This note documents the data sources, methodology, and known limitations behind the Value-Added Earnings Test interactive tool. The tool approximates the Workforce Pell value-added earnings calculation using publicly available data. It is not the official federal calculation.
The interactive tool lets users select a state and institution, then see which short-term certificate programs would pass a two-step test:
Users can override the pre-filled tuition with actual program costs to test eligibility interactively.
| Source | Provider | Version | What it supplies |
|---|---|---|---|
| Post-Secondary Employment Outcomes (PSEO) | U.S. Census Bureau | 2025Q4 | Median earnings at Years 1, 5, and 10 after completion. Graduate counts (graduates with earnings in LEHD). By institution, CIP code, and degree level. |
| Regional Price Parities (RPP) | Bureau of Economic Analysis | 2023 | State-level price index (all items) used to adjust earnings for regional cost of living. National average = 100. |
| Integrated Postsecondary Education Data System (IPEDS) | NCES, Dept. of Education | 2023-24 | In-state tuition and fees by institution. Used to pre-fill the tuition input. Matched to PSEO institutions by name and state. |
| Federal Poverty Guidelines | Dept. of Health and Human Services | 2026 | 150% of the Federal Poverty Level for a single individual: $23,940. |
| Workforce Pell Guidance (PDF) | Dept. of Education | 2025 | Official methodology for the value-added earnings test under proposed Workforce Pell rules. |
The tool includes PSEO records matching all of the following criteria:
This yields 10,025 program-institution records across 512 institutions in 25 states. PSEO coverage is not national; it depends on state participation in the LEHD program and data-sharing agreements with postsecondary institutions.
PSEO publishes four five-year graduation cohorts (2001-2005, 2006-2010, 2011-2015, 2016-2020) and a pooled all-cohorts aggregate. Individual cohorts have significant limitations:
The pooled cohort maximizes cell sizes, reduces suppression, and provides the widest timepoint coverage. The trade-off is that it blends earnings across labor market conditions from different periods. This parallels the Department of Education's own approach, which pools up to four years of completers to reach the 50-completer minimum.
This tool is an approximation, not a reproduction of the federal calculation. The following differences are material:
The Department of Education requires median earnings measured 3 years after program exit. PSEO does not publish a 3-year time point. This tool uses Year 5 median earnings as the closest available proxy. Because earnings generally grow over time, Year 5 figures likely overstate what the official Year 3 calculation would produce. Value-added results shown here should be interpreted as an upper-bound approximation.
PSEO publishes certificate earnings at the 4-digit CIP code level, which groups related programs together (e.g., CIP 51.39 covers all practical nursing programs). The actual Workforce Pell test operates at the 6-digit CIP or individual program level. Institutions may offer multiple distinct programs within a single 4-digit CIP with different earnings profiles and completer counts. The 4-digit aggregation is suitable for analysis and pattern recognition but does not represent the exact federal counting unit.
The "Graduates with Earnings" column shows the number of program completers whose earnings appear in the LEHD database, which covers over 96% of U.S. employment. LEHD is built from state unemployment insurance wage records, supplemented with Census surveys and other federal administrative records. This count is a subset of total completers. The Department of Education's 50-completer minimum threshold applies to actual completers, which may be higher than the LEHD-based count shown here.
IPEDS reports a single annual, institution-wide in-state tuition and fees figure. It does not break out costs by program or by credential length. For short-term certificates (under 1 year), the actual program cost is likely lower than the annual figure. Users should override the pre-filled value with the actual published tuition for the specific program when available.
PSEO and IPEDS use different institution identifiers. This tool matches them by normalized institution name and state, with fuzzy matching for minor name variations. Of 512 PSEO institutions with certificate data, 435 (85%) were successfully matched to IPEDS tuition records. The remaining 77 institutions show no pre-filled tuition; users can enter it manually.
The Census Bureau suppresses PSEO earnings and graduate counts when the protected cell count falls below 30 after applying differential privacy. This is a disclosure avoidance measure to protect individual privacy.
Suppression is the dominant feature of this dataset. Of 10,025 program-institution records, 6,638 (66%) have suppressed earnings. These programs cannot be tested for value-added and appear in the tool with "suppressed" labels. Every program with suppressed earnings also has a suppressed graduate count.
Suppression is not random. It correlates with program size, institutional scale, and state-level data infrastructure. Smaller programs at smaller institutions are more likely to be suppressed. This has significant implications for how the value-added test operates in practice, particularly for rural institutions and states with decentralized community college systems. See the related analysis below.
PSEO is an experimental data product. Its coverage depends on:
Results should not be interpreted as nationally representative. States and institutions not present in PSEO are absent due to data-sharing limitations, not because of program quality.
All data used in this tool is publicly available from the sources listed above. The Python build script (build_value_added.py) reads the raw PSEO CSV files, applies the value-added formula, matches IPEDS tuition by institution name, and generates the interactive HTML page with embedded data. No proprietary data or APIs are required. The BEA Regional Price Parities are embedded directly in the script as published values.