Netherlands in Numbers

What changed

Statistics get restated, sources get replaced, and we make mistakes. This page is the quarterly record of all three — including our own corrections, with what was wrong and why.

each entry is auto-drafted from the ingestion database, then edited by a person before publication — the pipeline cannot see its own blind spots

2026 Q3 — launch quarter

published 7 Jul 2026

The first entry covers the launch itself, so everything is new. Later entries will be quarterly and mostly report what the sources restated and what we corrected.

New

  • The site launched: 44 headline and domain indicators plus context series, ingested nightly from CBS, Eurostat, OECD and RIVM/CLO. Every observation is stored append-only with its retrieval timestamp, so future source revisions stay visible forever.
  • Real GDP per capita (X8, Eurostat) was added as a context series for the pairings page.

Corrections — what we got wrong and fixed

  • Median household income (H1) was presented as inflation-adjusted. The CBS table is nominal — it has no price-basis dimension, and the giveaway was +60% growth 2011–2024 against ~36% cumulative inflation. The indicator was renamed, and a CPI-deflated line (our calculation, published formula) was added: +15.1% real over that span, which matches CBS's own purchasing-power statistics.
  • The manufacturing headline (A4) was showing value added in euros on a page titled “jobs and hours”. Re-pinned to hours worked.
  • Labour productivity growth (A6) was titled “(5y avg)” but is the annual rate — the five-year mean would have the opposite sign in 2024. Renamed to “(annual)” rather than shipping a smoothed series nobody computed.
  • The “feeling unsafe” headline (K3) was the general question (37.2% in 2025), not the neighbourhood question (16.8%). The neighbourhood series is now the headline; the general one stays on the chart as a labelled companion.
  • The house price index (G1) was labelled “2015 = 100”, but the CBS/Kadaster table is based 2020 = 100 (2015 reads 69.5 in the source). The real (CPI-deflated) companion inherited the error and its values matched no base year at all. Both were corrected on 9 July: the unit now says 2020 = 100, and the real formula (v2) is anchored so it reads exactly 100 in 2020. The shape of both lines was always right; the base label was not.

Derived series — published formulas

  • Income left after housing costs (G5): 100 minus the median housing-cost share. An after-housing-costs amount in euros is deliberately not published — median income and median housing costs come from different distributions, and medians do not subtract.
  • Real median household income (H1) and real house prices (G1): the official nominal series divided by the CBS consumer price index (2015 = 100).

Peer comparisons

  • Housing costs (G5) got its first peer chart: the Eurostat share of housing costs in disposable income, replacing the planned 40%-overburden rate, which is a different statistic. The Eurostat series is the weighted mean while the Dutch series is the median — the charts say so and are never mixed.

Reading features

  • Every indicator page now opens with a plain-language block: the latest value said as a sentence, what the statistic counts, and why it is tracked — verified against the sources before shipping (that pass is what caught the A4, A6 and K3 corrections above).
  • “Since you were born”: enter a birth year once and every card and page re-anchors its comparison to it; links carry the year.
  • Pairings: four cases where two official numbers pull in different directions, shown side by side without a verdict.
  • Rank charts: where a harmonized peer chart exists, a bump chart below it shows the Netherlands' position among the same countries year by year — always ordered by value, highest first, never labelled “best”.

Corrections are content here, not embarrassments: a dashboard that never admits an error is one you cannot check.