The federal government released earnings data on U.S. colleges and universities for the first time last month. The new College Scorecard adds an important dimension to other government data collections on higher education, which do not include earnings in the labor market or other longer term outcomes associated with attending particular postsecondary institutions.
In a full report available here we identify five important limitations of the new Scorecard: First, the salary estimates are based only on students receiving federal student aid, and the proportion of such students varies considerably across institutions. This is likely to produce biased earnings estimates for many colleges and universities. Second, the salary data are credited to any institution a student attended with federal student aid, regardless of whether it was for a full degree program or only a single semester. This leads to problematic estimates for institutions in which many students transfer out or are non-completers. Third, the salary data are not provided at the level of program of preparation, which is likely a much more important driver of salaries than the overall institution. Fourth, the salary data are strongly correlated with students’ background upon admission but this is not taken into account in the Scorecard. This is likely to lead users to believe that the association between a college and the salaries of those who attended is the valued added by that institution (bang for the buck) whereas it really reflects to a significant degree the academic preparation and family background of the students admitted. And finally, the lack of plans for stewardship of the Scorecard effort presents the risk that it will be a one-time effort.
We propose two simple actions that solve four of these problems: The first is adding two boxes to the 1098T form that the Internal Revenue System already collects from colleges and universities on all of their students who pay tuition, regardless of whether they receive federal aid. By collecting information on graduation status and major field of study on the 1098T, the federal government through the IRS could easily produce annual earnings data for all students by program of study within institution, and report that at the aggregate level for program graduates vs. non-completers. These data could be further sliced and diced by broad demographic variables, including family income. Importantly this could be accomplished without developing or maintaining a new federal data system of individually identifiable higher education records, which was prohibited by Congress in the 2008 reauthorization of the Higher Education Act. The second action is turning stewardship of the Scorecard effort over to the National Center for Education Statistics, which can better carry this torch successfully through 2017 and beyond than political appointees in the White House and Department of Education.
The 1098T with information on major and graduation status does not directly solve the challenges of identifying which programs provide the most bang for the buck. But with the annual IRS reports in hand the federal government or other entities building college search sites using federal data could provide parents and prospective students with an individualized comparison of costs and outcomes among nominally identical programs of preparation at different institutions. Students willing to anonymously input information about their family background would have the benefit of having those outcomes displayed for graduates who are similar to them.
The new College Scorecard is a wonderful idea. Relatively minor tweaks in current data collections and the pathways the information takes as it makes its way to the federal government can make the Scorecard truly useful.