Happy Hour Electricity Pricing: Household Demand Response to Dynamic Discounts - Evidence from a Field Experiment
How to cite
Montfoort, Floris van.
2026. "Happy Hour Electricity Pricing: Household Demand Response to Dynamic Discounts - Evidence from a Field Experiment." AEA RCT Registry. AEARCTR-0018325.
https://doi.org/10.1257/rct.18325-2.0
Download:
BibTeX
·
RIS— for reference managers without a browser
translator. Zotero and most
others pick up this page's citation metadata directly.
The increasing share of renewable energy has led to more periods with abundant electricity supply and low wholesale prices. However, most households are on fixed-price energy-supply contracts and therefore do not adjust their electricity use in response to these conditions.
This study evaluates an electricity pricing program in which households receive temporary price discounts during hours with negative day-ahead wholesale market prices. The program is implemented as a randomized controlled trial, allowing us to study how households respond to dynamic price incentives. Using high-frequency smart meter data, we examine whether households shift their electricity use toward discounted hours and how total electricity consumption changes. We also explore how responses differ across households and whether real-time feedback on energy consumption and timing of notifications affects demand response. This study contributes to understanding how dynamic pricing can support a more flexible use of electricity in systems with a high share of renewable energy.
Intervention
Households are randomly assigned to a an electricity pricing program that provides a 100% discount on electricity consumption during selected hours when wholesale prices fall below a predefined threshold. Treatment varies in price thresholds, notification timing (day-ahead or intraday), and the provision of real-time feedback via a smartphone application and in-home device.
Experimental design
Randomized controlled trial with household-level assignment to four treatment groups and a control group. Treatment groups differ along two dimensions: the price condition triggering ‘Happy Hour’ discounts and access to real-time feedback via an in-home device (‘dongle’). In addition, notification timing varies within subjects across events (day-ahead versus intraday). All treatments are randomly assigned.
Design details
Randomization method
Eligible population: Households with a smart meter and a contract with the participating energy supplier.
Recruitment: Approximately 20,000 households are randomly selected from the eligible population and invited to participate through the supplier’s smartphone application. Participation is voluntary. Enrollment is capped at approximately 2,500 households, after which recruitment is closed.
Assignment: Participating households are randomly assigned at the household level to one of five groups: a control group or one of four treatment groups. Approximately 600 households are assigned to the control group, with the remaining participants allocated to the treatment groups. Access to real-time feedback via the in-home device (“dongle”) is implemented at a later point during the experiment. Approximately 500 households are randomly selected to receive a dongle, conditional on availability. The timing of provision depends on the readiness of the required software.
Randomization procedure: Assignment to groups is conducted immediately after enrollment using an automated computerized randomization procedure. Each participating household has a known, non-zero probability of being assigned to each group.
Randomization unit
Individual households
Clusters
Approximately 2,500 households
Observations
Approximately 2,500 households, observed at 15-minute intervals using smart meter electricity consumption data.
Arms
500 households control, 700 households Happy Hour below −5 €/MWh, 700 households Happy Hour below −10 €/MWh, 250 households Happy Hour below −5 €/MWh + dongle, 250 households Happy Hour below −10 €/MWh + dongle. Approximately 500 dongles are expected to be available in total.
Primary outcomes
We use 15-minute electricity consumption data to construct our primary outcome variables. Primary outcomes are: (i) hourly electricity consumption (kWh), (ii) intraday load shifting—i.e. electricity consumption during discounted (“Happy Hour”) periods relative to other hours, and (iii) total electricity consumption (daily or monthly kWh).
Secondary outcomes
Smartphone application use is constructed from app usage data, including the number of log-ins and engagement with push notifications (e.g. whether notifications are enabled and whether they are opened/clicked). Survey-based outcomes are constructed from questionnaire responses collected during the study, including self-reported responses, preferences, and household characteristics.
Cited by
No citation data available for this registration yet.