# Apiar Data

> Institutional statistics on the economy, labour, health, prices, trade, demographics,
> and research — sourced from primary statistical agencies (Eurostat, OECD, World Bank,
> IMF, national statistics offices) and published as 12,909
> individually documented series with provenance, methodology notes, and source links.

Free to browse. No account or key needed for anything described here.

## Machine-readable data

- [Series catalog](https://apiardata.com/statistics/index.json): all 12,909
  series in one JSON array — `slug`, `title`, `domain`, `geography`, `unit`,
  `frequency`, `temporal_coverage`, `publisher`, `last_updated`. Start here: it is
  the only complete listing carrying enough metadata to match against, and it turns a
  description like "Dutch long-term unemployment" into a slug in one request.
  It is ~3 MB uncompressed (far less on the wire, it is served compressed), so filter
  it rather than reading it whole.
- [Example series document](https://apiardata.com/statistics/ae-agriculture-value-added-gdp.json): what a catalog
  `slug` resolves to. Fill `https://apiardata.com/statistics/{slug}.json` for the full record, or
  `https://apiardata.com/statistics/{slug}/` for the human page.

The catalog's field names are a subset of the per-series document's, so a filter
written against one keeps working against the other.

Prefer the JSON over the HTML. The HTML table is formatted for humans — it rounds to one decimal,
adds thousands separators, and collapses a signed rounding-to-zero to the word
"unchanged" — and on single-series pages the chart values live in a `data-values`
attribute that tag-stripping extractors drop. The JSON carries the parsed source
values verbatim and typed, with missing observations preserved as `null` rather than
zero.

Each JSON document contains:

- Identity: `slug`, `url`, `title`, `description`, `domain`, `category`, `keywords`
- Provenance: `publisher`, `sources[]` (name + URL of each primary source),
  `measurement_technique`, `license`
- Coverage: `frequency`, `geography`, `temporal_coverage`, `unit`,
  `last_updated` (ISO 8601), `data_as_of`
- Data: `series[]`, `observation_count`, `observations[]` (one object per period),
  and `table` (columns + raw rows)
- Context: `kpis[]` and `qa[]` — the Q&A covers definitional caveats, source
  comparability, and known limitations, and is usually the fastest way to avoid
  misreading a series

Each HTML page also carries schema.org `Dataset` JSON-LD in its head, including a
`distribution` entry pointing at the JSON, plus a `rel="alternate"` link to it.

## Important caveat when comparing series

Series are sourced from different agencies with different definitions, survey
instruments, and population bases. Figures across two pages are frequently not
directly comparable even when the titles look equivalent. `measurement_technique`
and `qa` on each page state the specific caveats — read them before combining series.

## Domains

- [Economy & Macro](https://apiardata.com/statistics/macro): 1,889 series — GDP, central bank policy, credit, exchange rates, and government finance
- [Financial Markets](https://apiardata.com/statistics/markets): 992 series — Equity indices, derivatives markets, and asset prices
- [Prices & Inflation](https://apiardata.com/statistics/prices): 1,085 series — Consumer and producer price indices across major economies
- [Labour & Employment](https://apiardata.com/statistics/labor): 1,176 series — Employment, payrolls, wages, and unit labour costs
- [Trade & Flows](https://apiardata.com/statistics/trade): 1,069 series — Trade volumes, current account flows, and FDI
- [Demographics](https://apiardata.com/statistics/demo): 1,330 series — Population estimates, projections, and demographic breakdowns
- [Health](https://apiardata.com/statistics/health): 4,621 series — Life expectancy, mortality, health expenditure, and health system capacity
- [AI & Research](https://apiardata.com/statistics/papers): 747 series — R&D investment, AI adoption, research output, and frontier capability metrics

## Analysis

- [Comparative analyses](https://apiardata.com/analysis): 12 cross-country pieces
  reading several series together. Prose, not machine-readable — the underlying
  numbers are in the statistics JSON above.

## Search across sources

An MCP server indexes 16.6M datasets, statistical series, and publications from
primary sources — broader than the curated pages above, returning metadata plus
deeplinks to each original source rather than the data itself. Use it to find data
Apiar has not published a page for; use the catalog above for data it has.

- [MCP endpoint](https://search.apiardata.com/mcp): Streamable HTTP transport
- [Integration guide](https://apiardata.com/mcp): tools, auth, and client setup

## Optional

- [Full URL list](https://apiardata.com/sitemap-index.xml): every page as XML. Superseded for most
  purposes by the series catalog above, which covers the same statistics pages with
  the metadata needed to pick between them.