Methodology

I. Purpose and Scope

The project links school-district pension contributions to school-district spending for almost every operating public school system in the United States (2015-2023). The goals are:

• To measure how much of the K-12 budget is absorbed by employer pension contributions each year;
• To compare this burden across districts, states, and over time; and
• To test how required contributions would change under alternative discount-rate assumptions.

The analysis covers 95 distinct state and local defined-benefit retirement systems serving K-12 teachers and other school employees (state plans, statewide local-government plans, and large city systems).

II. Data Sources

Primary Sources

• NCES School District Finance Data (F-33, 2015-2023). Provides district-level revenues and expenditures by function (instruction, support services, operations, etc.) and by fund type. These data are the basis for all spending denominators and the weighting used in national aggregates.
• GASB 67/68 Pension System Reports. State and local pension plans' annual financial reports and employer allocation schedules, including: Schedule of Employer Allocations; Schedule of Pension Amounts by Employer; Notes on special-funding situations and state on-behalf contributions.
• District-Level ACFRs / Audit Reports / Budgets. For many states (e.g., Maryland, Pennsylvania, South Carolina, West Virginia, etc.) district CAFRs and state-published budget AFRs were needed to identify employer vs. state contributions when GASB tables or crosswalks were incomplete or withheld.
• Employer Crosswalks and Payroll Data. In several states, plan administrators provided confidential employer-code crosswalks or separate payroll files to align GASB codes with NCES LEAIDs and to allocate state contributions.

III. Coverage

We matched nearly all operating public school districts in the U.S., producing pension spending ratios for roughly 12,800 districts across 50 states and the District of Columbia. Most districts have a complete nine-year panel (2015-2023). Gaps are limited to a small number of districts with missing financials or pension disclosures (documented in each state tab). Apparent "holes" on the map usually reflect non-operating or tuitioning districts (entities with little or no operating spending in NCES) rather than unmatched pension data. The primary partial-coverage exception is Pennsylvania, where lack of state cooperation and limited district reporting prevented us from constructing a complete panel despite extensive imputation work.

IV. Defining Pension Contributions (Numerator)

For each retirement system, "pension contributions" are defined as cash employer contributions made on behalf of school employees in a given fiscal year, including:

• Regular (normal-cost) contributions; and
• Amortization payments toward unfunded liabilities, when reported separately.

Depending on the state and plan, these contributions can be: District employer contributions (paid directly by the LEA); State non-employer contributions on behalf of districts; or A combination, including unusual state "special funding" payments or side-account contributions.

Where GASB 68 employer schedules cleanly separate employer and non-employer amounts, we use those amounts directly. Where they do not (for example, when state contributions are reported only at the plan level), we allocate state totals across districts using plan-provided allocation factors or covered-payroll shares, as detailed in each state write-up.

V. Defining Associated Education Expenditures (Denominator)

We define "associated education expenditures" as the subset of district spending that corresponds to employee categories covered by the pension systems. In most states, this includes teachers, instructional staff, pupil services, school and central administration, operations, transportation, and other support personnel.

To capture these costs consistently, we sum the following NCES F-33 functional codes:

• TCURINST - Current instruction expenditures
• E17 - Pupil support services
• E07 - Instructional staff support
• E08 - General administration
• E09 - School administration
• E11 - Operations and maintenance of plant
• V40 - Student transportation
• V45 - Enterprise operations / food services where applicable
• V90 - Other support services

NCES technical documentation notes that state payments on behalf of LEAs for employee benefits (including pensions and OPEB) are embedded in these current-expenditure lines rather than reported as a separate item; the Census Bureau redistributes state "on-behalf" amounts across the appropriate functions when they are provided. This is why we use this broader nine-category denominator rather than, say, instruction alone: it aligns with where NCES actually records both district and state-paid pension costs.

In a small number of states (Connecticut, Massachusetts, Tennessee, and Washington, D.C.), the pension system covers only certificated instructional staff and administrators rather than all K-12 employees. For these states, the denominator is restricted to the corresponding NCES categories (typically TCURINST + E07 + E08 + E09 ± E17) to maintain numerator-denominator alignment. Specific category lists are documented in the per-state methodology pages.

VI. Pension Ratio and Weighting

For each district and year, we compute:

National and state-group weighted averages use associated education expenditures as the weight. This choice reflects the budget reality we care about: the share of the combined district + state operating education budget that is absorbed by pensions. We also report simple unweighted averages so that users can see the median district experience independent of size.

VII. Treatment of State Contributions and NCES "On-Behalf" Payments

Two layers of state contributions exist in the data:

Plan-Side State Contributions

These are explicit transfers from the state to the pension plan on behalf of districts. In our dataset, these appear in the numerator as "state (non-employer) contributions," allocated to districts using either explicit GASB special-funding schedules or proportional allocation factors (e.g., California TRS, North Dakota TFFR, Texas TRS, etc.).

NCES "State Payments on Behalf of LEAs" in F-33

Some states separately report state-paid benefits (including pensions) for school employees to the Census Bureau. NCES then redistributes these "state payments on behalf of the LEA" across the appropriate expenditure lines (instruction, support services, etc.), so they are already baked into the district-level F-33 expenditures used in our denominators. In other states, NCES imputes these on-behalf amounts from state totals or related data when LEA-level reporting is incomplete. Examples of states that directly report on-behalf payments include Illinois, Kansas, Vermont, and several others where Appendix C notes that "the state reported State Payments on Behalf of the Local Education Agency; the Census Bureau redistributed these data to the corresponding expenditure items." For other states, Appendix C flags that state-payment-on-behalf expenditures were imputed rather than directly reported, or that such payments are negligible.

Because NCES incorporates these on-behalf amounts directly into the functional expenditure lines that we use, our denominator already includes both district and state-paid benefits where NCES can identify them. Our plan-side work then ensures that the numerator (employer contributions) reflects the same combined obligation. We do not attempt to reverse-engineer NCES's internal on-behalf imputations; instead, we rely on NCES's redistribution to keep F-33 totals consistent and use plan-side data (GASB 67/68, CAFRs) to allocate state pension contributions across districts as accurately as possible.

VIII. Handling Missing or Inconsistent Data

Financial Gaps (NCES). When a district has pension contributions but is missing one NCES spending year, we typically impute the missing year using the average of surrounding years, or by carrying forward the closest year when only a year is missing. These cases are documented in the relevant state methodology and are concentrated in small rural districts.

Pension Gaps. When a district lacked pension data but had complete financials, we sometimes imputed contributions using: Ratios of district vs. state contributions in years where both exist (e.g., Pennsylvania); District-specific growth rates inferred from nearby years (e.g., Pennsylvania); or Plan allocation factors combined with covered-payroll estimates (e.g., North Carolina, Pennsylvania, West Virginia, and Illinois).

Consolidations and Name Changes. Where districts merged or where pension systems report "unified" entities that do not match NCES boundaries (e.g., Montana, Vermont, Mississippi, South Carolina), we built explicit consolidation maps, aggregated pre-consolidation NCES districts to the unified entity, and applied a consistent name/LEAID across the full study period so that trends are meaningful.

In all cases, imputations are transparent, limited in number, and generally concentrated in very small districts, so they have little impact on state or national aggregates but materially improve visual and analytical coverage.

Systematic Validation. Beyond these documented adjustments, we conducted systematic validation across all 50 states including cross-source reconciliation against GASB plan-level totals, longitudinal consistency checks to identify and investigate unusual year-over-year patterns, and enrollment plausibility reviews to flag districts where composition shifts materially affected per-pupil ratios. Where corrections were made or observations excluded, the rationale is documented in the relevant state methodology page.

IX. Sensitivity Analysis and Funding Metrics

Beyond historical contribution shares, the site provides:

• Discount-Rate Sensitivity Analysis based on GASB 67 liability and duration data, recalculating required contributions under lower assumed returns (-1, -1.5, -2 percentage points).
• Market-Value Liability (MVL) Benchmark using duration-matched U.S. Treasury yields to re-measure liabilities under risk-free discounting and compute the contribution share required to amortize those liabilities over 20 years.
• Funding Ratios (plan assets / accrued liabilities) using plan-reported GASB 67 data.

These calculations are done at the plan level, and they are then linked back to states and districts by applying the necessary increases required under different scenarios to the actual contribution district total. They assume a 20-year amortization of unfunded liabilities. In the MVL benchmark, liabilities are first re-valued using Treasury yields and then amortized over the same horizon, allowing direct comparison between actuarial and market-value stress scenarios.

X. Alignment and Limitations

Finally, we flag several structural limitations:

• Timing and Classification Differences. Pension systems and NCES do not always align perfectly in fiscal year timing, employer boundaries, or benefit classifications. Some plans smooth investment gains and losses (affecting reported contributions and liabilities) while F-33 records cash expenditures only.
• NCES On-Behalf Imputations. In states where NCES imputes on-behalf payments (rather than receiving full LEA detail), functional expenditures may embed approximated state-paid benefits. Our plan-side work uses actual contribution data wherever available, but some residual mismatch is unavoidable.
• Heterogeneous Accounting Practices. States differ in how they classify certain costs (e.g., food service, enterprise operations, central services). We use a common set of functions designed to capture the vast bulk of payroll-related expenditures, but there may still be minor cross-state comparability issues.
• Incomplete Coverage in a Few States. Pennsylvania is the main case where we could not obtain a complete, reliable panel of district-level contributions despite extensive efforts. A handful of small districts in other states also lack full data and are documented individually.

Despite these caveats, the combination of plan-side contributions, NCES functional spending, and careful crosswalks produces what we believe is the most comprehensive and methodologically transparent picture to date of how K-12 pension costs interact with district budgets across the United States.

XI. Contact & Acknowledgments

For questions, corrections, or data-related inquiries about this project, please use the form below. Given the size of the project, we recognize there may be errors across different states or districts. We welcome feedback from researchers, state and local officials, pension administrators, and members of the public who are interested in improving the accuracy and clarity of K-12 pension finance data.

Special Thanks

This project would not have been possible without the support, guidance, and expertise of the State and Local Governance Initiative at the Hoover Institution, whose team provided invaluable feedback and assistance throughout the construction of the dataset and website.