Energy Transition and Fiscal Preferences: A Survey Experiment (2)

We run an online survey experiment that targets United States residents in one of the top eleven fossil-fuel-rich states ( Alaska, Colorado, Kansas, Louisiana, Montana, New Mexico, North Dakota, Oklahoma, Texas, Utah, and Wyoming) and above the age of eighteen. The survey is designed in Qualtrics and administered on Prolific.

Respondents are first asked to report basic demographic information about themselves.

Respondents are then randomly assigned to one of two experiments. Experiment 1 provides respondents with information about Laramie, WY., and then gauges their i) interest in moving to Laramie, WY., and ii) their expected satisfaction associated with living there. Experiment 2 elicits i) people's stated willingness to leave their state and ii) people's expected life satisfaction of remaining in their state, under various conditions in which their state government has responded (hypothetically) in a certain way to declining fossil-fuel revenue.

After exposure to one of the two experiments described above, subjects are then asked to report their stated preference for various policy responses to declining fossil-fuel revenue in their state.

2026-03-06

2026-03-31

For Experiment 1:
i. Willingness to move to Laramie measured as a self-reported score from 0-100.
ii. Life satisfaction from moving to, and living in, Laramie, Wyoming measured as a self-reported score from 0-100.
iii. (Log) time subjects spend voluntarily collecting additional information about Laramie.
iv. Number of total clicks subjects make when collecting additional information about Laramie.
v. Whether or not a respondent clicked on any of the links providing real-world information about Laramie, WY.

For Experiment 2:

1. Willingness to leave your current state of residence measured as a self-reported score from 0-100.
2. Satisfaction of leaving your current state of residence measured as a self-reported score from 0-100.

Most of the primary outcome variables are directly measured. Time taken reading about Laramie with be transformed using the natural log function.

We will estimate heterogeneous treatment effects by interacting the treatment indicators with the following pre-specified characteristics:
1) the presence of dependents below the age of eighteen, 2) age (used to proxy for retirement status), 3) income, 4) employment status, 5) education, 6) political ideology (measured as a binary variable derived from a left/liberal and right/conservative Likert scale), 7) stated mobility (ranging from rooted, stuck, to mobile), and 8) whether the state a respondent currently lives in currently has an income tax (Alaska, Texas, and Wyoming do not).

In the in-migration experiment, we will additionally explore heterogeneities based on 1) intention to pursue higher education (given that the general information mentions the University of Wyoming), 2) holds a job that allows fully remote work, and 3) reports entrepreneurial intent, 4) is familiar with Laramie.

We will examine treatment effects for the full sample and for the subset of i) respondents who self-report that they are mobile.

At the end of the experiment, respondents will receive a question about their fiscal policy preference in the event of a fossil-fuel revenue decline. They can choose multiple options from different categories of tax raises and spending cut options, along with preferences over which income-groups should face the tax-burden.

For Experiment 1, treatments include i) baseline information about Laramie, WY., ii) information about high quality public goods in Laramie, iii) information about the low-tax climate in Laramie, iv) information about natural amenities in Laramie, or v) all of the above. Subjects are then asked to state their willingness to move to Laramie and their expected life satisfaction associated with living there.

Experiment 2 allows us to answer three main questions:
Q1: Do people respond more strongly to the implementation of a tax as opposed to an increase in an existing one
a. We will do this by comparing stated willingness to move (and expected life satisfaction) among subjects assigned to the “income tax raise” treatment to those in the “income tax implement” treatment. We will also estimate whether the effect of tax implementation is more significant among people who live in an oil-rich state without an income tax (Wyoming, Texas, and Alaska).
b. We will also explore whether people are more opposed to an income tax (regardless of whether it is framed as being “implemented” or “raised”) among people in states without an income tax (Wyoming, Texas, and Alaska) compared to oil-rich states with an income tax.

Test 2: Which type of tax do people prefer?
a. We will test whether people have preferences over how they “lose” $1,000. We do this by testing people’s responses (willingness to move and expected life satisfaction) to three informational treatments: a) a hypothetical situation in which they pay $1,000 more in income tax, b) they pay $1,000 more in sales tax, and c) they lose $1,000 from a wage cut (all as a result of declining fossil-fuel revenue).
b. We will also test whether people have heterogeneous preferences over different types of taxes and/or spending cuts:
a. Income tax increase
b. Sales tax increase
c. Education spending cut
d. Health spending cut

Q3: Do people respond differently to taxes if they are framed as being in response to a negative oil shock?
a. We will compare responses (stated willingness to move and stated life satisfaction) among subjects who are assigned to tax increase treatments that are framed as being in response to a negative energy shock to those among people who are assigned to tax increase treatments that are not framed as being in response to a negative energy shock.

Finally, we will explore people’s preferences for different fiscal responses to negative energy shocks using a within-subject design in which we simply ask people to choose their preferred response. Here, we will explore peoples preferences again based on political ideology, income, and education level.

Not available

Randomization using Qualtrics.

Individual

No

NA

3000

We will have roughly 1,500 respondents in Experiment 1, with roughly 300 respondents in each of the five treatments (including the control). We will also have roughly 1,500 respondents in Experiment 2, with roughly 130 respondents in each of the 11 treatments.

NA