Methodology

Every member of Congress, senior executive-branch official, and federal judge must file an annual financial disclosure. But these forms do not ask for exact dollar amounts. Instead, each asset and liability is reported as falling within a value bracket, for example “$1,000,001 – $5,000,000.”

Because of this, any net-worth figure (ours, OpenSecrets’, or a newspaper’s) is an estimate derived from ranges. We never present a single number as if it were precise.

Why some figures are negative. Disclosures require officials to report mortgages and loans as liabilities, but they exclude personal residences and other non-income-producing personal assets from the asset schedule. An official whose main asset is their home will show the mortgage as debt with no offsetting asset, so their disclosed net worth can be negative even though they are not actually underwater. We surface this caveat on any profile with a negative estimate rather than implying the person is millions in debt.

For each filing we:

1. Convert every asset and liability bracket to a [low, high] dollar range.
2. Sum the lows and the highs separately to get a total asset range and a total debt range.
3. Compute net worth as a range: net low = assets low − debts high, net high = assets high − debts low.
4. Report a single midpoint estimateas the headline number: the sum of each item’s bracket midpoint. This is the standard approach used across the field.

On charts, the shaded band is the disclosed range and the line is the midpoint. The spread between them is the irreducible uncertainty baked into the source data.

Judicial disclosures (AO-10) use these gross-value codes, which we map as follows:

† The top bracket is open-ended. When a holding falls in it, the high end is a floor (the true value may be higher), and we flag the estimate accordingly.

Each estimate carries a confidence level based on how completely we could parse a filing. When many line items lack a readable value code, the estimate is marked low confidence and likely understates the total.

Donation figures come directly from the Federal Election Commission. Campaign totals (receipts, small-dollar share, PAC share) are exact. The per-donor breakdowns (top PAC donors, employers, and outside spenders) show only the largest 15 per candidate, so a single donor’s total shown across the site is a lower bound, not their complete giving.

“PAC donors” are direct contributionsto the candidate’s committee (FEC transaction types 24K/24Z). Money a super PAC spent for or against a candidate is an independent expenditure, shown separately as outside spending, never as a donation to the candidate.

Two limits: only elected officials (Congress, and presidential/vice-presidential candidates) have FEC records, while appointed officials and Supreme Court justices do not; and contributions of $200 or less are not itemized by the FEC, so small-dollar donors appear only in aggregate.

On each profile we plot the official’s government salary for every year in office, derived from published US federal pay tables by role (rank-and-file member of Congress, President, Vice President, Associate/Chief Justice, Cabinet Level I). Salary figures are exact for known years and approximate where interpolated; they reflect base pay and don’t include leadership differentials.

We also show a hypothetical retirement benchmark: what an account would hold if the person invested 20%of each year’s salary, compounding at an assumed 10%annual return (roughly the S&P 500’s long-run nominal total return), from their first year in office. It is a deliberately simple “disciplined saver” yardstick to compare against actual disclosed net worth, nota claim about anyone’s real savings, and it ignores taxes, spending, outside income, and market variation.

The spending pages start with USAspending’s public Spending Explorer. It groups federal outlays by OMB budget function, then by agency. We refresh those rows monthly because the source updates on a monthly cadence.

We also show corporate tax breaks as a dashed red line. They are not checks the Treasury writes. They are revenue the government chooses not to collect through deductions, credits, and write-offs. The Joint Committee on Taxation publishes those tax-expenditure estimates, and budget analysts often compare them with direct spending because both move public money. We label them as revenue foregone every time.

“Services for immigrants” is deliberately narrow: refugee and entrant assistance, citizenship and immigration services, asylum/citizenship operations, and closely named federal assistance accounts. Immigration enforcement is not counted there. ICE, Customs and Border Protection, detention, border security, and immigration courts get their own line. Policing immigration is not the same thing as serving immigrants.

USAspending rows refresh automatically. OMB history and JCT tax-break estimates are hand-refreshed when those annual tables are published, so the source can be checked before the number moves.

This is a living record built source by source. Digitized disclosure coverage varies by branch (House from ~2008, Senate from ~2012, judicial via CourtListener through ~2021, executive filings are recent and shallow). Where a record has not yet been ingested, we say so plainly rather than implying zero. Corrections are welcome.

On Generational Wealth we price necessities in hours of after-tax work: the nominal price divided by take-home pay. Take-home is the gross wage (federal minimum, or BLS average hourly earnings) minus the Bottom-50% all-in effective tax rate from Saez-Zucman, the same series behind our Tax Files. It is all taxes (income, payroll, sales, property, state), so it runs a little heavier than paycheck withholding, and it is the same rate for both wage measures. A work-year is 2,080 hours. Prices are nominal, from Census/FRED (home, rent), NCES (college), NADA/BEA (cars), and CMS (health care).