The Role of Pay Transparency in Migrants’ Labor Market Integration

We collect data in an online survey experiment in which respondents are randomly provided with information on average wages in their 3-digit occupation, gender, age, education, and commuting zone, following a similar setup as in Jäger et al. (2024).
We study: (i) whether receiving the information treatment affects respondents' reservation wages, wage expectations, firm switching, and wage negotiation intentions; (ii) whether receiving the information treatment affects respondents' realized outcomes in the German social security data.
The experiment takes place in the second wave of a large survey of recent migrants to Germany, conducted by the Institute for Employment Research (IAB). We incentivize participation in the survey with a €10 voucher, which respondents receive at the end of the survey.

2026-07-29

2027-03-31

(1) Survey-based outcomes
(2) Realized labor market outcomes

(1) Survey-based outcomes: i) reservation wages (also as difference to pre-intervention reservation wages), (ii) self-reported probability of bargaining over wages within next twelve months, (iii) self-reported probability of switching firms within next twelve months, (iv) expected wage in twelve months
(2) Realized labor market outcomes: (i) actual wages, including difference between expected and actual wage, (ii) employment status (including whether individuals are still registered in the social-security data), (iii) establishment switches, (iv) geographic mobility, (v) occupation switches

(1) Survey-based outcomes from potential follow-up survey

(1) Survey-based outcomes from potential follow-up survey: (i) wage bargaining since intervention, (ii) job satisfaction, (iii) wage satisfaction, (iv) subjective integration, (v) intentions to stay in Germany, (vi) job search intensity, (vii) self-reported callback rate, including job offers received, (viii) training and development

Among respondents who are interested in receiving wage information, we randomly allocate half of the participants to each treatment arm:
(i) Treatment group: respondents receive information about the average wage of observationally similar workers.
(ii) Control group: respondents do not receive the information.
The intervention closely follows the experiment of Jäger et al. (2024). The information is visualized as a bar chart. Alongside the treatment, we also show respondents' guess of the wage of similar workers as well as their (reservation) wage. The control group only sees these two bars, while the treatment group also sees the bar with the actual average wage. Observationally similar workers are workers in the same 3-digit occupation who are full-time employed and have the same gender, age, education, and commuting zone. We clearly state this when providing the wage information.
We stratify by employment status (non-employed, part-time employed, full-time employed). The intervention shown to each group is identical except that (i) for non-employed workers, the wage of similar workers also conditions on having been non-employed for a similar period of time, and (ii) for non-employed and part-time employed workers, we do not show their current wage but rather their pre-intervention reservation wage, since no information on their full-time wage is available.

Not available

Randomization done by a computer

Individual. We stratify by the following three employment groups to ensure balanced
treatment assignment within each group: 1) full-time employed; 2) part-time employed, employed in a minijob, or with a signed employment contract but not yet working; 3) non-employed

No

2000 individuals (estimation based on wave 1 participation)

2000 individuals (estimation based on wave 1 participation)

1000 control individuals, 1000 treated individuals (estimation based on wave 1 participation)

We do not have any baseline data and therefore cannot provide minimum detectable effect sizes. In addition, the actual number of observations may differ depending on the attrition rate, the share of migrants actually moving to Germany between waves 1 and 2, and their willingness to receive wage information. These are factors that we cannot predict ex ante, given that the respondent sample is novel (migrants surveyed before and after moving to Germany).