Looking for a job as a (non-)minority teenager on the German job market. Does previous work experience help to find an apprenticeship? Field experimental evidence on discrimination against (underage) students when looking for a job in Germany

How to cite

Köhler, Ekkehard; Wiemann, Dilara. 2024. "Looking for a job as a (non-)minority teenager on the German job market. Does previous work experience help to find an apprenticeship? Field experimental evidence on discrimination against (underage) students when looking for a job in Germany." AEA RCT Registry. AEARCTR-0012833. https://doi.org/10.1257/rct.12833-2.2

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

Ekkehard Köhler* , Dilara Wiemann

* Primary investigator

Registry ID
AEARCTR-0012833
DOI
https://doi.org/10.1257/rct.12833-2.2
First registered
Last updated
Status
Completed
Trial dates
2024-01-21 – 2025-04-30
Countries
Germany (All States)
JEL codes
C93, D73, J15, H75, I24
Registry record
http://www.socialscienceregistry.org/trials/12833

Abstract

Numerous studies find discrimination in the hiring process of applicants from ethnic minorities (e.g. Bertrand and Duflo 2017, Carlson and Roth 2006; Kaas and Manger 2011, SVR 2014). Apart from an observational study from Norway (Helland and Støren 2006), little is known about pre-K-12 discrimination against students who apply for an apprenticeship or vocational training when leaving middle school. Although ten-thousands are enrolled in this type of vocational training with cooperation and businesses, it has not yet been researched whether prior work experience, grades or even certificates of prior economic knowledge can mitigate the disadvantages of preference-based discrimination (e.g. racism). We conducted field experiments to examine these research questions with empirical data and complemented this by an employer survey.

Pilot study (Campaign 1): In a randomized, controlled pilot study, starting in November 2022, we sent around 18,900 email inquiries to companies that had reported training positions to local job centers in German cities in three business sectors (public administration, industry and services). The study was designed using a block-randomized treatment approach, manipulating four key applicant characteristics: (1) migration background (German, Turkish, Russian), (2) gender (male, female), (3) academic performance (high GPA, satisfactory GPA after 10 years of school with the completion of the technical college entrance qualification), and (4) prior economic knowledge through the EBC*L certificate .

Main study: Campaign 2 (February until May 2024) expanded our approach with three waves of more than 21,200 email inquiries sent nationwide across the same three sectors. In this campaign, we modified our treatment to 3*2*2*2 = 24 dimensions to include: (1) migration background to German, Israeli, and Arab, (2) gender (male, female), (3) academic performance (high GPA, satisfactory GPA after 10 years of school with the completion of the technical college entrance qualification), and (4) prior economic knowledge through internships.

Between April and June 2024, we fielded a survey with employers, confronted them with our findings and asked about their perceptions why young migrant applicants receive less answers compared to German applicants in open-ended questions (Stancheva 2022). The survey was distributed to a stratified random sample of firms that were treated in the previous campaigns. We collected responses from 772 participants. The survey was designed to address three primary research questions:
1) Whether statistical discrimination mechanisms are reflected in respondents' beliefs about productivity and success rates; (2) whether taste-based mechanisms are evident in concerns about workplace integration; and (3) whether perceptions vary systematically across different ethnic groups in ways that align with observed discrimination patterns in our correspondence studies.
The participants reported significant differences in perceived perseverance across migrant backgrounds and varying levels of cultural distance. Both factors might attribute to estimate opportunity costs of hiring a migrant applicant.



Campaign 3 (February until March 2025): To test the robustness of these findings, we designed and implemented a field experiment in the spring of 2025.
We block-randomized the treatments into (2*2*2*3) 24 dimensions at the level of industries and federal states:
Building on insights from the previous campaigns and survey results, we narrowed the focus to Turkish and German candidates to examine specific mechanisms in greater detail and introduced voluntary extracurricular activities to either signal cultural affinity to Germans or the exact opposite as a new treatment dimension: We employed a 2*2*2*3 = 24 dimensions treatment design:
(1) migration background (German, Turkish), (2) gender (male, female), (3) academic performance (high GPA, satisfactory GPA after 10 years of school with the completion of the technical college entrance qualification), and (4) voluntary extracurricular activities. The fourth dimension consists of either participation in a German Turkish cultural association (signalling affinity to Turkish culture), participation in a school-based natural science club (signalling perseverance), or no volunteering experience (control condition).

Between July and August 2025, we fielded a second survey with employers, confronted them with our findings and asked about their perceptions why young migrant applicants receive less answers compared to German applicants in open-ended questions (Stancheva 2022). The survey was distributed to a stratified random sample of firms that were treated in the previous campaign. We collected responses from 572 participants.
The survey was designed to address three primary research questions:
1) Whether statistical discrimination mechanisms are reflected in respondents' beliefs about productivity and success rates; (2) whether taste-based mechanisms are evident in concerns about workplace integration; and (3) whether perceptions vary systematically across different ethnic groups in ways that align with observed discrimination patterns in our correspondence studies.
The participants reported significant differences in perceived perseverance across migrant backgrounds and varying levels of cultural distance. Both factors might attribute to estimate opportunity costs of hiring a migrant applicant.
In addition, we used a treatment that was randomly assigned to respondents as follows:
"Imagine that you have been responsible for selecting trainees at a medium-sized company for several years. Every month, you review dozens of applications and make preliminary decisions. In addition to school grades and resumes, your assessments are also based on personal impressions and operational experience.
Please answer the following questions from this perspective—that is, how you think someone in this role would typically think or act. This is not about your personal opinion, but about the perspective of a human resources manager.

Intervention

In a randomized, controlled study, starting in spring of 2024, we sent around 8 k email inquiries in three business sectors (public administration, industry and services) to companies that had reported training positions to local job centers in German cities. We block-randomized the treatments into (5*2*2*2 = 50) dimensions at the level of industries and federal states: First, we varied the migration background (German, Turkish, Russian, Arab and Jewish) of the applicants by choosing German, Turkish or Russian or Israeli or Arab given sounding names. Second, we varied the gender (female vs. male) of the applicants. Third, we varied the expected grade point average (very good vs. satisfactory) after 10 years of school with the completion of the technical college entrance qualification. The fourth treatment dimension varied previous economic knowledge acquired in an internship.

Experimental design

we sent around 8 k email inquiries in three business sectors (public administration, industry and services) to companies that had reported training positions to local job centers in German cities. We block-randomized the treatments into (5*2*2*2 = 50) dimensions at the level of industries and federal states: First, we varied the migration background (German, Turkish, Russian, Arab and Jewish) of the applicants by choosing German, Turkish or Russian or Israeli or Arab given sounding names. Second, we varied the gender (female vs. male) of the applicants. Third, we varied the expected grade point average (very good vs. satisfactory) after 10 years of school with the completion of the technical college entrance qualification. The fourth treatment dimension varied previous economic knowledge acquired in an internship.

Design details

Randomization method
Randomization is done in office by a computer.
Randomization unit
State level and sectoral level
Clusters
Between 160 clusters with respect to randomization unit (state level). 50 clusters with respect to treatment clusters (three binary dimensions * 5 dimensions for racial background) within randomization unit.
Observations
8k obs are expected.
Arms
8k
Primary outcomes
The primary outcome variable is responsiveness of the addressed business. Responsiveness is measured as follows. If we observe a non-automated response to our treatment email, we mark it as “1”. If we do not observe a non-automated response, we mark it as “0”. (i) We expect that the overall responsiveness towards inquirers without immigrant background is higher compared to sender's responsiveness towards inquirers with immigrant background. (ii) We expect that firm within districts that have a higher share of immigrants answer more frequently to inquirers with an immigrant background than firm in districts that have a low share of immigrants. (iii) We expect that firms are more responsive to male compared to female inquirers. (iv) We expect that firms are more responsive to inquirers with high GPA compared to inquirers with low GPA
Secondary outcomes
Apart from responsiveness we are interested in the tonality and the text length of the written answers to the students. We expect that the tonality also varies across the treatment dimensions. This will be an additional variable we will use once we have a sufficient number of observations. In addition, we are interested in categorizing the answers between helpful answers, rejections and other categories that exemplify good, bad or even ugly answer types. (i) We expect that migrant inquires receive non-helpful answers more often compared to non-migration inquires. (ii) We expect that that migrant inquires receive neutral or even negative answers more often compared to non-migrant inquires. (iii) We expect that non-migrant inquirers receive shorter answers compared to migrant inquires because respondents are spending more time to write non-helpful or negative answers compared to positive answers.

Cited by

No citation data available for this registration yet.

Keywords