Reauthorizing the Higher Education Act: The Role of Consumer Information in College Choice
May 6, 2015
U.S. Senate Committee on Health, Education, Labor & Pensions
Hearing on “Reauthorizing the Higher Education Act: The Role of Consumer Information in College Choice”
Much of my work at the American Institutes for Research and at College Measures involves tracking students completing their studies at colleges and universities into the workplace. The goal of this work is to identify differences in student earnings over time, to identify postsecondary credentials with high market value. This work has been with state governments and my state partners all share a commitment to put this information into the public sphere in a form that is usable by different audiences.
This work leads me to believe that we can and should organize consumer information into five different questions that students and their families need to ask (and answer) to better inform their decisions about where to enroll and what to study.
- Will I get in? (Selectivity)
- Will I get out? (Graduation rates)
- How long will it take? (Time to complete)
- How much will I pay? (Net price)
- How much will I make? (Post completion earnings)
Answering each of these questions has associated measurement problems, but none of the problems are insurmountable. We can make progress with the measures the federal and state governments already collect. But we can and should do better.
Below I will discuss some of the issues as we think about the data in each of these five categories, but first some issues that cut across all of these categories.
First, any efforts to develop consumer information about postsecondary education must include information about subbaccalaureate credentials, such as associate’s degrees and certificates, most of which are delivered by America’s community colleges.
One reason is evident by looking at trends in student enrollments: The number of subbaccalaureate degrees granted in the United States is growing more rapidly than the number of bachelor’s degrees granted (last year, subbaccalaureate awards granted almost equaled the number of bachelor’s degrees granted, although the bachelor’s degree still remains the most commonly granted college credential).
Second, the United States must break its “bachelor’s addiction.” Empirically, the bachelor’s degree is a good investment on average and in the long run. However, many students do not have the time, money or inclination to pursue this degree. There is consistent empirical evidence, much of it produced by College Measures, that subbaccalaureate credentials can lead to earnings that exceed those of bachelor’s graduates and that place students earning those credentials squarely in the middle class. The data also show that the subbaccalaureate credentials with the most market value produce students who know how to fix things (technicians) or how to fix people (health care). These credentials can help the nation fill the market for “mid-skilled” level workers, where some postsecondary training but not a bachelor’s degree are the usual requirements.
Third, in addition to battling our bachelor’s fixation, we need to battle the fixation on institution level measurement. We love league tables that rank campuses against one another. But the student outcomes vary more by program of study than by institution. In other words: What a student studies often is more important than where they study it. In turn, we need to deliver usable consumer information at the program level.
Finally, gathering good information is not enough—getting the data into the hands of consumers in a format that is useful, usable, and used is a challenge. It is not clear to me that the federal government, which has a unique capacity to gather the data, has the capacity to disseminate it.
With these thoughts in mind, I return to some of the issues in gathering data in each of the five categories noted above.
Will I get in?
We need to keep in that while the press and many parents are fixated on the competition for seats in the nation’s most prestigious universities, the bulk of our colleges and universities are broad or open access.
Because among these campuses, there are huge differences in student outcomes, any data collected about gaining admission should help students broaden their choice of schools, alerting them to the many options they have, and steering their selection process to include options that point them to schools that are higher on measures of student success.
Will I get out?
The limits on Federal graduation rate statistics, reported through IPEDS, are well-known. Most basically, they are still based on first-time, full-time, beginning students, a declining proportion of America’s college students. While the coverage of different student populations will expand in the next few years, graduation rate data will still be at the institution level.
We need to move to the collection of program level graduation rates. Given the number of students who change majors (and swirl through campuses), this will be difficult, but we need to start down this path.
How long will it take?
The time it takes to earn a degree is important. The longer a student is enrolled in pursuing a degree the more likely it is that “life happens”, derailing student progress. Moreover, each year spent enrolled is one more year of tuition paid out and one more year of foregone earnings. While it is possible to use IPEDS to estimate the average time to degree from institution-level graduation rates, we need to gather time to degree by program. Texas already reports these data and this is something that other states should be collecting and reporting.
How much will it cost?
Thanks to Congressional action embodied in the Higher Education Opportunity Act, the nation has made great strides in making more public the difference between sticker price and net price. However, according to recent work by Andrew Kelly at the American Enterprise Institute, most students still do not have good information about the true costs they will encounter—and they are far more likely to overestimate the cost of college, which can discourage attendance.
Any tools we develop to estimate costs must allow students to enter personalized information. For example, in the My Future Texas application College Measures built, students could enter personal information from their own financial aid letters to compare their likely costs to earn a degree (the application in the background took into account the time to degree for the program the student was interested in to generate a total expected cost).
How much will I make?
Yes, postsecondary education is about many more things than simply making money, and, yes, college graduates usually are healthier, live longer, and engage in our democratic processes at higher levels than non-graduates—but the path to all of these other rewards runs largely through success in the labor market. In addition, students themselves overwhelmingly say that they the prospect of good careers and strong earnings drives their desire for postsecondary education.
Not surprisingly, I believe that we need program level earnings data. Right now, these data come from states’ unemployment insurance (UI) wage data. Many states link these UI data to student level data about year of graduation, and program/institution of study, allowing detailed reporting of earnings of graduates as much as 10 years after graduation.
The problems with these state data are well-known: students who move across state lines to work are no longer found in the data system of the state where they earned their degree. In some states, such as Colorado, average match rates are in the mid-40 percent range. In big states like Texas and Florida, with booming economies, match rates are 20 percentage points higher. But these are still low—and we don’t know how systematic error is introduced as students choose to leave the state.
Match rates also vary across institutions, with match rates for graduates from state flagships lagging match rates from regional comprehensive campuses. Field of study also matters: match rates for teachers, where state certification matters for employment, are far higher than for engineers.
The Wage Record Interchange System (WRIS 2) held out some promise to ameliorate the problem of interstate movement of graduates. WRIS 2 is a consortium of over 30 states that theoretically agreed to search for UI wage data requested by other members of the consortium. If this system worked as planned, UI coverage would expand dramatically. However, some very large states are not in the WRIS 2 consortium—and states that are members often do not honor the requests from other states. Match rates hover in the single digits and about 1/3 of the states in the consortium do not run requested matches.
The alternative is federal tax collected by the IRS or SSA. I recognize and appreciate that using these data is fraught with privacy concerns. Nonetheless, I believe that there are sufficiently strong statistical procedures that can be employed to protect these data. With these protections in place, I believe the federal government should seek ways to allow matching data about students education with IRS/SSA earnings data.
I believe that since the federal government has a compelling interest in students receiving Title IV aid, a reasonable place to start would be to match FSA data with tax data (as is already being done for Gainful Employment and for the College Ratings system). However, to be useful the FSA data must be expanded to capture program of study. But I also believe that the federal government should allow state governments to match their much more complete student data with federal tax data.
Let us assume that the nation makes a commitment to expand the collection of consumer data. The next question is how to disseminate it in a way to that is useful. To ensure widespread use, like the perennial description of real estate, it is essential to keep in mind three fundamentals about data usage: audience, audience, audience.
We can distinguish at least three audiences for these data—and while the underlying data may be the same, the way in which the data are presented and which strands are highlighted will vary. Trying to satisfy all three audiences with the same data presentation application may not be possible.
Here is my view on the three most important target audiences:
Students, their families and the guidance counselors who help students find and choose schools
The data need to be tailored to help students find schools and programs that they are likely to complete—and that will give them a strong chance to enter the middle class. They need to be able to understand that a bachelor’s degree is not the only path into the labor market. And, given the well documented low levels of financial literacy among young adults, they may need targeted help in helping them to understand exactly how the return on investment (ROI) that underlies the five key questions structuring my approach to consumer information translates into outcomes they can easily grasp: such as what kind of car will I be able to afford or will I be able to live somewhere else but my mother’s basement.
We also have to recognize that guidance counselors can act as intermediaries helping students navigate the data to choose programs that will lead to more success following completion.
I am not at all sanguine that the federal government can produce applications that will be appealing to this consumer audience.
State policy makers
States invest large amounts of money in their postsecondary systems because these systems are viewed as human capital investment designed to help the state remain economically competitive. At the current time, states also “own” the student level data and the state UI wage data. This allows them to build far more powerful applications tracking labor market outcomes than the federal government can at present. Furthermore, most states are now constructing performance based budgeting systems to reward colleges and universities that are exceeding benchmarks. Most of these systems are dominated by measures of the flow of students through the institutions (e.g., retention or graduation rates of different types of students), but many states have already included or are considering the inclusion of wage data into these budget systems.
One motivation behind this movement is clear: the ROI to the taxpayer should play a role in state budget allocations. While most of the previous discussion was focused on student ROI, taxpayers also have the right to know about the ROI on their investment in their state’s colleges and universities. Clearly, the ROI to both students and taxpayers will be driven by the earnings of students post completion, but the returns can vary across these two audiences: while a high subsidy may increase the ROI to the student, it could lower the ROI to the taxpayer.
For taxpayer ROI to be better measured, we not only need to measure better student earnings, we also need better measures of the amount of government subsidies flowing into and through campuses: This ultimately will require far better information about how government money is actually spent when it gets to campuses. This may require either better state finance data tracking systems and likely a systematic overhaul of IPEDS finance data.
Federal policy makers
Congress controls the ability of the federal government to link student data to wage data. There is a compelling federal interest measuring both student and taxpayer ROI to Title IV student aid and despite legitimate privacy concerns, linking FSA data with IRS/SSA tax data is essential. FSA data would need to be modified to include information about the programs in which Title IV recipients are enrolled. These merged data would cover around 60% of all students in the nation. In addition or in lieu of federal action, the Congress should make it possible for states to commission the IRS or SSA to merge state held student data with federal income tax data. Again, these data would be returned to the states aggregated to the program level and subject to statistical methods, such as perturbation, to ensure privacy.
Clearly, the federal government is in the position of creating a post completion earnings data base that no state or private entity can come close to matching in terms of coverage and quality. As noted, I believe the Congress should create such a data base perhaps starting with FSA data and Congress should also clarify the ability of states to match their own student level data with IRS/SSA tax data.
I also believe that while the federal government can be the most powerful actor in the nation in creating these data bases, it is far from the best actor when it comes to disseminating the data in a usable form. Rather, the federal government should make the data base widely available to states and private actors, such as College Measures, and encourage the creation of applications based on these data. Many efforts will inevitably follow, as many entities experiment with different user interfaces, emphasizing different metrics. This competition will likely yield the best solutions to the need for better, more widely used consumer information about the large and growing number of postsecondary options available across the land.
Thank you for your time and for your consideration of these ideas.