Your STATS exposure map: triage your programs before February 2027

In February 2027, the U.S. Department of Education will publish the first official Earnings Premium results under STATS, at the 6-digit CIP level, for every Title IV program at every institution that accepts federal student loans. The list will be public. Prospective students will read it. Reporters will read it. Your trustees will read it.

The institutions that have already run the analysis on themselves will be making restructure and discontinue decisions privately, in the quiet months between now and the publication. The institutions that wait will be making the same decisions in public, under pressure, after a press release goes out.

PPD:2026, the program-level dataset ED released to inform the AHEAD negotiated rulemaking, is already on the street. The data isn't final, and the official results won't be calculated until 2027, but the methodology and the program universe are close enough to project against. That gap, between PPD:2026 today and ED's published results in February 2027, is the only window for quiet triage. Roughly eighteen months. The clock is running.

This piece is a four-step program-level playbook for IR teams who want to walk into 2027 with a map, not a surprise. The companion piece on the 50% institutional-level trigger covers the portfolio-rollup question; this one stays at the program layer. For background on how STATS, FVT/GE, and PPD:2026 fit together as policy, the explainer post is the right starting read.

STATS replaces the old debt-to-earnings test with a single, sharper measurement: the Earnings Premium. The OBBB-to-Earnings-Premium math walkthrough covers where this formula came from in statute and the 2-of-3 eligibility math in detail; the short version follows here.

The formula reads in one line. Median earnings of your working graduates, measured four tax years after they completed the program, minus the median earnings of working adults aged 25 to 34 with only a high school diploma in the same state. Positive number, you pass. Zero or negative, you fail. Three changes from FVT/GE are worth noticing: the earnings horizon moved from three years out to four, non-working completers are excluded entirely (they used to pull medians down), and graduate programs are now benchmarked against working bachelor's-degree holders in the same state and field, not high-school graduates. For master's programs in particular, that's a much higher bar than FVT/GE ever applied.

A single failure year is survivable. A program loses federal student loan eligibility only after failing in two out of any three consecutive years. The eligibility loss runs for two years, and during that window the institution can't quietly relaunch the same program under a new name; the rule explicitly blocks substantially similar programs sharing CIP code and occupation codes. That last constraint is the one most schools haven't priced in yet, and it's the reason Step 3 below matters this year, not next.

Start by running every program in your portfolio against the OBBBA-aligned Earnings Premium test in PPD:2026. (If your analyst is the one running this for the first time, the PPD:2026 data-wrangling walkthrough covers the join keys, benchmark matrix, and privacy-suppression rules they'll need to get right.) You're looking for three buckets: clear pass, clear fail, and at-risk (within 10 percent of the benchmark in either direction).

The published failure rates by program type tell you where to look first. The pattern is consistent: the rule hits hardest at short-term certificate programs in fields where earnings are modest and variable, and barely touches programs with strong labor-market returns. The chart below shows the OBBBA-aligned failure rates pulled from PPD:2026 estimates.

If you're a community college with a strong cosmetology or culinary program, that program is statistically more at risk than your engineering technology program, regardless of how much you've invested in either one. The score is a starting point, not a verdict. Some programs in high-fail-rate fields will pass comfortably; some in low-fail-rate fields will sit on the edge. The point of Step 1 is to know which is which before someone else does.

A single soft year doesn't lose eligibility. Two failures in any three consecutive years does. That distinction changes which programs need urgent attention and which need monitoring.

For each at-risk program, plot a five-year forward window and ask one question: how many of these years would have failed under the OBBBA-aligned test? A program that has been within 5 percent of the benchmark for the last three measurement periods is materially different from a program that failed once by 30 percent and has otherwise passed. The first is a structural problem and almost certainly will trip the 2-of-3 rule on its next cycle. The second is a cohort anomaly and probably won't.

The earnings horizon matters here. STATS measures earnings four years after program completion, and the data is pooled across two graduation cohorts. So today's intervention (improving job placement, tightening the curriculum, partnering with employers) won't show up in the federal data for five or six years. The cohort whose results land in February 2027 graduated in roughly 2022. If a program is at risk, the decision to address it is already late. The decision to plan around it is right now.

If a program fails the 2-of-3 test, it loses Direct Loan eligibility for two years. During that two-year window, the institution cannot launch a substantially similar program, defined as one sharing the same CIP code and overlapping occupational codes, to replace the failing one.

This is the rule that quietly closes the side door. Under FVT/GE, an institution could shut a failing program and stand up a renamed near-clone the following year. Under STATS, that path is blocked at the CIP-plus-occupation level. The substitution either has to genuinely change what's being taught (different CIP, different occupational outcome) or it has to wait out the full two-year ineligibility period.

For any program you've identified as at-risk in Step 1, work the substitution table below before the first failure year, not after.

In AIR's 2023 survey on FVT/GE implementation, 72 percent of responding members reported that their IR or IE office held institutional responsibility for the compliance reporting. The most-cited pain points were tight timelines and a lack of clear communication from the Department of Education. Those were Biden-era complaints about a regulation that ultimately never produced published results.

STATS is the version that will produce published results, and it will produce them in February 2027, at a more granular CIP level, with institutional-level consequences that didn't exist under FVT/GE. (Our analysis of all 8,796 NPRM public comments covers which final-rule softening the sector is asking for, and which is most likely to land.) The same IR teams that lived through the FVT/GE reporting cycle (October 1, 2026 is the last submission) are going to be asked to run program-by-program risk analyses, model substitution paths, and brief the cabinet on portfolio exposure. With the same headcount.

The teams that won't be overwhelmed are the ones that stop treating this as a one-off compliance exercise and start treating it as recurring portfolio analytics. The question "which of our programs would fail STATS today?" is not a once-a-year question. It's a quarterly question, asked at the program level, at the institutional rollup level, and by sector for benchmarking.

PPD:2026 is six Excel files and 58 columns of variable names like fail_obbb_cip2_wageb. Running Step 1 by hand (joining the files, applying the right benchmark column for your credential level, filtering on your OPEID) is a long week of analyst time before you produce a single chart. Re-running it next quarter, with a new filter or a new scenario, is another long week. The six-files walkthrough lays out exactly where those hours go.

Clema's STATS (FVT/GE) AI Agent reads the full PPD:2026 dataset directly. You ask in plain English (which of our programs fail under the OBBBA-aligned test, which sit within 10 percent of the benchmark, which would trip the 2-of-3 window under a flat-earnings scenario, which CIP-6 + SOC combinations are off-limits for the next two years) and it answers, with the underlying numbers cited from PPD:2026 variables. The same question, re-framed for a different scenario, is the next sentence. Not the next weekend.

If you're heading into the eighteen-month triage window without an answer ready, this is the path that gets you one in an afternoon. When you've mapped your program-level exposure, the 50% institutional trigger is the natural next read.

AIR, Program Accountability Reporting resource center. STATS Notice of Proposed Rulemaking, docket ED-2026-OPE-0100. U.S. Department of Education, Federal Student Aid Partners, FVT/GE knowledge center.