Energy Policy 140 (2020) 111385

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Attitudes towards carbon taxes across Europe: The role of perceived

uncertainty and self-interest
Resul Umit ∗, Lena Maria Schaffer
University of Lucerne, Frohburgstrasse 3, Postfach 4466, 6002 Lucerne, Switzerland

ABSTRACT

While using carbon taxes to reduce greenhouse gas emissions may well be effective, this has recently proved too unpopular to put into practice in a number of
countries. Yet, at a time when governments across the world are preparing their nationally determined contributions to the Paris Agreement, our knowledge of
whether and why people oppose these taxes originates from a single or small number of cases. Drawing on the European Social Survey (n = 44,387), this article
provides evidence on public attitudes towards increasing taxes on fossil fuels to reduce climate change from 23 countries, most of which have never featured
in the literature before. The results point to a widespread aversion to carbon taxes. On the one hand, this worsens with the perceived costs of taxes, such as
the case among consumers who depend highly on energy. On the other, it improves with political trust and external political efficacy—factors that help ease
the uncertainty around policy proposals. Our estimations suggest that the effect of changes in these factors alone would be large enough to reverse the public
resistance to carbon taxes in some countries. These results are robust to a number of alternative specifications and various checks.

1. Introduction State. The proposal to increase the taxes was abandoned in France as
well, amid the violent ‘yellow vest’ protests against the policy (Douenne
Carbon taxes offer an effective choice of instrument for climate and Fabre, 2019).
change policy (Stern and Stiglitz, 2017; Goulder and Parry, 2008). By
As a result, governments committed to fighting climate change
increasing the cost of activities detrimental to the environment, taxes
face a difficult decision between effectiveness and popularity, and—as
can lead to changes in the behaviour of masses for the better—towards
public opinion shapes policy (Shapiro, 2011)—the benefits of carbon
innovative and greener alternatives that reduce pollution (Aldy and
taxes might remain outweighed in the near future. Understanding the
Stavins, 2012). When the French government proposed to increase the
resistance to carbon taxes could offer other ways out of this problem,
taxes on fossil fuels in 2019, for example, the plan was to help reduce
and there is an urgency to know the determinants of individual atti-
carbon emissions dramatically and facilitate the transition to electric
tudes to carbon taxes. This urgency has led to many studies on the
cars. Moreover, the revenues raised from these taxes offer further
opportunities for governments, such as the ability to finance specific public attitudes to carbon taxes in the last decade or so,2 but the
environmental purposes (Kallbekken and Aasen, 2010) or to reduce existing evidence comes from a limited number of countries and/or
other taxes for a stronger and fairer economy (Ballard and Medema, from studies that do not seek nationally representative samples—such
1993). Hence there are good reasons why governments may consider as interviews, focus groups, and laboratory experiments. In this study,
using taxes to reduce greenhouse gas emissions while preparing their we heed the call to cover more cases (Fairbrother, 2017a, p. 6),
nationally determined contributions to the Paris Agreement. and contribute to this influential literature by analysing cross-national
Yet taxes are not a popular choice—at least, not as much as their individual preferences with data from 23 countries, most of which are
effectiveness would suggest. For example, as of April 2019, carbon yet to feature in the literature.
taxes are implemented at national level in as few as 24 countries The main results are three-fold. First, there is a widespread aversion
worldwide (World Bank, 2019). Many see the public resistance to taxes to carbon taxes in Europe. If governments are to introduce new taxes,
as the main reason behind the reluctance of governments to use this as things stand they will face public resistance almost everywhere—
policy instrument (Carattini et al., 2018b),1 pointing to the numerous although the unpopularity of taxes varies between individuals and
occasions where tax proposals were rejected in popular votes, such as countries. Second, we find the support for taxes improves significantly
in Switzerland (Thalmann, 2004; Carattini et al., 2017) or Washington with individuals’ political trust and efficacy. Third, however, it worsens

∗ Corresponding author.
E-mail addresses: resul.umit@unilu.ch (R. Umit), lena.schaffer@unilu.ch (L.M. Schaffer).
1
Negative attitudes are also common among the industry (Clinch and Dunne, 2006; Dresner et al., 2006; Klok et al., 2006), which contributes to the resistance
to taxes (Farrell, 2016).
2
For recent reviews of this literature, see Maestre-Andrés et al. (2019) and Carattini et al. (2018b).

https://doi.org/10.1016/j.enpol.2020.111385
Received 14 March 2019; Received in revised form 28 January 2020; Accepted 22 February 2020
Available online 9 March 2020
0301-4215/© 2020 Elsevier Ltd. All rights reserved.
R. Umit and L.M. Schaffer Energy Policy 140 (2020) 111385

among people who depend highly on energy. Substantively, their po- Another group of reasons behind the public resistance relates to
tential effects on the public attitudes to carbon taxes are large enough the perceived economic consequences of taxes. First and foremost,
to be decisive at least in some countries. people are worried that carbon taxes will increase their personal costs
The remainder of the article proceeds as follows. The following (Douenne and Fabre, 2019; Brännlund and Persson, 2012; Jagers and
section reviews the related literature, showing that concerns for the Hammar, 2009; Clinch and Dunne, 2006). Studies repeatedly show that,
environmental and economic consequences of taxes drive the public for example, the opposition to taxes is significantly higher among car
resistance. Section 3 then details five hypotheses based on a rational- owners (Hsu et al., 2008; Gaunt et al., 2007; Hammar and Jagers,
choice perspective. We briefly introduce the data, main variables, and 2006; Thalmann, 2004), who are more likely to see their costs increase
the methods in Section 4, leaving the fine details to Supporting Informa- as a result of carbon taxes. Similarly, there is a negative relationship
tion. Section 5 presents the descriptive analysis and multivariate tests between tax rates and public support (Baranzini and Carattini, 2017;
of the hypotheses. Finally, the paper concludes with remarks on why Gevrek and Uyduranoglu, 2015; Cherry et al., 2012; Kallbekken and
the results of these tests might be politically significant. Sælen, 2011; Thalmann, 2004) as people prefer taxes that would cost
them as little as possible. However, calculating the personal costs of
2. Literature review taxes is not always an easy task, and people seem to overestimate the
personal costs associated with these taxes (Douenne and Fabre, 2019;
There is a growing interest—scholarly or otherwise—on the public Schuitema et al., 2010)—unlike their environmental benefits which, as
attitudes to carbon taxes, which reflects the importance of social accep- discussed above, are often underestimated.
tance for this policy instrument to be successfully implemented. How- Then again, the worry about the economic consequences of taxes
ever, what we already know points to a challenge for decision makers might go beyond self-interest, and to a certain extent it is also based
looking to implement carbon taxes to reduce climate change: against on the consideration for ‘others’, such as low-income groups (Carattini
the overwhelming evidence of their superior effectiveness (Weitzman, et al., 2017; Kallbekken and Sælen, 2011). For example, Carattini et al.
2017; Baranzini and Carattini, 2017; Aldy and Stavins, 2012; Mankiw, (2017) show that the vote against the non-renewable energy tax in
2009; Metcalf, 2009; Goulder and Parry, 2008; Ballard and Medema, the 2015 Swiss referendum can be explained largely by its perceived
1993; Baumol and Oates, 1971), the public support for carbon taxes repercussions on low-income households and businesses. In general,
remain low (Douenne and Fabre, 2019; Carattini et al., 2017; Jagers the acceptability of carbon taxes decreases as they lead to regressive
and Hammar, 2009; Dietz et al., 2007)—lower than the level of support distributions (Berry, 2019; Gevrek and Uyduranoglu, 2015; Brännlund
for other policy instruments, such as subsidies (Heres et al., 2017; and Persson, 2012), posing a higher share of the burden to low-income
Cherry et al., 2012; De Groot and Schuitema, 2012; Kallbekken and rather than high-income earners.
Aasen, 2010; Steg et al., 2006) or regulations (Clinch and Dunne, 2006;
Deroubaix and François, 2006). Studies on the causes of this resistance
3. Theory and hypotheses
to carbon taxes suggest that environmental and economic consequences
of taxes, as perceived by the people, are to blame.
Our reading of the literature suggests that the uncertainty around
To begin with, people do not seem to agree with the scientific evi-
carbon taxes and the self-interest of taxpayers are among the key
dence that taxes can be a solution to environmental problems. In part,
determinants of personal attitudes towards carbon taxes. As a result, in
this is a question of effectiveness, and many believe that increasing
the following, we will formulate and test four hypotheses related to the
the cost of polluting would be ineffective to address the problems at
overall categories of uncertainty and self-interest. With the former—
hand (Baranzini and Carattini, 2017; Carattini et al., 2017; Kallbekken
uncertainty—we refer to the findings that people are uncertain about
and Aasen, 2010; Gaunt et al., 2007; Clinch and Dunne, 2006; Dresner
whether carbon taxes could solve the problem (e.g., Baranzini and
et al., 2006; Steg et al., 2006). Interviews and focus groups show that
Carattini, 2017), how the revenues from these taxes would be used
participants expect people would ‘pay to pollute’ even if governments
(e.g., Klok et al., 2006), and how much they would cost to individuals
introduce or increase taxes (Kallbekken and Aasen, 2010; Gaunt et al.,
(e.g., Douenne and Fabre, 2019). With self-interest, we refer to the
2007; Clinch and Dunne, 2006)—in other words, there is a widespread
findings that people oppose carbon taxes out of worries for their own
belief that the price elasticity of polluting activities is too low for taxes
purchasing power (e.g., Douenne and Fabre, 2019).
to change the behaviour of consumers. However, the level of support
for carbon taxes increases if people experience—for example, in trials We view the attitudes towards carbon taxes from a rational-choice
(Carattini et al., 2018a; Tiezzi and Xiao, 2016; Cherry et al., 2014; perspective, where individuals support policy instruments that are
Schuitema et al., 2010)—or simply believe (Hammar and Jagers, 2006) likely to minimise their costs and/or maximise their benefits (Stern
that these taxes work. et al., 1993). However, these cost and benefit calculations take place
In part, it is a question of trust. Increasing the cost of polluting amid uncertainty. Against this background, in the following subsections
may well be an effective deterrent, but people are then suspicious that we develop two further sets of hypotheses based on factors that (1)
this is why governments are keen to introduce carbon taxes (Clinch decrease the uncertainty around carbon taxes and that (2) make the
and Dunne, 2006; Hammar and Jagers, 2006). Instead, they worry cost of taxes clearer for consumers.
that governments use environmental problems ‘as a cover for obtaining
new revenues’ (Klok et al., 2006, p. 913), which may not be spent for 3.1. Uncertainty
the environment or redistributed back to the people (Hsu et al., 2008;
Clinch and Dunne, 2006; Dresner et al., 2006). Putting the two parts A first set of theoretical expectations is related to the uncertainty
together, in short, there is an uncertainty that taxes could or would around carbon taxes. As reviewed above, there is mounting evidence
possibly be used to address environmental problems. This is why, as that people are uncertain about whether taxes as policy instruments
many studies show, earmarking helps (Beiser-McGrath and Bernauer, could or would be used to address environmental problems, and this
2019; Carattini et al., 2019; Baranzini and Carattini, 2017; Carattini uncertainty hinders the support for carbon taxes. If this is true, factors
et al., 2017; Gevrek and Uyduranoglu, 2015; Kallbekken and Aasen, that decrease this uncertainty should at the same time contribute to the
2010; Deroubaix and François, 2006; Dresner et al., 2006; Steg et al., support. In fact, certainty likely improves the acceptance of any policy
2006): if governments designate the revenues, the level of support for instrument, but as a result of the direct and coercive nature of taxes
taxes increases significantly as this practice provides some certainty (Heres et al., 2017; Steg et al., 2006), we expect the improvement to
that taxes can and will be used to address certain environmental be particularly prevalent for as salient an issue as taxes. In this article,
problems. we consider two such factors—political trust and efficacy.

2
R. Umit and L.M. Schaffer Energy Policy 140 (2020) 111385

Political trust is a crucial element of social capital for governments Second, one special case of energy dependency emerges in rural
as ‘institutional trust will matter for the support or rejection of any areas, where people rely more heavily on energy for mobility (due to
government activity’ (Paxton and Knack, 2012, 174, emphasis in orig- relative lack of public transport amid larger distances) and heating (due
inal). Consequently, it has become a concept that features often in to characteristically less energy-efficient housing) than people living
studies over public attitudes—including environmental attitudes (Fair- in urban areas (Berry, 2019; Kallbekken and Sælen, 2011; Broz and
brother, 2017a), which are associated with political trust in a large Maliniak, 2010; Ewing and Rong, 2008). As a result, carbon taxes are
body of research (Baranzini and Carattini, 2017; Fairbrother, 2017b; likely to have more repercussions on rural areas. This effect is not
Dietz et al., 2007; Gaunt et al., 2007; Beuermann and Santarius, 2006; limited to those rural habitants who personally depend on energy. If
carbon taxes mean less money to remain in local economy, all residents
Hammar and Jagers, 2006). When a government proposes to address
would have to shoulder the burden of these taxes disproportionately.
as complex an issue as climate change with taxes, those who have
This leads us to hypothesise the support for carbon taxes to be lower
higher levels of political trust should be more likely to believe that
in rural than in urban areas.
the government proposal could and would work, and therefore have
a higher probability to support the proposal. This is why we expect Hypothesis 4: The level of support for carbon taxes is lower among
political trust to correlate positively with the level of support for carbon people living in rural areas.
taxes.
4. Data and design
Hypothesis 1: The level of support for carbon taxes is higher among
Our analysis draws on data from the European Social Survey (ESS).
people with high political trust.
Including for the first time a module on energy attitudes, Round 8 of
the ESS (2016) provides a unique opportunity to test the hypotheses
What is yet to establish itself in the literature on the acceptabil- above across 23 countries (n = 44,387). The majority of these countries
ity of carbon taxes is political efficacy—specifically, external political are in the European Union (EU) or European Free Trade Association
efficacy, which refers to beliefs about government responsiveness to (EFTA)—Austria, Belgium, Czech Republic, Estonia, Finland, France,
citizen demands (Craig et al., 1990). Irrespective of the political trust Germany, Hungary, Iceland, Ireland, Italy, Lithuania, Netherlands, Nor-
that governments have, their policy to tax carbon for environmental way, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland, and the
purposes might turn out to be a failure in the eyes of the citizens. Under United Kingdom—while the survey includes also Israel and the Russian
high external efficacy—i.e. where people (are perceived to) have a say Federation. This allows us to provide cross-national evidence for indi-
in what their governments do—the citizens can at least then expect vidual preferences for carbon taxes in a range of countries larger than
these policies to be reformed or dropped altogether in line with their it has ever been possible before.
demands. However, where governments are not responsive to citizen The data for our dependent variable comes from the following sur-
demands, policies are more likely to remain untouched once enacted vey question: ‘To what extent are you in favour or against ... increasing
taxes on fossil fuels, such as oil, gas, and coal ... in [country] to reduce
even if they prove unpopular in time. Hence we hypothesise the support
climate change?’ The response options were 1 = ‘strongly in favour’, 2 =
for carbon taxes to increase with external efficacy.
‘somewhat in favour’, 3 = ‘neither in favour nor against’, 4 = ‘somewhat
against’, and 5 = ‘strongly against’. We have reversed the original
Hypothesis 2: The level of support for carbon taxes is higher among scale so that higher values indicate increasing favour. To provide easily
people with high external political efficacy. interpretable results, we treat this dependent variable as continuous
in the main body of the article although the measure is ordinal. In
3.2. Self-interest the Supporting Information, however, we provide a robustness test on
this modelling choice by treating the dependent variable as ordinal.
A second set of expectations is related to the personal cost of Our conclusions remain the same. Finally, because our data has a
hierarchical structure (where respondents are nested in countries), we
carbon taxes. As already suggested above, one of the main reasons
use multilevel regression models.
behind the public resistance to carbon taxes is their perceived economic
For the independent variables of interest, we use data from the sur-
consequences to people,3 who purposefully try to minimise their costs.
vey items measuring (a) how much, on average, the respondents trust
For this assertion to hold, the resistance should be stronger among their country’s parliament, political parties, and politicians (Political
people who are more likely to be paying a higher cost of the taxes. Trust ), (b) how much they believe the political system in their country
Here we develop hypotheses based on two such groups of people. allows people like them to have a say in what the government does
First, as carbon taxes often aim at limiting the energy consump- (Political Efficacy), (c) how confident they are that they could use less
tion behaviour, heavy consumers of energy share a higher burden of energy than they do now (Energy Dependence), and finally (d) how
these taxes—a principle that has widespread support among the public rural is the area they live in (Rural Area). While Political Trust and
(Jagers and Hammar, 2009; Hammar and Jagers, 2006)—unless they Energy Dependence range from zero to ten, Political Efficacy and Rural
reduce their consumption levels. However, the elasticity of energy Area range from one to five. We treat these independent variables as
consumption is not the same for everyone, and consumers are more continuous.
likely to lose if they could not use less energy despite a new or in- To show that our results are not driven by inclusion or exclusion of
creased carbon tax. Therefore, we expect to find a negative relationship co-variates, we report models with and without control variables.4 At
between the energy dependency of individuals and their support for the individual level, these variables include the respondents’ age, gen-
taxes. der, left–right orientation, climate change worries, energy cost worries,
and energy saving behaviours (efficiency and curtailment). In addition,
whether there is already a carbon tax in place, average party position
Hypothesis 3: The level of support for carbon taxes is lower among
on green energy, and gross domestic product are the country-level
people with high energy dependency.
controls. For more details on the data and variables, see the Supporting
Information.

3
Perceptions about the economic consequences of carbon taxes may not
4
always match the reality. In fact, surveying a representative sample of the Note that this also allows us to provide evidence from all 23 countries
French population, Douenne and Fabre (2019) show that, as many as 89% of at least in some models, as the number of countries decreases to 18 with the
people overestimate how much carbon taxes would affect them personally. country-level control variables due to missing data.

3
R. Umit and L.M. Schaffer Energy Policy 140 (2020) 111385

Table 1
Multilevel linear regression models—summary results.
Basic Individual Country All
model controls controls controls
Political 0.068* 0.064* 0.074* 0.069*
Trust (0.003) (0.004) (0.003) (0.004)
Political 0.110* 0.079* 0.116* 0.080*
Efficacy (0.007) (0.008) (0.008) (0.009)
Energy −0.047* −0.028* −0.050* −0.029*
Dependence (0.002) (0.003) (0.003) (0.003)
Rural −0.074* −0.053* −0.074* −0.052*
Area (0.005) (0.006) (0.005) (0.006)
Individual-level ✗ ✓ ✗ ✓
Controls
Country-level ✗ ✗ ✓ ✓
Controls
Constant 2.669* 2.350* 2.304* 2.031*
(0.050) (0.068) (0.156) (0.163)
N-Observations 40972 29890 33079 24379
N-Groups 23 23 18 18
Log likelihood −64732.776 −46232.587 −52229.111 −37686.125
Wald 𝜒 2 2121.56 3492.43 1971.66 2939.02
R2 (Levels 1 / 2) 0.05 / 0.51 0.10 / 0.64 0.06 / 0.69 0.11 / 0.71

Notes: These results are from multilevel linear regressions, where individuals are
nested in countries, with standard errors in parentheses. The dependent variable is
the support for increasing taxes on fossil fuels to reduce climate change. Individual-
Fig. 1. Mean levels of public support for increasing taxes on fossil fuels to reduce level controls include age, gender, left–right orientation, climate change worries, energy
climate change. Notes: The values are based on the ESS Round 8 (ESS, 2016), cost worries, and energy saving behaviours (efficiency and curtailment). Country-level
weighted to account for sampling error, non-response bias, and differences in inclusion controls include whether there is already a carbon tax in place, average party position
probabilities. Horizontal bars indicate standard errors. See Table S2 in Supporting on green energy, and GDP. R2 values are calculated according to Bryk and Raudenbush
Information for the underlying values in table format. (1992). See Table S3 for complete results.
*𝑝 < 0.001.

5. Results
the unexpected direction. This suggests that countries with high rural
5.1. Country-level analysis populations are not necessarily the ones with low public support for
carbon taxes. Note that, however, our hypotheses are for the level of
To begin with analysis, Fig. 1 plots the mean levels of public individuals, which we analyse in the following sub-section.
support for increasing taxes on fossil fuels to reduce climate change,
in countries with and without existing carbon taxes. Overall, it shows 5.2. Individual-level analysis
that people are rather negative about increasing these taxes, with the
average preference (2.72, standard error = 0.01) under the neutral mid- Table 1 presents a summary of four multilevel linear regression
point of the scale—‘neither in favour nor against’. This is indeed the models, where the dependent variable is the support for increasing
situation in most countries; the public attitude towards this policy is on taxes on fossil fuels to reduce climate change. We report the com-
average negative in 17 out of 23 cases. The support turns only slightly plete models in Table S3 in Supplementary Information. To facilitate
positive in no more than five countries in the survey. In Germany, interpretation of the substantive results, Fig. 3 visualises the predicted
the public seems divided almost equally between being against and in levels of support, at the minimum and maximum values of our key
favour of increasing taxes on fossil fuels. independent variables. These predictions are based on the fourth model
At the time of data collection, carbon taxes existed in 13 of the 23 in Table 1, which includes the individual- and country-level control
countries in the survey. Fig. 1 shows that most of these countries are variables.
clustered at either ends of the level of support for increasing taxes on For the factors related to uncertainty, Table 1 shows that there
fossil fuels. On the one hand, we see that in all five countries where the are statistically significant, positive relationships between both Political
attitude towards increasing taxes are positive, carbon taxes are already Trust and Political Efficacy as independent variables and the support
in place. Notice also that, with the exception of Switzerland, these are for increasing taxes as the dependent variable. Moreover, as visible in
all Nordic countries (Sweden, Finland, Iceland, and Norway). On the Fig. 3, Political Trust is associated with a larger change in predictions
other hand, carbon taxes exist in Poland, Estonia, Portugal, and France than it is the case for any other variable of interest in our models. In
as well, but the public support for increasing taxes on fossil fuels are its absence (Political Trust = 0), we predict the average support for
at its lowest level in this set of countries. This suggests that, at least at increasing taxes to be 2.51—similar to the situation in Portugal, the
the country level, there is no linear relationship between having carbon country with third lowest score in the dataset. At the other end of the
taxes already in place and the public attitude towards these taxes. spectrum (Political Trust = 10), the cross-country result (mean = 3.20)
Before we move on to individual-level analysis, Fig. 2 plots the looks very much like in Iceland, the country with the fourth highest
average values of our dependent variable, support for increasing taxes score. With regard to Political Efficacy, while the average support is
on fossil fuels to reduce climate change, against the average values of predicted to be 2.69 among those who believe that people like them
our independent variables of interest, by each country in the dataset. have no say in what their government does (Political Efficacy = 1),
Overall, the results suggest that the indicators for perceived uncertainty we predict this average to be 3.01 (‘neither in favour nor against’
and self-interest are related to the mean levels of support for carbon increasing the taxes) for those who believe the complete opposite—
taxes in several countries. Here we find strong correlations for Political that they have a great deal of say (Political Efficacy = 5). These results
Trust (r = 0.9), Public Efficacy (r = 0.7), and Energy Dependence (r confirm our Hypotheses 1 and 2.
= −0.7)—all in the expected directions. The only exception is Rural For the factors related to self-interest, we find statistically signifi-
Area (r = 0.2), which has a much weaker correlation coefficient, in cant and negative relationships, confirming our respective Hypotheses

4
R. Umit and L.M. Schaffer Energy Policy 140 (2020) 111385

Fig. 2. Mean levels of public support for increasing taxes on fossil fuels to reduce climate change, by independent variables. Notes: All values are country means, based on the
ESS Round 8 (ESS, 2016), weighted to account for sampling error, non-response bias, and differences in inclusion probabilities.

3 and 4. As we expect, first, the support for taxes decreases with

respondents’ dependency on energy. All else being equal, as Fig. 3
shows, a switch from maximum (Energy Dependence = 10) to minimum
(Energy Dependence = 0) confidence that the respondents could use
less energy is associated with a 0.29 decrease on a five-point support
scale for taxes. Second, there is a likewise decrease in support on the
urban–rural divide. Here the effect size is smaller in comparison—
about a quarter of a point on a five-point scale—where the support for
increasing taxes on fossil fuels decreases significantly as the area that
the respondents live in switches from ‘a big city’ (Rural Area = 1) to ‘a
farm or home in the countryside’ (Rural Area = 5).
These results are robust to inclusion or exclusion of control vari-
ables, be it at the level of individuals or countries. Adding these groups
of variables to our analysis in separate models, we observe no change
in terms of sign or significance of the coefficients of interest, and Fig. 3. Predicted support for increasing taxes on fossil fuels to reduce climate change.
only small (ranging between 0.01 and 0.04, on a five-point scale) Notes: Predictions are based the fourth model in Table 1. The dependent variable is
the respondents’ support for increasing taxes for fossil fuels to reduce climate change.
changes in terms of size. We provide further robustness checks in Horizontal bars represent 95% confidence intervals. Other variables are set to observed
Supplementary Information. First, we run 13 separate regressions, by values.
adding one control variable to our basic model at a time. Our main
results remain the same in all these models. Second, we conduct a test
of multicollinearity, and find that there is no highly linear correlations 6. Conclusion and policy implications
among our independent variables. Similarly, we show that our results
In search for policy instruments to reduce climate change, many
replicate if we treat our dependent variable as ordinal, and therefore
governments face a hard choice between effectiveness and popularity.
run multilevel ordinal logistic regressions instead. Finally, repeating
One would think carbon taxes are the go-to instrument since they are
this regression analysis in each of the 23 countries in the dataset, we known for their effectiveness in changing behaviour in favour of the en-
show that our results are not driven by respondents in a minority of vironment. What is more, these taxes also generate additional revenues
countries. To the contrary, there is empirical support for our hypotheses for governments to improve the lives of their citizens—environmentally
in a large majority of the cases. or otherwise. Yet they are unpopular with the public. Indeed, we find

5
R. Umit and L.M. Schaffer Energy Policy 140 (2020) 111385

that there is no overall support for carbon taxes in most of the 23 References
countries in our dataset.
Aldy, Joseph E., Stavins, Robert N., 2012. The promise and problems of pricing carbon:
Multivariate analyses provide insights into the reasons behind the
Theory and experience. J. Environ. Dev. 21 (2), 152–180.
unpopularity of carbon taxes. On the one hand, the results show that Ballard, Charles L., Medema, Steven G., 1993. The marginal efficiency effects of taxes
the attitudes towards taxes improve significantly with higher political and subsidies in the presence of externalities: A computational general equilibrium
trust and efficacy. We argue that this is because the level of support for approach. J. Public Econ. 52 (2), 199–216.
Baranzini, Andrea, Carattini, Stefano, 2017. Effectiveness, earmarking and labeling:
taxes are particularly prone to uncertainty, which decreases with trust testing the acceptability of carbon taxes with survey data. Environ. Econ. Policy
and efficacy. On the other hand, the level of support for carbon taxes Stud. 19 (1), 197–227.
is significantly lower among people who depend highly on energy or Baumol, William J., Oates, Wallace E., 1971. The use of standards and prices for
protection of the environment. Swedish J. Econ. 73 (1), 42–54.
live in rural areas, for whom the economic cost of energy-based taxes
Beiser-McGrath, Liam F., Bernauer, Thomas, 2019. Could revenue recycling make
is likely to be higher. We see this as evidence that the self-interest is effective carbon taxation politically feasible? Sci. Adv. 5 (9), http://dx.doi.org/
one of the determining factors behind the attitudes to carbon taxes. 10.1126/sciadv.aax3323, https://advances.sciencemag.org/content/5/9/eaax3323.
These results can be considered as good as well as bad news for Berry, Audrey, 2019. The distributional effects of a carbon tax and its impact on fuel
poverty: A microsimulation study in the French context. Energy Policy 124, 81–94.
the future of carbon taxes. These are bad news because, as things Beuermann, Christiane, Santarius, Tilman, 2006. Ecological tax reform in Germany:
stand, people are on average against carbon taxes in many countries. handling two hot potatoes at the same time. Energy Policy 34 (8), 917–929.
If there are any governments looking into introducing taxes into their Brännlund, Runar, Persson, Lars, 2012. To tax, or not to tax: preferences for climate
policy attributes. Climate Policy 12 (6), 704–721.
climate change policy any time soon, they are likely to meet public
Broz, J. Lawrence, Maliniak, Daniel, 2010. Malapportionment, gasoline taxes, and
resistance like in France. However, our results can also be considered climate change. https://ssrn.com/abstract=1642499.
as good news at least in countries where the average support for taxes is Bryk, A.S., Raudenbush, S.W., 1992. Hierarchical Linear Models: Applications and Data
only slightly negative. If the governments in these countries take steps Analysis Methods. Sage.
Carattini, Stefano, Baranzini, Andrea, Lalive, Rafael, 2018a. Is taxing waste a waste of
to decrease the uncertainty around taxes, they can shift the attitudes time? Evidence from a supreme court decision. Ecol. Econom. 148, 131–151.
towards positive. Increasing political trust and efficacy of the citizens Carattini, Stefano, Baranzini, Andrea, Thalmann, Philippe, Varone, Frédéric,
could be a step in this direction. Vöhringer, Frank, 2017. Green taxes in a post-Paris world: Are millions of
nays inevitable? Environ. Resource Econ. 68 (1), 97–128.
This article extends the empirical evidence on public attitudes to Carattini, Stefano, Carvalho, Maria, Fankhauser, Sam, 2018b. Overcoming public
carbon taxes, contributing to a growing literature of mainly single- resistance to carbon taxes. Wiley Interdiscip. Rev. Clim. Change 9 (5), 1–26.
case studies with cross-national evidence from 23 countries in Europe. Carattini, Stefano, Kallbekken, Steffen, Orlov, Anton, 2019. How to win public support
for a global carbon tax. Nature 565, 289–291.
First, it puts the existing reports on the unpopularity of carbon taxes
Cherry, Todd L., Kallbekken, Steffen, Kroll, Stephan, 2012. The acceptabil-
from separate countries into a wider context. For example, we find ity of efficiency-enhancing environmental taxes, subsidies and regulation: An
that the overall public attitude towards carbon taxes is negative in experimental investigation. Environ. Sci. Policy 16, 90–96.
about two out of three countries. Second, it shows that factors that Cherry, Todd L., Kallbekken, Steffen, Kroll, Stephan, 2014. The impact of trial runs on
the acceptability of environmental taxes: Experimental evidence. Resour. Energy
decrease the uncertainty around carbon taxes or make the personal cost Econ. 38, 84–95.
of taxes clearer for consumers are among the important determinants of Clinch, J.P.eter, Dunne, Louise, 2006. Environmental tax reform: an assessment of social
public support for this policy instrument across the borders in Europe. responses in Ireland. Energy Policy 34 (8), 950–959.
Craig, Stephen C., Niemi, Richard G., Silver, Glenn E., 1990. Political efficacy and trust:
Changes in these factors alone can be decisive, at least in countries
A report on the NES pilot study items. Politic. Behav. 12 (3), 289–314.
where the levels of support for and opposition to carbon taxes are rather De Groot, Judith I.M., Schuitema, Geertje, 2012. How to make the unpopular popular?
close to each other. Policy characteristics, social norms and the acceptability of environmental policies.
Environ. Sci. Policy 19, 100–107.
Our claims rest on survey data from 44,387 respondents in 23
Deroubaix, José-Frédéric, François, Lévèque, 2006. The rise and fall of French Ecolog-
countries (ESS, 2016), serving our main motivation to conduct a study ical Tax Reform: social acceptability versus political feasibility in the energy tax
that goes beyond the national borders. However, further research on implementation process. Energy Policy 34 (8), 940–949.
this publicly available dataset can focus on one or more countries as Dietz, Thomas, Dan, Amy, Shwom, Rachael, 2007. Support for climate change policy:
Social psychological and social structural influences. Rural Sociol. 72 (2), 185–214.
well, addressing a limitation of our cross-national study. Although we Douenne, Thomas, Fabre, Adrien, 2019. Can we reconcile French people with
show that our main results hold in a large majority of countries in the the carbon tax? Disentangling beliefs from preferences. https://faere.fr/pub/
dataset, case-specific findings can further our understanding of public WorkingPapers/Douenne_Fabre_FAERE_WP2019.10.pdf, FAERE Working Paper.
Dresner, Simon, Jackson, Tim, Gilbert, Nigel, 2006. History and social responses to
attitudes to carbon taxes as well.
environmental tax reform in the United Kingdom. Energy Policy 34 (8), 930–939.
ESS, 2016. European Social Survey Round 8 Data, Data file edition 2.1 Norwegian
Centre for Research Data, Norway—Data Archive and distributor of ESS data for
Declaration of competing interest
ESS ERIC, https://www.europeansocialsurvey.org/download.html?file=ESS8e02_1&
y=2016.
Ewing, Reid, Rong, Fang, 2008. The impact of urban form on US residential energy
The authors declare that they have no known competing finan-
use. Housing Policy Debate 19 (1), 1–30.
cial interests or personal relationships that could have appeared to Fairbrother, Malcolm, 2017a. Environmental attitudes and the politics of distrust.
influence the work reported in this paper. Sociol. Compass 11 (5), 1–10.
Fairbrother, Malcolm, 2017b. When will people pay to pollute? Environmental taxes,
political trust and experimental evidence from Britain. British J. Politic. Sci. 1–22.
Acknowledgment Farrell, Justin, 2016. Corporate funding and ideological polarization about climate
change. Proc. Natl. Acad. Sci. 113 (1), 92–97.
Gaunt, Martin, Rye, Tom, Allen, Simon, 2007. Public acceptability of road user
This article is part of the project ‘Beyond Policy Adoption: Implica- charging: the case of Edinburgh and the 2005 referendum. Transp. Rev. 27 (1),
tions of Energy Policy on Parties, Publics, and Individuals’, funded by 85–102.
Gevrek, Z. Eylem, Uyduranoglu, Ayse, 2015. Public preferences for carbon tax
the Swiss National Science Foundation (PYAPP1-173642/1). attributes. Ecol. Econom. 118, 186–197.
Goulder, Lawrence H., Parry, Ian W.H., 2008. Instrument choice in environmental
policy. Rev. Environ. Econ. Policy 2 (2), 152–174.
Appendix A. Supplementary material Hammar, Henrik, Jagers, Sverker C., 2006. Can trust in politicians explain individuals’
support for climate policy? The case of CO2 tax. Climate Policy 5 (6), 613–625.
Heres, David R., Kallbekken, Steffen, Galarraga, Ibon, 2017. The role of budgetary
Supplementary material related to this article can be found online information in the preference for externality-correcting subsidies over taxes: A lab
at https://doi.org/10.1016/j.enpol.2020.111385. experiment on public support. Environ. Resource Econ. 66 (1), 1–15.

6
R. Umit and L.M. Schaffer Energy Policy 140 (2020) 111385

Hsu, Shi-Ling, Walters, Joshua, Purgas, Anthony, 2008. Pollution tax heuristics: An Schuitema, Geertje, Steg, Linda, Forward, Sonja, 2010. Explaining differences in
empirical study of willingness to pay higher gasoline taxes. Energy Policy 36 (9), acceptability before and acceptance after the implementation of a congestion charge
3612–3619. in Stockholm. Transp. Res. A Policy Practice 44 (2), 99–109.
Jagers, Sverker C., Hammar, Henrik, 2009. Environmental taxation for good and for Shapiro, Robert Y., 2011. Public opinion and American democracy. Public Opinion
bad: the efficiency and legitimacy of Sweden’s carbon tax. Environ. Politics 18 (2), Quart. 75 (5), 982–1017.
218–237. Steg, Linda, Dreijerink, Lieke, Abrahamse, Wokje, 2006. Why are energy policies
Kallbekken, Steffen, Aasen, Marianne, 2010. The demand for earmarking: Results from acceptable and effective? Environ. Behav. 38 (1), 92–111.
a focus group study. Ecol. Econom. 69 (11), 2183–2190. Stern, Paul C., Dietz, Thomas, Kalof, Linda, 1993. Value orientations, gender, and
Kallbekken, Steffen, Sælen, Håkon, 2011. Public acceptance for environmental taxes: environmental concern. Environ. Behav. 25 (5), 322–348.
Self-interest, environmental and distributional concerns. Energy Policy 39 (5), Stern, Nicholas, Stiglitz, Joseph E., 2017. Report of the High-Level Commission on
2966–2973. Carbon Prices. World Bank.
Klok, Jacob, Larsen, Anders, Dahl, Anja, Hansen, Kirsten, 2006. Ecological tax reform Thalmann, Philippe, 2004. The public acceptance of green taxes: 2 million voters
in Denmark: history and social acceptability. Energy Policy 34 (8), 905–916. express their opinion. Public Choice 119 (1–2), 179–217.
Maestre-Andrés, Sara, Drews, Stefan, van den Bergh, Jeroen, 2019. Perceived fairness Tiezzi, Silvia, Xiao, Erte, 2016. Time delay, complexity and support for taxation. J.
and public acceptability of carbon pricing: a review of the literature. Climate Policy Environ. Econ. Manag. 77, 117–141.
19 (9), 1186–1204. Weitzman, Martin L., 2017. Voting on prices vs. voting on quantities in a World Climate
Mankiw, N.G.regory, 2009. Smart taxes: An open invitation to join the Pigou club. Assembly. Res. Econ. 71 (2), 199–211.
Eastern Econ. J. 35 (1), 14–23. World Bank, 2019. Carbon pricing dashboard [dashboard]. https://
Metcalf, Gilbert E., 2009. Designing a carbon tax to reduce US greenhouse gas carbonpricingdashboard.worldbank.org/map_data.
emissions. Rev. Environ. Econ. Policy 3 (1), 63–83.
Paxton, Pamela, Knack, Stephen, 2012. Individual and country-level factors affecting
support for foreign aid. Int. Politic. Sci. Rev. 33 (2), 171–192.