Gold reserves by country data: two numbers, one table

Two different quantities answer to that phrase, and a model that conflates them is wrong in a way no amount of cleaning fixes. Official (above-ground) reserves are the holdings central banks and finance ministries report — tonnes sitting in vaults, the figure the IMF-style reserve statistics tradition measures. In-ground mineral reserves are geological estimates of what remains minable, published by country each year in the USGS Mineral Commodity Summaries. The first counts metal already owned; the second counts metal still buried. They share a unit (tonnes), a dimension (country) and almost nothing else.

Datadory keeps both under one roof so you never have to choose a publisher to choose a concept: World Gold Council GoldHub carries the official-holdings series with country-level breakdowns alongside demand, mine production and prices from 1978; BullionStar maintains central-bank reserve pages for the major national central banks as a secondary view beside its quote grid. Twenty-seven primary datasets sit on the Precious Metals & Minerals shelf, and only a handful carry the official-reserves series at all — which is precisely why it pays to know which number you are buying.

World Gold Council GoldHub is the only record on the shelf that bundles an official-reserves series with everything needed to interpret one. Fourteen named series ship as workbooks carrying up to roughly 48 years of observations apiece: daily-to-annual price averages in multiple currencies, a quarterly supply/demand balance, monthly ETF holdings and flows, mine production by country, and the reserves snapshots keyed by country. The reserves survey itself runs annually since 2018, which makes it young relative to the price history — plan your backtest around that asymmetry rather than discovering it mid-project.

The USGS Mineral Commodity Summaries answers the other question: how much gold is left. It carries Datadory's highest quality score on this shelf — a 10 — and for good reason: roughly ninety commodity chapters with dedicated gold coverage, world mine production and reserves for about 175 countries, released as a 226-page book beside an 8,886-row tidy CSV in which 112 rows are gold statistics. Each observation is one commodity × statistic × country × year cell, already tidy enough to load without reshaping.

The window is the part people trip on. Every edition reports a rolling five-year span — 2021 through 2025 in the current edition — and estimates are revised annually, so a figure is a statement about an edition, not about geology. Pin the edition year on every row and reach for the archived editions when a longer series matters; they extend back toward the 1990s.

Four records cover the reserves question from different angles, and the choice among them decides how far back your series runs:

Read the table as three units of account: tonnes of reported official holdings (GoldHub, BullionStar), tonnes of estimated in-ground metal (USGS), and annual production or trade tonnage (BGS). Only the first two belong in a “gold reserves by country” table — and they should never be summed. A national figure of X tonnes above ground says nothing transferable about Y tonnes below it.

The defensible shape is a two-panel table with a shared country key and separate value columns — never a merged column:

1. Panel A, official holdings. Country × tonnes × survey year from GoldHub's reserves series, with the survey year carried on every row.
2. Panel B, in-ground estimate. Country × tonnes × edition year from the Mineral Commodity Summaries reserves section, filtered to the 112 gold rows.
3. Pin vintages. Panel A dates to its annual survey; Panel B to its five-year edition window. Record both on every export.
4. Cross-check jumps. Where a national figure moves sharply between editions, check BullionStar's central-bank pages and the prior USGS edition before treating it as signal.
5. Keep the panels separate downstream. Ratios and gaps between panels are legitimate features; sums are not.

Datadory ships this as assembly work: name the countries and both windows in a sample request and the two panels arrive aligned on one country key, vintage-stamped, with the field dictionary beside them.

Macro and FX desks read shifts in official holdings as a sentiment signal on fiat currencies, using GoldHub's monthly ETF flows and quarterly supply/demand balance to explain price moves a reserves snapshot alone cannot. Policy researchers compare reported official holdings against in-ground estimates to flag countries whose mining endowment and reserve policy diverge. Journalists quote both series because each traces to a named publisher with a stable methodology.

Two more records deepen the picture. For deposit-level drill-down beneath any national total, USGS MRDS contributes 304,632 point site records worldwide, with gold appearing as a commodity in roughly 70,000 of them. For asset-level diligence behind a producing country's figures, Mining Data Online details ownership, stage, reserves, AISC and mine financials for thousands of named projects. The persona fit runs strongest for quantitative analysts — the same shelf appears in workflow order on the page for best data for precious metals minerals teams.

Native grains map onto tiers cleanly: realtime suits the quote grids beside BullionStar's reserve pages, daily fits GoldHub workbooks, annual covers the Mineral Commodity Summaries editions and BGS reference years, and fixed snapshots like MRDS's deposit register load once and stay put. Formats arrive as recorded upstream — tidy CSV for the commodity summaries, XLSX for balance workbooks, JSON for quotes — flattened and joined on request.

Reserves are one thread in a 32-dataset pool. The precious metals minerals data guide walks all of them — price benchmarks, physical-supply statistics and GIS layers included. For the production counterpart to these stock figures, see gold mine production by country; for the flows side, OEC gold trade profile HS 7108; for holdings behavior in funds rather than vaults, gold ETF holdings and flows data. Or start at the precious-metals-minerals data hub and request a sample cut to your countries.