Information Asymmetry in Job Search

Students in their senior year of college will receive a survey inviting them to participate in a study. If they agree, those randomly sorted into the treatment group will receive salary information that the control group will not. We will study the impact of the information intervention on job search outcomes by observing differences between the treatment and control group.

2022-08-15

2023-05-31

Our primary outcome is students' accepted salary offer; the ultimate goal of the project is to see if an information intervention can narrow demographic wage gaps.

We will measure final job outcomes as follows: students' final job base salary, salary including bonus, negotiation behavior, non-wage amenities provided in job, and location of job. We will also measure final job satisfaction via how students' believe their final salary compares to other students in their graduating year and in their school or major.

To account for mechanisms that would influence our primary outcome, we detail two additional sets of outcomes to measure: (1) initial post-treatment beliefs, salience, and job search effort, and (2) post-job search beliefs and application behavior.

1. Initial post-treatment beliefs, salience, and job search effort
- Initial post-treatment beliefs: We will measure if the treatment group internalizes the information, and changes their beliefs on salary median, 25th percentile, and 75th percentile.
- Salience and job search effort: We will measure if the treatment group says they are more likely to sign up for a career counseling meeting, and if they actually sign up for a career counseling meeting. This is meant to measure if students exert more effort in job search after seeing the infographic. We will also measure if those who signed up for a career counseling meeting attended a career counseling meeting before the end of job search.
- Salience and job search effort version 2: We will measure if the treatment group says they are more likely to sign up for email reminders. This is meant to measure if students exert more effort in job search after seeing the infographic. We will also measure if those who signed up for reminder emails opened the emails.

2. Post-job-search beliefs and application behavior
- Post-job-search beliefs: We will measure student beliefs on the median, 25th percentile, and 75th percentile of their major's salary in the second survey. Because the second survey occurs after job search, this measures whether or not the treatment group remembered the information, and also whether or not the control group eventually attained information about true salary distributions during job search.
- Application behavior: Which group had more applications, more interviews, and more offers? Which group rejected more offers? How long did students take to accept an offer?
- Search behavior: How did students behave on search? Did they seek out more information on typical salaries? Did they target a certain salary?

To test whether information access contributes to different job search outcomes, our project will (1) conduct an information intervention experiment to assess first whether students of different races and genders have different information about the labor market, and (2) study whether an information intervention could change their job search behavior and help bridge the gap in wages.

The experiment will center on college students in their senior year, who are about to enter the job search process. The treatment group will be provided with information about the starting salary distributions of students from prior years by major, along with general information about the Career Center website. The control group will only receive the information about the Career Center website. The project will then examine whether the treatment information changes job search behavior: in particular, the salary of the accepted job, the number of jobs applied to and interviewed at, the number of jobs rejected, negotiation behavior, valuation of non-wage amenities, and beliefs about how well students did on the job market. To account for heterogeneity across demographic groups, stratified random assignment across race, gender, and major will be used to split participants into treatment and control. This design not only allows us to control for confounding factors, but observe heterogeneity in the effectiveness of an information intervention across student demographics, student major choice, and other characteristics.

We will use stratified randomization across gender, race, and major. Randomization will be done within Qualtrics. Qualtrics software allows the randomization to occur after students have answered demographic questions on the survey, ensuring that stratified randomization occurs even if there is selection regarding who responds to original survey invitation.

Randomization occurs at the individual level.

No

N/A

We aim to enroll any adult undergraduate student over the age of 18 who is in their graduating year at UC Berkeley. The screening/assessment procedures will focus on whether the participant is graduating or not - only eligible participants can continue with the survey and be compensated. The maximum number of subjects is the maximum number of these individuals, which in 2020-21 was 9,024 according to the Berkeley Office of Planning and Analysis. This number for the relevant school year, 2022-2023, is likely to be similar. However, as we do not yet know response rates for our survey, we do not currently have a final observation number.

We have one treatment arm: approximately half the sample will be randomized into treatment, and the other half into control.