Prices vs. Quantities from a Citizen's Perspective

2022-11-30

2022-12-22

1. Willingness to support carbon taxes or emissions trading systems (ETS)
2. Views on the role of different actors for decarbonization (i.e., state, citizens, companies)
3. Perceived mitigation mechanism of carbon taxes or ETS
4. Perceived policy properties of carbon taxes and ETS (e.g., perceived environmental effectiveness, perceived incidence between consumers and companies)

Our analysis aims at identifying systematic variation in perceptions of economic and non-economic properties of carbon taxes and ETS, as well as their perceived mitigation mechanism. We further aim to test the hypothesis that support for carbon taxes or ETS increases when respondents believe the instrument targets the (in their perception) relevant actor for effective decarbonization (i.e., government, consumers, or companies). More specifically, we expect that carbon taxes are more popular with citizens, who believe that governmental efforts and consumer-driven change play a primary role for decarbonising society, and who believe that the relevant mechanism of carbon taxes consists in raising revenue for green investment or incentivising demand-side change. On the other hand, we expect emissions trading to be more popular among people adhering to the idea that decarbonisation relies primarily on the action of companies and seeing emissions trading as an instrument to primarily incentivise firms to decarbonise their production.

The survey is structured as follows: We first elicit people’s political attitudes, world views and more general beliefs about government, companies, and citizens, and which actor they think is most important in the transition to a climate-friendly economy.
In a second step, we test people’s knowledge of existing carbon pricing schemes in their countries, including whether people know if their countries currently have a national carbon tax or an ETS and at which level national carbon prices currently stand.

We then separate the sample in two randomized groups and provide them with short descriptive information and definitions of either carbon taxes or emissions trading. In a third step, we elicit their support for the group’s assigned instrument and proceed with questions aimed at their intuitive beliefs about the intended mechanism of carbon taxes and ETS, their perceptions of the policy’s incidence between consumers and companies, as well as their perception of environmental, economic, and distributional effects and further policy characteristics.
In a last step, we test people’s factual knowledge and understanding about their group’s assigned carbon pricing policy. To avoid priming at an earlier stage, at this stage we also inquire about respondents’ familiarity with the European Union Emissions Trading System.

As the field time of this survey is scheduled for winter 2022, we further elicit people’s concern about the energy crisis currently unfolding in the European Union as a result of the Russian war of aggression against Ukraine as an additional control variable.

Summary of variables:
(i) View on the role of different actors for decarbonisation: (1) State; (2) Consumers; (3) Companies
(ii) Support for carbon pricing (carbon tax/ ETS)
(iii) Assumed mechanism of carbon pricing: (i) Revenue-raising + green investment; (ii) demand-side change; (iii) supply-side change
(iv) Policy properties: (a) Incidence; (b) costs to self; (c) costs to companies; (d) fairness; I evasion potential; (f) revenue-raising potential; (g) economy-wide damage; (h) innovation leverage
(v) Beliefs about effectiveness: (A) mitigation effectiveness; (B) Pigouvian ignorance
(vi) Beliefs about increase in personal living costs
(vii) Controls: Knowledge, locus of control, environmental awareness, party affiliation, world views, trust, satisfaction with current climate policy ambition, further socio-economic variables, concern about the current European energy crisis

While most of the above variables will be elicited through five-point Likert scales, we plan to dichotomize key variables (e.g., instrument support, world views, view on the role of different actors) for our analysis.

Randomization done in office by a computer

Individuals

No

15,000 (6,000 in Germany, and 1,500 in the other six countries)

15,000 (6,000 in Germany, and 1,500 in the other six countries)

7,500 (3,000 in Germany, and 750 in the other six countries)

Most of our steps in the analysis involve only correlational analysis, which do not require a power analysis. In the following, we will thus provide a power analysis for our main hypothesis that the support for carbon pricing is higher if it is perceived in accordance with people’s views on which actor drives decarbonization. This is contingent on the hypothesis that carbon taxes and ETS are, in the perception of citizens, distinctly mapped to governmental efforts and consumer-side change on the one hand, and supply-side change on the other hand. To be precise, we model the dependent variable (acceptability) as binary. To this end, we build upon Baranzini and Carratini (2017) who show that labelling matters when it comes to the acceptability of a carbon pricing scheme. With standard assumptions on the desired significance level (t1) of 5% and power (t2) of 80%, we calculate the minimum detectable effect size as follows (Djimeu and Houndolo, 2016): δ=(t_1+t_2 )*√(P(1-P)/T(1-T)n), where P is proportion of the study population that support carbon pricing. In the basic setting of Baranzini and Carratini (2017) this amounts to 0.49. In our setting T equals 0.5 as we assign half of the respondents to the treatment group. Given our sample size of n=15,000, we get a minimum detectable effect size δ of 0.022. Conducting this analysis for each country with n=1,500 results in a MDE of 0.072, whereas it amounts to 0.036 for Germany where n=6,000. Our large sample size allows us also to test for a series of heterogeneities as outlined in Equation (2) regarding the views on the role of different actors for decarbonization and the perceived mitigation effect. The combination of the number of treatment groups (2), the number of actors (3) and our assignment into two groups regarding the mitigation effect results in 16 groups. Assuming that P and T do not vary over these groups and that respondents are equally distributed across the 16 groups, would result in an MDE of 0.09, which is the effect size we can test with our interaction terms. Unfortunately, we cannot conduct this detailed heterogeneity analysis on the country level because the sample size would get too small.