Cross-national analysis of attitudes towards fossil fuel subsidy removal

The study investigates the political feasibility of fossil fuel subsidy removal by analysing public attitudes in five developing countries where such subsidies are substantial. While carbon pricing is widely promoted for cost-effective emissions reductions, existing fossil fuel subsidies counteract its effects and impose heavy fiscal burdens. Public acceptability is crucial for implementing these reforms, given potential immediate costs for households and industry. The research asks: How do attitudes toward removing fossil fuel subsidies compare to attitudes toward introducing a carbon tax? Do attitudes differ by targeting subsidies for private versus industrial use? Does specifying how savings from subsidy removal will be used (revenue recycling) increase acceptance? Addressing these questions informs policy design aimed at enhancing public support for effective climate mitigation in diverse national contexts.

Previous research shows that attitudes toward climate policy are shaped by perceived fairness, effectiveness, political trust and climate concern. Experimental and survey studies in primarily developed countries suggest that policy design, particularly revenue recycling (for example, fee-and-dividend, earmarking for welfare or climate investments), can increase acceptability of carbon pricing, though results vary by type of recycling. Citizens tend to prefer policies aimed at industry over those imposing direct household costs, and prefer less coercive instruments. Policy feedback theories suggest that design choices can build supportive constituencies. However, few studies examine attitudes in developing countries or specifically target fossil fuel subsidy removal, leaving a gap this study addresses.

The authors conducted a pre-registered 1×7 factorial online survey experiment (YouGov) across Ecuador, Egypt, India, Indonesia and Mexico. The pooled sample included 6,636 respondents (≈1,000 in Ecuador; ≈1,400 in each of the other countries), recruited via quota sampling based on census demographics (gender, age, region). Respondents were randomly assigned to one of seven groups and rated policy support on a 0–10 scale. Treatments: (1) introduce a carbon tax; (2) remove fossil fuel subsidies for industrial use; (3) remove fossil fuel subsidies for private consumption; (4) remove private-consumption subsidies with revenues for general welfare (education/healthcare); (5) remove private-consumption subsidies with revenues for income tax reductions; (6) remove private-consumption subsidies with revenues for climate adaptation projects; (7) remove private-consumption subsidies with revenues for cash transfers to poor/most-affected households. Sociodemographic and attitudinal covariates included age, sex, education, income, urban/rural residence, vehicle ownership and climate concern. Hypotheses were tested using independent-sample one-sided t tests and OLS regressions with robust standard errors (P<0.05). For H3, group 3 was compared to the aggregate of groups 4–7. Country-specific exploratory analyses were also conducted. Data and code are available via Harvard Dataverse; preregistration at OSF.

• Overall support levels (0–10 scale): industrial-use subsidy removal M=6.22 (s.d.=2.57); private-consumption subsidy removal M=6.31 (s.d.=2.67); carbon tax M=6.33 (s.d.=2.77).
• H1 (subsidy removal less acceptable than carbon tax): Rejected. No significant difference between private-consumption subsidy removal and carbon tax (t(1,893.16) = -0.1604, P=0.4363).
• H2 (removing private-consumption subsidies less acceptable than industrial-use): Rejected. No significant difference between industrial-use and private-consumption subsidy removal (t(1,896.18)=0.6985, P=0.7575).
• H3 (specifying revenue use increases acceptance): Supported in aggregate. Specified revenue use M=6.49 (s.d.=2.59) vs non-specified M=6.31 (s.d.=2.67); t(1,428.25)=1.88, P=0.03. By category: welfare investments M=6.59 (s.d.=2.55) vs non-specified, t(1,893.68)=2.36, P=0.01; climate adaptation M=6.62 (s.d.=2.65) vs non-specified, t(1,887.99)=2.57, P=0.01; income tax reduction M=6.25 (s.d.=2.48), ns (t(1,884.34)=-0.50, P=0.69); cash transfers to poor/most-affected M=6.49 (s.d.=2.66), ns (t(1,894)=1.53, P=0.06).
• Cross-national patterns: Attitudes toward subsidy removal are broadly on par with attitudes toward carbon tax across countries; Egypt exhibits lower support (industrial-use M=5.4; private-consumption M=5.3) compared to averages in the other countries (≈6.2 and ≈6.3, respectively). Preferred revenue uses vary: climate adaptation most popular in Mexico and Ecuador; least popular in Egypt.
• Additional correlates: Higher climate change concern is a strong predictor of policy support; effects of revenue recycling differ by country.

Findings indicate that removing fossil fuel subsidies is not inherently more politically challenging than introducing a carbon tax in the studied developing countries. The similar support levels suggest that opposition to price-based climate instruments may be comparable across these policy options. Importantly, specifying how freed public funds will be used—particularly for welfare and climate adaptation investments—improves acceptance of subsidy removal, though monetary compensation mechanisms do not consistently increase support. Country-specific differences in preferred revenue uses highlight the need to tailor policy design to national contexts, potentially reflecting cultural norms, existing welfare systems and tax structures. The results reinforce theories that transparent, beneficial revenue recycling can mitigate perceived personal costs and bolster public buy-in for climate policy.

This study provides one of the first cross-national experimental assessments of public attitudes toward fossil fuel subsidy removal in the Global South and compares them to carbon taxation. It shows that subsidy removal can be as acceptable as carbon taxes and that specifying revenue recycling—especially toward welfare and climate adaptation—can increase support. Policymakers should pair subsidy reforms with clear, context-appropriate investments to enhance acceptability. Future research should expand to more countries, incorporate varying magnitudes and pacing of subsidy cuts, assess sensitivity to energy price fluctuations, improve public information about policy mechanisms and include control groups to better isolate treatment effects.

• Limited external validity: Only five countries studied; results may not generalize broadly.
• Survey sensitivity: Attitudes may depend on wording and sampling via online quotas.
• Policy design scope: No variation in size or timing of subsidy cuts was tested.
• Temporal context: Data collected before the sharp energy price increases linked to the Ukraine conflict.
• Experimental design: No pure control group for benchmarking treatment groups.