Carbon tax acceptability with information provision and mixed revenue uses

Carbon taxes are a key climate policy but often face public resistance, underscoring the need to better understand attitudes and improve acceptability. A central design question concerns how to use tax revenues—commonly for climate projects, as part of general funds, or as transfers to firms/households. Carbon taxes can be regressive, but targeted transfers to low-income households can mitigate inequity. Prior research frequently finds a preference for earmarking revenues to climate projects, though political context and information about emission reductions or distributional effects can shift preferences toward compensating inequities. Despite real-world mixes of revenue uses, most studies assess single-use scenarios. Moreover, people’s prior knowledge and information provision may shape acceptability, with suggestions that support for climate-project earmarking may stem from misunderstanding that carbon taxes reduce emissions regardless of revenue use. This study investigates: (a) how knowledge and information influence acceptability; (b) which single or mixed revenue uses maximize acceptability; and (c) how acceptability relates to perceived effectiveness and fairness, considering both personal and distributional (low-income) effects. The context is Spain (N=2004), where no carbon tax exists and public debate is limited.

Prior studies indicate that revenue use strongly affects acceptability. Many people prefer using revenues for climate projects over options aimed at distributional compensation, even though carbon taxes can be regressive. Information provision, political context, and communication about emission reduction or distributional impacts can alter preferences, increasing support for compensatory transfers. While actual policies often mix uses, research has mostly assessed single-use proposals. Public understanding is limited; some evidence suggests people conflate effectiveness with revenue earmarking. After implementation of environmental taxes, beliefs about effectiveness can adjust, but before implementation, information provision is crucial. Research has tested impacts of model-based information on tax effects, expert-stated emission reductions, and information on other policies’ effectiveness. This study extends the literature by simultaneously examining single and mixed revenue uses, the roles of prior knowledge and information provision, and the mediating perceptions of effectiveness and fairness.

Design: Online survey experiment among Spanish adults (N=2004) conducted in August 2019 via Netquest, with quotas on age, gender, and region to approximate national representativeness. Respondents were randomly split: information treatment (N=1004) received a brief explanation of how a carbon tax works; control (N=1000) did not. No tax rate was specified to focus on general acceptability independent of specific costs. Measures: Respondents first reported self-perceived knowledge (5-point scale) and completed six true/false items assessing knowledge about carbon taxation (subject, mechanism, effects, revenue use). A Mokken scale aggregated assessed knowledge; one weak item was dropped, yielding a 5-item scale with overall scalability coefficient 0.370 (weak but acceptable). Perception and acceptability outcomes were elicited on 5-point Likert scales for: perceived effectiveness (emissions reduction), perceived fairness, personal effects ("fairness to me"), effects on low-income households (distributional fairness), and acceptability. These were first asked for a carbon tax with unspecified revenue use. Next, five revenue-use scenarios were presented: (i) Climate projects (Climate); (ii) Equal transfers to all households (AllHH); (iii) Transfers to low-income households (PoorHH); (iv) 50% Climate + 50% AllHH (AllHH&Climate); (v) 50% Climate + 50% PoorHH (PoorHH&Climate). Respondents rated each scenario on the same perception and acceptability scales. Subsequently, respondents allocated 100% of revenues across three categories (Climate, PoorHH, AllHH) and indicated whether effectiveness or fairness weighed more in their allocation decision. Sample and administration: 3415 invited; 2534 accessed; 530 excluded (failed controls, incomplete, quota filled), yielding 2004 completes (response rate 58.68%). Average completion time ~15 minutes; informed consent obtained; ethics approval from Autonomous University of Barcelona. Statistical analysis: Nonparametric tests (pairwise Mann-Whitney for within-person comparisons across revenue uses; Kruskal-Wallis with Bonferroni correction for information-treatment effects within revenue uses). Ordered logit regressions modeled acceptability and perceptions as outcomes with revenue-use indicators (unspecified baseline), information provision, assessed knowledge, self-perceived knowledge, interactions of assessed knowledge×revenue use and information×revenue use, and controls (age, gender, education, climate concern, political orientation, income, trust in politicians, household size, car use). Additional ordered logit models linked acceptability to perceptions and fairness to its components (personal and low-income effects). Predictors of acceptability were also ranked using gradient boosting machines with 10-fold cross-validation. Missing values for political orientation (n=263) and income (n=432) were handled via listwise deletion in main models; robustness checks omitted these controls to retain all observations.

• Knowledge and information:
  • Assessed knowledge and self-perceived knowledge are positively but modestly correlated (rank correlation 0.33, p<0.01). Many who strongly reject the tax overestimate their knowledge; knowledge gap (self-perceived minus assessed, z-scored) is negatively related to acceptability (β = −0.0385, p = 0.0479).
  • Information provision increased acceptability for unspecified revenue use, especially among those with moderate-to-high assessed knowledge (levels 3–4; Kruskal-Wallis p≈0.018 and 0.059, respectively). No effect at the lowest knowledge level.
• Revenue use effects:
  • Spending all revenues on climate projects yields the highest acceptability (mean 3.88 vs all others <3.5). Mixed uses that include climate projects (PoorHH&Climate; AllHH&Climate) are more acceptable than unspecified and more acceptable than the non-climate single-use transfers.
  • Perceived effectiveness and fairness are highest for Climate, followed by PoorHH&Climate. Transfers to households (AllHH or PoorHH) are perceived as least effective for emissions reduction. Climate is seen as relatively less favorable for poor households compared to PoorHH, yet respondents report better personal effects under Climate.
  • Information provision raises acceptability, perceived fairness, and effectiveness mainly for unspecified revenue use (ordered logit interactions: Information×Unspecified OR for acceptability 1.40 [1.15–1.70]; effectiveness 1.30 [1.08–1.56]; fairness 1.36 [1.12–1.66]). Effects for other revenue schemes are small or not significant, with a weak positive effect for AllHH.
• Regression results (ordered logit; unspecified as baseline):
  • Acceptability odds ratios (95% CI): Climate 3.57 (2.41–5.29); PoorHH&Climate 3.36 (2.28–4.96); AllHH&Climate 3.07 (2.08–4.52); PoorHH 2.86 (1.93–4.21); AllHH 2.58 (1.76–3.80). Climate also most increases perceived effectiveness (2.22 [1.52–3.26]) and fairness (3.13 [2.12–4.63]).
  • Knowledge interactions: Assessed knowledge×Climate increases acceptability (1.31 [1.22–1.42]), effectiveness (1.20 [1.11–1.29]), and fairness (1.25 [1.17–1.35]); positive but smaller for PoorHH&Climate. Assessed knowledge×AllHH and ×PoorHH reduce acceptability and perceived effectiveness/fairness. Self-perceived knowledge relates negatively to acceptability and perceptions (e.g., OR 0.86 [0.81–0.91] for acceptability).
• Free allocation task:
  • Over half prefer combining all three uses (Climate, PoorHH, AllHH), allocating on average a larger share to Climate. Reported averages: ~45% to Climate overall; ~41% among initial opponents; ~50% among initial supporters. Among those endorsing all three uses, Climate allocation differs significantly by acceptability group (Kruskal-Wallis χ²=58.08, df=2, p≈0), with supporters allocating more to Climate than opponents or indifferent respondents.
• Drivers of acceptability:
  • Acceptability correlates with both perceived effectiveness and personal effects and less with low-income effects. When modeled jointly, perceived fairness is a stronger predictor of acceptability than perceived effectiveness, indicating distributional considerations play a pivotal role.

Earmarking carbon tax revenues to climate projects maximizes acceptability and boosts perceptions of effectiveness and fairness. Mixed uses that include climate projects, particularly coupled with transfers to low-income households, also perform well, suggesting people value both environmental and equity objectives, with a stronger emphasis on environmental earmarking. Information about how carbon taxes work raises acceptability mainly when revenue use is unspecified, implying that clarifying the regulatory effectiveness of the tax can substitute for, but does not strongly complement, specifying revenue allocation. Prior knowledge shapes preferences: better-informed individuals favor climate earmarking and are less supportive of pure transfers, aligning with the notion that understanding the tax’s incentive mechanism increases perceived effectiveness and fairness of climate-linked uses. Importantly, perceived fairness has a stronger association with acceptability than perceived effectiveness, highlighting that distributional and personal outcome considerations are central to public support. These insights suggest policy communication should emphasize fairness dimensions (e.g., compensations for vulnerable groups) alongside clear explanations of how carbon taxes reduce emissions independently of revenue use.

This study contributes by jointly analyzing single and mixed revenue uses, the roles of prior knowledge and information provision, and the mediating perceptions of effectiveness and fairness for carbon tax acceptability. It shows that dedicating revenues to climate projects yields the highest acceptability and perceived performance; mixes that include climate projects, especially with support to low-income households, are also well-received. Information about tax functioning increases acceptability when revenue use is unspecified, indicating that effective communication can improve support. Fairness perceptions are more influential for acceptability than effectiveness. Future research should test communication strategies that integrate policy-functioning information with detailed, equity-focused revenue allocations to look for synergies, and employ varied experimental designs to further validate findings across contexts and policy settings.

• Potential order effects: Perception items followed the information treatment, and revenue-use scenarios were evaluated sequentially, possibly priming or anchoring later responses. The survey provided an initial overview of all options to mitigate this, and results align with external literature, but residual order effects cannot be ruled out.
• Knowledge scale strength: The assessed knowledge Mokken scale was weak (overall scalability coefficient 0.370) though acceptable; measurement error may attenuate knowledge-related estimates.
• Context-specific generalizability: The study focuses on Spain, where no carbon tax exists and public debate is limited; responses may differ in jurisdictions with existing carbon pricing or different political contexts.
• Missing data on some controls (political orientation, income) required listwise deletion in main regressions; robustness checks suggest stability but some bias cannot be excluded.