Title: Understanding the climate responsibility associated with elections
Authors: Seth Wynes1,2*, Matthew Motta3, Simon D. Donner1
1Department of Geography, The University of British Columbia, Vancouver, British Columbia, Canada
2Department of Geography, Environment and Planning, Concordia University, Montreal, Quebec, Canada
3Department of Political Science, Oklahoma State University, Stillwater, Oklahoma, United States
*Corresponding Author: firstname.lastname@example.org
Summary: Elections represent infrequent, high leverage opportunities for everyday individuals to
contribute to climate change mitigation. In this Perspective, we present two ways of thinking about the
climate impact of voting in elections. The first, “emissions responsibility”, intuitively apportions
emissions to voters according to popular principles in carbon accounting and can be calculated for
elections where there is a clear difference between major candidates (as in the 2019 Canadian federal
election). The second “expected emissions value” approach is more probabilistic and can be used to
investigate the rationality of participating in an election for climate-motivated voters. Building on these
ideas, we discuss the possibility that “political carbon offsets” (donations to pro-climate politicians)
could constitute a more effective and more equitable alternative to traditional, voluntary carbon offsets.
Main text: Climate change, like voting in a democracy, is a collective action problem. Individuals may
hesitate to expend resources and effort to reduce their carbon footprints knowing that the marginal
impact on the global carbon budget is close to zero. Likewise, the probability of casting the deciding vote
in an election is low; in one United States presidential election, the probability was estimated at one in
sixty million.1 Given the low likelihood of casting a decisive ballot, researchers have argued over the
efficacy and even rationality of voting.2 Yet elections certainly influence collective action on problems
like environmental pollution and climate change; Green party strength is associated with lower air
pollution on a national scale,3 and US states that vote for greener candidates have lower growth in
carbon dioxide emissions4.
While we know the relative impact of individual actions involving air travel, personal vehicles and diet,5
the impact of political actions like voting is difficult to assess. Fortunately, in the vast majority of
situations, there is little conflict between choices that lead to measurable reductions (cycling, planting
trees) and choices where the causal chain is murky and seemingly unquantifiable (attending a climate
strike or canvassing for a pro-climate politician). But on occasion, circumstances require tradeoffs. A
climate organization could gather pledges to not fly for one year or collect the requisite number of
signatures needed to force a state-wide vote for a renewable energy standard. A philanthropist could
opt to buy a parcel of land and protect it from logging, or they could finance an election campaign for a
politician with strong climate credentials. These dilemmas can be meaningful in scale: voluntary carbon
markets were estimated at a total value of $191.3 million in 2016,6 and Amazon founder Jeff Bezos
alone pledged $10 billion to “scientists, activists (and) NGOs” in the effort to fight climate change.7 How
should such funds and efforts be allocated to maximize impact on climate mitigation?
The desire to quantify certain climate outcomes is justified insofar as it enables informed decision
making. It is because of Life Cycle Assessments that an informed consumer can judge that an electric
vehicle will produce less emissions over its lifetime than a conventional vehicle. Similarly, it is because
the carbon sequestered by trees can be quantified that companies are able to leverage climate guilt for
the emissions of a flight into verifiable carbon offsets. But since we have not quantified political actions
in the same way, they cannot be assessed on the same terms, and may not be prioritized accordingly.
Understandably, there have been expert calls to more explicitly measure the relationship between
climate activism and reductions in greenhouse gas emissions.8 While the large number of factors at play
may prevent us from assessing the impact of an action like attending a climate demonstration in terms
of greenhouse gas emissions, the path from voting for a pro-climate candidate to the implementation of
policy that achieves emissions reductions is more quantifiable.
In this perspective, we explore two ways of thinking about the climate impact of voting in elections and
note the implications for individual decision-making and motivating climate action. First, drawing on
carbon accounting literature, we introduce the concept of the “emissions responsibility” associated with
voting and present methods of quantifying that responsibility using a case study of the 2019 Canadian
federal election. Second, drawing on the political science literature, we describe a probabilistic method
for calculating the “expected emissions value” of casting an additional vote and use this method to
discuss the rationality of voting. We then bring these ideas together to consider the merits of “political
carbon offsets”: donating to pro-climate politicians as a way to reduce emissions. Finally, we address the
limitations of these approaches and how they could be improved in the future.
Emissions responsibility of voting
We propose emissions responsibility as a simple carbon accounting approach for calculating the
emissions associated with an individual’s votes. It follows the logic of carbon calculators which aim to
link local decisions to global greenhouse gas emissions.9 For instance, emissions from government
activities like health care or road building are sometimes divided equally amongst a population,9
whereas emissions from multi-person travel are divided amongst passengers based on the fraction of
space that passengers occupy in the vehicle or aircraft.10, 11 The thinking behind carbon footprint
calculations can be traced to Life Cycle Assessments whose methods have been codified by various
professional groups and government agencies.12, 13 While there is no standard way to apportion
emissions responsibility for political decisions, the concept of dividing responsibility for group emissions
amongst members of that group is common and uncontroversial.
Per capita emissions therefore offer an entryway to the discussion of emissions responsibility. We often
approximate both a nation’s and an individual’s responsibility for emissions by looking at per capita
emissions in that nation. While it is true that every individual in a nation requires additional energy and
resources, we would not ascribe responsibility to infants for their emissions. There are also many
individuals in a democracy with little to no say over their own emissions since their lifestyle choices (e.g.,
drive or take public transit) are constrained by the decisions made by others in the voting booth (e.g.,
the winning party’s support for new public transit initiatives). In terms of responsibility for emissions in a
democratic country, it is therefore more informative to examine emissions per voter. If we calculate
emissions per voter instead of emissions per capita, each voter takes on roughly twice the responsibility
(Figure 1), since the emissions which we previously attributed to adolescents or the disenfranchised are
now assigned to those with political agency.
Figure 1 Emissions responsibility per voter. The ten most populous Annex I nations with Democracy
Scores of six or greater shown above. Emissions responsibility per actual voter is represented by blue
bars, emissions responsibility per registered voter is represented by black points and emissions per
capita is represented by diamonds.
Quantifying emissions responsibility of a single vote
It seems reasonable that voters have higher responsibility for emissions than those who cannot vote,
but climate policy may not always be “on the ballot”. The 2019 Canadian federal election offers an
instructive case study for quantifying emissions responsibility because there was a clear divide between
the climate policies of the major political parties and because climate change was a top issue for voters
14. Leading up to the election, four of the five major parties, including the ruling majority Liberal Party,
presented platforms that would lead to modeled or stated reductions in greenhouse gas emissions. The
Conservative Party of Canada, the largest opposition party, proposed removing existing policies,
including the federal carbon tax, and enacting other policies that were widely criticized by climate
experts. Because the Liberal Party was re-elected to the most seats in Parliament and was expected to at
minimum maintain the climate policies which the second place Conservative Party promised to revoke,
we can calculate the emissions potentially saved by this electoral outcome and suggest a few ways that
responsibility for those saved emissions could be distributed among voters.
An analysis of the two front-running party platforms concluded that, by the year 2030, Canada would be
emitting 100 MtCO2e per year less under a Liberal-led government, which represents roughly 14% of
current emissions.15 Assuming linear reductions, as projected by government modelling,16 this would
result in a cumulative sum of 612 MtCO2e by 2030 (Table S2). Since the next election could either
overturn or enhance the relevant policies, we take only those projected emission reductions that would
be achieved over 3.4 years (the average duration of an elected government in Canada between the last
ten elections) and attribute them to this election (a total of 192 MtCO2e).
A cautious or conservative approach to estimating emissions responsibility of voters is to equally
apportion the emissions total to the 338 electoral districts that elect Members of Parliament, and then
give equal responsibility to every registered voter in those districts. This results in a median value of 6.7
tCO2e in reduced emissions per registered voter. This estimate is conservative because it assigns
responsibility to individuals who were registered but did not vote and to individuals whose chosen
candidate did not win in their district.
Figure 2 Distributing emissions responsibility in the 2019 Canadian federal election. Emissions
allocated only to “winning voters”, see Table S1. Values for lifestyle choices taken from Wynes and
Alternatively, we could apportion the projected emissions reductions from the election only to “winning
electoral districts” – those electoral districts where either the Liberal Party or another party favoring
progressive climate policies was successful. We then distribute emissions in each winning electoral
district only to individuals who voted for the successful candidate (Figure 2). In this approach, the
median winning voter is responsible for 34.2 tCO2e reductions, while voters in the most influential
electoral district are responsible for 228.7 tCO2e of reductions (Table S3). These estimates may be more
intuitive because they only attribute responsibility for policy decisions to voters who supported a party
advocating those policies. A wide range of values will occur in any electoral system, like that of Canada
or the United States, which has an uneven distribution of voters and or multiple parties splitting the
There are various other ways one could apportion responsibility, though most would result in values
somewhere between those we have presented. Regardless of which method we choose, for this
particular election, the potential climate responsibility of voting is higher than most individual lifestyle
decisions that the average person has the opportunity to make.5
The actual emissions impact of this or any other election cannot be known until years later, and changes
in emissions may never be directly attributable to the election’s outcome. For example, the elected
party may alter or fail to enact their platform, the proposed policies may not have the estimated impact,
or other factors may influence emissions and negate, inflate or obscure the impact of the policies. In this
case, the emissions impact of the election may be larger than those suggested by the policy proposals
announced during the campaign: in late 2020, the elected Liberal-led government released a more
aggressive set of climate initiatives, including a linear increase in the carbon tax to CDN$170/ton in
2030, which may lead to twice the emissions reduction indicated in our analysis.17 Nevertheless,
elections in which the platforms of the leading candidates could result in very different climate
outcomes have been common in recent years, including the national elections in the United Kingdom in
2019, Brazil in 2018, and the United States in 2016 and 2020. Note that in the case of the United States
in 2020, we did not attempt an analysis due to differences in the political systems that create large
uncertainties concerning the Biden administration’s ability to pass climate legislation (these include the
greater likelihood of court challenges and the antimajoritarian US legislative system).
The benefit of engaging in this type of thinking is that it provides an intuitive way for people to
understand their share of climate responsibility during an election. Some people may believe that all
eligible voters share equal responsibility for an outcome, and some may believe that only a subset of
voters with a controlling influence over the result (winning voters) share responsibility. But if the
emissions responsibility is comparable to that of high-impact lifestyle choices in both cases then we
should regard the responsibility of voting to merit the same level of moral consideration.
Individuals may have a greater responsibility for the climate impact of their vote than for the climate
impact of their consumer choices, but that does not necessarily indicate that their vote causes more
emissions than their consumer behavior. The emissions responsibility approach is similar to that of a
standard air travel carbon calculator. If an air traveler opts to stay home they would actually reduce
emissions less than estimated by a carbon calculator, because the flight is likely to still proceed, and the
effect of their absence on fuel use is small or negligible. Similarly, if a single voter opts to stay home,
their missing vote is likely to have less of an emissions impacts than estimated by the emissions
responsibility calculation. A more marginal approach would ask, “what is the emissions impact of this
entire election, and what is the probability that my vote happens to be the one that causes one party to
The next section examines how our thinking about elections changes if we approach the problem from
this instrumentalist perspective. By synthesizing evidence from political science and carbon accounting,
we describe a way to evaluate the causal value of a pro-climate voter’s choice. We begin by applying this
thinking to the question of the rationality of voting and then consider whether political donations are a
viable alternative to carbon offset projects for certain pro-climate individuals.
Is it rational for a pro-climate citizen to vote?
There has been considerable debate as to why citizens spend time and energy casting votes that result
in very little personal benefit. Voters may incur costs to cast a ballot because of a sense of civic duty,18
because of pride and social norms19, 20 or because of a desire for self-expression.21 Setting these motives
aside, scholars are undecided as to whether even self-interested citizens can rationally engage in the act
Often a financial test is used to help evaluate whether the choice is rational.22, 23 The idea makes use of
the expected value calculation. In gambling if the expected value of a game is positive, then it is rational
for a gambler to play because over a large number of attempts, the gambler can expect to gain money.24
Lotteries are an interesting example of expected value in gambling because, though the odds of winning
are astronomically low, the payoff is also quite substantial.25 We can use the same math to evaluate
whether participation in a lottery is advisable, finding that, except in rare cases, it is not.26-29 Instead, the
expected value of a lottery ticket is usually about half the cost of participating.30 Elections are similar to
lotteries in that the likelihood of one person’s vote “making a difference” by serving as the deciding vote
is also quite low but the consequences of an election can be considerable.
Based on this logic, scholars have argued that voting is a rational act for those voters who wish to
increase social well-being instead of maximizing selfish gains.22, 31-33 In larger elections where a voter is
less likely to cast a pivotal vote, there is greater potential for social good that can compensate for these
lowered odds.34 The same relationship might be true for climate mitigation – the larger the number of
voters, the larger the potential difference in terms of greenhouse gas emissions. Since it is possible to
estimate both a range of probabilities for a pivotal election, and a range of potential outcomes for the
difference in emissions attributable to an election, we can ask whether an ardent climate activist is
rationally justified in voting.
Let us assume that there is a group of voters who are willing to vote altruistically, and whose sole aim in
voting is to mitigate climate change. Such voters may engage in actions that have climate costs, such as
driving to a voting booth, hoping that the costs will be paid off in the event that they cast a deciding
vote which results in more ambitious climate legislation. We can adapt an expected value calculation to
see whether it is a rational choice for a voter to drive to the polling station in order to cast such a vote. A
basic model of this expected value calculation can be described by:
where R represents the reward gained from voting, B represents the benefit gained by the preferred
candidate winning, P represents the probability of casting a deciding vote and C represents the cost of
voting.35 For a climate voter seeking to maximize emissions reductions, R would represent the balance of
emissions, where a positive value would justify the act of voting, B would represent the benefit in terms
of the expected emissions reduced through the policies of the preferred candidate winning, P is the
probability of casting a decisive vote, and C is the cost, in emissions, of casting a vote. Next, we consider
ways to estimate P.
Estimating the likelihood of a pivotal election
Most actual estimates of P have focused on the United States, either at the national or the state level,1,
36, 37 but methods for calculating this probability vary. If a forecast based on opinion polling data is
available, one can use that data to estimate the likelihood that the election will be decided by a single
vote. Assuming a forecast with low standard error for a very close election (where two candidates are
expected to receive an equal share of the vote), an upper bound for P has been estimated at 20/n,
where n is the number of voters37. Since most elections have much larger margins between candidates,
we can expect most probabilities to be substantially lower.
In fact, researchers have looked at historical records to give an indication of how often pivotal votes
occur. One analysis found that of 16,577 US elections, only one was decided by a single vote and in
40,036 US state elections, two elections were tied and seven were decided by only one vote36. These
databases can be used to crudely approximate the probability of a pivotal vote by taking the empirical
frequency, x/N , where x is the number of elections that the winner’s vote tally exceeds the loser’s by no
more than a single vote, and N is the number of elections in the database. Because the number of
pivotal elections is so rare, researchers can improve on these estimates by incorporating additional data
such as the frequency of close but not pivotal elections, as in those within 10 or 100 votes36; for
example, in the 2020 US election, one congressional race with almost 400,000 votes cast was decided by
only six votes38. Using such methods, the probability of a pivotal election is estimated to be
proportionate to the size of the electorate according to n-1 or more precisely, n-0.9.39 As might be
expected, the larger the electorate, the less likelihood there is of a single individual casting a decisive
vote. We assume the more conservative relationship of n-1 and use Equation 2 from Gelman, King and
Boscardin37 and Margolis40 which allows us to take the size of the electorate (n) the forecast share of the
vote (μ) and the standard deviation (τ) to approximate the probability of casting a decisive vote.
Figure 3 shows the likelihood of casting a decisive vote in a range of election scenarios for an electorate
with size n=100,000.
Figure 3 Probability of casting a pivotal vote for a generic election with n=100,000 voters. Blue line
represents high level of forecast certainty (SD=0.02), grey is medium certainty (SD=0.03) and yellow is
low certainty (SD=0.05).
In order for a voter to rationally cast a ballot in hopes of lowering net greenhouse gas emissions, the
cost of casting a ballot (in terms of greenhouse gas emissions) must be lower than the probability of
casting a decisive ballot multiplied by the expected emissions reduced by the preferred candidate
winning. Setting the reward, R, to 0 and rearranging (1) we have:
Even in a rural riding, it would be reasonable to expect a voter to be within 10 miles of a voting
precinct.41 Taking a few vehicles that exemplify a range of carbon intensities (a Tesla, a Honda Civic and
a Ford F-150) and multiplying their fuel efficiencies by the range we find the cost in emissions (C) (Table
We can then take an electorate size equal to n and solve for the amount of emissions (Bmin) that would
need to be reduced by the preferred candidate’s victory and subsequent policies in order to make the
act of voting rational for our climate voter. Doing so shows that the scale of the emissions required by
various example electorates are achievable in most close elections (Figure 4). For instance, even in the
case of a blowout election where the forecast share of votes is 60% for one candidate (roughly Joe
Biden’s share of the vote in New York State in 2020), the total Bmin necessary to make voting rational in a
highly polluting pickup truck is just 0.09%, 0.3% and 1.5% of one year of Cleveland’s, Chicago’s and
France’s respective emissions. To give an idea of the magnitude of change engendered by a typical
climate policy, a single new national climate law can be expected to reduce emissions by 0.78% per unit
of GDP, with that number growing after three years.42 Since policies usually last for multiple years, there
is a good chance that it is rational to vote in elections that are not blowouts provided that one candidate
or party offers to implement meaningful climate policy while the other does not. However, for extreme
cases, such as an election where one candidate is forecast to receive 80% of the vote with high
certainty, the improbability of casting a decisive vote becomes so high that voting is no longer rational
from an instrumentalist perspective. Note that Bmin is less sensitive to distance travelled than to
forecasted vote share since Bmin scales linearly with emissions from the drive but increases exponentially
as the forecasted share of the vote diverges from 50% (see sensitivity analysis Figures S1 to S3).
Figure 4 Emissions benefit of election (Bmin) required to rationally justify voting versus forecasted
share of the vote for the leading candidate. Analysis is based on a return trip of ten miles each way to
the voting booth, and an election forecast with reasonable certainty (SD=0.03). Red dashed line shows
0.1% of annual emissions for each jurisdiction.
Nations with electoral systems that assign unequal voting power to different individuals may hold more
elections where participation is not rational. For instance, the United States Electoral College devalues
the voting power of individuals in states far from the national median partisanship.1 This makes the
likelihood of casting a pivotal vote for a Presidential candidate vanishingly small for many voters, and
disproportionately likely for others (Table 1). However, in the case of US elections, participation may still
be rationalized due to the opportunity to vote for candidates in other races or in ballot initiatives.
Table 1: Emissions reductions needed to justify voting in US States
Necessary reductions from winning candidate (MtCO2e)
Note: US emissions in 2017 totaled 5742 MtCO2e. Probabilities based on a forecast by the Economist
taken 15 days prior to the election. Shows a range of states including most and least likely to cast a
pivotal vote. Data for all states can be found in Table S5.
Not every election will offer a clear range of choices to our hypothetical climate voter. The leading
candidates may have identical climate policies, or a set of policies whose emissions cannot be easily
projected or even guessed. Still, as noted previously, there have been many recent elections where
voters can choose between a candidate with rigorous climate policies and a candidate without. In such
instances it is likely that even driving a high-emitting vehicle to cast a vote for a climate leader would be
a rational decision.
The viability of political carbon offsets
For climate-concerned actors who are not constrained by a corporate directive to remain apolitical,
there are strong arguments in favour of expending time and resources to change systems rather than
focusing on discrete projects. For instance, a philanthropist could purchase a large plot of land and plant
trees on it, or they could work to elect a pro-climate government who could plant the same number of
trees, but also implement an electric vehicle incentive and a price on carbon. This reasoning, if applied
to the average citizen, might imply that donating to a political campaign could lead to larger emissions
reductions than purchasing carbon offsets. The question is whether campaign contributions influence
election outcomes efficiently enough to outweigh purchasing offsets.
We can examine this question using both the emissions responsibility and the expected emissions value
approaches to thinking about the climate impact of voting. First, we previously reported in the emissions
responsibility case study that the reelection of the Liberal Party in Canada in 2019 is expected to result
in savings of 192 MtCO2e compared to the next most likely alternative. Table 2 shows the cost of 192 Mt
of emissions using carbon prices from a variety of different approaches, including the cost of offsets, a
carbon tax, a cap and trade scheme and modelled estimates of the cost of the impacts of carbon
emissions on society.
Table 2: Different measures of cost per tonne of CO2
Cost per tonne
Cost of 192 MtCO2e
(in billion USD)
Offsetters General Portfolio
Canada carbon tax
US social cost of carbon
The maximum spending limit on all political parties combined for the 2019 Canadian federal election
was $175 million43, about 1/15th of the cost of the carbon emissions outcome by the lowest measure
(Table 2). While spending huge sums on an election does not guarantee a victory, there is strong
evidence that candidates who spend more increase the likelihood of winning.44, 45 This suggests that for
elections where large quantities of emissions are at stake, as they were in Canada in 2019, and where
campaign contributions are able to shift outcomes, political donations could be competitive with offsets
in terms of efficacy.
But to answer this question using the expected emissions value of a vote approach, we would ideally
want to know how much emissions can be saved through electing climate leaders and the relationship
between campaign spending and increased likelihood of winning an election. At present, neither of
these values can be determined with great precision. Still, to provide an initial exploration of the
question we take the approach of using conservative measures and comparing these to the alternative
of using carbon offsets.
A review of the “get out the vote” literature found that campaigns can increase voter participation at
the cost of anywhere from $18-$137 per added vote.46 But, from the perspective of a donor, some
fraction of a donation will end up going to administration instead of advertising or get out the vote
efforts. For generic campaign spending, the cost across four studies of adding one vote to a candidate’s
final vote margin for an incumbent was $15-$367 and $12-$110 for a challenger.44, 47 We take $110 per
vote as the maximum value from the campaign spending literature for challengers and convert into
present day dollars ($175). The choice to use challenger spending is justified since some of the previous
research on incumbent spending assumes that incumbents optimize vote share instead of probability of
re-election,44 which would lead to higher cost per vote for incumbents. The $175 estimate is also
relatively close to the upper estimate from the get out the vote literature, which has the advantage of
drawing on a broader range of elections (from local to national). It also assumes that the donation does
not prompt additional fundraising or spending from the opposing candidate.
Next, as an example of how we could apply the expected emissions value calculation in the real world,
we consider the Washington 1631 Ballot Initiative, where 3 million voters decided whether to enact a
carbon tax in their state. Though ballot initiatives operate under circumstances where the effects of
campaign financing are less understood than in typical elections,48, 49 we assume that the cost of
generating a vote is the same to take advantage of other useful features of the case study: voters knew
the issue was centered on climate change and there is a strong causal link between the voters’ choices
and the implementation of a reasonably effective climate policy.
Applying Equation 2 to polling data preceding the election50 we find the probability of casting a decisive
vote was 2.7x10-6 (see Supplemental Experimental Procedures). We can then compare the $175/vote
cost of adding another vote to a generic campaign with the cost of carbon offsets using an expected
value calculation (similar to Equation 1), which relates the emissions reductions associated with a single
vote (R) to the total emissions benefit of a candidate or party winning (B) and the probability of casting a
deciding vote (P). The expected reward from voting (R) must be at least 12.5 tCO2e to be comparable to
a carbon offset, since $175 would purchase that much from the low-cost carbon offset (Table 2):
Substituting for the probability of a pivotal election and rearranging we find that the ballot initiative
would need to result in reductions of 4.6 MtCO2e over the course of its implementation. To put this into
perspective, Washington State emits 97.5 MtCO2e every year, so achieving 4.6 MtCO2e would require
reducing statewide emissions by one percent for five years. Studies of carbon prices implemented in
other jurisdictions have found reductions of 2%,51 3.8%52 and 5-15%53 annually, suggesting that such a
reduction is feasible. If we assume the minimum reported effectiveness of the carbon price (2%) and
that it is in place for only five years, the expected emissions value of a winning vote is 25 tCO2e, and a
donation to the winning campaign would have an expected efficiency of 7$/tonne, less than half the
price of the cheapest carbon offset (Table 2).
This outcome, though specific to the ballot initiative, is reasonably robust to changes in the forecast (see
Figure S4 for a sensitivity analysis) and demonstrates the potential value of political donations in
elections where a large quantity of emissions is at stake and the forecast is either close or the forecast
uncertainty is high. In such cases, the only other influential variable is the size of the electorate. If the
electorate is larger, then the chance of influencing the election with a donation is lower, but the impact
of the election in terms of total emissions will likely be greater.
The Washington I-1631 ballot initiative was not successful and those who donated to that cause did not
directly reduce emissions with their efforts. But the concept is probabilistic: a large amount of donations
in a large amount of elections will reduce emissions (barring a “gambler’s ruin” situation where there
are too few pro-climate electoral victories to recoup costs in a limited number of election opportunities
28). And while our estimates of efficacy are subject to large uncertainty, so too are carbon offsets;
development projects do not always achieve the expected results54 and forest carbon offsets may fail to
generate real, verifiable emissions reductions for a host of reasons55. Since the goal of these donations is
to impose policies that reduce emissions in one’s own nation, “political carbon offsetting” also avoids
the moral quandaries of paying less wealthy jurisdictions to make emissions cuts on behalf of polluting
industries and individuals in developed nations.56
The additional benefits of electing climate leaders are substantial since changes wrought by government
policy can extend beyond calculable emissions reductions. For example, the policies which stimulated
market growth of photovoltaics were responsible for worldwide reductions in the cost of solar energy,57
allowing for further growth of that market and more emissions reductions than what were achieved
through the original policy. On the international level, when high-emitting countries achieve ambitious
reductions it increases the willingness of other countries to contribute.58 And targeting elections where
agreed upon social tipping elements such as fossil fuel or renewable subsidies are at play increases the
likelihood of achieving transformational change.59
Given these perspectives on the climate impact of voting and political action, how should everyday
citizens or motivated philanthropists make effective political donations? All else being equal, investing in
races in which the polling is close should be more advantageous since there is more chance of the
donation swinging the election. Additionally, early research found incumbent spending to be less
advantageous than challenger spending,47, 60, 61 though later studies44, 62 have convincingly questioned
conventional wisdom on this point. Still, incumbents are more likely to use their finances to pay off
debts or support other candidates45, so challengers remain a more efficient investments from the
viewpoint of donors.
Donors would also be wise to avoid the diminishing returns associated with giving to already well-
funded campaigns. Evidence from senate elections62, state house elections63, judicial elections64, and
ballot initiatives65, confirms the logic that initial spending which raises awareness adds more value than
repeated contact with voters already exposed to campaign outreach. Based on the advantages of
supporting underfunded challengers, donating to candidates attempting to supplant climate laggards in
United States Democratic Primaries might be one way to scale up ambition for climate change in the
Some of the criteria listed above for an efficient political donation are correlated, making it less likely
that all conditions for an ideal donation can be satisfied. For instance, politicians increase spending in
close elections,66 so it may be difficult for a donor to identify close elections where spending is low. For
small donors hoping to give effectively, identifying races that fit the multi-dimensional criteria suggested
above could therefore be challenging. Instead best practice for political carbon offsetting might simply
be achieved by giving to organizations who aim to elect climate-concerned legislators and relying on
their expertise to allocate funding. This would also reduce the burden of evaluating which candidates
are genuinely intending to reduce emissions. Donors can also add value to their efforts by openly
discussing their campaign contributions, as this type of giving has been demonstrated to be contagious
.67 Finally, volunteering can be viewed as a substitute for donations, since the efforts of a volunteer at a
get out the vote drive would be equivalent to a campaign paying for that same labor.
By drawing on the literature on carbon accounting and political science, this Perspective introduces new
ways of conceptualizing the climate impact of voting. The case studies serve as illustrative examples and
should not be seen as precise estimates of the climate impact of voting. The cost of generating a vote in
a campaign will always vary based on the specifics of the campaign in question. In the future we may
have a better understanding of how that value changes in different circumstances. The case studies
nonetheless suggest that the responsibility that individuals have for emissions under elections can be
comparable to the responsibility that they may incur for emissions in their everyday lives.
Each of the two proposed ways of conceptualizing the climate impacts of voting has its strengths and
weaknesses. One potential critique of the emissions responsibility of voting as a form of personal carbon
accounting is that it will result in double-counting. An individual might consider themselves to be
responsible for saving a certain amount of emissions by voting in favour of a renewable energy standard,
and then later credit themselves for using less home energy emissions now that their electricity has a
lower carbon intensity. One could avoid the double counting issue by treating emissions responsibility
for voting as an alternative metric, and not a complementary metric to carbon footprints from lifestyle
change. For instance, individuals could be motivated to consider the active reductions they can
engender in the world (“carbon handprint”) as opposed to their lifestyle emissions (carbon footprint).
We acknowledge that introducing a concept of emissions responsibility for voting could lead to a type of
moral licensing where individuals would then feel emboldened to increase emissions elsewhere, but it is
not the same thing as a carbon accounting error with verifiable consequences. On that note, there is
reason to question the utility of introducing these ideas into the public. Individualizing group action may
make certain behaviors appear less important and thereby actually decrease motivation. Conversely,
since some members of the public do not consider voting to be an effective way to reduce emissions,68
communicators could motivate voters with an emissions responsibility analysis that demonstrates what
is “at stake”. Future climate communication research should test the efficacy of emission responsibility
analyses in application to voting behavior.
The more probabilistic approach to calculating the emissions associated with voting suggests that
climate voters are rationally justified in participating in many elections, and also that those wanting to
fight climate change with donations could give serious consideration to political efforts, instead of
limiting themselves to traditional carbon offsets. Introducing this type of probabilistic thinking to those
with the wealth to contribute to climate mitigation projects offers a way to justify investing in
candidates with the leverage to achieve greater results than private investments can create on their
Further information and requests for resources should be directed to and will be fulfilled by the
corresponding author, Seth Wynes (email@example.com).
This study did not generate new unique materials
Data and code availability
The datasets and code underpinning this research are available in the Supplemental Information or in
the Concordia University Dataverse: https://doi.org/10.5683/SP2/9B1IO3
Thanks to George Hoberg and Ludovic Rheault for helpful suggestions on an earlier draft of this paper.
S.W. conceived of the presented idea, performed calculations and drafted the manuscript. M.M. and
S.D.D. contributed to the final manuscript. S.D.D. supervised the work.
1. Gelman, A., Silver, N., Edlin, A. (2012). What is the probability your vote will make a difference?
Econ Inquiry. 50, 321-326.
2. Dowding, K. (2005). Is it rational to vote? Five types of answer and a suggestion. The British
Journal of Politics and International Relations. 7, 442-459.
3. Neumayer, E. (2003). Are left-wing party strength and corporatism good for the environment?
Evidence from panel analysis of air pollution in OECD countries. Ecolog Econ. 45, 203-220.
4. Dietz, T., Frank, K.A., Whitley, C.T., Kelly, J., Kelly, R. (2015). Political influences on greenhouse
gas emissions from US states. Proceedings of the National Academy of Sciences. 201417806.
5. Wynes, S., Nicholas, K. (2017). The Climate Mitigation Gap: Education and Government
Recommendations Miss the Most Effective Individual Actions Environmental Research Letters. 12.
6. Ecosystem Marketplace. Unlocking potential: State of the voluntary carbon markets 2017.
Washington, DC: Forest Trends; 2017. https://www.forest-trends.org/wp-
7. Luscombe, R. Amazon's Jeff Bezos pledges $10bn to save Earth's environment. The Guardian.
8. Fisher, D.R., Nasrin, S. (2021). Climate activism and its effects. Wiley Interdisciplinary Reviews:
Climate Change. e683.
9. West, S.E., Owen, A., Axelsson, K., West, C.D. (2016). Evaluating the use of a carbon footprint
calculator: communicating impacts of consumption at household level and exploring mitigation options.
Journal of Industrial Ecology. 20, 396-409.
10. BEIS. (2016). Greenhouse gas reporting - Conversion factors 2016.
11. ICAO. ICAO Carbon Emissions Calculator Methodology Version 11. International Civil Aviation
Organization; 2018. https://www.icao.int/environmental-protection/CarbonOffset/Pages/default.aspx.
12. Wiedmann, T., Minx, J. (2008). A definition of ‘carbon footprint’. Ecological economics research
trends. 1, 1-11.
13. Pandey, D., Agrawal, M., Pandey, J.S. (2011). Carbon footprint: current methods of estimation.
Environ Monit Assess. 178, 135-160.
14. Bricker, D. (2019). Four Weeks In, Climate Change is Fastest-Moving (29%, +4), but Health Care
(35%) Still Top Issue to Make a Difference at the Ballot Box. Ipsos. https://www.ipsos.com/en-ca/news-
15. Jaccard, M. (2019). Emissions will rise under Conservative climate plan. Policy Options.
16. Environment and Climate Change Canada. (2020) Canada's 4th Biennial Report to the United
Nations Framework Convention on Climate Change (UNFCCC)
17. Environment and Climate Change Canada. (2020). Modelling and analysis of a healthy
environment and a healthy economy.
18. Blais, A., Gidengil, E., Nevitte, N. (2004). Where does turnout decline come from? Eur J Polit Res.
19. Gerber, A.S., Rogers, T. (2009). Descriptive social norms and motivation to vote: Everybody's
voting and so should you. The Journal of Politics. 71, 178-191.
20. Panagopoulos, C. (2010). Affect, social pressure and prosocial motivation: Field experimental
evidence of the mobilizing effects of pride, shame and publicizing voting behavior. Political Behavior. 32,
21. Brennan, G., Hamlin, A. (1998). Expressive voting and electoral equilibrium. Public Choice. 95,
22. Edlin, A.S., Gelman, A., Kaplan, N. (2008). Vote for Charity's Sake. The Economists' Voice. 5.
23. Usher, D. (2014). An alternative explanation of the chance of casting a pivotal vote. Rationality
and Society. 26, 105-138.
24. Bjerg, O. (2010). Problem gambling in poker: Money, rationality and control in a skill-based
social game. International Gambling Studies. 10, 239-254.
25. Ariyabuddhiphongs, V. (2011). Lottery gambling: A review. J Gambl Stud. 27, 15-33.
26. Matheson, V.A. (2001). When are state lotteries a good bet (revisited)? Eastern Econ J. 27, 55-
27. Thaler, R.H., Ziemba, W.T. (1988). Anomalies: Parimutuel betting markets: Racetracks and
lotteries. J Econ Perspect. 2, 161-174.
28. Chernoff, H. (1981). How to beat the Massachusetts numbers game. The Mathematical
Intelligencer. 3, 166-172.
29. MacLean, L.C., Ziemba, W.T., Blazenko, G. (1992). Growth versus security in dynamic investment
analysis. Management Science. 38, 1562-1585.
30. Clotfelter, C.T., Cook, P.J. (1990). On the economics of state lotteries. J Econ Perspect. 4, 105-
31. Jankowski, R. (2002). Buying a lottery ticket to help the poor: Altruism, civic duty, and self-
interest in the decision to vote. Rationality and Society. 14, 55-77.
32. Fowler, J.H. (2006). Altruism and turnout. The Journal of Politics. 68, 674-683.
33. Jankowski, R. (2007). Altruism and the decision to vote: Explaining and testing high voter
turnout. Rationality and Society. 19, 5-34.
34. Edlin, A., Gelman, A., Kaplan, N. (2007). Voting as a rational choice: Why and how people vote to
improve the well-being of others. Rationality and society. 19, 293-314.
35. Riker, W.H., Ordeshook, P.C. (1968). A Theory of the Calculus of Voting. Amer Polit Sci Rev. 62,
36. Mulligan, C.B., Hunter, C.G. (2003). The empirical frequency of a pivotal vote. Public Choice. 116,
37. Gelman, A., King, G., Boscardin, W.J. (1998). Estimating the probability of events that have never
occurred: when is your vote decisive? J Amer Statistical Assoc. 93, 1-9.
38. Foley, R.J. Iowa board certifies 6-vote Republican win in US House race. Associated Press. 2020.
39. Gelman, A., Katz, J.N., Bafumi, J. (2004). Standard voting power indexes do not work: an
empirical analysis. British Journal of Political Science. 34, 657-674.
40. Margolis, H. (1977). Probability of a tie election. Public Choice. 31, 135-138.
41. Gimpel, J.G., Schuknecht, J.E. (2003). Political participation and the accessibility of the ballot
box. Political Geography. 22, 471-488.
42. Eskander, S.M., Fankhauser, S. (2020). Reduction in greenhouse gas emissions from national
climate legislation. Nature Climate Change. 1-7.
43. Elections Canada. (2020). Final Election Expenses Limits for Registered Political Parties.
44. Gerber, A.S. (2004). Does campaign spending work? Field experiments provide evidence and
suggest new theory. Am Behav Sci. 47, 541-574.
45. Schuster, S.S. (2020). Does Campaign Spending Affect Election Outcomes? New Evidence from
Transaction-Level Disbursement Data. The Journal of Politics. 82, 1502-1515.
46. Green, D.P., Gerber, A.S. (2019) Get out the vote: How to increase voter turnout (Brookings
47. Levitt, S.D. (1994). Using repeat challengers to estimate the effect of campaign spending on
election outcomes in the US House. J Polit Economy. 102, 777-798.
48. Stratmann, T. (2010) Campaign spending and ballot measures. In Financing referendum
campaigns (Springer) pp. 9-22.
49. Smith, D.A. (2010). US States. In Financing referendum campaigns (Springer) pp. 39-61.
50. Cascade Public Media (2018), Crosscut Ellway Poll.
51. Bruvoll, A., Larsen, B.M. (2004). Greenhouse gas emissions in Norway: do carbon taxes work?
Energy Policy. 32, 493-505.
52. Bayer, P., Aklin, M. (2020). The European Union emissions trading system reduced CO2
emissions despite low prices. Proceedings of the National Academy of Sciences. 117, 8804-8812.
53. Murray, B., Rivers, N. (2015). British Columbia’s revenue-neutral carbon tax: A review of the
latest “grand experiment” in environmental policy. Energy Policy. 86, 674-683.
54. Wang, Y., Corson, C. (2015). The making of a ‘charismatic’carbon credit: clean cookstoves and
‘uncooperative’women in western Kenya. Environ Planning A. 47, 2064-2079.
55. St-Laurent, G.P., Hagerman, S., Hoberg, G. (2017). Barriers to the development of forest carbon
offsetting: Insights from British Columbia, Canada. J Environ Manage. 203, 208-217.
56. Hyams, K., Fawcett, T. (2013). The ethics of carbon offsetting. Wiley Interdisciplinary Reviews:
Climate Change. 4, 91-98.
57. Kavlak, G., McNerney, J., Trancik, J.E. (2018). Evaluating the causes of cost reduction in
photovoltaic modules. Energy Policy. 123, 700-710.
58. Sælen, H., Hovi, J., Sprinz, D., Underdal, A. (2020). How US withdrawal might influence
cooperation under the Paris climate agreement. Environ Sci Policy. 108, 121-132.
59. Otto, I.M., Donges, J.F., Cremades, R., Bhowmik, A., Hewitt, R.J., Lucht, W., et al. (2020). Social
tipping dynamics for stabilizing Earth’s climate by 2050. Proceedings of the National Academy of
Sciences. 117, 2354.
60. Jacobson, G.C. (1985). Money and votes reconsidered: congressional elections, 1972–1982.
Public Choice. 47, 7-62.
61. Jacobson, G.C. (1978). The effects of campaign spending in congressional elections. Amer Polit
Sci Rev. 72, 469-491.
62. Gerber, A. (1998). Estimating the effect of campaign spending on senate election outcomes
using instrumental variables. Amer Polit Sci Rev. 401-411.
63. Stratmann, T. (2006). Contribution limits and the effectiveness of campaign spending. Public
Choice. 129, 461-474.
64. Bonneau, C.W., Cann, D.M. (2011). Campaign spending, diminishing marginal returns, and
campaign finance restrictions in judicial elections. The Journal of Politics. 73, 1267-1280.
65. De Figueiredo, J.M., Ji, C.H., Kousser, T. (2011). Financing direct democracy: Revisiting the
research on campaign spending and citizen initiatives. The Journal of Law, Economics, & Organization.
66. Erikson, R.S., Palfrey, T.R. (2000). Equilibria in campaign spending games: Theory and data. Amer
Polit Sci Rev. 595-609.
67. Traag, V.A. (2016). Complex contagion of campaign donations. PLoS One. 11, e0153539.
68. Wynes, S., Zhao, J., Donner, S.D. (2020). How well do people understand the climate impact of
individual actions? Clim Change. 162, 1521-1534.