Identification
Survey ID Number
EGY_LMD_Audit_Study_V1.0
Title
Egypt Labour Market Discrimination Audit Study 2023 (LMD Audit Study)
Country
| Name | Country code |
|---|---|
| Egypt | EGY |
Study type
Randomized controlled audit study of online job posting in Egypt
Series Information
Egypt's labor market is highly segmented, with women concentrated in specific occupations. Women typically earn less than men in the private sector, which is often interpreted as discrimination. This research is Egypt's first audit study, randomizing gender and marital status on applications and resumes to assess the degree of employer discrimination against women in Egypt.
Abstract
Egypt's labor market is highly segmented, with women concentrated in specific occupations (Assaad, Krafft, Rahman, & Selwaness, 2019; El-Hamidi & Said, 2014). In the private sector, women typically earn less than men, even after accounting for differences in their characteristics (El-Hamidi & Said, 2014; Said, Galal, & Sami, 2022). This wage differential is often interpreted as a sign of discrimination against women but could also be driven by unobserved characteristics of women or selection into specific jobs. Norms that prioritize jobs for men over women, when jobs are scarce (Keo, Krafft, & Fedi, 2022), may also lead to employer discrimination in challenging economic times. Egyptian women frequently work in the private sector in advance of marriage and leave private sector work at marriage (Assaad, Krafft, & Selwaness, 2022; Krafft, Assaad, & Keo, 2022; Selwaness & Krafft, 2021). The degree to which exits are driven by women choosing to leave work versus employer discrimination against married women are unknown but have very different implications for interventions to improve women's employment. Generally, understanding the role of employer discrimination and the labor demand side of the market in Egyptian women's low and declining employment rates is a critical but under-researched area. There has been little research on employer discrimination and its impact on hiring in Egypt. This research is Egypt's first audit (correspondence) study, randomizing the gender and marital status on applications and resumes in order to assess the degree of employer discrimination against women in Egypt.
===================================================================================
For more details please see:
Krafft C., (2023). Do Employers Discriminate Against Married Women? Evidence from A Field Experiment in Egypt. International Labor Organization and Economic Research Forum, SWP 2023_1.
https://erf.org.eg/publications/do-employers-discriminate-against-married-women-evidence-from-a-field-experiment-in-egypt/
===================================================================================
===================================================================================
For more details please see:
Krafft C., (2023). Do Employers Discriminate Against Married Women? Evidence from A Field Experiment in Egypt. International Labor Organization and Economic Research Forum, SWP 2023_1.
https://erf.org.eg/publications/do-employers-discriminate-against-married-women-evidence-from-a-field-experiment-in-egypt/
===================================================================================
Kind of Data
Clinical data [cli]
Unit of Analysis
Individual-job posting level. There are four individual observations for each job posting - one single man, one single woman, one married man, one married woman.
Version
Version Description
V1.0: Version 1 of the Egypt Labour Market Discrimination Audit Study (LMD Audit Study).
Production
Date
2023-11
Scope
Notes
Scope of the questionnaire (or gathered information) and the main topics covered:
- Job posting information (job characteristics)
- Personal information of applicants (marital status, gender, age & military status)
- Education of applicants (highest degree obtained, majors & specialization)
- Skills of applicants (soft and technical)
- Languages of applicants
- Employers’ responses (callbacks) to applications
- Job posting information (job characteristics)
- Personal information of applicants (marital status, gender, age & military status)
- Education of applicants (highest degree obtained, majors & specialization)
- Skills of applicants (soft and technical)
- Languages of applicants
- Employers’ responses (callbacks) to applications
Coverage
Geographic Coverage
We randomly sampled the job postings from a range of 13 online job platforms. These online platforms posted jobs available in Egypt. These 13 online platforms were the most commonly used in Egypt with enough details on the job postings. We de-duplicated postings that we found on multiple sites.
Universe
The number of jobs taken from each website was proportional to the number of jobs on the website posted on that day. Different percentages from each platform were calculated depending on the number of jobs posted daily, and we used those percentages to create the weight variables.
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| Economic Research Forum | ERF |
Producers
| Name |
|---|
| Economic Research Forum |
Funding Agency/Sponsor
| Name | Abbreviation |
|---|---|
| The International Labour Organization | ILO |
Sampling
Sampling Procedure
Sampling and weights documents: Since this is an experiment an alternative will be information about the experiment design, number of vacancies and applications (expired not expired) for example and number of responses.
- Experiment design:
This audit study randomized the following characteristics of applicants:
Gender (male/female) & Marital status (unmarried/married)
We collected data on the occupations, industries, and educational requirements of jobs. For positions on websites that required a login and profile, first, we tried to locate the job elsewhere (e.g. on another job search website that does not require a profile, on the employer's website or elsewhere). Second, we tried to find an email for HR to send the resume. If we could not find the job anywhere that does not require a profile and cannot find an email for HR, we recorded the job characteristics, this outcome, and moved to the next position.
Four resumes (one single male; one single female; one married male; one married female) were randomly generated that match the position requirements, but with the specifics randomly different (e.g. name, university attended, etc.). In order to have a manageable number of phones to answer by name, we used only sixteen first names (four for each identity combination). We selected eight common male and eight common female first names (no names that are common for both men and women). Common last names were also selected. First names were randomized across marital statuses and last names across sex and marital status. Resumes included photos. We used artificially generated (composite) photos. Women were shown wearing the hijab (photoshopped onto the generated pictures), since 95% of women aged 15-29 in Egypt wear the hijab (Population Council, 2011). Within gender, photos were randomized in creating resumes. We carefully matched the photos across gender in terms of similar apparent age, skin tone, etc., to avoid any confounders. Photos have neutral backgrounds and avoid any markers of socioeconomic status as much as possible (e.g. in hairstyle or dress). Photos were selected to plausibly cover ages 18-29.
The resumes were designed to be well-qualified for the positions in order to maximize the power of the experiment and the chances of callback (and thus be better able to assess discrimination). The resumes were sent from corresponding email addresses and with specific phone numbers for the randomly generated identity.
- Experiment design:
This audit study randomized the following characteristics of applicants:
Gender (male/female) & Marital status (unmarried/married)
We collected data on the occupations, industries, and educational requirements of jobs. For positions on websites that required a login and profile, first, we tried to locate the job elsewhere (e.g. on another job search website that does not require a profile, on the employer's website or elsewhere). Second, we tried to find an email for HR to send the resume. If we could not find the job anywhere that does not require a profile and cannot find an email for HR, we recorded the job characteristics, this outcome, and moved to the next position.
Four resumes (one single male; one single female; one married male; one married female) were randomly generated that match the position requirements, but with the specifics randomly different (e.g. name, university attended, etc.). In order to have a manageable number of phones to answer by name, we used only sixteen first names (four for each identity combination). We selected eight common male and eight common female first names (no names that are common for both men and women). Common last names were also selected. First names were randomized across marital statuses and last names across sex and marital status. Resumes included photos. We used artificially generated (composite) photos. Women were shown wearing the hijab (photoshopped onto the generated pictures), since 95% of women aged 15-29 in Egypt wear the hijab (Population Council, 2011). Within gender, photos were randomized in creating resumes. We carefully matched the photos across gender in terms of similar apparent age, skin tone, etc., to avoid any confounders. Photos have neutral backgrounds and avoid any markers of socioeconomic status as much as possible (e.g. in hairstyle or dress). Photos were selected to plausibly cover ages 18-29.
The resumes were designed to be well-qualified for the positions in order to maximize the power of the experiment and the chances of callback (and thus be better able to assess discrimination). The resumes were sent from corresponding email addresses and with specific phone numbers for the randomly generated identity.
Weighting
Weights were generated for each platform based on the sampling rate for that platform. Weights were the inverse of the sampling probability times the number of workers required for the posting.
Data Collection
Dates of Data Collection
| Start | End | Cycle |
|---|---|---|
| 2022-06-07 | 2022-10-05 | 1 |
| 2022-12-08 | 2023-02-01 | 2 |
Data Collection Mode
Job Posting information are gathered online. Employers' responses are recorded by phone and by email
Data Collection Notes
ERF seeks IRB (Institutional Research Board) approval for all its data collection exercises to ensure the protection of the rights and welfare of human subjects participating in the research project. All interviews are conditioned upon receiving informed consent (which spells out the respondent's rights) from respondents.
Data were collected on all vacancies sampled, but some were excluded from the experiment. The following exclusion criteria were used:
1. Position is in the public sector or a state-owned enterprise
2. Position is a job working outside of Egypt
3. Position is for non-Egyptians only
4. Position is a volunteering position (unpaid)
5. Position requires a minimum of more than five years of experience
6. Position is senior, executive level
7. Position has extremely specific technical requirements that are beyond our understanding to be able to create a fake resume
8. Position age range does not include 18-29
9. Position or application requires a license or certification be provided (e.g. a medical license)
10. Application requires upload of documents other than a resume and/or cover letter (e.g. a writing sample)
11. Position is at organization requiring a profile and is NOT posted elsewhere and no HR email
12. The company name is not mentioned/confidential
Batch 1:
Total number of vacancies = 2,420 vacancies
Included vacancies = 749 vacancies
Vacancies excluded (for the reasons above) = 1,671 vacancies
Batch 2:
Total number of vacancies = 2,113 vacancies
Included vacancies = 365 vacancies
Vacancies excluded (for the reasons above) = 1,748 vacancies
Total:
Total number of vacancies = 4,533 vacancies
Included vacancies = 1,114 vacancies
Vacancies excluded (for the reasons above) = 3,419 vacancies
- Expired/not expired:
Some vacancies expired before applications could be completed.
Batch 1:
Expired vacancies= 359 vacancies
Not expired= 390 vacancies
Batch 2:
Expired vacancies= 65 vacancies
Not expired= 320 vacancies
Total:
Expired vacancies= 424 vacancies
Not expired= 710 vacancies
- Submitted Applications:
Four resumes were submitted per posting, except when there was a particular gender requirement for that position.
Batch 1:
Submitted 1,470 resumes to 390 job vacancies.
Batch 2:
Submitted 1,206 resumes to 320 job vacancies.
Total:
Submitted 2,676 resumes to 710 job vacancies.
- Callbacks
Batch 1:
The number of callbacks is 145, distributed among six different scenarios (described below). The number of callbacks distributed among the positive scenarios 1, 2, 3 & 4 is 95.
Batch 2:
The number of callbacks is 250, distributed among six different scenarios (described below). The number of callbacks distributed among the positive scenarios 1, 2, 3 & 4 is 244.
Total:
The number of callbacks is 395, distributed among six different scenarios (described below). The number of callbacks distributed among the positive scenarios 1, 2, 3 & 4 is 339.
S1. Scheduling an interview
S2. Asking for additional info
S3. Accepted without interview
S4. Instant interview
S5. Rejection
S6. Not able to get reviewed (this is a scenario where the RAs used to receive a reply from the employer to send the CV using the online profile of the platform, where the job position is advertised on. OR the RAs received an automatic email that the mailbox of the employer is full.)
Data were collected on all vacancies sampled, but some were excluded from the experiment. The following exclusion criteria were used:
1. Position is in the public sector or a state-owned enterprise
2. Position is a job working outside of Egypt
3. Position is for non-Egyptians only
4. Position is a volunteering position (unpaid)
5. Position requires a minimum of more than five years of experience
6. Position is senior, executive level
7. Position has extremely specific technical requirements that are beyond our understanding to be able to create a fake resume
8. Position age range does not include 18-29
9. Position or application requires a license or certification be provided (e.g. a medical license)
10. Application requires upload of documents other than a resume and/or cover letter (e.g. a writing sample)
11. Position is at organization requiring a profile and is NOT posted elsewhere and no HR email
12. The company name is not mentioned/confidential
Batch 1:
Total number of vacancies = 2,420 vacancies
Included vacancies = 749 vacancies
Vacancies excluded (for the reasons above) = 1,671 vacancies
Batch 2:
Total number of vacancies = 2,113 vacancies
Included vacancies = 365 vacancies
Vacancies excluded (for the reasons above) = 1,748 vacancies
Total:
Total number of vacancies = 4,533 vacancies
Included vacancies = 1,114 vacancies
Vacancies excluded (for the reasons above) = 3,419 vacancies
- Expired/not expired:
Some vacancies expired before applications could be completed.
Batch 1:
Expired vacancies= 359 vacancies
Not expired= 390 vacancies
Batch 2:
Expired vacancies= 65 vacancies
Not expired= 320 vacancies
Total:
Expired vacancies= 424 vacancies
Not expired= 710 vacancies
- Submitted Applications:
Four resumes were submitted per posting, except when there was a particular gender requirement for that position.
Batch 1:
Submitted 1,470 resumes to 390 job vacancies.
Batch 2:
Submitted 1,206 resumes to 320 job vacancies.
Total:
Submitted 2,676 resumes to 710 job vacancies.
- Callbacks
Batch 1:
The number of callbacks is 145, distributed among six different scenarios (described below). The number of callbacks distributed among the positive scenarios 1, 2, 3 & 4 is 95.
Batch 2:
The number of callbacks is 250, distributed among six different scenarios (described below). The number of callbacks distributed among the positive scenarios 1, 2, 3 & 4 is 244.
Total:
The number of callbacks is 395, distributed among six different scenarios (described below). The number of callbacks distributed among the positive scenarios 1, 2, 3 & 4 is 339.
S1. Scheduling an interview
S2. Asking for additional info
S3. Accepted without interview
S4. Instant interview
S5. Rejection
S6. Not able to get reviewed (this is a scenario where the RAs used to receive a reply from the employer to send the CV using the online profile of the platform, where the job position is advertised on. OR the RAs received an automatic email that the mailbox of the employer is full.)
Data Collectors
| Name | Affiliation |
|---|---|
| The Abdul Latif Jameel Poverty Action Lab Middle East and North Africa | J-PAL MENA |
Access policy
Access authority
| Name | Affiliation | URL | |
|---|---|---|---|
| Economic Research Forum | ERF | erfdataportal@erf.org.eg | www.erf.org.eg |
Confidentiality
To access the micro-data, researchers are required to register on the ERF website and comply with the data access agreement. The data will be used only for scholarly, research, or educational purposes. Users are prohibited from using data acquired from the Economic Research Forum in the pursuit of any commercial or private ventures.
Access conditions
Licensed datasets, accessible under conditions.
Citation requirements
The users should cite the Economic Research Forum as follows:
"OAMDI, 2023. Egypt Labour Market Discrimination Audit Study (LMD Audit Study) 2023, http://www.erfdataportal.com/index.php/catalog. Version 1.0 of the licensed data files; LMD Audit Study, Egypt: Economic Research Forum (ERF).”
"OAMDI, 2023. Egypt Labour Market Discrimination Audit Study (LMD Audit Study) 2023, http://www.erfdataportal.com/index.php/catalog. Version 1.0 of the licensed data files; LMD Audit Study, Egypt: Economic Research Forum (ERF).”
Disclaimer and copyrights
Disclaimer
The Economic Research Forum has granted the researcher access to relevant data following exhaustive efforts to protect the confidentiality of individual data. The researcher is solely responsible for any analysis or conclusions drawn from available data.
Copyright
(c) 2023, Economic Research Forum
Metadata production
DDI Document ID
EGY_LMD_Audit_Study_V1.0
Producers
| Name | Abbreviation |
|---|---|
| Economic Research Forum | ERF |
Date of Metadata Production
2023-11
DDI Document version
Version 1.0