Biden’s Flawed Attempt to Stem Rising Tuition Costs

The proposed ‘gainful employment’ rule would deny funding to some low-value higher education programs, but there would be collateral damage.

Key Points

• An effort to hold institutions accountable for poor outcomes. The Biden administration has proposed a framework to restore the Gainful Employment rule (GE), which aims to revoke federal funding for certain higher education programs with poor outcomes.
• Biden’s rule is imprecise, and imprecision is problematic. FREOPP’s estimates of return on investment (ROI) show that GE’s targeting of programs without financial value is imprecise. That imprecision is problematic, given the severity of the penalties that GE imposes.
• Case in point: GE’s treatment of medical assistance programs. Almost 70 percent of certificate programs in medical assisting would fail GE, despite producing real financial value for their students.
• GE provides little accountability for public and private nonprofit institutions. GE only applies to certificate programs and for-profit colleges, but not public or private non-profit colleges. If GE were applied to all programs, it would fail to hold most low-value master’s degrees accountable.

Executive Summary

The Biden administration has proposed a framework to restore the Gainful Employment rule (GE), a federal regulation aimed at holding postsecondary certificate programs and for-profit colleges accountable for student outcomes. GE terminates programs’ eligibility for federal aid funding if students in those programs have a high debt-to-earnings ratio or fail to realize earnings above the median high school level. Without federal funding, most programs would not continue operating, meaning GE has the potential to reshape American higher education.

This report compares programs’ performance on GE to FREOPP’s estimates of return on investment (ROI) to evaluate whether GE effectively targets programs that fail to produce a financial return for their students. Most programs that pass GE have positive ROI and vice versa. However, there are exceptions. Nearly 70 percent of certificate programs in medical assisting would fail GE, even though most of these programs have positive ROI. Fortunately, some straightforward changes to the GE framework would go a long way toward addressing this problem.

Some have proposed applying GE to all higher education programs — “GE for all” — including the degree programs at public and private nonprofit institutions that are currently exempt. This would be a significant step forward for higher education accountability. However, many master’s degree programs with negative ROI would survive “GE for all” even as hundreds of positive-ROI certificates at trade schools fail. “GE for all” would also adversely affect professional programs in law and medicine.

While no accountability system would be perfect, policymakers could improve the design of “GE for all” by replacing its termination enforcement mechanism with a system of graduated financial penalties. Programs with worse outcomes would pay larger fines. While this would result in de facto termination for the worst programs, others with less extreme outcomes would face a financial incentive for improvement. Finally, GE should not be a substitute for sensible limitations on the scope of federal student aid, such as capping or eliminating federal loans for graduate school.

Policy Recommendations

1. Reduce the earnings threshold for the GE rule by 15 percent. Doing so will reduce the number of programs incorrectly failing GE that have positive student outcomes.
2. Eliminate the discretionary earnings test for graduate programs. Gross earnings is a better measure of student outcomes than discretionary earnings, which assumes that income greater than 150 percent of the poverty level is “discretionary.”
3. Hold schools accountable for institution-level outcomes in addition to program-level outcomes. The proposed GE rule only uses data on students who complete their programs, but schools should also be held accountable for non-completers.
4. Release GE outcomes data for public and private nonprofit institutions. The Education Department argues that they are only legally allowed to penalize for-profit colleges and certificate-granting institutions, which creates a discriminatory policy. At the very least, the ED should release GE outcomes data for these non-profit institutions, because transparency alone can serve as a powerful accountability tool.
5. “GE for all” should introduce a graduated financial penalty. Congress should amend the Higher Education Act so that accountability applies to all institutions: “GE for all.” In addition, Congress should empower the Education Department to apply graduated financial penalties for programs below an improved performance benchmark. In this way, failing programs would have the opportunity to improve before being shut down.
6. Change the amortization period for calculating graduate degrees’ annual loan payments. Professional degrees provide lifetime benefits over a very long period, whereas master’s degrees are meant to generate an immediate rise in earnings. The GE rule should account for this by adjusting the timeframes used to calculate loan payments.
7. End federal loans for graduate school, and use the private sector to hold these programs accountable. Studies show that most graduate students could secure private-sector loans at lower interest rates than those on federal loans. And private lenders are much better equipped than the government to provide better-targeted accountability to graduate programs.

Introduction

The Higher Education Act, the primary law governing the federal role in postsecondary education, has not undergone a comprehensive reauthorization since 2008. With the prospect of major legislation during the current Congress dimming, the Biden administration is actively considering higher education reforms that it could enact without Congressional authorization. One of these is the Gainful Employment rule (GE), a federal regulation which aims to terminate the right of certain low-value higher education programs to receive federal student aid funding.

The Obama administration was the first to promulgate a version of GE, in 2014, but the Trump administration reversed its predecessor’s rules before any sanctions could take effect. The Biden administration has signaled its intention to resurrect and strengthen the rule. Though the Department of Education (ED) has not yet formally proposed regulatory text, an issue paper released during negotiated rulemaking in March 2022 details a likely framework for the eventual regulation. The version of GE apparently under consideration would be much more stringent than the previous rule.

This report evaluates whether the Biden administration’s GE framework would effectively target programs that fail to provide financial value for their students. FREOPP’s estimates of return on investment (ROI) for higher education provide the measure of each program’s financial value. The FREOPP ROI estimates compare the expected lifetime earnings of the median student in each program to a counterfactual earnings profile that estimates lifetime earnings for that same student if she had never gone to college. Expected earnings minus counterfactual earnings equals the increase in lifetime earnings that the median student can anticipate from her degree or certificate; subtracting the costs of higher education from this figure provides an estimate of ROI.

ROI is the most comprehensive measure of a higher education program’s financial value. However, it is impractical to calculate on an ongoing basis. The need to consider lifetime earnings requires ROI to be calculated with a severe lag or with projections of students’ future earnings, not hard data. These considerations mean that policymakers would be ill-advised to use ROI directly for purposes of higher education accountability. However, estimates of ROI such as FREOPP’s can help evaluate other potential accountability metrics that are more practical to calculate.

This report uses FREOPP’s ROI estimates to evaluate the metrics proposed in the Biden administration’s draft GE rule. The accountability metrics which the administration has proposed include two debt-to-earnings tests which compare program completers’ student loan payments to their median earnings three years after graduation. The administration has also proposed a metric which compares completers’ median earnings to the earnings of early-career high school graduates in their state. FREOPP’s estimates of ROI can inform whether these measures effectively proxy for ROI, and how policymakers can use them if their goal is to terminate negative-ROI programs while allowing positive-ROI programs to continue receiving support.

Granted, ROI is not the only higher education outcome that policymakers might value. The debt-to-earnings ratio in GE has value as a measure of whether students’ loan payments are affordable, and ED might wish to pursue a regulatory agenda based on the debt-to-earnings ratio for reasons independent of its correlation with ROI. But for most students, increasing their earnings power is the main motivation for pursuing higher education, and therefore ROI should be a principal consideration in designing accountability policy.

Biden’s Gainful Employment framework

The Education Department released an issue paper for its March 2022 negotiated rulemaking describing a potential framework for the GE rule it eventually intends to promulgate. No rule has yet been officially proposed, so it is possible that the structure of the final GE rule will change relative to the framework in the issue paper. (ED will first issue a proposed rule, likely in 2023, and receive feedback through public comment. Later on, it will issue a final rule.) But the March issue paper is likely to be the baseline on which ED constructs its eventual regulation. Moreover, the ideas therein could form the basis for future policy proposals.

ED’s stated statutory authority for GE comes from the Higher Education Act, which stipulates that institutions of higher education that award certificates must “prepare students for gainful employment in a recognized occupation.” The Act also applies the “gainful employment” standard to proprietary institutions awarding either degrees or certificates. No such requirement applies to public and private nonprofit institutions awarding degrees of at least two years of study. In the Education Department’s view, ED only has the legal authority to apply “gainful employment” regulations to certificate programs at any institution, along with degree programs at for-profit colleges. Degree programs at public and private nonprofit schools are therefore exempt from GE under the proposed framework.

The two-part test in GE

The framework proposes two tests to evaluate whether a given program prepares its students for gainful employment in a recognized occupation. The debt-to-earnings test considers the ratio of students’ loan payments to their median earnings. The earnings threshold test compares students’ median earnings to those of early-career high school graduates. If a program fails either test in two out of any three consecutive years, it will lose eligibility to continue participating in federal student aid programs including Pell Grants and student loans. The tests are applied at the program level, so an institution may continue to receive funding for passing programs even if one of its programs fails GE.

The first prong of GE, the debt-to-earnings test, has two sub-tests. To run the debt-to-earnings test, ED first calculates the median annual loan payment of program graduates. ED does not use students’ actual payments, but calculates an estimated annual payment based on students’ median debt, including any debt their parents took on for their education, plus certain private loans. For undergraduate certificate programs and associate degrees, loan payments are amortized over a 10-year period using the average undergraduate student loan interest rate during the three years prior to the cohort’s completion date. Bachelor’s and master’s degrees are amortized over 15 years. Doctoral and professional degrees are amortized over 20 years using graduate loan interest rates.

ED then divides the annual loan payment by the median annual earnings of program completers three years after graduation. If the ratio exceeds eight percent — meaning expected annual loan payments are more than eight percent of earnings — the program fails the first part of the debt-to-earnings test. Next, ED divides the annual loan payment by the discretionary earnings of program completers, which is equal to median annual earnings minus 150 percent of the poverty line in the earnings measurement year ($18,735 in 2019). If this ratio exceeds 20 percent, the program fails the second part of the debt-to-earnings test. A program need only pass one of these two sub-tests in order to pass the debt-to-earnings test overall.

The second part of GE is the earnings threshold test. This test compares program completers’ median annual earnings three years after graduation to the median annual earnings of individuals aged 25 to 34 in the state where the institution is located who have only a high school diploma or GED. (If fewer than 50 percent of students at the institution come from in-state, ED substitutes the median earnings of young high school graduates nationally.) If the program’s median earnings do not exceed those of the relevant high school graduate comparison group, the program fails the earnings threshold test.

If a program fails the earnings threshold test in two of three consecutive years, it loses eligibility for federal aid even if it passes the debt-to-earnings test. Likewise, a program which passes the earnings threshold test loses its eligibility if it fails the debt-to-earnings test in two of three consecutive years.

A loss of federal aid is a death sentence for most programs. Colleges shut down hundreds of programs which failed the Obama administration’s original GE rule even though the rule’s sanctions had not yet taken effect. This is a strong signal that colleges do not consider most of their programs viable without federal funding, and therefore GE has considerable potential to change the landscape of higher education. Fortunately, existing data sources provide a good idea of which programs could lose federal aid under GE and whether those terminations are justified based on the programs’ ROI.

Does Gainful Employment effectively target low-ROI programs?

On March 15, 2022, ED released GE information rates which provide the estimated annual loan payments and median graduate earnings associated with thousands of higher education programs that will be subject to GE, should it become part of the code of federal regulations. The ED dataset indicates whether each program would pass or fail GE and each of its sub-tests based on currently available information. While the information rates are not the final word on which programs will pass GE (a program must fail for at least two years to lose access to federal aid, but the information rates contain only one year’s worth of data), they nonetheless provide a strong indication of which programs are likely to be terminated under GE.

It is possible to match this data to FREOPP’s estimates of ROI for bachelor’s degrees, graduate degrees, and sub-baccalaureate credentials. Though there are hundreds of thousands of higher education programs receiving federal funding, ED does not calculate informational data for programs exempt from GE and those with fewer than 30 completers. There are 4,192 programs with data in the information rates; FREOPP has estimates of ROI for 3,836 of these (92 percent). While FREOPP has estimates of ROI for over 60,000 programs in total, the vast majority of those are degree programs at public and private nonprofit universities, which are exempt from GE.

This section compares programs’ performance on the outcomes metrics employed in GE (the debt-to-earnings ratios and the high school earnings threshold) to FREOPP’s estimates of these programs’ ROI. Since GE uses debt and earnings outcomes for program completers only, it is fair to assume that the intention behind GE is to hold programs accountable for the value of their credentials in the labor market rather than these programs’ completion rates. (Separate regulations may be in order to address low completion rates.) The initial analysis therefore uses FREOPP’s estimates of ROI for on-time program completers rather than the more comprehensive version of ROI which adjusts for completion rates.

Annual earnings rate vs. ROI

The annual earnings rate, one of two prongs of GE’s debt-to-earnings test, compares graduates’ median annual loan payments to their median earnings three years after graduation. If this ratio exceeds eight percent, the program fails the annual earnings test. Programs need only pass one of the annual earnings test and the discretionary earnings test in order to pass the debt-to-earnings component of GE.

There is a moderate negative correlation (-0.37) between the annual earnings rate and ROI, weighting programs by cohort size. A one-percentage-point increase in the annual earnings rate is associated with a $43,630 decline in lifetime ROI. This suggests that programs where the debt-to-earnings ratio is higher yield a significantly lower lifetime payoff for students.

But the annual earnings rate is not a perfect proxy for ROI. For many programs, especially bachelor’s degrees, earnings rise sharply during the early career. Earnings shortly after graduation might therefore understate these programs’ lifetime financial value, meaning a program could have a high annual earnings rate and high ROI. By contrast, many inexpensive schools where debt levels are low might nevertheless offer credentials with limited labor market value. These programs would have a low annual earnings rate, but low ROI.

As it happens, the annual earnings test in GE misclassifies hundreds of programs, as shown in the following chart. Programs in the upper right quadrant are “incorrect fails,” meaning they have positive ROI but nonetheless fail the debt-to-earnings test. These include several bachelor’s degrees at proprietary colleges. Programs in the lower left quadrant are “incorrect passes,” meaning they have negative ROI but still pass the debt-to-earnings test. These include several certificate programs at community colleges where typical debt at graduation is $0. However, these programs are still negative-ROI because graduates realize low earnings and must spend time out of the labor force to earn their credentials, even if they don’t take on debt.

Discretionary earnings rate vs. ROI

The discretionary earnings rate is the second prong of the debt-to-earnings test. It compares graduates’ median annual loan payments to their discretionary earnings, defined as median earnings minus 150 percent of the poverty level for a single person ($18,735 in the GE information rates). A program fails the discretionary earnings test if this ratio exceeds 20 percent. If a program fails this test but passes the annual earnings test, or vice versa, it passes the debt-to-earnings test overall.

The correlation between the discretionary earnings rate and ROI is weak (-0.06), and it appears to add little value to the annual earnings rate as a proxy for ROI. While the annual earnings rate and discretionary earnings rate together explain 12.6 percent of the variation in ROI, the annual earnings rate alone explains 12.4 percent of the variation.

Since programs must only pass one of the two sub-tests in order to pass the debt-to-earnings test overall, the discretionary earnings test could serve as an escape hatch for positive-ROI programs that fail the annual earnings test. But only five percent of programs pass the discretionary earnings test while failing the annual earnings test, and nearly half of those have negative ROI. The discretionary earnings test appears to “save” almost as many negative-ROI programs as positive-ROI programs.

The following chart relates the discretionary earnings rate to ROI.

Earnings threshold vs. ROI

The earnings threshold test compares the median earnings of program completers three years after graduation to the median earnings of early-career high school graduates in the same state. If program earnings exceed the high school earnings threshold, the program passes the test. But a program which fails the earnings threshold test will fail GE overall, even if it passes the debt-to-earnings test.

There is a strong correlation (0.64) between ROI and “excess earnings” — the amount by which program earnings exceed the passing threshold. The earnings threshold test remedies one of the weaknesses of the debt-to-earnings test by failing a number of negative-ROI certificate programs where debt is low but earnings are nonetheless mediocre.

However, the benchmark of median high school earnings is largely too weak to catch negative-ROI programs above the undergraduate certificate level. The earnings threshold test fails just nine of 252 graduate programs subject to GE, even though 148 of these programs have negative ROI. Among bachelor’s degrees, just 17 out of 336 programs fail the earnings threshold test. Selection bias is the main reason for the low fail rate of negative-ROI programs above the certificate level: students who enroll in bachelor’s or master’s degree programs have preexisting earnings potential that is well above the high school level, so programs with moderately high earnings can still have negative ROI.

This phenomenon is apparent on the chart below, which relates ROI to the amount by which program earnings exceed the minimum passing level. Negative numbers indicate programs which fail the earnings threshold test. Almost all graduate programs, along with the vast majority of bachelor’s degree programs, escape sanction under the earnings threshold test. However, the test succeeds in penalizing almost all negative-ROI certificate programs. In this it may even do its job too well: almost 400 certificate programs with moderately positive ROI nevertheless fail the earnings threshold test.

This finding may seem odd. A program with positive ROI should increase its graduates’ earnings well above the level of high school diploma holders, so programs with positive ROI but earnings below the high school level should be rare. Yet programs at the passing threshold have an average ROI of nearly $93,000.

But as Kristin Blagg of the Urban Institute points out, the median earnings of all high school graduates may not be the appropriate counterfactual. High school diploma holders in the labor force are 61 percent male, while many certificate programs are predominantly female. Women tend to earn less than men within all educational strata, so a program which appears not to raise graduates’ earnings above the median level for all high school graduates may still provide a significant earnings boost relative to a counterfactual that includes more female high school graduates.

FREOPP’s estimates of ROI compare certificate holders’ earnings to those of a demographically similar group of high school graduates, allowing identification of programs with seemingly weak earnings outcomes that still have positive ROI. Many certificate programs with a predominantly female student body fall into this category.

Overall GE results vs. ROI

In theory, combining the debt-to-earnings test with the earnings threshold test allows each to cover the other’s weaknesses. The earnings threshold test catches many low-value certificate programs with low median debt levels that allow them to pass the debt-to-earnings test. Meanwhile, the debt-to-earnings test fails many bachelor’s and graduate degrees which yield earnings above the high school level but nonetheless have poor ROI due to the programs’ high costs.

To measure how well the two halves of GE work together, I calculate the minimum level of graduate earnings necessary for each program to pass both components of GE, holding annual loan payments and other factors constant. Subtracting this figure from actual earnings yields the amount by which each program’s earnings exceed the minimum passing level. “Earnings relative to passing level” (EPL) are negative for failing programs.

The correlation between ROI and EPL is 0.71, which is higher than the correlation between ROI and any of the individual components of GE. This suggests that the components of GE work in concert to target low-value programs better than any component could individually.

However, there is evidence that GE does its job too well. Regression analysis reveals that a program with EPL of $0 — i.e., a program that just barely passes — has an average ROI of $134,000. This means that GE is failing hundreds of programs which are producing real though modest financial value for their students.

How GE classifies higher education programs

The following chart relates EPL to ROI. Programs may be divided into four categories based on their ROI and whether they pass GE. “Correct pass” programs (top right quadrant) are those which pass GE and have positive ROI, while “correct fail” programs (bottom left quadrant) fail GE and have negative ROI. However, “incorrect pass” programs (top left quadrant) pass GE despite having negative ROI, while “incorrect fail” programs (bottom right quadrant) fail GE despite having positive ROI.

Overall, GE classifies 76 percent of programs correctly, as measured by ROI. But 17 percent of programs are incorrectly failed while seven percent are incorrectly passed.

Incorrectly failed programs are disproportionately undergraduate certificates, with a smattering of associate degrees and bachelor’s degrees. Nearly half of the certificate programs which fail GE despite having positive ROI are in medical assisting. Graduates of medical assistant programs are overwhelmingly female, so comparing their earnings to the median earnings of (mostly male) high school diploma holders is likely to produce an understatement of these programs’ financial value. When enshrined into policy, this misguided comparison leads to hundreds of positive-ROI programs failing GE.

How GE affects key certificate programs

If GE becomes regulation as is and current debt and earnings figures hold, almost 70 percent of certificate programs in medical assisting will lose federal funding. Most of those terminations will be unwarranted, if policymakers’ goal is to continue funding positive-ROI programs. Along with medical assisting, 76 percent of certificates in health and medical administration will fail GE, as will 60 percent of certificates in dental support services. In both cases a large portion of the programs failed have positive ROI.

To be sure, many program terminations under GE will be justified. Just 12 out of over 650 cosmetology programs will survive GE as written. Cosmetology is also one of the lowest-value fields of study in our analysis (nearly 90 percent of programs have negative ROI). Conversely, most high-ROI certificate programs will survive GE. Over 90 percent of certificate programs in licensed practical nursing, precision metal working, and vehicle maintenance and repair will pass GE; incorrect fails among these programs are rare.

Earnings misreporting is one potential reason for the low pass rate of cosmetology programs. Beauticians may receive part of their income in cash tips which go unreported to the IRS, the source of GE’s earnings data. The upshot is that cosmetology programs may have higher median earnings than those reported here. Stephanie Cellini and Kathryn Blanchard estimate that cosmetologists’ incomes are underreported by 8%, which is significant but not large enough to substantially alter these programs’ performance on GE. Still, misreporting strengthens the case for using graduated financial penalties rather than termination as GE’s enforcement mechanism; the recommendations at the end of this report will expand on this point.

While most of the errors in the overall GE framework are incorrect fails, around seven percent of programs are incorrect passes. The overwhelming majority of incorrectly passed programs are master’s degrees, especially MBAs. As explained further in FREOPP’s report on the ROI of graduate degrees, MBAs and many other master’s degrees tend to have high earnings in an absolute sense, but this is mainly a function of the types of students they enroll. Enrolling students with high preexisting earnings potential allows graduate programs to claim high earnings without necessarily having high value-added. As a result, these programs tend to pass GE despite having low ROI.

GE’s effect on higher education by sector

For-profit institutions make several objections to GE. First, the rule exempts degree programs at public and private nonprofit colleges, while degree programs at proprietary schools are held accountable. Although certificate programs in all sectors are subject to GE, a second objection is that the debt-to-earnings test stacks the deck in favor of public institutions. Public community colleges receive direct appropriations from state and local governments which defray the costs to students, while for-profit schools must raise almost all their money through tuition. Student debt will naturally be higher at for-profit schools, even if their programs are of equivalent value.

While FREOPP’s preferred estimates of ROI compare the financial value of a program to the cost of tuition, it is also possible to calculate ROI with respect to underlying education-related spending per student at each institution. Education-related spending measures how much it costs society — students and government together — to produce a particular degree or certificate. Calculating ROI with respect to spending rather than tuition eliminates the bias in favor of public institutions that government subsidies introduce.

The following chart compares each program’s performance on GE to its estimated ROI, calculated with respect to spending rather than tuition. For-profit colleges do have a point: 19 percent of certificate programs at for-profits are incorrectly failed, compared to just seven percent at public institutions. Meanwhile, three percent of certificate programs at public institutions are incorrectly passed, compared to less than one percent at for-profits.

However, GE’s bias in favor of public institutions cannot explain the majority of program failures at for-profit schools. While 19 percent of for-profit certificates are incorrectly failed, more than twice as many — 41 percent — are correctly failed. Meanwhile, 79 percent of certificate programs at public schools are correctly passed. Even when accounting for the additional subsidies they receive, the majority of certificate programs at public institutions have positive financial value.

As discussed further in FREOPP’s report on the ROI of sub-baccalaureate credentials, certificate programs at public institutions are concentrated in high-value fields such as licensed practical nursing and precision metal working. Meanwhile, the most common certificate offered at for-profit colleges is cosmetology, which is almost always negative-ROI. While government subsidies do tilt the playing field in favor of community colleges, most of their superior performance on GE is due to the better mix of certificate programs they offer.

GE results vs. ROI, adjusting for completion rates

The data points which GE relies on — including median debt and earnings — reflect program completers only. This is a necessary concession to the realities of data collection. If a student drops out before declaring a major, it is usually impossible to classify her to a single program. Program-level outcomes metrics, such as those in GE, must usually rely on completer-only data.

This shortcoming of the data means that GE may pass programs which register strong results for completers but fail to graduate most of their students. Worse, the design of the rule could create a perverse incentive for schools not to help students with low earnings potential finish their degrees, since dropouts’ earnings are not included in GE medians. But dropping out of school is almost always worse than finishing; dropouts tend to have lower earnings and higher loan default rates than completers.

Therefore, it is worth examining whether GE effectively targets programs where ROI is lower due to a high dropout rate, even though that is not the intent of GE. In addition to ROI for on-time completers, FREOPP has also calculated estimates of ROI which account for the risk that students do not finish their programs. A program with high ROI for completers but a low graduation rate will have lower ROI under this “completion-adjusted” metric.

As expected, the metrics employed in GE have a weaker relationship with completion-adjusted ROI. While the correlation between earnings relative to passing level and ROI for completers is 0.71, the correlation between EPL and completion-adjusted ROI is 0.60. The share of programs that GE correctly classifies remains unchanged at 76 percent, but the breakdown is different. A greater share of misclassified programs are now incorrect passes rather than incorrect fails.

However, adjusting for completion rates reduces the proportion of incorrectly failed certificate programs only marginally, since certificate programs tend to have high completion rates. The share of medical assistant certificate programs which are incorrectly failed drops from 45 percent to 41 percent. Adjusting for completion rates largely does not address GE’s high rate of misclassification among predominantly female certificate programs in health care. This remains one of the principal shortcomings of the GE framework as written.

Policy Recommendations: Improving the GE rule

Nearly a quarter of programs subject to GE would be misclassified if the rule went into effect as written. ED would terminate a large number of positive-ROI programs from federal grant and loan aid, while allowing a smaller number of negative-ROI programs to continue receiving taxpayer funding.

Virtually any accountability system would misclassify some programs, so GE should not be measured against the unattainable standard of perfection. Making GE more lenient might allow more positive-ROI programs to survive, but at the cost of continuing to fund negative-ROI programs. Conversely, a stricter accountability system would catch more bad programs, but it might make collateral damage of good programs. At a certain point, policymakers will be forced to choose not between correct classification and incorrect classification, but between the two types of incorrect classification.

For the certificates that comprise the bulk of programs currently subject to GE, policymakers should lean towards incorrectly passing programs rather than incorrectly failing them. Certificate programs often serve students from low-income backgrounds for whom the certificate may be one of the few avenues for economic mobility available. Moreover, postsecondary certificates can be stepping stones towards more valuable degrees. While extremely low-quality or scam certificate programs should have their federal funding revoked, policymakers should be more lenient towards programs at the margin. Incorrectly passing some moderately negative-ROI programs may be the price of ensuring that most positive-ROI programs continue to receive support.

This calculus changes somewhat for graduate programs. By definition, all potential graduate students have bachelor’s degrees, meaning they are far less likely to be low-income than high school-educated certificate students, and therefore have more opportunities for economic advancement. But the financial risks, to both students and taxpayers, are greater: graduate students have access to unlimited federal loans and tend to accumulate much greater debts than undergraduate students. These debts are often not paid back in full. It follows that policymakers should seek to avoid incorrectly passing graduate degrees, even if this comes at the cost of incorrectly failing some programs.

Fortunately, ED could make several changes to GE that would increase the share of programs being classified correctly. These relatively simple changes would reduce the number of incorrectly failed certificate programs and the number of incorrectly passed graduate programs. Implementing the following recommendations would improve GE’s performance as an accountability tool.

Recommendation 1: Reduce the earnings threshold for the GE rule by 15 percent. Because of demographic differences between the two populations, the median earnings of high school graduates are not the correct benchmark against which the earnings of certificate holders should be judged. I calculate that the optimal earnings threshold is instead 85 percent of its current level. Reducing the earnings threshold to 85 percent of early-career high school diploma holders’ median earnings reduces the share of programs incorrectly failing GE by nearly half, from 17 percent to nine percent. Under the modified earnings threshold, the vast majority of certificate programs in medical assisting with positive ROI pass GE. The share of programs incorrectly passing GE rises from seven percent to ten percent, but the reduced number of incorrect fails outweighs this increase. Overall, the share of programs correctly classified rises from 76 percent to 82 percent.

Recommendation 2: Eliminate the discretionary earnings test for graduate programs. Almost half of graduate degrees currently subject to GE pass the rule despite having negative ROI. The discretionary earnings test (median earnings minus 150 percent of the poverty level for a single person) is responsible for much of this incorrect classification by allowing an underserved escape hatch to programs with high debt-to-earnings ratios. Yet the discretionary earnings rate has only a weak correlation with ROI, and it explains almost none of the variation in ROI after controlling for the annual earnings rate. It follows that the annual earnings rate alone is the best way to judge programs based on their debt-to-earnings ratios. While ED could comfortably eliminate the discretionary earnings test for all programs, the graduate level has the highest prevalence of incorrectly passing programs. I estimate that eliminating the discretionary earnings test increases the share of graduate programs being correctly classified from 49 percent to 63 percent. Another option to reduce the number of incorrect passes is to lower the amortization period for master’s degrees from 15 to 10 years.

Recommendation 3: Hold schools accountable for institution-level outcomes in addition to program-level outcomes. GE only uses data on program completers to calculate median debt and earnings. But noncompletion is a major risk for students when they begin a program, and schools that have decent ROI for graduates but abysmal completion rates should be held accountable. However, data limitations may not allow the identification of noncompleters with a particular program. One solution is to calculate median debt and earnings for all students who attended a particular institution for a significant period and did not transfer out; this cohort would include both completers and noncompleters. The institution itself would then be held accountable under GE for its aggregate outcomes in addition to its program-level outcomes. At schools with high completion rates, aggregate outcomes would closely track program-level outcomes. However, aggregate outcomes might significantly lag program-level outcomes for schools where dropout rates are high.

Recommendation 4: Release GE outcomes data for programs not subject to GE. While ED may be constrained by the law from applying GE to degree programs at public and private nonprofit institutions, ED could still release outcomes data for these programs to evaluate whether or not they would pass GE. Transparency alone can be a powerful accountability tool. “Naming and shaming” programs that would fail GE if they were subject to the rule might lead students to go elsewhere, even if ED could not legally revoke their federal funding.

The changes to GE recommended in this section are within the power of ED to make on its own. But fixing GE’s biggest problem — its limited scope — might require action by Congress. Moreover, expanding the scope of GE presents a whole new set of considerations, which the next section considers.

How would ‘Gainful Employment for all’ affect higher education?

GE’s exemption of degree programs at public and private nonprofit institutions is its greatest weakness. Just 23 percent of graduates earn a degree at a for-profit school or a certificate anywhere, meaning the rule leaves 77 percent of graduates unprotected. More than two-thirds of negative-ROI programs, weighted by student counts, are not covered under GE.

ED maintains that it is limited by the statutory authority underpinning GE, which applies the gainful employment requirement only to certificate programs and for-profit institutions. But many have urged Congressional action to apply GE to all institutions and programs, regardless of sector or credential type. Scholars have also pointed to Section 454(a)(4) of the Higher Education Act, which allows the Secretary of Education to implement a “quality assurance system” for institutions accessing federal direct student loans. This provision may give ED the legal authority to apply GE across the board, without the need for an act of Congress.

It is important to consider how a hypothetical policy of “GE for all” would affect higher education programs not currently subject to the rule, and whether a GE rule applied to all programs would effectively target those with negative ROI.

ED has not calculated GE information rates for programs not subject to the rule, but program-level data on median debt and earnings is available in the College Scorecard. However, there are important differences between how these variables are defined in GE versus how they are reported in the Scorecard. For instance, GE median debt includes all recipients of Title IV federal aid regardless of whether they received loans; Scorecard median debt includes only borrowers. This makes a minor difference for most programs and but a major difference for some, especially community colleges where many students receive Pell Grants but do not borrow. Still, Scorecard data allows a rough sense of how “GE for all” would affect programs that receive federal aid.

GE results vs. ROI for all programs

Using Scorecard data, I estimate that 21 percent of programs, weighted by student counts, would fail GE if the rule were applied to all programs and current outcomes held. Failures would be concentrated among undergraduate certificate programs and associate degrees, though the failure rate for these credentials at public community colleges may be overstated due to the data problems mentioned earlier. Around 22 percent of doctoral and professional degrees would also fail GE, thanks to these programs’ astronomical median debt levels.

It is possible to calculate earnings relative to passing level (EPL) for all programs to measure how close each program comes to passing or failing GE. Overall, the correlation between EPL and ROI (not adjusting for completion rates) holds up well when GE is applied to all programs. The correlation is 0.71, almost the same as the correlation for programs currently subject to GE. However, the relationship weakens substantially for graduate programs, which tend to have higher debt loads and more variable early-career earnings.

The following chart shows the clear positive relationship between EPL and ROI. Graduate degrees (marked in gray, light blue, and yellow) are less clustered than undergraduate degrees, meaning they are more difficult to classify correctly using the GE framework.

GE for all would incorrectly classify 21 percent of programs; slightly more of these are incorrect passes rather than incorrect fails. GE’s performance is strongest for bachelor’s degrees. Just 13 percent of bachelor’s degrees are incorrectly classified, with most of those being incorrect passes. Associate degrees fare slightly worse: 23 percent of two-year degrees are incorrectly classified, roughly evenly split between incorrect passes and incorrect fails.

How ‘GE for all’ would affect graduate programs

GE’s effectiveness breaks down at the graduate level. Less than ten percent of master’s degrees fail GE, even though more than 40 percent of master’s degrees have negative ROI. Thirty-six percent of master’s degree programs incorrectly pass GE. Since master’s degree programs enroll only college graduates, the earnings of master’s degree holders tend to be elevated even though the degree itself does not add much to students’ earnings power. This phenomenon causes many negative-ROI master’s degrees to pass GE by virtue of their misleadingly high earnings. Nowhere is this more evident than among MBAs, which tend to have high earnings but negative ROI thanks to their students’ strong preexisting earnings potential.

The opposite problem affects professional graduate degrees, including law and medicine. ROI is extraordinarily high for most professional degrees. Though most of them pass GE, 20 percent incorrectly fail. A quarter of law programs and over a third of medical programs would fail GE despite having positive ROI.

This is a function of these programs’ extraordinarily high debt loads ($119,000 for law and $137,000 for medicine) and the relatively modest earnings that many of their graduates can expect in the early career. Medical students, for instance, must complete residencies that pay salaries much lower than they will eventually earn as fully fledged doctors. This inflates medical programs’ debt-to-earnings ratios and causes them to fail GE, even though their graduates will enjoy far higher salaries later on in their careers.

GE addresses this somewhat by using earnings six years after graduation in the calculation for programs requiring residencies, up from the usual three years. Though I incorporate estimated sixth-year earnings into my “GE for all” estimates, it appears that sixth-year earnings are still not a good representation of the full financial value of many medical programs. Given that many observers already consider the United States to have a shortage of doctors, the high number of incorrectly failed medical programs presents a serious concern with the GE framework.

Improving ‘GE for all’

FREOPP has proposed seven principles that should inform the design of accountability policy in higher education. These are:

• Accountability should be based on outputs, not inputs.
• Accountability should consider both benefits and costs.
• Accountability should be sector- and major-neutral.
• Accountability rules should use graduated penalties rather than termination as enforcement.
• Accountability should consider program-level outcomes.
• Accountability rules should not exempt low-ROI but socially valuable programs.
• Accountability should be proactive rather than reactive.

If it were implemented, a policy of “GE for all” would align with several of these principles, making a significant improvement over the status quo. By focusing on student earnings and comparing debt to earnings, GE relies on student outcomes and considers both benefits and costs. The rule also looks at program-level rather than institution-level outcomes, a major improvement over prior attempts at accountability. Presently, GE makes no exemptions for low-ROI programs considered to be “socially valuable.”

The most apparent shortcoming of the current rule is that it is not sector- or credential-neutral. GE applies strict accountability rules to for-profit institutions and certificate programs, but completely exempts degree programs at public and private nonprofit colleges. “GE for all” would fix this issue by applying the same set of rules to programs currently exempt from GE.

However, a universal application of GE would fall short of the ideal accountability system in multiple ways. First, there are many undergraduate certificate programs which provide their graduates with positive ROI, but nonetheless fail GE as currently constructed. Policymakers could partially fix this problem with one of the changes outlined in the previous section: lower the earnings threshold to 85 percent of its current level. But fiddling with the thresholds can only do so much. Making GE still more lenient would pass more positive-ROI programs, but at the cost of allowing more negative-ROI programs to continue receiving funding.

Policymakers could address this by embracing one of the other principles FREOPP outlines: use graduated financial penalties rather than program termination to enforce accountability rules. GE is currently all-or-nothing. A program just below the passing threshold will completely lose access to federal aid, while a program just above will face no penalties whatsoever. Thus the costs of misclassifying a program under GE are high.

Recommendation 5: “GE for all” should introduce a graduated financial penalty. If Congress pursues a “GE for all” policy, it should replace the rule’s all-or-nothing threshold system with a graduated financial penalty for programs below a new performance benchmark. Institutions would pay a fine based on how many of their students are in a failing program and how far below the benchmark that program’s outcomes are. For instance, if the benchmark were an annual earnings rate of five percent, a program with an annual earnings rate of nine percent would pay twice as much per student as a program with a rate of seven percent. Positive-ROI programs only slightly below the benchmark would pay a small fine, which is not the end of the world. The same logic applies to programs where reported earnings are slightly below actual earnings; minor reporting errors are unlikely to be ruinous. But programs far below the benchmark would pay a much larger fine, making continued participation in federal aid financially prohibitive and thus be de facto termination.

Congress could also make “GE for all” more proactive by requiring institutions to purchase insurance against the risk that their programs would be subject to financial penalties. Insurance companies would charge higher premiums to programs at a greater risk of incurring penalties; this would provide a front-end incentive for institutions to improve their outcomes before ED levies a single fine.

Another problem with “GE for all” is more difficult to solve: the relationship between GE metrics and the true financial value of each program breaks down at the graduate level. The issue is twofold: low-value master’s degrees pass GE at high rates, while high-value professional degrees fail more often than they should. Adjusting GE thresholds to make the rule more strict or more lenient usually ameliorates one problem while exacerbating the other. However, one parameter of GE can be made more strict and more lenient at the same time: the amortization period for calculating annual loan payments.

Recommendation 6: Change the amortization period for calculating graduate degrees’ annual loan payments. Given that master’s degrees are one- and two-year programs, the amortization period should be closer to that of associate degrees than that of bachelor’s degrees. Reducing the amortization period for master’s degrees from 15 years to 10 years lowers the share of incorrectly passing master’s programs from 36 percent to 31 percent. Meanwhile, professional degrees which provide truly lifetime benefits should have a longer amortization period. Increasing the amortization period for doctoral and professional degrees from 20 to 25 years reduces incorrect fails among professional programs from 20 percent to 16 percent.

While adjusting amortization periods for graduate programs marginally improves GE’s targeting, it stops short of fully solving the problem. The relationship between ROI and metrics such as debt-to-earnings ratios is complex at the graduate level: master’s degrees tend to have lower ratios but lower ROI, while professional degrees tend to have higher ratios and higher ROI. Simple tests based on variations of debt-to-earnings ratios may not be best suited to hold graduate programs accountable for their outcomes.

Recommendation 7: End federal loans for graduate school, and use the private sector to hold these programs accountable. The justification for a federal role in student lending is that young adults have little credit history and would thus have a difficult time securing private education loans at reasonable interest rates. But this argument largely does not apply to graduate students, who are older and have already demonstrated their ability to complete a bachelor’s degree. In fact, studies show that most graduate students could secure private loans at lower interest rates than those on federal loans. Private lenders, however, would be better equipped than ED to assess the financial value of graduate degrees and strategically deny financing to programs unlikely to pay off. Eliminating the federal graduate loan programs would allow the private sector to step in and fund positive-ROI graduate degrees, but with a more sophisticated system of underwriting that would provide better-targeted accountability.

Conclusion

The Biden administration’s proposed Gainful Employment framework would move higher education accountability in the right direction. GE would terminate federal funding for hundreds of negative-ROI programs, protecting students from taking on debt for programs that will not pay off and likely saving taxpayers money as well.

However, the framework as proposed would also terminate funding for several programs that do have positive ROI, especially undergraduate certificates in medical assisting and dental support services. GE would likely shut down nearly 70 percent of medical assisting certificate programs, even though most of these yield a modest but real earnings gain for their students. Some simple changes to the framework, such as lowering the high school earnings threshold to 85 percent of its current level, would help address this problem.

Technical changes to the framework would not address GE’s greatest shortcoming: its limited scope. Many reformers have proposed applying GE to all programs, including the degree programs at public and private nonprofit institutions that are currently exempt. This is welcome, but if the GE framework were applied to all programs as written, ED would unfairly terminate hundreds of good programs at trade schools while giving most low-value master’s degrees a pass. Further changes to the framework are warranted before GE is applied across the board.

While a robust proposal, GE should not be the last word in higher education accountability and reform. Policymakers should consider a system of graduated financial penalties to hold schools accountable in lieu of GE’s current all-or-nothing design. Moreover, Congress could consider supplementing GE with other outcomes metrics such as student loan default and repayment rates; future research will explore how well these metrics proxy for ROI. Finally, Congress should explore leveraging the private sector to aid in higher education accountability by requiring insurance for federally dependent institutions and ending the government role in graduate school lending. These changes would leave America’s higher education system stronger and better positioned to advance economic mobility for all students.