Virginia is the state everyone points to when they want to say workforce data can be done right. It spent a decade building FastForward, a state-funded system of noncredit credentials with wage tracking built into the funding model since 2016.
Then Virginia ran its own 250 credit and noncredit workforce programs through Workforce Pell's length, completion, and earnings filters. Six were eligible.
Why do so few workforce programs qualify for Workforce Pell? Workforce Pell requires a program to document a completion rate of at least 70 percent, a job placement rate of at least 70 percent, and a value-added earnings result showing graduate earnings clear the cost of the program. The binding constraint for most institutions is not program quality, it is measurement. State longitudinal data systems match graduates to jobs using unemployment insurance wage records, which code employment by the employer's primary industry rather than the graduate's actual occupation, and which cannot see employment across state lines. Virginia, widely regarded as the national model, found only six of 250 programs eligible when it applied the federal screens. In North Carolina, system president Jeff Cox estimated roughly 4 percent of existing short-term credential programs were likely to qualify.
If the honor student clears the bar six times out of 250, the test is measuring something other than whether the programs work.
The cybersecurity graduate who works in retail
Randall Stamper, Virginia's associate vice chancellor for workforce programs, gave the cleanest illustration of the problem in Bruno V. Manno's reporting for Washington Monthly.
A Virginia student earns a cybersecurity credential. They land a cybersecurity job at Walmart. Real job, real salary, exactly the field they trained for. The state's employment data records them as working in retail, because the code follows the employer's industry classification rather than the graduate's role.
If that same graduate takes a job across the state line, Virginia's system does not see the placement at all. It does not record an out-of-state hire. It records nothing.
Multiply that across fifty states with fifty different data architectures and the numbers meant to separate strong programs from weak ones stop functioning as a scorecard. A good program with a border-adjacent labor market and graduates hired by large diversified employers can fail on paper while placing nearly everyone.
This is not a reporting glitch. It is the difference between a program getting funded and a good program looking like it failed.
For Career Services Teams at Universities
Your placement rate is only as good as the record behind it
State wage records will not tell you that your cybersecurity graduate took a cybersecurity job, or that eleven of your completers crossed a state line for work. Prentus captures employer, job title, and start date directly from students as part of ongoing career services engagement, then verifies it, so in-field placement is something your institution can document instead of something a state agency approximates.
Virginia is the ceiling, not the floor
Virginia's data exists because of a policy accident of good design. The state built a pay-for-performance funding model in 2016 that tied reimbursement to credential attainment, which forced systematic statewide collection as a side effect years before Workforce Pell was written. A peer-reviewed study this year found FastForward credentials raise quarterly earnings by roughly $818, about a 10 percent gain over pre-enrollment pay, recouping program cost in a little over half a year.
That is a genuinely strong return, produced by a genuinely strong data system, and it still yielded six eligible programs out of 250.
Other states are working from much less. North Carolina's community college system president told the state's Governor's Council on Workforce and Apprenticeships that once length, completion, and placement screens were applied, roughly 4 percent of existing short-term credential programs were likely to qualify. As of late July, twenty-six states had an operational system in place to enroll students at all.
The Education Department wrote the rules and handed enforcement almost entirely to the states, without funding the infrastructure the rules require. New America's Iris Palmer called it an unfunded mandate. Carrie Warick-Smith of the Association of Community College Trustees put the timeline plainly: July 1 was not a floodgate, it was the start point of a marathon.
Now the same test comes for degree programs
Most institutions watching this treat it as a community college story. It is not, and the date is on the calendar.
Starting July 1, 2027, the earnings benchmark reaches most regular degree programs. Programs whose graduates fail the earnings test in two out of three years can lose access to federal student loans. The rules and the mechanics are laid out in our breakdown of the earnings rule, and the underlying measurement problem is the same one described in the first-destination data gap.
So the missing placement data that is disqualifying workforce programs this year is the same missing placement data that will sit underneath degree program funding next summer. Except most four-year institutions are measuring outcomes with an annual graduate survey that a fifth of alumni open, compiled into a spreadsheet by one person in institutional research.
There is one real reprieve worth knowing. The stricter requirement that a graduate's job be in the field they trained for does not take effect until the 2029-30 award year. That is time to build occupation-level matching. It is not time to do nothing.
What institutions can control
No single college is going to fix its state's unemployment insurance data architecture. Washington State required employers to report each worker's occupation starting in 2022, and researchers are only now connecting that to postsecondary records. Lumina is working with a dozen states on similar infrastructure. That work will take years.
What an institution can control is whether it holds its own record of where graduates went:
- Occupation, not industry. Capture the job title and function the graduate actually holds, so a cybersecurity hire at a retailer reads as a cybersecurity hire.
- Placements that leave the state. Out-of-state hires are invisible to state wage matching and have to be collected directly or they count as zero.
- Continuous, not annual. Data collected as part of how students already use career services survives a deadline. A survey sent six months after graduation does not.
- Verified, not self-reported alone. A placement claim a school can substantiate is worth more under federal review than a number it cannot source.
Institutions that own that record are not at the mercy of how their state codes a job. They can show a regulator, an accreditor, or a legislature what actually happened to their graduates, and they can correct a state number that undercounts them.
If you are working out how to capture career outcomes without adding another annual survey to the pile, we would welcome the conversation.
Schools that start now can lead on career outcomes. The ones that wait will spend next summer digging through five years of spreadsheets named FINAL_v7.
