The Role of Long-Term Unemployment in the Hiring Process: Evidence from a correspondence study in Italy

How to cite

Gatta, Arianna; Spiezio, Mario. 2025. "The Role of Long-Term Unemployment in the Hiring Process: Evidence from a correspondence study in Italy." AEA RCT Registry. AEARCTR-0015852. https://doi.org/10.1257/rct.15852-1.0

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

Arianna Gatta* , Mario Spiezio

* Primary investigator

Registry ID
AEARCTR-0015852
DOI
https://doi.org/10.1257/rct.15852-1.0
First registered
Last updated
Status
Completed
Trial dates
2019-06-01 – 2020-07-31
Countries
Italy ()
JEL codes
J64
Registry record
http://www.socialscienceregistry.org/trials/15852

Abstract

Employers play a critical gatekeeping role in the labour market, yet much research on labour market discrimination overlooks how employers interpret candidates’ employment histories when they intersect with experiences of unemployment, welfare use and interrupted careers. Furthermore, the intersection between career histories and aspects such as gender and ethnicity is still poorly understood. Existing studies often rely on workers' outcomes to infer employer behaviour, obscuring the mechanisms employers use to assess employability.
To investigate these dynamics, this study employs a field experiment using an online correspondence design. Fictitious job applications are sent to real job advertisements in the fields of Marketing, Administration, and HR. Each employer/vacancy receives four CVs that differ randomly on characteristics such as unemployment duration, participation in Active Labour Market Policies, gender and ethnicity. We then record the callback rate for each resume and test whether employers' hiring behaviour is shaped by employment history and individual characteristics. This experiment will inform multiple publications on the topic of labour market discrimination.

Intervention

Fictitious resumes were sent in response to online job ads, advertised online between September 2019 and June 2020. The dependent variable of this study is the callback rate from the resumes. Resumes were randomised to display 0, 6, 14 or 22 months in unemployment from the last job, male or female-sounding names, as well as northern-Italian, southern-Italian and Romanian-sounding surnames. Resumes were randomised to show either one or two jobs, and either a regular internship,a subsidised internship or an unemployment spell. We also randomised whether the candidate attended high school in the same city where the job was advertised or in a different one. We also randomized four additional information treatments that provided the following information with a 50% chance each: (1) motivation statement, (2) job-specific IT skills, (3) ongoing training, (4) volunteering.

Experimental design

Fictitious resumes were sent in response to online job ads, advertised online between September 2019 and June 2020. The dependent variable of this study is the callback rate from the resumes. Resumes were randomised to display 0, 6, 14 or 22 months in unemployment from the last job, male or female-sounding names, as well as northern-Italian, southern-Italian and Romanian-sounding surnames. Resumes were randomised to show either one or two jobs, and either a regular internship,a subsidised internship or an unemployment spell. We also randomised whether the candidate attended high school in the same city where the job was advertised or in a different one.
For this experiment, we did not use a paired design. We also randomized four additional information treatments that provided the following information with a 50% chance each: (1) motivation statement, (2) job-specific IT skills, (3) ongoing training, (4) volunteering.

Design details

Randomization method
Applications are generated using a customised program that populates templates with real job experience descriptions using text that is randomly assigned.
Randomization unit
Treatments on each resume were randomized independently of other resumes sent to the same and other employers. By doing this we avoided spillover effects that might arise as a result of keeping the pool of candidates fixed . Therefore the unit of randomization is the single resume.
Clusters
We targeted 1041 employers/vacancies, which received about 4 resumes each.
Observations
We sent out 4066 resumes to 1041 employers. Each employer received about 4 resumes. A minority of employers received less than 4 resumes. In most cases this happened because the job post was not available anymore.
Arms
We do not have treatments arms in this study. Resumes were randomized within and between employers
Primary outcomes
The dependent variable of this study is the employers' callback rate from the resumes.

Cited by

No citation data available for this registration yet.

Keywords