Consumer Acceptance of Residential Electricity Tariff Reform

The study employs a sequential modular design consisting of two distinct components. To ensure that structural preferences are elicited without bias from choice-architecture interventions, respondents first complete a Discrete Choice Experiment (DCE), followed by a Randomized Decision-Structure Experiment.

[Intervention 1: Discrete Choice Experiment (DCE)]
The DCE evaluates latent preferences for residential electricity tariff designs. We use a fractional factorial design to construct systematic choice sets. Respondents are randomly assigned to one of two blocks, each containing five choice tasks. In each task, participants choose between the status quo and hypothetical tariff profiles defined by five attributes:
1.Seasonality: Peak pricing season adjustments.
2.Time-of-Use (ToU) Structure: The specific configuration and frequency of peak vs. off-peak blocks.
3.Price Differential: The magnitude of the price ratio between peak and off-peak periods.
4.Adoption Incentive: Short-run monetary benefits provided upon switching.
5.Bill Variation: The expected change in the monthly bill, presented as percentage changes ranging from −10% to +10% relative to the respondent’s self-reported baseline.
Across respondents, the reference option (the status quo against which alternatives are compared) is randomized at three levels to test for reference dependence: current progressive tariff, current tariff +5% increase, and current tariff +10% increase. The magnitude of the price ratio between peak and off-peak periods. Additionally, we elicit the "inertia range"—the specific bill-increase threshold below which a respondent remains passive (i.e., would not alter consumption or switch tariffs).

[Intervention 2: Decision-Structure and Information Experiment]
Following the DCE, we implement a 2 by 2 between-subjects factorial design embedded within a within-respondent pre-post design. This component identifies how choice architecture influences the acceptance of price normalization.
1.Baseline (Ex-ante): We measure the respondent's baseline acceptance of a general electricity price increase on a Likert scale of 10, without offering any mitigating alternatives.
2.Randomized Intervention: respondents are offered a specific alternative tariff option. They are randomly assigned to one of four treatment arms, varying by Default Setting (Opt-in vs. Opt-out) and Information Framing (Neutral vs. Positive):
-T1 (Control, Opt-in): Participant must actively switch; neutral description.
-T2 (Control, Opt-out): Alternative is set as the default; neutral description.
-T3 (Treatment, Opt-in): Participant must actively switch; framing emphasizes recommendations and system efficiency.
-T4 (Treatment, Opt-out): Alternative is set as the default; framing emphasizes social norms and economic rationality.
3.Post-Treatment (Ex-post): We re-measure the respondent's acceptance of the general price increase. This allows us to calculate the acceptance increase due to the causal effect of providing choice, defaults, and framing on the political feasibility of price reform.

2025-12-22

2025-12-31

[Discrete Choice Experiment: Tariff Choice Behavior]
The primary outcome of the discrete choice experiment is individual tariff choice behavior, recorded as a binary indicator across paired choice tasks. Using conditional and mixed logit specifications, we utilize these choices to estimate the marginal utility weights of key tariff attributes, specifically identifying the trade-offs consumers make between short-run adoption incentives and longer-run expected monthly bill stability.

[Decision-Structure and Information Experiment: Acceptance Lift]
The primary outcome is the Average Treatment Effect (ATE) on the change in stated acceptance of electricity price normalization. This is measured as the within-subject difference on a 10-point Likert scale. This measure identifies the "Acceptance Lift" generated by the introduction of an actionable alternative, directly testing the hypothesis that instrumental agency reduces the political cost of reform.

[Discrete Choice Experiment: Structural Parameters]
We will estimate consumers’ willingness to pay (WTP) for specific tariff attributes (e.g., the premium consumers are willing to pay for reduced seasonality). We will also examine inertial thresholds by measuring the degree of status-quo stickiness. Specifically, we identify an inertia range—defined as the interval of bill increases over which the probability of switching remains statistically indistinguishable from zero. Finally, we assess reference dependence by analyzing shifts in choice probabilities conditional on randomized reference baselines (status quo, +5%, or +10% bill increases).

[Decision-Structure and Information Experiment: Behavioral Mechanisms]
For the decision-structure experiment, secondary outcomes capture the mechanisms of behavioral response. First, we measure the tariff adoption rate, the binary outcome of whether the respondent "signs up" for the alternative tariff under the assigned condition (T1–T4). This measures the power of defaults (opt-out) versus active choice (opt-in). Second, we calculate the proportion of respondents who shift from opposition (baseline) to support (post-intervention). The proportion of "Reform Opponents" (defined as baseline scores ≤ 4) who transition to "Reform Supporters" (post-intervention scores ≥ 6). This distinguishes between marginal shifts in opinion and categorical changes in political stance.

This study employs a sequential experimental design comprising two integrated modules: a discrete choice experiment followed by a decision-structure experiment, using a nationally representative sample of 2,200 adults in South Korea. We utilize a nationally representative sample of 2,200 South Korean adults, recruited via a professional survey firm and stratified by age, gender, and geographic region.

[Module 1: Discrete Choice Experiment (DCE)]
In the first module, participants complete a Discrete Choice Experiment (DCE) designed to elicit structural preferences over alternative residential electricity pricing schemes. The DCE consists of 10 choice sets generated via a fractional factorial design to maximize statistical efficiency while maintaining orthogonality. To minimize cognitive burden, participants are randomly assigned to one of two blocks, completing five binary choice tasks each. Each choice task presents two alternative pricing profiles that vary systematically across five attributes: seasonality, structure, price differential, adoption incentive, and bill variation. Crucially, the expected monthly bill change attribute is personalized for each participant. It is dynamically pivoted based on the respondent’s self-reported average monthly electricity bill and presented as a percentage change within a predefined range (e.g., -10% to +10%). This design feature ensures that the trade-offs presented are financially salient and realistic for each household.

[Module 2: Decision-Structure and Information Experiment]
This module constitutes the core RCT testing the drivers of policy acceptance. Participants are randomly assigned at the individual level to one of four experimental groups in a 2 by 2 between-subjects factorial design, varying decision structure (opt-in vs. opt-out default) and information framing (neutral vs. positively framed information). All participants first report their baseline acceptance of residential electricity price increases in the absence of any alternative pricing option. Participants are then exposed to an alternative electricity pricing scheme under their assigned decision-structure and information condition and report post-intervention acceptance. Treatment effects are identified through changes in acceptance between the pre- and post-intervention measures and comparisons across experimental groups.

Not available

Computer-based stratified is implemented via the online survey platform’s algorithm. To ensure covariate balance across the four experimental arms, the randomization sequence is stratified by key demographic characteristics. Specifically, the algorithm balances the sample based on gender, region of residence, and age groups. Within each stratum, respondents are independently assigned to one of the four treatment conditions with equal probability.

Individual. Randomization occurs at the level of the individual survey respondent, conditional on the demographic strata defined by age, gender, and region.

No

N/A. individual level randomization

2,200 individual adults’ participants

The target sample size is N = 2,200, providing approximately 550 participants per arm. This sample size is powered to detect small-to-moderate treatment effects in both the adoption rates and the shift in Likert-scale acceptance.
- Control opt-in (approximately 550 participants)
- Control opt-out (approximately 550 participants)
- Treatment opt-in (approximately 550 participants)
- Treatment opt-out (approximately 550 participants)
Stratified randomization is employed to ensure balanced representation of age groups, gender, and regional residence across all four arms. All 2,200 participants in the main sample also complete the DCE module.

This study is designed to achieve 90% statistical power at a 5% significance level (two-tailed test). With approximately 550 participants per experimental group, the study is powered to detect small-to-moderate effect sizes in the primary outcome—changes in acceptance of residential electricity price increases measured on a 10-point Likert scale. The exact minimum detectable effect size (MDE) depends on the variance of the outcome measure and the correlation between pre- and post-intervention responses.