Methodology

How CauseComp builds its benchmarks

CauseComp runs two independent, data-driven engines from authoritative public sources — one for executive pay, one for the broader workforce. Both report a full percentile range (25th / 50th / 75th), so you see the market spread, not just a single number.

Executive Compensation 990

Data source

Officer, director, trustee, and key-employee pay reported on IRS Form 990 (Schedule J and Part VII, Section A) and Form 990-EZ — the public filings U.S. nonprofits submit annually, sourced directly from the IRS's bulk e-file releases. This is real, disclosed compensation, not survey self-report.

Drawn from 2M+ cleaned compensation records across the most recent IRS e-file releases; 1,000,000+ meet the modeling quality standards that power the benchmarks — curation, not raw volume, is what makes an estimate worth putting in front of a board.

Model

• A gradient-boosted quantile-regression model estimates the shape of the percentile curve (10th, 25th, 50th, 75th, 90th) for a role and organization profile.
• Subscribers also see the compensation broken into components — base, bonus/incentive, other reportable pay, retirement & deferred compensation, and nontaxable benefits — alongside total. Component detail comes from Schedule J; peers filing only Part VII or 990-EZ report totals and show n/d for components.

Minimum comparables

We hold a standard for how thin a cohort can be before we speak: we'd rather tell you we don't have enough comparables than invent a number. Where no comparable filings back a role, sector, and size, CauseComp declines to serve a benchmark and instead suggests better-covered roles. Where only 1 to 9 comparable filings are available, the benchmark is served but carries a visible low-confidence flag; ten or more is our standard for serving cleanly.

§4958 comparables

For board documentation, CauseComp surfaces the actual comparable organizations behind a benchmark — the specific public 990 filings — supporting the IRS §4958 rebuttable presumption of reasonableness for setting executive pay.

Published medians for individual roles, drawn from the same filings and the same minimum-n rule described here, are on the nonprofit salary benchmark pages.

If you are new to these rules, our plain-English guides to §4958 and board pay governance cover the process this data supports. For the excise tax that sits alongside them, see how the expanded §4960 rules reach further than most boards expect.

Broad-Based & Workforce BLS OEWS · O*NET

Data source

The U.S. Bureau of Labor Statistics Occupational Employment and Wage Statistics (OEWS) program — the official national wage survey — covering 850+ roles mapped to standard SOC occupation codes, using the most recent BLS OEWS release.

Calibration chain

• Starts from the official national median for the occupation, then layers geographic (state and metro-level, ~390 MSAs) and industry (NAICS) calibration, all computed against a consistent national reference base.
• Industry calibration runs in both directions. It is credibility-weighted by measured employment in each industry, and the data decides the direction: sectors whose employers genuinely pay above the all-industry level (for example education, health, and philanthropy) adjust upward, while sectors that pay below it (for example human services and the arts) adjust downward. There is no blanket "nonprofit discount."
• Wages are aged to the current period using the BLS Employment Cost Index (ECI).
• Incentive and benefits loads are derived from BLS Employer Costs for Employee Compensation (ECEC) and O*NET composite scalars, producing total-cash and total-compensation figures.
• A Nonprofit / Private-sector toggle switches the reference population and industry mix. Private-sector figures reflect BLS OEWS market wage data across all ownership types, with private-ownership-specific estimates where BLS publishes them (notably education and hospitals).
• Seniority tiers (entry / mid / senior) are derived from within-state wage distributions — holding geography fixed so the tier premium reflects experience rather than re-importing regional differences — calibrated against published experience-premium evidence and constrained to increase monotonically from entry to senior.

Coverage standard

Where BLS does not publish a reliable wage for an occupation, CauseComp does not benchmark it — the role returns a clear "not benchmarked" explanation rather than a nearest-relative guess. Legacy occupation titles are aliased to their current SOC successors, so a served figure always reflects an occupation BLS actually measures.

Role context

Each role links to an O*NET profile — tasks, knowledge, skills, and work context — so you can confirm you're benchmarking the right job, not just a matching title.

Reading the numbers

• P25 / Median / P75 describe where pay falls across comparable organizations — a market range, not a recommendation.
• Every benchmark shows its data vintage and confidence so you know how current and how well-sampled it is.
• Figures are for informational purposes and don't constitute legal, tax, or compensation advice.