NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Back to History
The Effects of Earnings and Cost Disclosure on College Enrollment Decisions
Initial registration date
June 16, 2015
June 16, 2015 10:50 AM EDT
University of Chicago
Other Primary Investigator(s)
Additional Trial Information
This project studies the effects of the provision of information on degree-specific costs and labor market outcomes on college choice, college persistence, and labor market outcomes. Collaborating with the Chilean Ministry of Education, we conduct a large-scale informational intervention in which we provide institution and major- (henceforth degree-) specific data on tuition costs and labor market outcomes for past graduates of degree programs to applicants for federal student aid who are considering applying to those programs. We track students' enrollment decisions to determine how the intervention affected matriculation choices. Over the long run, we will consider the effects of our intervention on college persistence and labor market success as well. We frame our results using a simple model of choice under uncertainty, and pay special attention to differential effects by student socioeconomic status, preferences and beliefs elicited in our short baseline survey, and loan access.
Hastings, Justine, Christopher Neilson and Seth Zimmerman. 2015. "The Effects of Earnings and Cost Disclosure on College Enrollment Decisions." AEA RCT Registry. June 16.
Randomly-selected federal student loan applicants in Chile received information about institution- and major-specific earnings and debt outcomes as part of the 2013 online loan application process. In addition, we conduct a survey asking students six questions on enrollment plans and expectations about earnings and cost outcomes. Survey responses were collected from treatment and control groups. Survey response data supplements administrative data and allows us to test predictions about the impact of the information treatment based on a model of enrollment choice with limited information. We use administrative data to track students in the treatment and control groups and identify whether and where they chose to enroll.
We administered the survey and field experiment in partnership with the Ministry of Education (MINEDUC). Directly following the submission of student loan applications, students were sent an email from MINEDUC requesting that they log into a secure website to fill out an additional set of questions. Applicants logged in, accepted an informed consent statement, and were asked six questions. These included questions about enrollment plans, questions about own earnings and tuition cost expectations at the degree programs where the student was considering enrolling, and questions about expected earnings for typical students at those degree programs. 49,166 students completed the online survey. Upon completing the survey, randomly-selected students continued to two additional web pages designed to provide information about and prompt search for degrees with higher returns for past graduates. Our web application used prior survey responses to display personalized information for each applicant based on our back-end database linking educational and tax records for past graduates. The first page displayed information on earnings gains (relative to no tertiary enrollment) in monthly terms for the participant’s first-choice degree, tuition costs in monthly payments, and a “net value” which was the difference between monthly gains and payments in pesos. Costs and benefits were calculated over the 15 year student loan repayment term, and converted into monthly gains and monthly debt payments. To encourage search, the page also displayed how much more net value the applicant could receive by enrolling in an alternative institution offering the same major, or in a different major in the same broad field of study (e.g. nursing vs. nutrition). Potential gains were drawn from degrees relevant to respondents based on the selectivity of their planned enrollment choices. The second page consisted of a searchable database that allowed students to select a major and enter an entrance exam score. Based on that information, the page populated a table of degrees admitting students with similar scores, sorted in descending order by net value. Students were told they could save up to ten search tables and could re-login to view them any time. The web program recorded all searches made.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Over the short run we will focus on matriculation outcomes for treated and control students. These include the decision to matriculate in any degree program as well as financial and academic outcomes (graduation rates, earnings, costs, and earnings net of costs) for past graduates of and enrollees in the degree programs chosen by students in the sample universe.
Over the longer run, we will in addition consider outcomes such as college persistence, college graduation, and, if data become available, labor market outcomes such as employment and earnings.
Primary Outcomes (explanation)
Constructed outcomes include the measures of predicted earnings, costs, and earnings net of costs conditional on graduation that we presented to students as part of the intervention. These values are fixed prior to data collection and cannot be altered ex post; we describe their construction in detail in our reports.
Constructed outcomes also include measures of predicted earnings and costs conditional on enrollment that we did not provide to students. These supplement the outcomes listed above. Please see the attached document for a description of how we compute these values. Matriculation, persistence, graduation, and labor market outcomes are observed directly and not constructed.
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Treatment and control groups were assigned as follows. Students in the 2012 graduating high school cohort and all other registrants for the 2013 administration of the Chilean college entrance exam (including those from older high school cohorts) were pre-assigned to treatment and control groups. Treatment status was assigned at the high school for current high school seniors, and at the individual level for registrants who had graduated in prior cohorts. Treatment assignment was stratified by school type (public, private, voucher) fraction of entrance exam takers, and average entrance exam score for current high school seniors. Treatment assignment was stratified by observed prior test scores for previous graduates.
Experimental Design Details
As described above, we used school-level randomization for current high school students, and individual-level randomization for older graduates.
Was the treatment clustered?
Sample size: planned number of clusters
For current high school seniors, 2851 schools. For older graduates, 12055 students.
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
For current high school seniors, 1415 treated schools and 1436 control schools. For older graduates, 5916 treated students and 6139 control students.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
INSTITUTIONAL REVIEW BOARDS (IRBs)
National Bureau of Economic Research
IRB Approval Date
IRB Approval Number