What if your partner works from home more often? A vignette experiment on labor supply

How to cite

Lent, Max van; Meekes, Jordy; Koopmans, Pim; Dur, Robert. 2025. "What if your partner works from home more often? A vignette experiment on labor supply." AEA RCT Registry. AEARCTR-0015637. https://doi.org/10.1257/rct.15637-1.0

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

Max van Lent* , Jordy Meekes , Pim Koopmans , Robert Dur

* Primary investigator

Registry ID
AEARCTR-0015637
DOI
https://doi.org/10.1257/rct.15637-1.0
First registered
Last updated
Status
In development
Trial dates
2025-04-11 – 2025-05-30
Countries
Private
JEL codes
D1, J2, J3, M5
Registry record
http://www.socialscienceregistry.org/trials/15637

Abstract

Gender gaps in the labor market are persistent. One reason for this phenomenon is that anticipation of unexpected events (such as care for family members, animals, and chores) keeps one partner (often women) from working more hours. In this project we study whether people wish to work more hours if their partner works from home more.

We develop a vignette experiment in which respondents choose between jobs that vary in wage, total work hours, and work from home (WFH) hours. Next, we experimentally vary the (fictitious) WFH hours of the partner and analyze the impact of partner's WFH hours on the job choices of the respondent.

Our results will show whether and to what extent partners' WFH hours affect labor supply decisions of an individual.

Intervention

Our intervention is a randomized (fictitious) change in the WFH hours of the respondent's partner, while keeping everything else the same (such as total work hours and partner's earnings).

Experimental design

We develop and implement a vignette experiment. Our sample consists of people that have a partner who has a job, and is about 1200 people.
We ask the respondent several times to state which of two jobs they prefer. Jobs vary by (hourly) wage, paid work hours, the option to choose one’s start and end times of the work day, and number of WFH hours. We randomize the vignettes using a D-efficient design.
After 6 (out of 12) choices we have our treatment. The treatment states that the number of hours the partner works from home has changed (in a randomized way). Then the respondent makes another 6 choices.
Our experiment wants to study the impact of a change in WFH hours of the partner on labor supply. We therefore compare the choices made using the current WFH of the partner and the experimentally randomized increased or decreased hours of WFH from the partner.

Design details

Randomization method
We randomize the vignettes that people get using a D-efficient design. We randomize the WFH hours of the partner by randomly increasing or decreasing the number of hours the partner works from home, in steps of 8 hours.
Randomization unit
We randomize at the individual level, not stratified.
Clusters
1200 respondents
Observations
1200 respondents
Arms
We have 2 treatment arms, one where the WFH of the partner increases, one where it decreases. The size of the increase or decrease varies in steps of 8 hours bounded by 0 and the maximum total number of work hours. Everyone is also in the control group (i.e. first half of the vignettes).
Primary outcomes
The difference in job choices for a respondent when their partner has different WFH hours. This is: number of work hours, WFH hours, and whether the job allows to choose start and end times of the work day.

Cited by

No citation data available for this registration yet.

Keywords