The Role of Information about Worker Skills in Hiring Decisions

Last registered on April 01, 2026

View Trial History

Pre-Trial

Trial Information

General Information

Title

The Role of Information about Worker Skills in Hiring Decisions

RCT ID

AEARCTR-0017993

Initial registration date

March 25, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

April 01, 2026, 9:49 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

France

Region

Primary Investigator

Name

Mitia Oberti

Affiliation

CREST (ENSAE, Institut Polytechnique Paris)

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Marion Brouard

PI Affiliation

ifo - LMU

Contact Investigator

PI Name

Roland Rathelot

PI Affiliation

ENSAE - Institut Polytechnique de Paris

Contact Investigator

Additional Trial Information

Status

Completed

Start date

2025-01-30

End date

2025-06-30

Keywords

Gender, Labor

Additional Keywords

hiring, skills, competencies, experience, education, interviews, screening, applications, information frictions

JEL code(s)

J23, J24, J62

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

Labour markets differ widely in candidate availability and quality across occupations. In tight occupations, firms face few qualified applicants per vacancy and may struggle to identify productive candidates, especially when applicants come from distant occupations where skill signals are less clear. In slack occupations, recruiters face many applicants and screening becomes more selective. These differences may shape not only hiring decisions, but also the value of additional information about job seekers' skills. We study these mechanisms using an incentivised resume-rating (IRR) experiment with recruiters. Each recruiter evaluates synthetic CVs tailored to their hiring occupation. We experimentally vary education, work experience (in terms of occupational distance between experience and hiring occupation), skill certificates, as well as gender and ethnicity signals within recruiters. We also randomise the quality of the applicant pool between recruiters. We study the impact of these signals on the way recruiters assess productivity, and on their callback decisions.

External Link(s)

Registration Citation

Citation

Brouard, Marion, Mitia Oberti and Roland Rathelot. 2026. "The Role of Information about Worker Skills in Hiring Decisions." AEA RCT Registry. April 01. https://doi.org/10.1257/rct.17993-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

We conduct an incentivised resume-rating (IRR) experiment with professional recruiters. Each recruiter evaluates a pool of 20 synthetic CVs tailored to their hiring occupation.

Within the 20 CVs, we experimentally vary candidate attributes: (i) education, (ii) work experience (in terms of occupational distance between experience and hiring occupation), (iii) gender and ethnicity signals, and (iv) skill certificates.

In addition, we vary the quality of the pool of applicants across recruiters by changing the composition of the 20 CVs.

Intervention (Hidden)

Intervention Start Date

2025-01-30

Intervention End Date

2025-06-01

Primary Outcomes

Primary Outcomes (end points)

- Callback decision in the pool choice task (dummy): whether the candidate is selected for callback when the recruiter chooses among the 20 CVs in the experimental pool.
- Callback decision in the standard-pool task (dummy): whether the recruiter would call back the candidate if the CV were part of their usual applicant pool for similar vacancies (elicited CV-by-CV).
- Perceived productivity (graded by recruiters between 0 and 10).

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

- Perceived retention probability (graded by recruiters between 0 and 10).
- Perceived probability that the candidate would accept an offer if made (graded by recruiters between 0 and 10).
- Weights allocated to CV attributes (education, experience, soft skills, others)
- Interest for skill certification services

Secondary Outcomes (explanation)

Relevant heterogeneity dimensions (non-exhaustive list) : tightness (pool quality), occupational distance, gender and name ethnicity (of candidates and recruiter), relationship with the skill certification provider (PES)

Experimental Design

We conduct a clustered randomised trial in which recruiters are the randomisation clusters and CV evaluations are the observation units.
- At the individual level, we first randomise assignment to 6 kinds of profiles defined as combination of education (either ideal education for hiring occupation, or another diploma of same level but targeted for another occupation), experience (either in the hiring occupation or in a different one, with two groups of occupational distance), and skill certificates. Independently from the profiles, we randomly assign combinations of gender and ethnicity (French sounding names and North-African sounding names). Finally, we randomly assign home address and CV templates.
- At the cluster level, there are three treatment arms, corresponding to three levels of quality for the pool of candidates. In practice, we vary the number/share of profiles with education and work experience matching the target occupation.

Experimental Design Details

Randomization Method

- Cluster level: deterministic function of meaningless identifier.
- Individual level: pseudo random number generator in statistical software.

The exact randomization design is presented in the additional randomization_notes document.

Randomization Unit

- Cluster: recruiter
- Individual: application to a given recruiter.

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

Around 100 recruiters.

Sample size: planned number of observations

Around 2000 (100 recruiters with 20 applications each).

Sample size (or number of clusters) by treatment arms

- Cluster level: around 33 per treatment arms.
- Individual level:

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

We target a sample of 100 recruiters, each evaluating 20 synthetic CVs (2,000 CVs total). Recruiters are randomized into three pool quality arms (weak/medium/strong, 33 each) that vary the share of high-quality profiles. Within each recruiter's batch, CV-level treatments (work experience, skill certificate) are randomized across profiles. All calculations use a two-sided test at the 5% significance level with 80% power. Baseline parameters are estimated from 5 pilot interviews. The primary outcome is a binary callback decision. The baseline callback rate among control profiles (wrong education, wrong experience, no skill certificate) is p0 = 0.50, with a unit SD of 0.50. The recruiter-level ICC is 0.71. 1. Effect of relevant work experience : within-cluster comparison of Profile 3 (wrong education, right experience) vs P4/5 (wrong education, wrong experience). MDE = 0.063 (6.3 pp, or 12.7% of baseline). 2. Effect of skill certificate. Within-cluster comparison of Profile 6 (wrong education, wrong experience, skill certificate) vs P4/5 (wrong education, wrong experience, no certificate). MDE = 0.044 (4.4 pp, or 8.9% of baseline). 3. Effect of pool quality : between-recruiter comparison (strong vs weak pool). The unit of analysis is the recruiter-level callback rate. MDE is 0.30 (all CVs) and 0.33 (skill certificates profiles only)

Supporting Documents and Materials

IRB