Take-up of social benefits

Last registered on December 06, 2023


Trial Information

General Information

Take-up of social benefits
Initial registration date
November 21, 2023

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
December 06, 2023, 7:44 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.


There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

ETH Zürich

Other Primary Investigator(s)

PI Affiliation
University of Zurich
PI Affiliation
ETH Zurich

Additional Trial Information

On going
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
In a randomized controlled study, we will investigate potential barriers in the application process for health insurance subsidies. For this purpose, all eligible individuals are randomly divided into seven groups. In each treatment group, a barrier in the application process will be addressed and reduced. In the control group, the same procedure as before is implemented. The aim of this study is to understand which measures are most effective in reducing the barrier to take-up of health insurance subsidies and the downstream consequences of these measures.
External Link(s)

Registration Citation

Hangartner, Dominik, Flavia Hug and Michel Marechal. 2023. "Take-up of social benefits." AEA RCT Registry. December 06. https://doi.org/10.1257/rct.10548-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


The experiment is conducted by a public institution in Switzerland. Each year, they send out a letter and an application form to approximately 30% of the households who are likely eligible for the health insurance subsidy. Our field experiment is implemented with 381,582 tax households. The treatment groups will receive different light-touch interventions that aim to reduce administrative barriers and behavioral frictions.

Based on tax data the households are categorized as likely eligible for health insurance subsidies. They are then randomly assigned to one of seven experimental groups: either to the control condition (i.e., business-as-usual) or to a treatment condition (i.e., one of six different flyers). The control condition serves as a baseline for subsidy take-up. Households assigned to one of the treatment conditions will receive the same mailing as the control condition, plus additional information in the form of a flyer. Each flyer will target a combination of frictions to increase the applications and thus increase take-up of the subsidy.

The households receive the subsidy if they submit the application in time and if they are
eligible in that year. True eligibility can only be determined ex-post with the final tax assessment (2-3 years later), thus, the analysis will distinguish between submitting an application and final awarding of the subsidy.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Our primary outcome is whether tax households apply for health insurance subsidies. We will assess this outcome for the entire study population. We will also evaluate the application behavior for the subsample that is deemed eligible. We will rely on the data available at the time and consider data from previous years to predict people’s final eligibility.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
We will further study whether individuals were granted the health insurance subsidy. This depends on their application and their eligibility based on the tax assessment in 2024.

The downstream financial outcomes are whether an individual has an entry in the debt registry and subsequently the health insurer’s certification for losses. These outcomes encompass defaults on various types of bills from the health insurance and are not limited to premium payments.

Long-term outcomes are available from the census data. They include, but are not limited to, labor market outcomes, family constellations (e.g., marriage, relocation, having children), and naturalizations
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
There are six experimental groups and one control group. The control group follows the business-as-usual procedure. They receive a mailing just like in previous years with a letter and an application form. All experimental groups will receive a flyer as a supplement.

Flyer 1 (“Simplification”) reduces complexity by avoiding legal jargon and using comprehensive language and illustrative examples instead. The backside is blank. All other flyers are built on this Simplification flyer.

Flyer 2 (“Cost-Benefit”) adds a distinct message mentioning the financial benefit from receiving the subsidy and the low time costs of applying.

Flyer 3 (“Stigma”) also builds on the Simplification flyer and adds a social proof to reduce stigmatization and by emphasizing a person’s right to social benefits.

Flyer 4, Flyer 5, and Flyer 6 add a back-side to Flyer 1, Flyer 2, and Flyer 3, respectively. The backside is the same for all three flyers and explains in English where to find translations of the German documents into the 11 most spoken languages. Additionally, it repeats some information from the front side in English. This setup creates a 2x3 factorial design.
Experimental Design Details
Not available
Randomization Method
Based on tax data, the project partner classified tax households as potentially eligible or ineligible for health insurance subsidies. Tax households may consist of single individuals, married couples with or without underage children, or adult children. Consequently, a residential household, defined by the living unit, such as an apartment or house, can encompass multiple distinct tax households.

We randomized the potentially eligible tax households to one of seven experimental groups. The randomization is clustered on residential households to reduce spillovers. Residential households were defined based on Swiss federal building and apartment identifiers.

The randomization was conducted by the research team, using the software R. It involved three major steps: First, we created clusters for tax households (e.g., an individual, married couple, or parent with a minor) belonging to the same residential household (e.g., shared flats or young adults living with their parents) to reduce spillovers. Second, we stratified based on whether all individuals in the household are Swiss or not. We do this because Swiss households could differ from mixed and all-foreign households in their attitudes and beliefs regarding social benefits. Third, we randomly created 6 equally sized treatment groups and one control group twice as large.

A balance check of the pre-randomization characteristics suggests that the randomization was successfully implemented.

These groups are then used to create batches that were sequentially forwarded for printing. The envelopes included letters informing tax households about their potential eligibility and the application form. The envelopes from the treatment groups additionally contained the corresponding flyer. The envelopes were assembled within these treatment batches to prevent mistakes. However, everything was dispatched on the same day.
Randomization Unit
Residential households based on federal building and apartment identifier
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
297,631 residential households
Sample size: planned number of observations
381,582 tax households
Sample size (or number of clusters) by treatment arms
Treatment Group 1: 47,681 tax households in 37,204 residential households
Treatment Group 2: 47,589 tax households in 37,204 residential households
Treatment Group 3: 47,849 tax households in 37,204 residential households
Treatment Group 4: 47,792 tax households in 37,204 residential households
Treatment Group 5: 47,508 tax households in 37,204 residential households
Treatment Group 6: 47,721 tax households in 37,203 residential households
Control Group: 94,442 tax households in 74,408 residential households
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Institutional Review Boards (IRBs)

IRB Name
OEC Human Subjects Committee, Faculty of Business Economics and Informatics, University of Zurich
IRB Approval Date
IRB Approval Number
OEC IRB # 2023-034