Extended unpaid vacation time

How to cite

Bloom, Nicholas; Linxi. 2026. "Extended unpaid vacation time." AEA RCT Registry. AEARCTR-0018268. https://doi.org/10.1257/rct.18268-1.1

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

Nicholas Bloom* , Linxi

* Primary investigator

Registry ID
AEARCTR-0018268
DOI
https://doi.org/10.1257/rct.18268-1.1
First registered
Last updated
Status
In development
Trial dates
2026-04-15 – 2028-10-31
Countries
Private
JEL codes
J0
Registry record
http://www.socialscienceregistry.org/trials/18268

Abstract

We collaborate with Trip.com to pilot a "Flexible Personal Leave" policy within the firm: Employees may optionally apply for additional unpaid personal leave without providing specific justification, subject to completed work handover (unpaid leave, benefits unchanged, pilot cap: 45 days/year per person). Trip wants to introduce this as they believes it will attract employees and improve retention as many employees will find this extremely valuable. We aim to assess the take-up and effects of this policy, in particular, how take-up varies across demographics, employee seniority and position, etc., and how the policy affects worker productivity and promotion, firm recruitment and retention, etc.

Intervention

Trip is randomizing in an A/B test offering employees up to 45 days of additional unpaid leave within a 12 month time frame. There are some constraints on this, particularly managers being OK that work is not suffering, but otherwise it should be allowed. Employees can take 1 day a week or perhaps a couple of longer breaks. Data on employee performance, retention, recruitment, promotions and surveys will be collected and analyzed.

Experimental design

6,000 employees at Trip.com will be asked to complete a survey gauging their interest in the flexible personal leave policy. Employees are then randomized into two groups by odd/even birthdays, and the winning group gets the option to take flexible personal leave as specified while the losing group does not (they remain as now - they have regular company leave, national holidays etc. but not additional option for extra unpaid leave). The firm will pilot this for a year (May 1, 2026-April 30, 2027). After the trial, we will compare the treated and control groups' policy take-up, worker and management perception of policy, worker productivity and promotion, recruitment and retention. We hope to evaluate this over the next 12+ months (in an earlier WFH experiment with Trip we collected 18 months of follow-up data).

Design details

Randomization method
Employees are randomized into two groups by odd or even birthdays. Trip.com's former founding CEO and current Chairman James Liang or another authority figure will publicly draw a ball to determine whether odd or even birthdays are treated, so everyone can see it is fair and random.
Randomization unit
Individual employee
Clusters
About 6000 Individual employees - Trip are leading the experiment and we believe they are including all the employees of four divisions.
Observations
About 6000 Individual employees.
Arms
3000 employees treated (able to take flexible personal leave), 3000 employees control (not able to take flexible personal leave).
Primary outcomes
policy take-up, worker and management perception of policy, worker productivity and promotion, firm recruitment and retention, and survey data

Cited by

No citation data available for this registration yet.

Keywords