DAAF-Contribution-Community/daaf / .claude/skills/education-data-source-pseo
education-data-source-pseo
VerifiedPSEO — Census data linking graduates to employment via LEHD wage records. Earnings percentiles at 1/5/10 years post-graduation by institution, degree, and CIP. Use for graduate earnings analysis. Coverage changes by release; Census V4.14.1 (2026Q2, posted 2026-08-25) lists 1,117 institution rows and partners in 37 states plus D.C. (plus online-only WGU).
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 22.3 KB |
| references/earnings-data.md | markdown | 10.4 KB |
| references/geographic-flows.md | markdown | 10.5 KB |
| references/industry-flows.md | markdown | 12.6 KB |
| references/lehd-methodology.md | markdown | 9.7 KB |
| references/state-coverage.md | markdown | 8.3 KB |
| references/variable-definitions.md | markdown | 15.7 KB |
references/lehd-methodology.md
View on GitHub# LEHD Methodology
How the Longitudinal Employer-Household Dynamics (LEHD) program produces PSEO tabulations.
## Contents
- [LEHD Program Overview](#lehd-program-overview)
- [Data Sources](#data-sources)
- [Data Matching Process](#data-matching-process)
- [Employment Coverage](#employment-coverage)
- [Differential Privacy Protection](#differential-privacy-protection)
- [Historical Context](#historical-context)
## LEHD Program Overview
The Longitudinal Employer-Household Dynamics (LEHD) program is part of the U.S. Census Bureau's Center for Economic Studies. LEHD creates public-use information by combining multiple administrative data sources:
- **Local Employment Dynamics (LED) partnership**: State unemployment insurance wage records
- **Federal employment data**: Office of Personnel Management records
- **Census Bureau surveys**: Demographic and business data
- **Partner institution data**: University transcript records
PSEO is one of several LEHD data products:
| Product | Description |
|---------|-------------|
| QWI (Quarterly Workforce Indicators) | Employment, earnings, job creation by industry/geography |
| LODES (Origin-Destination Employment Statistics) | Home-work commuting patterns |
| J2J (Job-to-Job Flows) | Worker transitions between employers |
| PSEO | Graduate employment outcomes |
## Data Sources
### Transcript Data (Education Input)
Higher education systems and individual colleges and universities submit graduation files containing:
| Field | Description |
|-------|-------------|
| Social Security Number | Submitted with the graduation record (PSEO Coalition methodology page); the PSEO Technical Documentation names only a "personal identifier"; graduates whose identifier is of poor quality (less than one percent) are omitted from the published statistics |
| Date of birth | Submitted with the graduation record |
| Graduation year | When credential was awarded |
| Degree/credential type | Certificate, Associate's, Bachelor's, Master's, Doctoral |
| Major/discipline | CIP code (Classification of Instructional Programs) |
| Institution | Institution of award (Census tabulates by 6-digit OPEID) |
### Employment Data (LEHD Input)
State unemployment insurance (UI) wage records include:
| Field | Description |
|-------|-------------|
| Worker identifier | SSN or state equivalent |
| Employer identifier | State Employer ID (SEIN) |
| Wages | Quarterly earnings |
| Industry | NAICS code from employer |
| Location | State and county of establishment |
## Data Matching Process
```
┌─────────────────────┐ ┌─────────────────────┐
│ Transcript Data │ │ LEHD Jobs Data │
│ (from partners) │ │ (UI wage records) │
└──────────┬──────────┘ └──────────┬──────────┘
│ │
└───────────┬───────────────┘
│
▼
┌─────────────────┐
│ Person Record │
│ Linkage │
│ (personal ID) │
└────────┬────────┘
│
▼
┌─────────────────┐
│ Employment │
│ Histories │
│ by Graduate │
└────────┬────────┘
│
▼
┌─────────────────┐
│ Tabulation & │
│ Aggregation │
└────────┬────────┘
│
▼
┌─────────────────┐
│ Differential │
│ Privacy Noise │
└────────┬────────┘
│
▼
┌─────────────────┐
│ Published │
│ Statistics │
└─────────────────┘
```
### Key Matching Steps
1. **Standardize identifiers**: Clean and format the personal identifiers submitted with each graduation record (the PSEO Technical Documentation names only a "personal identifier"; exact matching keys are not published here)
2. **Link to Person Universe Infrastructure (PUI)**: Census's master person file
3. **Attach employment histories**: All jobs in LEHD system for matched graduates
4. **Calculate post-graduation outcomes**: Earnings and employment 1, 5, 10 years out
5. **Apply labor force attachment filters**: Remove marginal workers
6. **Aggregate to tabulation cells**: By institution, degree, CIP, cohort
7. **Apply privacy protection**: Differential privacy noise injection
## Employment Coverage
### What's Covered
LEHD covers **over 96%** of U.S. employment (PSEO Technical Documentation):
**Private sector** (covered by state UI):
- All wage and salary workers
- Most corporate officials and executives
- Part-time workers
- Workers on paid leave
**State and local government**:
- Most employees (varies by state)
- Elected officials often excluded
**Federal government** (from OPM data):
- Most civilian federal workers
- Coverage: 2000-2020 per the PSEO Technical Documentation (updated June 12, 2023); later releases may extend this
### What's NOT Covered
| Category | Reason |
|----------|--------|
| Self-employed (unincorporated) | Not covered by UI |
| Independent contractors | Not covered by UI |
| Railroad workers | Separate railroad UI system |
| Military | Not in OPM data |
| U.S. Postal Service | Not in OPM data |
| Some family employees | State-specific exemptions |
| Certain farm workers | State-specific exemptions |
| Some non-profit workers | State-specific exemptions |
| National security agencies | Security exclusions |
### UI Coverage by State
States have discretion in UI coverage rules. LEHD's state coverage grew over time (PSEO Technical Documentation, updated June 12, 2023):
- **Early 1990s**: about 10 states
- **Late 1990s**: about 40 states
- **2010**: Massachusetts, the last state, entered the LEHD system
## Differential Privacy Protection
PSEO uses **differential privacy** to protect individual confidentiality, a state-of-the-art method from computer science.
### Why Differential Privacy?
Traditional suppression (hiding small cells) has limitations:
- Differencing attacks can reveal suppressed values
- External data can be combined to re-identify individuals
- Universities have access to their own transcript data
Differential privacy provides **mathematically provable** privacy guarantees regardless of external information.
### How It Works
**For earnings tabulations:**
1. Categorize individual earnings into predefined histogram bins
2. Add noise (geometric mechanism) to each bin count
3. Construct empirical cumulative distribution function (CDF)
4. Calculate percentiles from the protected CDF
5. Sum bin counts for protected cell counts
**For flows tabulations:**
- Apply geometric noise mechanism directly to counts
### Suppression Rules
Earnings cells with protected count < 30 are suppressed due to low quality (flows counts are instead noise-infused and post-processed):
- Status flag = 5 indicates suppression
- Suppression is for data quality, not privacy (differential privacy already protects)
### Privacy-Accuracy Tradeoff
- More noise = more privacy, less accuracy
- Small cells have higher relative noise
- Aggregating across institutions/cohorts improves accuracy
## Historical Context
### Timeline
| Year | Milestone |
|------|-----------|
| 2018 | First PSEO data released (March) |
| R2022Q4 | Partners in 24 states (Census pseo_all_partners.txt for that release) |
| V4.13.0 (2025Q4) | 952 institutions; partners in 32 states plus D.C. |
| V4.14.1 (2026Q2) | 1,117 institution rows; partners in 37 states plus D.C. |
PSEO coverage changes by release. The V4.14.1 count was verified on 2026-10-19 from the Census latest-release institution, partner, and version files (V4.13.0, now the prior release, was first counted on 2026-07-21); those files should be rechecked before reporting current coverage. No national graduate-coverage percentage is asserted here.
### PSEO Coalition
The **PSEO Coalition** "collaborates with LEHD at the U.S. Census to link postsecondary education data with national unemployment insurance wage data" (PSEO Coalition methodology page). Its members are state and institutional representatives who are, or plan to be, LEHD partners; they meet monthly, agree on priorities and requests for LEHD, share data-literacy practices, and meet Census staff quarterly (PSEO Coalition Toolkit, Chapter 1, January 2024). Data-sharing agreements are made with the Census Bureau (PSEO Technical Documentation).
## Technical References
- [PSEO Technical Documentation (PDF)](https://lehd.ces.census.gov/doc/PSEOTechnicalDocumentation.pdf)
- [PSEO Protection System Appendix (PDF)](https://lehd.ces.census.gov/doc/pseo_appendix_tech_doc.pdf)
- [LEHD Public Use Schema](https://lehd.ces.census.gov/data/schema/latest/lehd_public_use_schema.html)
- Abowd, J. M., Stephens, B. E., Vilhuber, L., Andersson, F., McKinney, K. L., Roemer, M., & Woodcock, S. (2009). The LEHD Infrastructure Files and the Creation of the Quarterly Workforce Indicators. In *Producer Dynamics: New Evidence from Micro Data* (NBER Chapters, pp. 149–230). National Bureau of Economic Research.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |