The 2026 Nonprofit Compensation Report: What the Newest IRS Filings Show — and How Far Behind the Market They Already Are

Finding 1 — The 990 lag, in dollars

Every 990-based compensation source shares a structural problem: by the time a filing is public, the compensation in it is years old. This report prices that lag. Each organization's chief executive compensation from its newest filing was projected to 2026 using federal wage-growth data (BLS Employment Cost Index) blended with observed 990 pay trends for the organization's own role, sector, and budget size. The gap between filed and projected, by filing vintage:

The curve is the finding: the gap shrinks monotonically as filings freshen, exactly as the lag mechanism predicts — and it never reaches zero. Across the full cohort (36,221 organizations, as of the July 2026 corpus), the median chief executive figure in an organization's newest 990 runs 6.5% — $14,523 — below the same cohort projected to 2026 ($222,852 filed vs $237,375 projected). Because Schedule J reports calendar-year compensation inside a fiscal-year filing, the true lag is up to a year larger than stated; these figures are conservative.

What it means for boards: a compensation decision anchored to raw 990 comparables is anchored to a market that no longer exists. The correction is knowable and quantifiable — this is it.

FY2020 26.5% (292 orgs)

FY2021 22.3% (426 orgs)

FY2022 16.0% (680 orgs)

FY2023 11.7% (4,699 orgs)

FY2024 7.8% (12,923 orgs)

FY2025 3.8% (16,872 orgs)

Percentage the filed chief executive figure runs behind the same cohort projected to 2026

Vintage rows publish at n ≥ 200; the degenerate FY2026 vintage row (n = 20) is suppressed.

Finding 2 — What size does to the corner-office paycheck

Median chief executive total compensation by budget size, from the newest filing of 99,809 organizations (Form 990 Part VII and Schedule J combined — see methodology for why the combined basis matters):

The curve is smooth, steep, and monotonic: from the median $54K at the smallest organizations (a band that includes many part-time executives) to $1.32M at $500M+ systems — a 24-fold spread that no single "nonprofit CEO pay" number can honestly summarize.

<$500K $54,154 (29,754 orgs)

$500K–$1M $86,416 (14,892 orgs)

$1M–$2.5M $114,019 (18,651 orgs)

$2.5M–$5M $149,526 (11,711 orgs)

$5M–$10M $185,013 (8,855 orgs)

$10M–$25M $234,062 (7,910 orgs)

$25M–$50M $299,859 (3,674 orgs)

$50M–$100M $402,421 (2,095 orgs)

$100M–$500M $577,238 (1,760 orgs)

$500M+ $1,316,042 (499 orgs)

Median chief executive total compensation, newest filings, combined Part VII + Schedule J basis

Medians from each organization's newest filing; the <$500K band includes many part-time executives.

Finding 3 — CEO pay tracked inflation almost exactly — but not for everyone

Holding organizations constant (same organization, same disclosure source, fixed endpoints), median chief executive pay rose 21.8% from FY2019 to FY2024 — against 21.4% compound growth in the Employment Cost Index over the same years. In aggregate, nonprofit CEO pay tracked the broad labor market almost exactly.

Underneath, a U-shaped divergence by size: 24.4% cumulative growth at the smallest organizations (<$500K), a trough of 19.1% in the middle ($2.5M–$5M), and 27.7% at $50M–$100M organizations. Year-over-year growth accelerated from roughly 3% per year in the FY2018–21 pairs to roughly 5% in FY2022–24, and the newly complete FY2024→FY2025 pair reads 3.2%–6.1% across size bands — the post-inflation-spike deceleration has not fully arrived at the top.

All organizations +21.8%

Employment Cost Index +21.4% (benchmark)

<$500K +24.4%

$2.5M–$5M +19.1% (trough)

$50M–$100M +27.7%

Cumulative growth in median chief executive pay, same-organization endpoints

Holding organizations constant: same organization, same disclosure source, fixed FY2019/FY2024 endpoints.

Finding 4 — State rankings, adjusted for what states actually contain

Raw state medians mostly measure each state's mix of small and large nonprofits. This report standardizes: each state's pay is scored at the national budget-size mix, then ranked. All 50 states + DC qualify under the publication gates (state n≥100; every cell n≥10; ≥80% weight coverage).

Presentation (ruled): exact ranks for the top 5 and bottom 3 only — the set whose order held across the latest corpus refresh; quartile bands for every other state (rank order there moves with corpus refreshes and is not durable enough to print, in table or prose); the size-adjusted index AND the unadjusted median printed for every state, because the index is a comparison score, not "typical pay."

Top 5 (size-adjusted): DC $195,275 · VA $166,706 · NY $152,642 · NJ $152,397 · MA $150,766. Bottom 3: MT $109,500 · WV $94,898 · WY $86,129.

State	Rank / quartile	Size-adjusted index	Unadjusted median	Organizations (n)
DC	1	$195,275	$209,750	2,471
VA	2	$166,706	$122,148	2,829
NY	3	$152,642	$134,424	7,178
NJ	4	$152,397	$128,558	1,769
MA	5	$150,766	$135,822	3,122
CA	Q1	$148,239	$127,010	11,584
MD	Q1	$146,568	$117,454	2,141
CT	Q1	$144,308	$126,894	1,296
IL	Q1	$142,517	$113,000	3,819
GA	Q1	$141,379	$96,000	2,241
FL	Q1	$138,628	$109,717	3,687
TX	Q1	$136,840	$96,787	5,516
CO	Q1	$136,411	$102,153	2,835
MN	Q2	$136,291	$105,262	2,556
WA	Q2	$135,748	$103,222	3,036
RI	Q2	$135,386	$115,938	482
OH	Q2	$134,309	$97,804	3,376
MO	Q2	$134,114	$101,082	1,833
AZ	Q2	$131,789	$103,802	1,616
NV	Q2	$131,309	$107,068	482
MI	Q2	$130,555	$106,310	2,430
PA	Q2	$130,108	$109,144	4,218
WI	Q2	$129,423	$98,754	2,164
NE	Q2	$129,046	$99,431	713
NC	Q2	$127,678	$88,000	2,800
SC	Q2	$126,949	$89,078	1,126
OR	Q3	$125,468	$95,766	2,210
IN	Q3	$123,742	$95,676	2,089
TN	Q3	$123,218	$90,000	1,964
KS	Q3	$121,434	$92,165	1,038
OK	Q3	$121,178	$84,100	1,017
AL	Q3	$120,551	$85,638	1,062
ND	Q3	$118,949	$104,759	414
NH	Q3	$118,663	$93,758	629
IA	Q3	$117,941	$95,000	1,055
HI	Q3	$117,281	$108,880	511
KY	Q3	$117,206	$94,974	1,062
LA	Q3	$115,251	$100,002	1,024
AK	Q3	$113,778	$103,769	493
UT	Q4	$112,929	$110,191	620
ME	Q4	$112,890	$91,187	828
NM	Q4	$112,654	$86,603	725
AR	Q4	$111,578	$85,970	652
ID	Q4	$110,847	$85,800	505
SD	Q4	$110,715	$91,274	409
DE	Q4	$110,508	$106,000	363
MS	Q4	$110,250	$87,166	537
VT	Q4	$109,574	$84,348	622
MT	49	$109,500	$82,778	853
WV	50	$94,898	$80,580	589
WY	51	$86,129	$77,291	337

Rankings are nominal — no cost-of-living adjustment. The size-adjusted index is a comparison score, not "typical pay" — the unadjusted median is the "typical" number. Exact ranks are printed for the top 5 and bottom 3 only — the set whose order held across the latest corpus refresh; every other state displays its quartile (Q1–Q4 by rank) because rank order there moves with corpus refreshes.

Finding 5 — The expanded §4960 excise tax (published July 2026)

CauseComp's standalone analysis — the first data-based estimate of the reach of the expanded Section 4960 excise tax — found the 2025 law change reaches an estimated 734 to 1,302 additional employees earning over $1 million (a 25–28% increase over the old top-five rule) while adding essentially zero newly exposed organizations, concentrated at hospitals, universities, and $500M+ budgets. Full analysis, methodology, and caveats: causecomp.org/research/section-4960-expanded-reach. (That analysis is pinned to its own recorded data vintage and is not restated here.)

Methodology (summary — full methodology publishes with the report)

• Corpus: 2,083,580 compensation records from IRS Form 990 e-filed disclosures (Part VII and Schedule J) across 163K+ organizations, certified vintage v2-2026pub (July 2026), including the first wave of May 2026 filings; newest filing per organization; person-level dedupe; "former"-titled rows excluded.
• Combined disclosure basis for descriptive findings: Schedule J has a $150K disclosure threshold, so Schedule-J-only medians overstate pay at small budget bands by as much as 59%. Findings 2–4 therefore use the combined Part VII + Schedule J basis; the Schedule-J-only curve is retained internally as a bias exhibit, not published as "CEO pay."
• Cohort-constant growth: every growth statistic compares the same organizations to themselves (consecutive-filing pairs; fixed 2019/2024 endpoints). Growth prices the position, not incumbent raises.
• Model use is confined to Finding 1 (the 2026 projection: BLS ECI compounding blended with observed 990 trend for the organization's role/sector/size cell, conservative — current asking-salary data is not applied). Findings 2–4 are raw-panel descriptive statistics.
• Minimum-n gates on every published cell (band cells n≥10 minimum, vintage rows n≥200, state gates as above); cells below gate are suppressed, not estimated. Suppressions this edition: the degenerate FY2026 vintage row (n=20) and the two early-wave FY2025→26 trend cells.
• Vintage discipline: the overall lag figure is stated as of the July 2026 corpus and will move with each refresh (by design — it measures the mix of filing vintages); the per-vintage curve is stable across refreshes and is the durable exhibit.
• Every figure derives from committed, deterministic analysis scripts. Aggregates only — this report names no organizations and no individuals.

Caveats log

1. (k) F1: cohort is Schedule-J basis; small-band selection bias does not materially bite a within-cohort ratio (the same orgs appear in numerator and denominator).
2. (l) F1: fiscal-year labeling: Schedule J reports calendar-year comp for the CY ending within the fiscal year, so a "FY2024" filing's comp is typically CY2023 dollars — the true lag is up to a year LARGER than stated; the gap is conservative.
3. (m) F1: the 2026 endpoint uses the ECI table's 2026 estimate (3.2%) until BLS actuals land; a ±0.5pp error moves the headline by ~0.5pp.
4. (n) F1: trend cells inherit the engine's org_type-default and thin-cell caveats (see the model-path and methodology notes on this page).
5. (o) F3/F5: Part VII and Schedule J comp totals are near- but not perfectly-identical definitions; the per-person Schedule-J-preferred pick and the same-source rule for growth pairs prevent definition mixing inside any single statistic.
6. (p) F3/F5: no hours filter — the <$500K and $500K–$1M cells include part-time EDs (visible in P25 $33.6K); a "typical full-time ED" framing needs that caveat or an hours-gated variant.
7. (q) F3/F5: revenue_band is the engine's revenue_for_model banding.
8. (r) F3/F5: F5 composite-index interpretation + nominal (no COL adjustment), (see the model-path and methodology notes on this page).
9. (s) F3/F5: territories excluded by the state gate, listed in f5_excluded_states.csv.

Vintage-sensitivity flags (delta note, July 2026 corpus refresh)

• F1 overall lag: VINTAGE-SENSITIVE by construction — it is a mix-weighted average over filing vintages; every corpus refresh will move it. Date-stamp it.
• F1 by-band gaps: VINTAGE-SENSITIVE (inherits each band's FY mix). Date-stamp or drop.
• F5 state rankings: levels robust, mid-table RANKS churn with cell composition — 36/51 states changed rank on a corpus refresh with adjusted-index moves of −5.2%..+8.2%; top-5 and bottom-2 order held. Mid-table single-rank precision is not defensible across vintages.
• F5 WY (near-gate): weight coverage 92.1% → 84.0% (6→5 usable cells) against the 80% floor — one more thin cell and WY is excluded. The ranking's tail is vintage-sensitive.

About CauseComp

CauseComp publishes compensation benchmarks for U.S. nonprofit organizations, built from IRS Form 990 filings and federal wage surveys — 2M+ compensation records. This report is free to cite with attribution and a link. The interactive version of every exhibit — benchmarks by role, budget size, sector, and state, projected to the current market — is the CauseComp product at causecomp.org.

Frequently asked questions