Calculating the ROI of College: Whose Approach Is Better?
In Monday’s Inside Higher Ed, Nic Ducoff (co-founder of Edmit) penned an opinion piece questioning the approach of some organizations that have attempted to calculate the ROI of college. Mr. Ducoff writes that most approaches include cost and earnings, but how those variables are determined impacts the result and how the result is presented to prospective students impacts the influence it will have on their decision making. I could not agree more.
He describes the approaches taken by four different entities. One of these approaches is from the Center on Education and the Workforce at Georgetown that projects future earnings based on current earnings and compares that to the upfront investment for a degree. I wrote about the CEW report and methodology a number of times in 2019 and 2020:
- A First Try at ROI: Ranking 4,500 Colleges
- Reviewing the Methodology Behind New ROI Rankings for 4,500 Colleges
- ROI of Liberal Arts Colleges
- Seeking Stories…From Liberal Arts Graduates
- Graduate Degrees – Worth the Time and Expense?
- The Overlooked Value of Certificates and Associate Degrees
- Buyer Beware: Examining Earnings by Degree and College Debt
Payscale has created an ROI analysis similar to the CEW, but it does not include all colleges and universities, limits the ROI to the projected 20-year future earnings, and has options with and without financial aid.
Brookings created a value-added approach in 2015. Its approach used College Scorecard data, Payscale data, the amount of financial aid offered by an institution, the completion rates of an institution, the share of graduates prepared to work in STEM fields, and the average labor market value of skills listed on alumni resumes.
Third Way published another methodology in early 2020 that calculates a price-to-earnings premium by using the average net cost to attend an institution divided by the net of the post-graduation earnings of the student, less the typical salary of a high school graduate. This calculation derives the number of years to recoup net cost.
Mr. Ducoff writes that the existing approaches are a good start but none, in his opinion, are ready for prime time. His criticisms of the four existing methodologies are as follows:
- Payscale uses the total published cost of college attendance (tuition, fees, room and board, and books and supplies). They acknowledge that a student’s actual net cost may vary (and it does, particularly at non-elite private colleges that provide tuition discounts).
- Brookings only looks at earnings and does not take cost into account at all.
- Third Way uses total net price using the average net price from the College Scorecard.
- CEW does not take into account what a student would earn if they didn’t attend college.
All of the approaches use average earnings at the institutional level, whereas it would be better to use program-level earnings data. Brookings and Third Way evaluate ROI based on earnings 10 years after enrollment, Payscale looks at 20 years, and CEW evaluates the ROI over 10, 15-, 20-, 30-, and 40-year periods.
According to Mr. Ducoff, incidental expenses should also be included, such as travel if a student is going to a college far away from home. At the same time, a portion of room and board should be subtracted, since students who enter the workforce after high school without attending college need to live somewhere as well as eat.
None of the approaches take debt into account, and since two-thirds of college students borrow to pay for their costs, interest payments on the debt should be included in the costs. None of the approaches account for the risk of not graduating, even though 40 percent of students who start college do not finish and many students who graduate take longer than four years.
The ideal approach would use a close approximation of a student’s actual net cost, plus incidentals and expected interest payments, but would exclude living expenses that they would incur if that student was not attending college. It would also incorporate the risk of not graduating and include costs associated with a longer time period to graduate before working full-time. It would consider starting salary, mid-career salary, and lifetime income.
Mr. Ducoff concludes by acknowledging that calculating ROI is complicated as it involves many variables and numbers. An ideal approach needs to provide sufficient user education, so that students may be able to make more informed decisions about cost and earnings. He writes that his company, Edmit, is working on this idea and that he welcomes feedback from readers on what it should include.
There are a number of good points that Mr. Ducoff makes. If I were designing an interactive tool for prospective students to review, I would include the annual tuition [populated from a database updated by the institution].
I would also add:
- Annual fees [populated from a database updated by the institution]
- Annual room and board [populated from a database updated by the institution]
- Incidentals such as travel [populated from a separate screen that calculates travel costs based on distances from home]
Furthermore, I would subtract:
- Tuition discount/merit scholarship [populated from a database]
- Annual room and board if a student lived at home [user provides an estimate]
- Interest expense [populated from a database or the user can override]
I would have the database present the user with the average earnings for all graduates of the institution at the degree level. I would give the prospective student the opportunity to select a specific major, and the average earnings for that specific program would appear if available (note – due to privacy concerns, the College Scorecard does not present earnings for specific programs if there are less than 20 graduates for the most recent period evaluated).
When ROI is calculated, I would display earnings one year after graduation, career midpoint, and lifetime earnings. (It is important to note that the earnings data, which currently comes from the College Scorecard, will need to include ALL students. The College Scorecard only provides earnings data for students who borrow loans.) The latter two numbers would be based on the prospective student’s age at graduation, which would require the prospective student to input his/her existing age and would assume a normal time to graduate. I would also calculate the ROI based on the college’s earnings data less the average earnings for high school graduates (agreeing with Mr. Ducoff).
Interestingly, these options do not take into account some of Mr. Ducoff’s concerns and some that I have expressed in my critique of the College Scorecard and the CEW. Mr. Ducoff makes a great point that 40 percent of those who start college do not graduate. I think anyone who builds such a tool would have to present that as a disclosure item rather than adjust the calculations.
He also correctly states that many people do not graduate on time. I would consider building in the report an option to find out the ROI if it takes longer to graduate.
Mr. Ducoff states that one-third of students change majors. My experience as a student, parent, and college president is that the number is higher than that, but perhaps Mr. Ducoff’s number reflects changes after sophomore year when a “final” choice of major is usually required. I changed my major of interest four times as an undergrad, all of those changes before my junior year. That’s why I would use the average institutional earnings for graduates and give prospective students an option to drill down further at the program level.
Then, we have the issues that Mr. Ducoff does not address. How do we look at students who transfer? Nearly 50 percent of undergrads transfer at least once. Nearly one-third transfer twice. Do we need to build the tool for first-time freshmen or build it for transfer students as well?
If we build it for transfer students, we need to input cost and loan data from their first college (or first and second colleges or first, second, and third colleges). We also need to know how many credits transferred from the first institution to the next in order to calculate the cost to graduate.
There are certain undergraduate degrees that most likely will require additional training or education. How do we account for that? When I earned my MBA two years after graduating with a BA in History, I was fortunate that my MBA was funded by a full scholarship. That scholarship made my decision easy. It might not be an easy decision for someone who has $35,000 in loans from an undergraduate degree and is looking at borrowing another $75,000 for an MBA.
At the graduate level, how do we account for the cost of the degrees earned before the graduate degree of interest? None of the existing models do that. In fairness, they need to have an offset for that cost.
Mr. Ducoff is on track when he writes that “an ideal approach needs to provide sufficient user education.” As Mr. Ducoff and his colleagues at Edmit build that education, I would hope that they realize there are many more situations not covered by any of the four models, Mr. Ducoff’s proposed changes, or my suggested additions.
Above all, let’s continue to educate and continue to collect data. Ultimately, the student will be the winner. Isn’t that what we want?