Employer Preferences for Local vs. Non-Local Workers: A Resume Audit

Last registered on February 13, 2023

Pre-Trial

Trial Information

General Information

Title
Employer Preferences for Local vs. Non-Local Workers: A Resume Audit
RCT ID
AEARCTR-0009904
Initial registration date
February 10, 2023

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
February 13, 2023, 11:32 AM EST

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

Locations

Region

Primary Investigator

Affiliation
Carnegie Mellon University

Other Primary Investigator(s)

PI Affiliation
Carnegie Mellon University
PI Affiliation
Carnegie Mellon University

Additional Trial Information

Status
In development
Start date
2022-07-22
End date
2023-07-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
There is significant and persistent heterogeneity in unemployment rates and labor earnings across labor markets in the US, contributing to overall economic inequality. Developments since the 1990s indicate that migration has declined even as job posting websites have expanded access to and awareness of employment opportunities beyond a worker’s location of origin. This project explores the importance of workers' geographic location when applying to online job postings to proxy for access to job opportunities for job seekers who are local and non-local. We examine how job applicant characteristics, specifically a worker's current location, implied gender, and implied career stage/age affect employer contact/callback rates across several occupations and cities in the USA. Methodologically, this study conducts a resume correspondence experiment, randomly assigning the resumes of fictitious candidates with different geographical locations to employers who are actively seeking employees to determine if employers' have a location preference for their new hires. We also address potential mechanisms for why employers may prefer local applicants and engage in place-based discrimination.
External Link(s)

Registration Citation

Citation
Kovak, Brian, Ashley Orr and Lowell Taylor. 2023. "Employer Preferences for Local vs. Non-Local Workers: A Resume Audit." AEA RCT Registry. February 13. https://doi.org/10.1257/rct.9904-1.0
Sponsors & Partners

Sponsors

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 systematically vary the distance between a fictitious job applicant and the labor market the applicant searches for jobs in.
Intervention (Hidden)
We systematically vary the distance between a fictitious job applicant and the labor market the applicant searches for jobs in. To do this, we randomly select addresses for job applicants within a random selection of mid to large (populations greater than 100,000) US cities. We then randomly select zip codes to apply for jobs in within three groups: local and easily commutable (zip codes less than 5 miles from the applicant), non-local but commutable (zip codes between 5 and 50 miles from the applicant), and non-local and far/requiring migration (zip codes 125-500 miles from the applicant). We measure employer contact/callbacks for interview of each fictitious applicant i to job j. We examine how callback rates vary by applicant distance between their home and the jobsite, their implied gender, and their implied stage of their career. We explore multiple cities/labor markets in the US and multiple occupations.
Intervention Start Date
2022-07-25
Intervention End Date
2023-07-31

Primary Outcomes

Primary Outcomes (end points)
The primary outcome of interest is the job application outcome.
Primary Outcomes (explanation)
Generally, job application outcomes are the employer’s contacting the candidate for interview- a categorical variable we define with four categories. One implies the employer is keen to interview the candidate, two implies a callback with ambiguity such as a request for more information, zero implies no response or no contact of the candidate from the employer, and -1 implies the employer contacted the candidate to reject or decline their application. This variable is constructed by monitoring the candidates phone voicemail, text message, job posting site messaging inboxes, and email inboxes.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary job application outcomes which we collect and may consider in our analysis includes duration to callback (a truncated time variable), means of contact (text message, voicemail message, direct message, or email), and strength of contact (neutral or strong). We also record whether the contact content acknowledges the applicants' location relative to the jobsite location or asks about whether they are moving or are willing to move.
Secondary Outcomes (explanation)
The duration to callback is a truncated time variable constructed by counting the time between when the application was sent and when the employer replied (if they replied). The means of contact is how the employer/hiring manager contacted the fictitious candidate: text message, voicemail message, direct message, or email. Strength of contact (neutral or strong) is implied from text analysis of message content- for instance, messages which ask for a return call to set up an interview may be interpreted as neutral, whereas messages which offers the candidate the job or invites them to a training may be interpreted as strong. The location acknowledgement variable is a binary variable that takes the value of one when the contact content mentions or acknowledges the worker's location, asks whether or not they are moving, or asks if they are willing to move.

Experimental Design

Experimental Design
In this resume audit, we consider the effect of the distance between fictitious applicants and job opportunities on the likelihood the employer contacts the applicant back for an interview for the job.

To do this we randomly selected residential addresses in random mid-large US cities (more than 100,000 people) and then randomly select zip codes for the fictitious applicants to apply for jobs in within three distance categories- local/close, local/far commutes, and non-local/requires migration. Our resumes were randomly constructed using the text randomizer from Lahey and Beasley (2009) and our fictitious applicants include randomly selected names which are implied female and implied male from younger and middle aged birth generations in the USA. We consider several occupational titles/job titles and several US cities.

With the assistance of a team of research assistants, we apply to open job postings available on publically available job boards, recording the date, time, and the text from the job posting. We then will monitor the fictitious candidates' email inboxes, voicemail boxes, job site direct message chats, and text messages and record which employers contact them for an interview.

We seek to compare callback rates for candidates by distance, gender, gender and distance, career stage cohort, career stage cohort and gender, and, if possible, age cohort, gender, and distance.
Experimental Design Details
Randomization Method
Randomization occurs by a computer in several aspects of the project. For instance, we randomly select items for the fictitious worker resume construction employing the process/program developed by Lahey and Beasley 2009. Worker names, addresses (within random cities), and experience profiles were randomly selected. Then, when applying for jobs, we randomly select the zip codes to search for jobs in within each of the three distance criteria and randomly select which jobs to apply for on each of the job websites (position of the posting on the page).
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1
Sample size: planned number of observations
10,000 job applications
Sample size (or number of clusters) by treatment arms
1/3 of applications will be close, 1/3 will be middle distance, 1/3 far (migration category)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Unit: firm contact/callback incidence/percentage (15% on average in the pilot- conduct late summer, early Fall 2022); Pilot Sample Standard deviation: .12; Alpha=.05, two sided hypothesis tests, thus with probability .95 we don't reject the null when the null is true and with probability .05 we commit a type I error or a false positive and reject the null when the null is true; Power: 80%, so there is a 20% chance of a type II error- or failing to reject the false null hypothesis of no effect of distance; The Minimum detectable effect size with that sample is: .02 or 2 percentage points difference between close and far candidates, put another way we can detect a 13% reduction in firm contact rates with our proposed sample size.
Supporting Documents and Materials

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

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
Carnegie Mellon University Institutional Review Board
IRB Approval Date
2020-03-06
IRB Approval Number
STUDY2019_00000341
Analysis Plan

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

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials