A visual guide to the OB3A accountability rule that measures whether college programs deliver economic value to their graduates.
01
The Core Question
🎓➔💰
How much do graduates earn 4 years after leaving?
$60,000
This is the median earnings across all testable programs, measured 4 years after graduates leave school. It comes from IRS tax records linked to federal student aid records. Graduates with zero earnings — those who are unemployed, self-employed without W-2 wages, or out of the labor force — are excluded. This makes the number more favorable to programs than if non-earners were counted.
A note on what this number leaves out: This zero-earner exclusion compounds with the ACS measurement issue covered in the next section. The program side (IRS) excludes non-workers, which biases earnings up. The benchmark side (ACS) is self-reported and includes all workers, including the self-employed, which can bias in either direction. For a field like cosmetology, where many graduates are self-employed (booth renters, independent stylists), the test creates a double mismatch: the program earnings miss their self-employment income (IRS captures W-2 only), while the benchmark includes self-employed high school grads. Both effects work against those programs.
02
The Benchmark
The earnings test doesn't compare programs to an arbitrary number. It asks: do graduates earn more than people with less education? The specific benchmark depends on two things: the degree level and where students come from.
What type of program?
Undergraduate (Associate / Bachelor's)
Mostly in-state students? Compare to: median earnings of high school graduates in that state
Mostly out-of-state students? Compare to: median earnings of high school graduates nationally
Graduate (Master's / Doctoral / Professional)
Mostly in-state students? Compare to the lowest of three benchmarks:
• Same-state, same-field BA holders
• Same-state BA holders (any field)
• National same-field BA holders (Uses whichever is lowest — most favorable to the program)
Mostly out-of-state students? Compare to the lower of two benchmarks:
• National same-field BA holders
• National BA holders (any field) (Uses whichever is lower)
A note on measurement: The benchmarks come from the American Community Survey (Census Bureau), which is based on self-reported income. Program earnings, by contrast, come from IRS tax records — actual W-2 wages reported to the government. These two data sources don't measure income the same way. Self-reported survey earnings tend to be less precise (people round up or down, estimate, or omit income), while IRS data captures only W-2 wages and misses self-employment income. The result: the test compares administrative tax data against survey-based estimates, which can introduce subtle bias in either direction depending on the field and income level.
03
The Test
Select an example to see how different programs compare against their benchmark.
Registered Nursing (BSN) — Ohio University, OH
Program Earnings (IRS W-2, 4yr post-exit)
$87,308
Benchmark (OH high school grad median — ACS)
$34,808
✔
PASS
Graduates earn $52,500 more than the benchmark.
04
What Happens When Programs Fail
📊
Program is flagged as failing
The Department publishes the results. The program's median earnings fall below its benchmark — graduates are earning less than people with less education.
⏳
Grace period
First sanctions don't begin until Academic Year 2028-29. Institutions have time to improve outcomes or close underperforming programs.
⚠️
Program-level sanctions
Failing programs face restrictions on federal student aid eligibility — limiting their ability to enroll students who depend on grants and loans.
🏛
Institutional trigger
If 50% or more of an institution's testable programs fail, the entire institution faces provisional certification — a major regulatory and reputational consequence.
05
The Catch — What We Can't See
The earnings test can only measure programs with enough graduates to produce a statistic. When a program has fewer than 16 graduates in the measurement window, the IRS suppresses the earnings data to protect privacy.
Pass
Fail
Suppressed
Only ~24% of programs have enough graduates to measure
The remaining ~76% are invisible to the test. The 2,880 failing programs identified in the data are likely a lower bound — most programs simply can't be evaluated.