Behavioral and Economic Impacts of Tariff Innovations in the Brazilian Electricity Market: Evidence from a Randomized Controlled Trial

How to cite

Meloni, Luis Eduardo Negrão; Lucinda, Claudio; Moita, Rodrigo; Oliveira, Lucas Lyrio de; Pereira, Luan Michel Soares. 2025. "Behavioral and Economic Impacts of Tariff Innovations in the Brazilian Electricity Market: Evidence from a Randomized Controlled Trial." AEA RCT Registry. AEARCTR-0015658. https://doi.org/10.1257/rct.15658-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Luis Eduardo Negrão Meloni* , Claudio Lucinda , Rodrigo Moita , Lucas Lyrio de Oliveira , Luan Michel Soares Pereira

* Primary investigator

Registry ID
AEARCTR-0015658
DOI
https://doi.org/10.1257/rct.15658-1.0
First registered
Last updated
Status
On going
Trial dates
2024-04-01 – 2026-09-30
Countries
Private
JEL codes
D03, D12, D91, L94, O13, Q41, C93, H42, H53
Registry record
http://www.socialscienceregistry.org/trials/15658

Abstract

We evaluate the impact of three innovative electricity tariff modalities—Time-of-Use (ToU), Dynamic Tariff, and Prepaid Tariff—on energy consumption behavior, non-technical losses, and default rates in low-voltage consumer segments in Brazil. The interventions are carried out through randomized controlled trials across three regional utility companies of the Energisa Group. In addition to economic incentives, we test behavioral interventions such as opt-in vs. opt-out default settings and different framing strategies in consumer communication. The experiment aims to inform future regulatory changes and improve tariff design using both economic and behavioral insights.

Intervention

This study evaluates the impact of two innovative electricity tariff modalities implemented by the Energisa Group in Brazil: Time-of-Use Tariff (ToU, known as “Tarifa Melhor Hora”) and Dynamic Tariff. These pricing schemes aim to shift electricity consumption away from peak hours, reduce system overload, and promote more efficient energy use.

The intervention covers three utility concession areas: Energisa Sul Sudeste (ESS), Energisa Paraíba (EPB), and Energisa Tocantins (ETO). In each region, eligible low-voltage consumers (residential and small commercial) were randomized into treatment and control groups separately for each tariff type.
• ToU Tariff introduces fixed off-peak and peak prices, communicated in advance to participants.
• Dynamic Tariff uses variable pricing, with day-ahead price signals sent to consumers, incentivizing real-time demand response.

Participants in the treatment group receive the new tariff along with targeted communications from the utility provider explaining how the pricing works and how they can benefit by adjusting their consumption patterns. Communications are tailored by consumption class and region and may include energy-saving tips, example bills, and visual cues.

Control group participants continue under the default tariff, without any changes to pricing or communication.

Smart meters were installed in all selected units prior to the intervention to allow for high-frequency monitoring of consumption. The intervention will last for at least 12 months, with additional monitoring post-intervention.

Randomization was conducted at the unit (UC) level, stratified by class (residential vs commercial) and historical energy usage quintiles.

Experimental design

This study uses a randomized controlled trial (RCT) to evaluate the behavioral and economic impacts of two electricity tariff innovations—Time-of-Use (ToU) and Dynamic Tariff—implemented by the Energisa Group across three utility concession areas: Energisa Sul Sudeste (ESS), Energisa Paraíba (EPB), and Energisa Tocantins (ETO).
Within each concession area and for each tariff type, a set of eligible low-voltage electricity consumers (residential and small commercial) was identified based on consumption history and administrative criteria (e.g., exclusion of consumers with distributed generation, social tariff beneficiaries, excessive disconnections, etc.).
The experimental design involved:
1) Separate randomizations for each tariff and region (6 total experiments: 3 ToU and 3 Dynamic Tariff);
2) Stratification by consumption quintile and customer class;
3) Random assignment of units to treatment or control groups, with a 1.3:1 treatment-to-control ratio;
4) Unit of randomization: electricity consumption unit (UC);
5) Randomization method: performed using R and Python scripts;
6) Sample selection included oversampling: six UCs were selected for each sampling unit to increase the chance of successful installation. One UC was chosen by the utility from each set based on practical considerations (same class, same municipality, same quintile).

Control group consumers continue under the standard tariff. Treatment group consumers receive the new pricing scheme, plus targeted communication encouraging demand response. High-frequency metering data is collected for all units to allow precise estimation of treatment effects.

Design details

Randomization method
Randomization was performed using R and Python scripts.
Randomization unit
The unit of randomization is the electricity consumption unit (UC), which corresponds to an individual metering point (typically one household or small commercial consumer). Randomization was conducted separately for each tariff type (ToU and Dynamic) and across the three concession areas of the Energisa Group (ESS, EPB, and ETO). Each UC was randomly assigned to either treatment or control, stratified by customer class (residential vs. commercial) and historical consumption quintiles.
Clusters
Not applicable. The treatment is individually randomized at the electricity consumption unit (UC) level.
Observations
31,063 electricity consumption units (UCs) Breakdown by tariff and region (limited by available smart meters): • ToU – ESS: 1,063 UCs • ToU – EPB: 3,000 UCs • ToU – ETO: 3,000 UCs • Dynamic – ESS: 6,000 UCs • Dynamic – EPB: 6,000 UCs • Dynamic – ETO: 6,000 UCs
Arms
Each regional experiment (ToU or Dynamic) follows a 1.3:1 treatment-to-control allocation. Approximate numbers by treatment arm: • ToU – ESS: ~603 treatment / ~460 control • ToU – EPB: ~1,730 treatment / ~1,270 control • ToU – ETO: ~1,730 treatment / ~1,270 control • Dynamic – ESS: ~3,429 treatment / ~2,571 control • Dynamic – EPB: ~3,429 treatment / ~2,571 control • Dynamic – ETO: ~3,429 treatment / ~2,571 control
Primary outcomes
1. Monthly electricity consumption (kWh): Measured at the consumer level through smart meters. This is the primary behavioral outcome, used to assess the effect of each tariff modality on overall energy use. 2. Peak-period consumption (kWh): Total energy usage during utility-defined peak hours. This outcome captures behavioral response to price signals and the effectiveness of shifting consumption.
Secondary outcomes
1. Bill amounts and payment regularity: Total bill value (R$) and the number of on-time payments. This captures the financial burden of the intervention and potential risk of default. 2. Household perceptions and behavioral responses: If implemented, a follow-up survey will collect self-reported data on consumption habits, perceived understanding of the tariff, satisfaction with service, and ownership of household appliances. These outcomes will help assess mechanisms of behavioral change and heterogeneous impacts.

Cited by

No citation data available for this registration yet.

Keywords