Methodology

The almanac is a dated annual reference: each edition is built once from that year's releases, stamped, and later archived at its own address rather than overwritten. And every page is a distribution, not a number — pay is published at the 10th, 25th, 50th, 75th and 90th percentile wherever the source releases them.

Three rules govern every figure. Published figures appear exactly as released — never interpolated, modelled or blended; the draggable marker interpolates between stops for its readout only, and survey top-codes are shown as top-codes. Derived figures are marked and rounded — price-adjusted and inflation-adjusted values round to the nearest hundred so they cannot be mistaken for release values. Suppressed data stays suppressed — where a statistical office withholds an estimate, we never fill the gap; the module, or the whole page, is omitted instead.

US occupation pages require the full five-percentile set and estimated employment of at least 100 in that occupation and metro — 9,974 pages across 214 occupations and 359 metro areas clear it, and every metro a comparison table references gets its own page only when it clears the same gate. Employer figures require three filings to be stored, five to be shown, and ten before a company gets a page (5,424 do). Career ladders require three seniority levels of five filings each with strictly rising medians — omitted, not smoothed, otherwise. UK pages (35) exist nationally only, because ONS suppresses distributions at occupation × region level; Swedish pages (361) carry the full national spread with sector splits.

Buying power. Adjusted figures divide the nominal wage by the metro's Regional Price Parity (US average = 100): what would this salary buy at national-average prices? Real trends. The five-year figure restates each year's median in current dollars using CPI-U annual averages, so nominal growth and real growth are shown side by side. Seniority. Job titles on visa filings are classified into entry, mid, senior and lead by explicit markers; industry-ambiguous titles such as "staff" are excluded rather than guessed. Employer pay. Certified full-time H-1B filings are annualised, mapped from worksite ZIP to metro, and aggregated under normalised employer names — offered base salary, excluding equity and bonus, skewed toward larger employers: read every employer figure as a floor, not an average. Employer pages describe a company's visa filings, not its whole workforce.

Taxonomy. All of this joins on one internal occupation taxonomy — US SOC 2018, UK SOC 2020, Swedish SSYK 2012 and free-text filing titles map into the same occupations, maintained as versioned, tested data.

Figures are shown as each country quotes pay — US annual dollars, UK gross annual pounds (full-time employee jobs), Swedish monthly kronor (full-time equivalents) — and are never converted across currencies. Comparisons only set side by side what the data honestly supports: the same job in two US metros; occupation against occupation nationally in the UK; private against public sector in Sweden. Contains public sector information licensed under the Open Government Licence v3.0 (UK); Swedish data from Statistics Sweden open data.