Energy Policy Preferences and Compensation Policies in Germany

Voluntary members of Respondi's repondent pool in Germany were contacted and asked to complete an online survey from their computer. The survey contained three experiments that tested how individuals' climate policy preferences change with information on compensation policies.

For clarity, we assign numbers to the experiments in our pre-analysis plan. However, the order in which experiments appeared to survey respondents was randomized. The three experiments contained the following interventions:

Experiment 1:
The first experiment is a conjoint-experiment that presents individuals with a choice between two hypothetical candidates for the upcoming federal election. The experiment randomly varies 5 candidate characteristics: party, gender, position on energy policy, position on social policy, and position on migration policy.

Experiment 2:
The second experiment uses a vignette design to test the effect of different forms of compensation on support for the energy transition. Respondents are asked to watch short videos that present them with information on the exit from coal. Videos for the treatment groups contain additional information on different forms of compensation.

Experiment 3:
The third experiment combines a vignette experiment with a conjoint experiment. First, all respondents are asked to read a short text on carbon pricing. The treatment groups each read an additional paragraph on compensation or competitiveness measures. After reading the information on carbon pricing, respondents are presented with a conjoint experiment, which asks them to pick one of two hypothetical climate policy plans. The plans randomly vary along four dimensions: cost, effectiveness, compensation measures and competitiveness measures.

2021-09-20

2021-09-24

Experiment 1:
In experiment 1, our key outcome of interest is the set of individuals' multidimensional preferences for electoral candidates. We are particularly interested in the effect of a candidate's energy policy position on electoral support.

Experiment 2:
In experiment 2, the key outcome of interest is support for a complete and rapid exit from coal energy. This is measured in two outcome questions, which we use to construct a total of four outcome variables, as detailed below.

Experiment 3:
In experiment 3, we are interested in the set of individuals' multidimensional preferences on climate policy plans. We further test how information received in the vignette treatments interacts with conjoint characteristics.

Experiment 1:
In this conjoint experiment, respondents are presented with five pairs of hypothetical candidates for the 2021 federal elections. These candidates vary along five characteristics: party, gender, position on energy policy, position on social policy, and position on migration policy. Each of the characteristics is randomly drawn from an underlying set of possible values. Respondents are then asked, in each instance, whether they prefer candidate 1 or candidate 2. This binary outcome will be used to construct, at the level of an individual candidate, our main outcome variable of Candidate Favored.


Experiment 2:
The second experiment has two outcome questions, which are asked right after respondents have watched the control video or one of the treatment videos. The first outcome question asks individuals to indicate their general support for a complete exit from coal on a 5-point scale. The second outcome question asks individuals to indicate their "ideal" year for the coal exit on a slider scale that ranges from 2021 to 2050.

Using these two questions, we create four outcome variables:
1. Year earlier than max (yem): continuous variable calculated as (2050 - ideal year indicated)
2. Year earlier or later than currently anticipated date of policy exit (2038) (yb): binary variable with value 1 of ideal year indicated is before 2038, 0 otherwise
3. General support (gs): continous variable that takes values 0-5, where 0 indicates absolutely no support for a full coal exit, and 5 indicates complete support for a full coal exit
4. Binary General support (bgs): Binary variable that takes the value 1 if respondents express complete support or support for a full coal exit, and 0 if respondents express no support or absolutely no support for a full coal exit

Experiment 3:
After reading the vignette texts of experiment 3, respondents are presented with five pairs of hypothetical climate policy plans. The plans randomly vary along four dimensions: cost, effectiveness, compensation measures and competitiveness measures. Each of the characteristics is randomly drawn from an underlying set of possible values. Respondents are then asked, in each instance, whether they prefer plan 1 or plan 2. This binary outcome will be used to construct, at the level of an individual plan, our main outcome variable of Plan Favored.

The survey contained three experiments with three interventions (the order in which the experiments appear to participants is randomized):

Experiment 1:
The first experiment is a conjoint-experiment that presents individuals with a choice between two hypothetical candidates for the upcoming federal election. The experiment randomly varies 5 candidate characteristics: party, gender, position on energy policy, position on social policy, and position on migration policy. We are particularly interested in the effect of energy policy positions on electoral support. Energy policy positions are drawn from a set of four possible positions: (1) candidate wants to slow down the energy transition to more renewable sources of energy, (2) candidate wants the energy transition to proceed at its current speed, (3) candidate wants the energy transition to proceed at its current speed and demands government compensation for low-income households, (4) candidate wants the energy transition to proceed at its current speed and demands compensation for energy-intensive industry to ensure competitiveness. The possible factors for all five levels are presented in the attached pre-analysis plan. Each respondent repeats the comparison process of this experimental design five times.

Experiment 2:
The second experiment, we use a vignette design to test the effect of different forms of compensation on support for the energy transition. In this experiment, we focus on the transition away from coal, as this particular policy goes hand in hand with a variety of compensation measures. Respondents are asked to watch short videos that present them with information on the exit from coal. First, all participants receive information on Germany’s current plans to exit coal. Among other things, the text mentions all the societal groups that could be considered losers of the exit from coal – consumers, workers, investors, and those living in coal regions. The control group receives no further information. Three treatment groups receive information on different forms of compensation: compensation for consumers (Treatment 1), compensation for investors (Treatment 2), compensation for workers (Treatment 3) and compensation for coal regions as a whole (Treatment 4). Links to the videos and English translations of their transcripts can be found in the attached pre-analysis plan.

Experiment 3:
The third experiment tests individuals’ willingness to pay for climate policy. Respondents are randomly assigned to the control group (C), or one of two treatment groups: the compensation treatment (T1) or the competitiveness treatment (T2). First, all respondents are asked to read a short text on carbon pricing. The treatment groups each read an additional paragraph. After reading the information on carbon pricing, respondents are presented with a conjoint experiment, which asks them to pick one of two hypothetical climate policy plans. The plans randomly vary along four dimensions: cost, effectiveness, compensation measures and competitiveness measures. The exact wording of vignette treatments and values of the conjoint experiment can be found in the attached pre-analysis plan. While individuals only read one excerpt at the beginning of the experiment, the conjoint task is repeated five times.

Computer software

individual respondents

No

2000 individuals

2,000 for experiment 2. As respondents repeat each of the conjoint comparisons 5 times, the number of observations for experiments 1 and 3 will be 20,000 (2000 individuals, 5 comparisons between 2 plans).

Experiment 1: 2000 respondents

Experiment 2: 400 respondents in control group, 400 respondents in Treatment 1, 400 respondents in Treatment 2, 400 respondents in treatment 3, and 400 respondents in Treatment 4.

Experiment 3: 667 respondents in the Control group, 667 respondents in Treatment 1, 667 respondents in Treatment 2.

PAP_Germany_Energy_Transition.pdf