Perceived Productivity and the Choice of Flexible Work Arrangements: Evidence from a Randomized Experiment in China

How to cite

Shi, Xinzheng; Liu, Chunru; Lu, Yao; Zhou, Yujun. 2026. "Perceived Productivity and the Choice of Flexible Work Arrangements: Evidence from a Randomized Experiment in China." AEA RCT Registry. AEARCTR-0017718. https://doi.org/10.1257/rct.17718-1.0

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

Xinzheng Shi* , Chunru Liu , Yao Lu , Yujun Zhou

* Primary investigator

Registry ID
AEARCTR-0017718
DOI
https://doi.org/10.1257/rct.17718-1.0
First registered
Last updated
Status
Completed
Trial dates
2024-11-20 – 2024-12-24
Countries
China (Beijing)
JEL codes
J24, J33, D83
Registry record
http://www.socialscienceregistry.org/trials/17718

Abstract

We conduct a randomized experiment to investigate whether job seekers have a perception of their work performance in flexible work arrangements and whether providing information on the scientific findings regarding work performance can improve their preference for flexible work arrangements. We find that job seekers have a clear perception of their work performance in flexible schedule work arrangements but not in flexible workplaces or a combination of both. We further find that providing information on scientific findings concerning work performance in flexible work arrangements can increase job seekers’ probability of choosing such arrangements. The impact is stronger among less experienced job seekers and applicants to positions requiring less face-to-face communication. Finally, structural estimates show that information interventions shape preferences by significantly updating beliefs about productivity under flexible work arrangements.

Experimental design

We collaborate with a financial consulting company and a job-posting platform to conduct a randomized experiment in China. The company provides eight positions for recruitment. We combine an information intervention experiment and a discrete choice experiment. Specifically, the company provides three flexible work arrangements, including a flexible workplace (i.e., workers can choose any place to work from 9 AM to 5 PM), a flexible schedule (i.e., workers need to work in the company but can choose a work schedule for eight hours between 7–10 AM and 3–6 PM), and a flexible workplace and schedule (i.e., workers can choose any place to work for eight hours between 7–10 AM and 3–6 PM). In addition, the company provides the traditional fixed work arrangement (i.e., workers need to work in the company for eight hours from 9 AM to 5 PM).
For each position, we divide eligible job seekers into three random groups, corresponding to the three flexible work arrangements. Each job seeker in these three groups is sent a job advertisement with two options (i.e., a fixed work arrangement and one of the three flexible work arrangements). The fixed work arrangement is accompanied by a market wage, but the wage for the flexible work arrangement is randomly drawn from a set of wages ranging from 80% to 120% of the fixed work arrangement wage. This is a typical setting of a discrete choice experiment.
In addition, we provide three interventions. One intervention involves adding a performance–pay compensation package to the job advertisement (the performance–pay treatment). This intervention is used to test whether job seekers perceive flexible work arrangements as a way to improve work performance. If job seekers believe that flexible work arrangements can improve their performance, those receiving an advertisement with a performance–pay compensation package would be more likely to choose flexible work arrangements. Another intervention is to add to the job advertisement a statement that scientific research finds that flexible work arrangements can improve workers’ performance (work productivity treatment). The last intervention is a combination of the above two interventions (combined treatment). To conduct the intervention experiment, we randomly divided each of the three groups (corresponding to the three flexible work arrangements) into four subgroups. The above three interventions are assigned to three groups, and we have a group without any intervention as the control group.

Design details

Randomization method
The randomization was done in office by a computer.
Randomization unit
In each job position, we firstly divide job seekers randomly into three groups. Then, in each group, we randomly divided the job seekers into four groups.
Clusters
Eight positions.
Observations
5,538 job seekers
Arms
Roughly one quarter of the total observations control, and the three fourth as treatment.
Primary outcomes
Job seekers' choice of flexible work arrangements

Cited by

No citation data available for this registration yet.

Keywords