A Field Experiment on the Hiring Gap in AI Skills and Gender in China’s Campus Recruiting Labor Market

How to cite

Li, Yaping; Ma, Rong; Xu, Zhufeng. 2024. "A Field Experiment on the Hiring Gap in AI Skills and Gender in China’s Campus Recruiting Labor Market." AEA RCT Registry. AEARCTR-0014640. https://doi.org/10.1257/rct.14640-1.0

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

Yaping Li* , Rong Ma , Zhufeng Xu

* Primary investigator

Registry ID
AEARCTR-0014640
DOI
https://doi.org/10.1257/rct.14640-1.0
First registered
Last updated
Status
On going
Trial dates
2024-10-31 – 2025-06-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14640

Abstract

The rapid development of AI technologies in recent years is reshaping the labor market. This study investigates how AI skills affect employment outcomes and whether they interact with gender. A large-scale resume field experiment will be conducted in China, sending thousands of fictitious resumes to potential employers. The resumes will vary randomly in terms of applicants’ AI skills and gender. We also explore the potential heterogeneity in outcomes based on applicant characteristics and job attributes.

Experimental design

We will implement a resume audit experiment. For each selected job posting, a collection of four fictitious resumes will be submitted. Randomization will occur across these four resumes, categorized as follows:
• Group 1: Men with AI skills
• Group 2: Women with AI skills
• Group 3: Men without AI skills
• Group 4: Women without AI skills
The study aims to examine the causal effects of AI skills based on the callback rates observed across these four groups, and whether they interact with gender.

Additional resume details will be sourced from a pre-constructed resume bank and randomly assigned to each resume. Every resume will be proofread by the researchers to ensure authenticity. This study will also explore outcomes variations based on applicant traits and job attributes.

Resumes will be submitted to 2,500 public job postings, totaling 10,000 applications.

Design details

Randomization method
Randomization will be done in office by a computer.
Randomization unit
Resumes level randomization
Clusters
2,500 public job postings, 4 fictitious resumes for each job posting.
Observations
10,000 fictitious resumes
Arms
2,500 fictitious resumes representing men possessing AI skills, 2,500 fictitious resumes representing women possessing AI skills, 2,500 fictitious resumes representing men lacking AI skills, and 2,500 fictitious resumes representing women lacking AI skills. The fictitious resumes without AI skills are treated as the control groups.
Primary outcomes
Response rates to job application in China

Cited by

No citation data available for this registration yet.

Keywords