be overlooked if economists targets. Nat. Clim. Change 10, 781–789 (2020).
continue to dominate the debate. Yet, how to structure such interaction 22. Nordhaus, W. Projections and uncertainties about climate change
within “larger-scale, participatory and deliberative, integrated scien- in an era of minimal climate policies. Am. Econ. J. Econ. Policy 10,
tific assessment processes”, as remarked by one philosopher (P17), is 333–360 (2018).
an avenue for consideration in future research and policy. 23. The Green Book (HM Treasury, 2020).
24. Cost–Benefit Analysis and the Environment: Further Developments
Online content and Policy Use (OECD, 2018).
Any methods, additional references, Nature Portfolio reporting sum- 25. Groom, B., Drupp, M. A., Freeman, M. C. & Nesje, F. The future,
maries, source data, extended data, supplementary information, now: a review of social discounting. Annu. Rev. Resour. Econ. 14,
acknowledgements, peer review information; details of author contri- 467–491 (2022).
butions and competing interests; and statements of data and code avail- 26. Howard, P. & Schwartz, J. A. Valuing the future: legal and
ability are available at [Link] economic considerations for updating discount rates. Yale J.
Regul. 39, 595–657 (2022).
References 27. Christensen, P., Gillingham, K. & Nordhaus, W. Uncertainty in
1. Arrow, K. et al. Determining benefits and costs for future forecasts of long-run economic growth. Proc. Natl Acad. Sci. USA
generations. Science 341, 349–350 (2013). 115, 5409–5414 (2018).
2. Weitzman, M. L. Gamma discounting. Am. Econ. Rev. 91, 28. Rennert, K. et al. Comprehensive evidence implies a higher social
260–271 (2001). cost of CO2. Nature 10, 687–692 (2022).
3. Aldy, J. E., Kotchen, M. J., Stavins, R. N. & Stock, J. H. Keep 29. Nordhaus, W. D. A Question of Balance (Yale Univ. Press, 2008).
climate policy focused on the social cost of carbon. Science 373, 30. Stern, N. The Economics of Climate Change: The Stern Review
850–852 (2021). (Cambridge Univ. Press, 2007).
4. Wagner, G. et al. Eight priorities for calculating the social cost of 31. Colyvan, M., Cox, D. & Steele, K. Modelling the moral dimension
carbon. Nature 590, 548–550 (2021). of decisions. Nous 44, 503–529 (2010).
5. Standards of Performance for New, Reconstructed, and Modified 32. Zuber, S. & Asheim, G. B. Justifying social discounting: the
Sources and Emissions Guidelines for Existing Sources: Oil and rank-discounted utilitarian approach. J. Econ. Theory 147,
Natural Gas Sector Climate Review (US Environmental Protection 1572–1601 (2012).
Agency, 2022). 33. Sterner, T. & Persson, U. M. An even sterner review: introducing
6. Groom, B. & Hepburn, C. Reflections—looking back at social relative prices into the discounting debate. Rev. Environ. Econ.
discounting policy: the influence of papers, presentations, Policy 2, 61–76 (2008).
political preconditions, and personalities. Rev. Environ. Econ. 34. Drupp, M. A. & Hänsel, M. C. Relative prices and climate policy:
Policy 11, 336–356 (2017). how the scarcity of nonmarket goods drives policy evaluation.
7. Dasgupta, P. Discounting climate change. J. Risk Uncertain. 37, Am. Econ. J. Econ. Policy 13, 168–201 (2021).
141–169 (2008). 35. Emmerling, J., Groom, B. & Wettingfeld, T. Discounting
8. Broome, J. Efficiency and future generations. Econ. Phil. 34, and the representative median agent. Econ. Lett. 161,
221–241 (2018). 78–81 (2017).
9. Groom, B. & Maddison, D. New estimates of the elasticity 36. Fleurbaey, M. & Zuber, S. Discounting, risk and inequality: a
of marginal utility for the UK. Environ. Resour. Econ. 72, general approach. J. Public Econ. 128, 34–49 (2015).
1155–1182 (2019). 37. Gollier, C. Pricing the Planet’s Future: The Economics
10. Nesje, F. Cross-Dynastic Intergenerational Altruism Working Paper of Discounting in an Uncertain World (Princeton Univ.
No. 9626 (CESifo, 2022). Press, 2013).
11. Beard, S. The dilemma of discounting: the impossibility of 38. Cropper, M. L., Freeman, M. C., Groom, B. & Pizer, W. A. Declining
setting a context independent Ramsey discount rate for human discount rates. Am. Econ. Rev. 104, 538–543 (2014).
wellbeing. Rerum Causae 3, 11–21 (2011). 39. Valuing Climate Damages: Updating Estimation of the Social Cost
12. Broome, J. Discounting the future. Phil. Public Aff. 23, of Carbon Dioxide Tech. Rep. (National Academies of Sciences,
128–156 (1994). Engineering, and Medicine, 2017).
13. Caney, S. Climate change and the future: discounting for time, 40. Newell, R. G., Pizer, W. A. & Prest, B. C. A discounting rule
wealth, and risk. J. Soc. Phil. 40, 163–186 (2009). for the social cost of carbon. J. Assoc. Environ. Resour. Econ. 9,
14. Greaves, H. Discounting for public policy: a survey. Econ. Phil. 33, 1017–1046 (2022).
391–439 (2017). 41. Emmerling, J. et al. The role of the discount rate for emission
15. Heilmann, C. Values in time discounting. Sci. Eng. Ethics 23, pathways and negative emissions. Environ. Res. Lett. 14,
1333–1349 (2017). 104008 (2019).
16. Kelleher, J. P. Pure time preference in intertemporal welfare 42. Pindyck, R. S. The use and misuse of models for climate policy.
economics. Econ. Phil. 33, 441–473 (2017). Rev. Environ. Econ. Policy 11, 100–114 (2017).
17. Medvecky, F. Valuing environmental costs and benefits in an 43. Sunstein, C. R. On not revisiting official discount rates:
uncertain future: risk aversion and discounting. Erasmus J. Phil. institutional inertia and the social cost of carbon. Am. Econ. Rev.
Econ. 5, 1–23 (2012). 104, 547–551 (2014).

Nature Climate Change

Article [Link]

Publisher’s note Springer Nature remains neutral with the author(s) or other rightsholder(s); author self-archiving of the
regard to jurisdictional claims in published maps and accepted manuscript version of this article is solely governed by the
institutional affiliations. terms of such publishing agreement and applicable law.

Springer Nature or its licensor (e.g. a society or other partner) holds © The Author(s), under exclusive licence to Springer Nature Limited
exclusive rights to this article under a publishing agreement with 2023

Nature Climate Change

Article [Link]

Methods and we illustrate the median model run. Note that in the comparison
Sampling, respondents, survey text and response bias to the economists’ views in Fig. 3, we do not represent the view of
We sampled economists on the basis of keyword searches for various the median economist run as a pair of parameters because they are
forms of the term ‘(social) discounting’ in 100 leading economics time-varying21.
journals to determine pertinent publications. This process yielded
627 potential expert authors, 197 of whom participated in the survey, Calibration of the SRR
which was administered between May and November 2014. Further The SRR can be calibrated in different ways. The approach taken
details are provided in ref. 20. For replication, see ref. 44. recently by the EPA5 in its latest interim estimates of the SCC (based
For philosophers, we took a broader approach to search for articles on refs. 28,40) is to work from the production side of the economy,
on ‘discounting’, ‘discount rate’ and ‘social discount*’ in the abstract r. Such an approach is supported by Carleton and Greenstone48, who
or main text of philosophy papers stored at PhilPapers, PhilIndex and “recommend continuing to rely on existing asset markets to guide the
Scopus. This was accompanied by a purposive search of philosophers discount rate choice”, as well as Recommendation 6-2 of the National
who work in the area based on our readings and recommendations Academy of Sciences39 report, which stated that parameter values
from well-known academics in the field. We then checked, case by case, should reflect the consumption rate of interest. An alternative, which
whether there was any evidence in their publications for expertise on we take here following ref. 20, is to directly apply expert recommen-
‘(social) discounting’. We identified 46 potentially relevant experts, 29 dations concerning the appropriate parameter values. These will
of whom participated in the survey, which was administered between reflect normative considerations of what these parameters ought to
October 2016 and April 2017. While the number of philosophers engag- be from an ethical perspective, as well as perceived production-side
ing with social discounting is lower than for economists, our survey opportunity costs.
design identifies a more complete account of the population of phi-
losophers who can be deemed as ‘expert’ for our purposes. We updated Reporting summary
the PhilPapers–PhilIndex–Scopus search in 2019, additionally picking Further information on research design is available in the Nature Port-
up 10 potentially relevant experts in an attempt to expand the sample folio Reporting Summary linked to this article.
size, but none of them chose to participate.
The invitation e-mail and survey text is provided in Supplementary Data availability
Information. We pretested the survey to ensure that it was understand- The data that support the plots in this paper and other findings of this
able to both expert groups. For the philosophers, we provided addi- study are available at the following repository: [Link]
tional clarification of some of the economic terminology. zenodo.7920803.
Overall, many more economists (n = 627) were identified in ref. 20
as potential experts on social discounting than philosophers were for Code availability
this survey (n = 46). However, we received a higher response rate from All code used to produce the analysis is available at the following reposi-
philosophers (63%, n = 29) than in ref. 20 (31%, n = 197). Thus, while the tory: [Link] The details of imple-
absolute number of responses is lower for philosophers, we captured a mentation can be found in Methods.
more complete account of the expert population on social discounting
in the field of philosophy than in economics. References
A core difference between the two expert groups is the relative 44. Nesje, F., Drupp, M., Freeman, C. M. & Groom, B. Replication
types of responses they were willing to provide. Only 52% (n = 15) of files for ‘Philosophers and economists agree on climate policy
philosophers provided any quantitative responses, with some ques- paths but for different reasons’. Zenodo [Link]
tions receiving as few as 10 answers (Supplementary Table 1). A much zenodo.7920803 (2023).
higher proportion of economists, 94% (n = 185), were prepared to give 45. Armstrong, J. S. & Overton, T. S. Estimating nonresponse bias in
numerical values, with each question receiving at least 173 answers. mail surveys. J. Mark. Res. 14, 396–402 (1977).
By contrast, all philosophers, bar one, provided open-ended qualita- 46. Necker, S. Scientific misbehavior in economics. Res. Policy 43,
tive comments, against only 51% of economists (n = 100). Thus, even 1747–1759 (2014).
though the number of quantitative responses by philosophers is 47. Johnson, T. P. & Wislar, J. S. Response rates and nonresponse
comparable to some recent expert surveys on key intergenerational errors in surveys. JAMA 307, 1805–1806 (2012).
aspects27,28, here we place greater weight on analysing the textual 48. Carleton, T. & Greenstone, M. A guide to updating the US
responses of both philosophers and economists, alongside the quan- government’s social cost of carbon. Rev. Environ. Econ. Policy
titative results. 16, 196–218 (2022).
We did not find evidence for potential non-response and strate-
gic response biases in the economists sample20. The Supplementary Acknowledgements
Information reports these standard tests also for the philosophers We thank our many survey respondents; seminar audiences at LSE,
sample45–47, with a specific focus on characterizing the larger share of Copenhagen, EAERE 2021 and Miljøøkonomisk konference 2021; and
philosophers that provided qualitative responses. G. Asheim, A. Blok, J. Broome, H. Greaves, K. Mintz-Woo, K. Nyborg,
For the illustration of economically optimal climate policy paths, P. G. Piacquadio and P. B. Sørensen for helpful comments. We also
we relied on an updated version of the IAM DICE22 by Hänsel et al.21, who thank M. Hänsel for providing an additional run from the updated
showed implications of the views of economic experts only. The model DICE model and M. Lustig for excellent research assistance. We
was updated to account for recent scientific evidence on the carbon acknowledge support from the LSE, STICERD, CREE funded by
cycle, the energy balance model and climate damage estimates. It also Norwegian Research Council (grant no. 209698); NATCOOP funded
included non-CO2 forcing in line with lower temperature scenarios, as by the European Research Council (grant no. 678049); and the
well as technical requirements on the availability of negative-emissions DFG under Germany’s Excellence Strategy (EXC 2037 and CLICCS)
technologies and the speed of decarbonization. In addition, for the pre- project no. 390683824, contribution to the Center for Earth System
sent paper, we considered the views among the ten philosophers who Research and Sustainability (CEN) of Universität Hamburg. B.G.
provided complete pairs on the two key normative policy parameters: δ is the Dragon Capital Chair of Biodiversity Economics, funded by
and η. We also considered the median view among these respondents, Dragon Capital.

Nature Climate Change

Article [Link]

Author contributions Correspondence and requests for materials should be addressed to

F.N., M.A.D., M.C.F. and B.G. designed the research. F.N., M.A.D. and Ben Groom.
B.G. analysed the data. F.N., M.A.D., M.C.F. and B.G. wrote the paper.
Peer review information Nature Climate Change
Competing interests thanks Lisa Rennels and the other, anonymous,
The authors declare no competing interests. reviewer(s) for their contribution to the peer review of
this work.
Additional information
Supplementary information The online version contains supplementary Reprints and permissions information is available at
material available at [Link] [Link]/reprints.

Nature Climate Change

 nature portfolio | reporting summary
Corresponding author(s): Ben Groom
Last updated by author(s): Apr 20, 2023

Reporting Summary
Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.

A description of all covariates tested

A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.

Software and code

Policy information about availability of computer code
Data collection NA

Data analysis STATA 17

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and
reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
March 2021

- Accession codes, unique identifiers, or web links for publicly available datasets
- A description of any restrictions on data availability
- For clinical datasets or third party data, please ensure that the statement adheres to our policy

All data and code is available in the Supplementary Information (line275, page 8)

1
 nature portfolio | reporting summary
Human research participants
Policy information about studies involving human research participants and Sex and Gender in Research.

Reporting on sex and gender sex or gender not analysed in the study.

Population characteristics Sample of experts. nothing to add.

Recruitment On the basis of expertise, specifically publications on a particular topic. All explained in the paper.

Ethics oversight LSE

Note that full information on the approval of the study protocol must also be provided in the manuscript.

Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see [Link]/documents/[Link]

Behavioural & social sciences study design

All studies must disclose on these points even when the disclosure is negative.

Study description Study of expert opinions on social discounting and intergenerational ethics. some quantitative analysis and some qualitatitive.

Research sample Sample of expert economists and philosophers

Sampling strategy Sampled on the basis of expertise via publications: keywords and abstracts containing terms relating to social discounting.

Data collection Online survey. Survey monkey.

Timing May - November 2014

Data exclusions None

Non-participation Philosophers: 29/46, Economists: 220/603

Randomization No treatments. no randomisation.

Reporting for specific materials, systems and methods

We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

Materials & experimental systems Methods

n/a Involved in the study n/a Involved in the study
Antibodies ChIP-seq
Eukaryotic cell lines Flow cytometry
March 2021

Palaeontology and archaeology MRI-based neuroimaging

Animals and other organisms
Clinical data
Dual use research of concern