Shocking Electricity Prices and Carbon Tax Aversion

I send out survey invitations by post to households living in municipalities in which the two most southern electricity zone borders cross. The sampling is random with the exception that households in close proximity to the border are assigned a higher probability to receive the invitation letter. The individual to whom the letter is addressed to is a random member of the household that is 18 years of age or older. The invitation letter contains a web address and a QR code that both links to the online survey. The survey itself contains random information treatments.

2023-04-23

2023-07-23

Support levels for a carbon tax increase and support levels for a carbon tax decrease. In addition to this, primary outcomes include the respondents’ beliefs regarding carbon taxation, namely whether carbon taxation negatively impacts one’s household’s purchasing power, whether it mitigates emissions, whether it disproportionately affects low-income households, and whether it harms rural areas the most.

To measure the support for a carbon tax increase and a carbon tax decrease respectively, I will construct indicator variables using the respondents’ answers to questions that ask them to report whether they are in support or not of a transport sector carbon tax increase reform as well as a carbon tax decrease reform. The other primary outcome variables concerning beliefs will also be constructed from questions asked after the respondent has been presented with the carbon tax increase or decrease. These questions will ask whether the respondent believes that their household will win or lose purchasing power from the policy (and how much), if the policy would reduce emissions, hurt low-income households, and hurt rural areas more than other areas.

In addition to the primary outcomes, I am also interested in the heterogeneity of the potential biases in beliefs around carbon taxation and how they may differ between households that have been exposed to relatively high versus low electricity prices. In addition to this, if space permits, I will also examine outcomes by sub-groups as defined by sympathizing with the Swedish protest movement against transport sector carbon taxation, political orientation, and low levels of trust.

I survey households that have been exogenously treated with relatively high (treatment) and low (control) electricity prices. In the online survey, respondents are randomly assigned to information treatments.

The first experimental intervention is that of a natural experiment. Sweden introduced a new electricity market model in November 2011 which divided Sweden into four electricity zones (SE1 to SE4) based on geographical transmission constraints in the grid. These energy zones aimed to identify the need for new transmission capacity or increased energy generation in the southern zones with higher demand and lower levels of energy generation. In practice, these energy zones became separate energy markets with prices determined by transmission constraints between them. Since the implementation, there have barely been any price disparities between the energy zones. This changed in 2020 as a result of shutdowns of nuclear power reactors in one of the southern zones and the price differences between zones were the following years exacerbated by higher electricity prices on the European continent after lowered electricity production in some countries and high natural gas prices following Russia’s invasion of Ukraine. Thus, the residents of the different zones have been exogenously treated with different electricity price levels, which is what I will utilize to identify the effect of the difference in electricity prices on attitudes and beliefs regarding carbon taxation. I only include survey households from the three most southern electricity zones (near the two most southern borders) since the electricity price differences have been the largest between those.

In addition to this natural experiment, I induce a number of randomized survey treatments for the purpose of identifying the provided information’s effect on beliefs and to be able to identify the causal relationship between beliefs and carbon tax support. The treatment on the belief that the policy will negatively impact the household’s purchasing power involves suggesting alternative carbon tax reforms that are funded differently – randomly leaving the respondent in or out of having to be a part of funding the policy through their tax bill or not (or get tax relief in the case of a carbon tax increase). The treatment of this belief also involves a within-survey estimation of the effect the policy will have on the respondent’s household’s purchasing power – and then presenting this estimate as feedback to the respondent. The other belief treatments involve written statements with relevant information for the belief in question that are presented to a random share of the respondents just above the question that asks for the belief.

The randomizations of the information treatments are done by the survey software.

The survey treatments are randomized on the individual level.

No

18 municipalities.

I will send out a total of 50,000 invitation letters. The final sample size will depend on the response rate.

Out of the 50,000 letters, 25,000 will be sent out to households just south of the borders (treatment) and 25,000 will be sent out to households just north of the borders (control). I thus expect the treatment and control groups to be similar in size. The information treatments in the survey are all randomized to 50% of the respondents with the exception of the feedback treatment, which is randomized to 67% of the respondents.