How to Read a Form 990 for Executive Pay — Without Being Misled

Every figure a board needs to understand a peer organization's executive pay is on the public Form 990 — but the form was designed by tax administrators, not compensation analysts, and it punishes casual reading. This guide covers where the pay data actually lives, what each number does and does not include, and the three traps that most often mislead boards comparing themselves against another organization's filing.

Start at Part VII, not the summary page

Part VII, Section A is the compensation table. The organization must list, with pay from the organization and from related organizations shown separately:

• every officer, director and trustee, regardless of pay (unpaid board members appear with zeros);
• every key employee — broadly, a non-officer over $150,000 with organization-wide responsibility or control (the Form 990 instructions set a three-part test);
• the five highest-compensated other employees over $100,000.

"Reportable compensation" in this table is taken from tax forms — for employees, the amount in box 1 or box 5 of Form W-2, whichever is greater; for non-employee directors and contractors, box 6 of Form 1099-MISC or box 1 of Form 1099-NEC. Column (F), "estimated amount of other compensation," adds retirement, deferred and nontaxable benefits above the instructions' reporting thresholds. It is an estimate, and it is where two organizations paying the same salary can look different.

Then go to Schedule J for anyone over $150,000

When the sum of a listed individual's Part VII Section A columns (D), (E) and (F) exceeds $150,000 — disregarding decreases in the actuarial value of defined benefit plans — the organization must also file Schedule J, which breaks total pay into base, bonus and incentive, other reportable, retirement and deferred, and nontaxable benefits. For benchmarking, Schedule J — not Part VII — is the table to compare, because it separates the components a board actually sets. Our Schedule J explainer walks through each column.

The three traps

1. Calendar year versus fiscal year

Part VII and Schedule J report compensation for the calendar year ending with or within the organization's fiscal year — a rule from the Form 990 instructions that exists so W-2 figures can be used directly. A June-2025 fiscal year return therefore carries calendar-2024 pay. Compare two organizations without checking their fiscal years and you can be comparing pay from different years without knowing it.

2. The filing lag

Returns arrive up to a year — often more, with extensions — after the year they describe. Any 990-derived dataset's newest year is structurally incomplete. That is why CauseComp's published figures state the filing-year span they rest on rather than claiming a single current year.

3. Part IX is not Part VII

Part IX (the statement of functional expenses) carries a line for officer and key-employee compensation, and it will rarely match a total built from Part VII: it is accrual-based, covers the fiscal year rather than the calendar year, and allocates differently. Boards comparing a Part IX line against a Part VII figure are comparing two different accounting constructions.

A worked illustration

Illustrative only — Riverbend Youth Services is invented and every number below is synthetic. Suppose Riverbend's fiscal year ends June 30, 2025, and its executive director shows $150,000 of reportable compensation and $30,000 of other compensation in Part VII. What a careful reader knows: that is calendar-2024 pay; the $30,000 is an estimate spanning retirement and benefits; and if Schedule J shows $20,000 of the total as deferred compensation earned this year but payable later, a naive W-2 comparison will misstate it. The number a board should carry into a comparison is the Schedule J column (E) total, on a stated year, against a cohort of similar organizations — not a single figure lifted from a summary page.

Where to check the organization's own process

Part VI, line 15 asks whether the organization's process for setting its CEO's pay involved independent persons, comparability data, and contemporaneous substantiation — the same three elements as the §4958 rebuttable presumption. A "No" on a peer's return tells you something; so does one on your own. See how the rebuttable presumption works.

When the question turns from reading one filing to comparing many — is our own executive's pay defensible? — the work becomes cohort construction, and that is what CauseComp automates: published medians by role and state on the salary benchmark pages, and full budget-, sector- and location-matched analysis in the Executive benchmarking tool.