← back

How we worked this out

Short version: the spending data is real and checkable. The tax figure is an estimate. Here is exactly where the line falls.

The spending side

Every spending number on this site is generated by a script, tools/build_data.py, straight from the sources below. Nothing is typed in by hand, and the script refuses to write anything if its checks fail. You can read it and see precisely which line of which document became which number on the page.

Three judgement calls

Real budgets don't fit neatly into a picture. Three decisions had to be made, and you should know about all three:

Two federal categories were negative. In FY2025, Commerce and Housing Credit came to −$28.1bn and Undistributed Offsetting Receipts to −$150.3bn. You can't draw a negative slice of a positive pie, so both are left out and the percentages are taken over the $7.189tn of positive spending rather than the $7.011tn net total.
City utilities are excluded. Seattle's General Fund is $2.0bn of a $7.3bn operating budget. The rest is water and power, paid for by your bills, not your taxes.
"Everything else" always sorts last, however big it gets. It's a leftover, and putting a leftover at the top would look like a finding when it's just an artefact of grouping.

Where a source genuinely has no deeper breakdown — Social Security and Medicare are each a single line in the federal accounts — the category simply doesn't open up. Inventing a plausible-looking split would be the one thing this site can't do.

The tax side, and how far to trust it

What's modelled: 2026 federal income tax brackets and the standard deduction, FICA up to the $184,500 wage base, state income tax, city income tax where one exists, sales tax, and property tax.

What isn't: itemised deductions, tax credits, dependants, capital gains and self-employment. If any of those apply to you, your real bill differs.

The two softest numbers. Sales tax assumes a share of your income goes on taxable goods, falling as income rises — that's an assumption, not a measurement. Property tax is built from a typical local home value or rent, so if your housing is unusual, that line will be off.

Housing is capped against income. Rent is assumed to be the lower of the local median and 40% of what you earn; an owned home, the lower of the local median and twelve times what you earn. Without that cap the model charged everyone the median rent regardless of income, and reported a 315% effective tax rate on an income of $1,000. The cap stops mattering once you earn enough that the local median rent is under 40% of your income — around $43,500 in Houston, $63,000 in Seattle, $96,000 in Boston and $105,000 in New York.

One known bias: the median rents here are closer to market asking rents than to the median rent people actually pay, which is lower in cities with long tenancies and rent stabilisation. If anything, the property tax line for renters is on the high side.

The point of this site is proportion, not precision. It will tell you reliably that debt interest costs you more than your fire service. It won't tell you your exact tax bill, and it doesn't pretend to.

Plain English

Budget documents are written for accountants. "Income Security & Safety Net" is shown here as "Help for people who need it"; "Consumer and occupational health and safety" becomes "Food and workplace safety". Only the wording changes — never the number, and the official budget label is always there on hover so anything can be traced back.