Methodology
Invokely republishes official statistics without changing them, except where an indicator is explicitly marked as an Invokely calculation.
Ingestion
One connector per source checks for new publications on a schedule, downloads the published series, and stores each value with its source URL, retrieval date and version. A revised value keeps its previous versions.
Harmonization
Places are matched by official codes (INE, Idescat, NUTS, ISO), never by name alone. Periods are normalized to years, half years, quarters and months. Each canonical indicator has one definition, and only source series that measure that definition are mapped to it.
Which source is served
When several sources publish the same indicator for a place, Spanish national sources come first, then regional sources, then Eurostat. A series always comes from one source. Comparisons look for a source that covers every place.
Quality checks
Values outside a plausible range are rejected. Large jumps between periods, duplicate periods, unknown place codes and format changes open an issue for review. Failing sources are retried with backoff and flagged.
Derived indicators
Registered unemployment rate
Registered unemployment published by SEPE for the month, divided by the resident population aged 16 to 64 on 1 January of the same year (INE annual census), times 100. When the population for the current year is not yet published, the latest available year is used and the value is marked estimated. Province, region and Spain totals are sums of municipal figures; municipalities published as fewer than 5 are left out of the sums and those totals are marked estimated.
This rate is comparable across all Spanish municipalities and levels. It differs from the survey-based unemployment rate (EPA) and from the Catalan registered unemployment rate, which divides by unemployed plus Social Security affiliates.
Other calculated indicators
Status flags
provisional: the publisher may revise the figure. estimated: estimated by the publisher or, for derived indicators, computed with a denominator from an earlier year. break: the series has a methodological break at this point.