Experimental Design
The experiment proceeds as follows:
*Survey part*
In the first part, participants are asked a series of general questions on climate acceptability and support.
*Experimental part*
In this second part, participants face a specific belief module.
STEP 1: GENERALITIES
First we present participants general information about the Fit-for-55 climate policy package, which aims at reducing net greenhouse gas emissions across the EU by 2030. Participants are randomly provided with very general information about one of the two following policies:
- Policy 1: Fit-for-55 without Carbon Dividend
- Policy 2: Fit-for-55 with Carbon Dividend
We then ask all respondents how much they are opposed to or in favor of the policy described. We use a 7-point Likert scale ranging from “Strongly Opposed” to “Strongly in Favour”.
STEP 2: PRE-TREATMENT BELIEF ELICITATION
After this first step, we proceed to elicit the pre-treatment beliefs about the potential impacts of the policy on emissions, the economy, and households’ finances.
Specifically, each participant is asked a series of questions about the impact of the policy on three different variables:
1. Either the net greenhouse gas emissions of the EU by 2030 (variable 1.A) or the economy of their country by 2030 (variable 1.B) (random assignment)
2. Either the country-specific median income (variable 2.A), poorest 10% (variable 2.B), or richest 10% (variable 2.C) households’ financial situation by 2030 (random assignment)
3. Their own household’s financial situation by 2030 (variable 3)
Each series of questions has the following elicitation structure (belief elicitation):
1. We ask participants their general opinion about the impact of the policy on the variable of interest using a 5-point Likert scale (question 1)
2. We ask participants to specify their guest guess (in %) about the impact of the policy on the variable of interest (question 2)
3. We ask participants to specify their most pessimistic and most optimistic estimates (using sliders) about the impact of the policy on the variable of interest (question 3)
4. We ask participants to allocate 100 points among five potential events concerning different possible impacts of the policy on the variable of interest. The partitioning of the event space is contingent on the participant’s answer in the previous question (i.e., we split the interval between the most optimistic and most pessimistic estimates into five events of equal length), (question 4)
5. We ask participants how confident they are about the point allocation they have indicated using a 5-point Likert scale (question 5)
While the belief elicitation procedure over variable 3 contains the full set of questions (1-5), the belief elicitation of variables 1 and 2 is restricted to questions 1-3.
STEP 3: INFORMATION TREATMENT
We then give participants additional and more specific information on the policy (“Policy 1” or “Policy 2” depending on what was provided in Step 1). Specifically, we present the expected quantitative impacts of the policy package on EU emissions and GDP, and on country-specific household finances (see Appendix A for how the quantitative impacts are derived). We make it clear that those results represent a summary of the evidence coming from scientific studies.
We ask a very simple question (Have you carefully read the provided information? This question aims to verify your attention. Please select the answer ‘Other’ in all cases.) to check participants’ attention, and then re-ask participants how much they are opposed or in favor of the policy now described with more details. As before, we use a 7-point Likert scale ranging from “Strongly Opposed” to “Strongly in Favour”.
STEP 4: POST-TREATMENT BELIEF ELICITATION
Finally, we proceed to re-eliciting the beliefs about the potential impacts of the policy on the same variables.
The belief elicitation procedure is exactly the same as in step 2, except that we do not ask question 1.
APPENDIX A: Computation of data for the information treatment
The economic effect on GDP for the EU of the Fit-for-55 policies has been estimated at about 2.1% of GDP by 2030 (Chateau et al., 2023). This is the main anchor point of estimated economic impacts.
Regarding household’s finance impacts, the median household is expected to experience a welfare loss of similarly 2.1% (Weitzel et al., 2023). To compute the impacts at different income levels, we first divided households by the income threshold of deciles based on EU-SILC data for 2022 (also used for the sociodemographic income question). For the 10th decile, the 95% percentile value is used as the median income in this class. We use the total household income of people aged between 18-74 after all tax deductions and social insurance contributions.
For each country (France and Italy), we use the data on policy costs from the Carbon Pricing Incidence Calculator (CPIC) from Steckel et al. (2023). We estimate a log-log model to compute income elasticities, which results in an income elasticity of 0.82 for both France and Italy. Based on these income elasticities, the financial impact for the different deciles is computed, resulting in values comparable to (Weitzel et al., 2023), regressive in between 3.0% and 1.5%.
In order to compute the impact of the Carbon dividend, we first compute the value of a carbon dividend by 2030 for the EU. Based on EEA data, and assuming that the 55% target is met, this implies GHG emissions of about 2,538 MtCO2 by 2030. At an average carbon price of 150 EURO per tCO2, and a projected population for the EU27 of 453 million (OECD), this implies a carbon dividend of 840 EURO per capita on average across countries, per year. We assume the dividend is uniform across EU countries.
We the add the redistribution of this carbon dividend to households on top of the expected income loss computed above. Thereby we assume that half of the total dividend is paid by each decile in proportion to its income (or that other taxes need to be raised to ensure fiscal balance), while the other half is simply added to the household budget. This results in a positive impact for lower deciles similar to existing studies (Budolfson et al., 2021, Emmerling et al., 2023, Weitzel et al. 2023), while for the richest deciles the positive impact is very small.
References:
Jean Chateau & Antonela Miho & Martin Borowiecki, 2023. "Economic effects of the EU’s ‘Fit for 55’ climate mitigation policies: A computable general equilibrium analysis," OECD Economics Department Working Papers 1775, OECD Publishing. https://doi.org/10.1787/f1a8cfa2-en
Steckel, J., Missbach, L. and Schiefer, T. (2023). The global Carbon Pricing Incidence Calculator (CPIC) (Version 1.0). http://www.cpic-global.net.
Weitzel, Matthias, Toon Vandyck, Luis Rey Los Santos, Marie Tamba, Umed Temursho, and Krzysztof Wojtowicz. “A Comprehensive Socio-Economic Assessment of EU Climate Policy Pathways.” Ecological Economics 204 (February 1, 2023): 107660. https://doi.org/10.1016/j.ecolecon.2022.107660.
Budolfson, Mark, Francis Dennig, Frank Errickson, Simon Feindt, Maddalena Ferranna, Marc Fleurbaey, David Klenert, et al. “Climate Action with Revenue Recycling Has Benefits for Poverty, Inequality and Well-Being.” Nature Climate Change, November 29, 2021, 1–6. https://doi.org/10.1038/s41558-021-01217-0.
Emmerling, Johannes, Pietro Andreoni, and Massimo Tavoni. “Global Inequality Consequences of Climate Policies When Accounting for Avoided Climate Impacts.” Cell Reports Sustainability 0, no. 0 (November 29, 2023). https://doi.org/10.1016/j.crsus.2023.100008.