How to Download and Use the Program Performance Data for Your Institution

You can already see your own institution’s numbers without touching the raw data. The interactive earnings-test dashboard shows where every program with reported earnings stands against its benchmark, and you can search it by institution or state to see which of your programs pass, fail, or sit close to the line. If you just want the answer, start there. It is ready to use.

This guide is for when you want more than the summary. If your institutional research team needs to link these federal results to your own programmatic data, enrollment, cost, outcomes, or anything else you track, or you simply want to check the underlying figures and run your own cuts, you can recreate the full dataset from the source. Here is how to download it, clean it, and read your programs’ pass or fail results.

For background on what the test measures and what happens to programs that fail, start with How the New Federal Earnings Test for Colleges Works. This guide is the hands-on companion: it assumes you know roughly what the test does and you just want the data in front of you.

What the PPD Is

The 2026 Program Performance Data (PPD:2026) is the federal dataset behind the STATS Earnings Accountability framework, enacted through the Working Families Tax Cuts Act (originally the One Big Beautiful Bill Act). It covers 209,321 programs at 5,096 institutions and reports, at the program level:

• Who enrolls, including demographics and Pell share
• Who completes, and who does not
• What students borrow across subsidized, unsubsidized, Grad PLUS, and Parent PLUS loans
• What graduates earn, linked to IRS records, four years after completion
• Whether the program passes or fails the new earnings test

If you work in higher education strategy, policy, enrollment management, or institutional finance, this is the most consequential dataset released in a decade. It is also, in its raw form, a little unfriendly. The steps below get you from the download page to a clean institutional profile.

Step 1: Download the Files

Where to Get Them

Go to the Department of Education negotiated rulemaking page:

ed.gov/laws-and-policy/higher-education-laws-and-policy/higher-education-policy/negotiated-rulemaking-for-higher-education-2025-2026

Click on the AHEAD session and scroll to the Session 2 Materials section. The data files are listed under the session materials distributed by the Department.

What to Download

There are eight files across three levels of granularity:

Each file name above links directly to the download. Start with files 1 and 6. File 1 tells you what the programs are. File 6 tells you whether they pass or fail. You can add the others as you need them.

One version note before you download. The Department released an initial version of this data in December 2025 and then posted a corrected update dated January 8, 2026. The files linked above are the current January release (the codebook and technical appendix are stamped update_01.08.2026, and the codebook itself flags a correction to how the master pass/fail variable was documented). If you have an older copy saved from December, replace it with these, along with the four supplemental files below, which were added in the update.

The column names inside each spreadsheet use short codes like opeid6, cip4, and md_earn_wne_p4. Those are the field identifiers referenced throughout this guide, and the Variable Codebook defines every one of them.

Also Download the Documentation

You will want these open in another window while you work:

Optional: Supplemental Data Files

Alongside the eight core files, the Department posted four supplemental datasets that were not in the original release. They join to the core files on the same keys and are useful when you are building a fuller exposure picture. The Data Supplements summary above explains what each one adds.

Step 2: Understand the Structure

How the Files Fit Together

Files 1 through 6 are the core dataset. They all have exactly 209,321 rows, one row per program, where a program is defined by the combination of three keys:

• opeid6, the institution, identified by its 6-digit OPEID
• cip4, the field of study, a 4-digit CIP code (for example, “5200” is Business)
• credlev, the credential level (for example, “Bachelor” or “Master’s”)

Files 1 through 6 merge one to one on these three keys. They were split into separate files only because of size limits. Conceptually they are one big table.

File 7 is at the CIP6 level, which is more granular (for example, “520201” for Accounting rather than “5200” for all Business). File 8 has a completely different structure and is mostly useful for program length information.

The Key Identifier: OPEID6

The PPD uses OPEID6, the 6-digit Office of Postsecondary Education ID, not the IPEDS UNITID you may be used to. File 1 includes a unitid column that lets you crosswalk between the two.

Most institutions have one OPEID6 per UNITID, but there are exceptions such as multi-campus systems and branch campuses. When in doubt, use the unitid column in File 1 to crosswalk.

Step 3: Handle the “Privacy Suppressed” Trap

This is the single most common mistake people make with this data, so handle it before you do anything else.

To protect student privacy, the Department hides any figure drawn from too few students, and it does this in two different ways depending on the file. Knowing which is which will save you a lot of confusion.

• In File 1, and in the demographic and pre-program fields, a hidden value appears as the literal text "privacy suppressed" sitting in a column that should hold a number. In Excel these cells look filled but break your formulas; in any tool they turn the whole column into text.
• In File 6, the earnings and test columns do not use that string at all. A hidden value is simply left blank. So a blank earnings cell is not missing by accident, it means the cohort was too small to report.

The threshold varies by measure: roughly fewer than ten students for most program-level counts, fewer than 30 for the FAFSA-based fields, and a small IRS cohort for the earnings figures. The practical rule is the same in every case: before you calculate anything, replace every "privacy suppressed" string with an empty cell, and remember that an empty earnings cell is a suppressed value, not a zero.

How Much Is Suppressed

Of the 209,321 program rows, only about 49,900 have a reportable median-earnings figure, and therefore a pass or fail result. The other three quarters are blank, either because the cohort was too small or because the program is not subject to the test. Suppression is most common in small programs, graduate programs, newer programs, and arts and humanities fields.

Keep this in mind: a suppressed cell is not a zero and it is not a pass. It means the Department did not have enough students to report a reliable number.

Step 4: Build Your Institution’s Profile

Filter Files 1 and 6 down to your institution however you normally identify it, by opeid6, unitid, or instnm, and join them on the three program keys from Step 2. These are the columns worth pulling first:

Step 5: Read the Earnings Test Result

A program passes if its median graduate earnings meet or exceed a benchmark. The benchmark is a median-earnings figure for 25-to-34-year-olds, and which one applies depends on the credential level:

The exact benchmark chosen for a given program is named in which_test_cip2_wageb and its dollar value is in ge_earn_test_bm_algn. The columns that carry the result:

Use the Right Pass/Fail Column

File 6 contains two columns that look like they both give the pass/fail result. Use mstr_obbb_fail_cip2_wageb. The codebook describes it as the flag for whether a program “fails the earnings test described in OBBB,” and it applies to all programs regardless of credential level.

The other one, fail_obbb_cip2_wageb, is an earlier program-level result computed before the Department’s process for aggregating small programs. It marks only 1,220 programs as failing. The master column marks 2,880, and that larger figure is the one the Department uses. Where both columns have a value they never disagree, but the master column assigns a result to about 5,800 programs the other one leaves blank. If you filter on fail_obbb_cip2_wageb, you will undercount your failing programs by more than half.

The benchmark pairs the same way. The master flag compares md_earn_wne_p4 against ge_earn_test_bm_algn, not against earn_bnchmrk_cip2_wageb. The two benchmark columns hold identical dollar values wherever both are present, but ge_earn_test_bm_algn is filled in for every tested program, so it is the one to use for margin math. In every one of these columns, 1 is a failing program, 0 is a passing program, and a blank is a program that could not be tested.

Flag Tipped-Occupation Programs

The final rule (91 FR 40136, published July 1, 2026), at 34 CFR 668.402(c)(4), delays accountability sanctions for programs tied to occupations where the IRS has determined that at least half of workers receive tip income. This is a timing provision, not an exemption. The Department still computes and publishes each program’s earnings and pass or fail result, so nothing changes in the data you pull. What changes is that a listed program is not counted as passing or failing for sanction purposes until its graduates’ earnings are measured in a “No Tax on Tips” year (tax year 2026 or later), which buys these programs at least a one-year delay. It is still worth flagging them so you read their results in the right light.

The rule defines the delay with a specific list of 20 six-digit CIP codes, given in Table 5.22 of the final rule, on page 40236 of the PDF. Be warned that the table is published as an image, so searching the PDF text for these codes will not find them. They are:

There is a catch when you apply this to the PPD. The list is at the six-digit CIP level, but the core files (1 through 6) identify programs only at the four-digit CIP, so you cannot separate a tipped six-digit code from its non-tipped siblings. The 20 codes fall into five four-digit families:

Filtering the core files to these five cip4 families is the closest you can get, and it matches how the Department handled the delay in its own impact analysis. Just know that it over-includes non-tipped programs in 1204, 1205, and 5209, since only part of each of those families is on the list. If you need an exact match to the 20 codes, use six-digit CIP, which is available in File 7 (the program earnings file).

Read More

Once you have your institution’s numbers, the analysis pieces put them in context. Which Programs Don’t Pay Off? breaks down the programs that fail under the new rule, and How the New Federal Earnings Test for Colleges Works explains what a failure actually triggers.

Questions? Get in touch! We’re happy to help.

← All writing