USGS Mineral Commodity Summaries - Gold

It is the United States' official one-sitting read on gold supply, delivered here as typed rows instead of a typeset spread. USGS Mineral Commodity Summaries - Gold compresses the federal picture onto two pages: a Salient Statistics - United States table running 2020 through 2024e across mine production, primary and secondary refinery output, imports and exports for consumption, reported consumption, year-end Treasury stocks, the average Engelhard price in dollars per troy ounce, and mine-and-mill employment; a World Mine Production and Reserves table pairing 2023 and 2024e production with reserves for 17 producer countries plus a world total; and tight narratives on where American gold actually comes from - more than 40 lode mines across 12 states, Nevada alone around 70 percent of US output - how much returns as scrap (about 45 percent of reported consumption in 2024), who ships doré and bullion into the country and in what proportions, which HTS lines and duty treatments apply, and where the government stockpile and depletion allowances sit.

The compression is the feature. One page hands you a demand-side weighting almost no price feed bothers to carry: estimated world consumption excluding ETFs splits jewellery 45 percent, central banks 21 percent, physical bars 19 percent, coins and medals 7 percent, electronics 6 percent. Another page ranks the producers and sizes what is left in the ground. And because every edition back to 2000 survives intact, the same table shapes rebuild into a quarter-century series without schema drift.

Within Datadory's catalog of 1,744 datasets across 159 viable industries, this slice scores 9/10 - the ceiling for a fundamentals set, held off perfection only by its coarse annual grain.

Exactly as they land after cleaning - one row per country in the production-and-reserves table, selected salient-statistics metrics, and the end-use split:

# World Mine Production and Reserves - metric tons of gold content
country=United States   prod_2023=170      prod_2024e=160      reserves=3,000
country=China           prod_2023=375      prod_2024e=380      reserves=3,100
country=Russia          prod_2023=313      prod_2024e=310      reserves=12,000
country=Australia       prod_2023=296      prod_2024e=290      reserves=12,000
country=World total     prod_2023=3,250    prod_2024e=3,300    reserves=64,000

# Salient Statistics - United States (selected)
metric=Mine production          value=160      unit=metric tons     year=2024e
metric=Avg price (Engelhard)    value=2,400    unit=USD per oz      year=2024e

# Estimated world consumption by end use (excluding ETFs)
jewellery=45%  central_banks=21%  physical_bars=19%  coins_medals=7%  electronics=6%

Read what those rows already say. The United States slipped from 170 to 160 tonnes while the world climbed from 3,250 to 3,300, so America's share keeps eroding toward 5 percent. Reserve cover is arithmetic away: at 2024e output, China's stated reserve base runs roughly eight years, while Russia's and Australia's 12,000-tonne bases stretch toward four decades - a gap that says everything about how differently the three countries have surveyed and disclosed what they hold.

Seven documented field groups, definitions written against the printed table headers rather than inferred. The salient-statistics table alone spans nine metric families; the world table adds three columns per country; and the narrative sections contribute import-source shares, tariff lines and end-use percentages.

The dictionary below is the verified core. Anything deeper confirmed during sample preparation - the full five-year grid beyond the endpoint years shown, footnote and revision language, the substitutes and world-resources text - folds under additional fields on request rather than being promised blind.

Geography: the United States in depth - production, refining, trade, consumption, employment - plus 17 leading producer countries and a world total on the supply side, and supplying-country shares on the import side. Nevada's roughly 70 percent weight inside US output makes even the national picture effectively a one-state story with a long tail.

Temporal: five years of salient statistics per edition (2020 through 2024e currently), two production years plus reserves in the country table, and import-source shares averaged over 2020-23. Editions reach back to 2000 unchanged in shape, so a twenty-five-year reconstruction is a mechanical exercise rather than a research project.

Granularity: annual and national throughout - one value per metric per year, one row per country per table. There is no mine-level or monthly layer here by design; that depth lives in different federal series, and this one stays honestly shallow and completely consistent.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your stack already speaks. The rows arrive identical either way - cleaned, typed, and documented against the field dictionary above, with estimate flags preserved so provisional cells stay distinguishable from settled ones.

Every delivery ships with the full field dictionary and sample rows for validation. Because the underlying tables hold steady between editions, your schema does not churn underneath you.

• Investors and quant researchers turn reserves into cover ratios - 12,000 tonnes in Russia and Australia against eight years of runway in China - and use the production table to sanity-check any producer-country exposure story before it reaches a deck.
• Market researchers and consultants adopt the end-use split (jewellery 45 percent, central banks 21 percent, bars 19 percent) as ready-made demand weights, citing federal numbers instead of assembling shares from four commercial reports.
• Competitive-intel and product teams in mining and metals track country-level output swings year over year, and anchor the US side with Nevada's 70 percent concentration when sizing jurisdiction risk.
• Procurement and supply-chain analysts at refineries, jewellers and electronics firms read the import-source shares (doré led by Mexico at 38 percent) alongside the 45-percent recycled share to plan both mined and secondary feedstock.
• Journalists and academics cite a federal statistical series with a named author on every chapter - traceable provenance rather than an aggregator's silent restatement.
• Sales teams calling on miners and refiners open with the recipient's own numbers - who produces what, who ships what to whom - which converts a cold outreach into a calibration conversation.

Deeper persona guidance lives on the investors & quants, market researchers and journalists & academics pages for gold.

The gold slice splits cleanly along a price-versus-fundamentals line. Kitco owns the live board - bid, ask, day range, twelve currencies - and Stooq carries daily XAUUSD history reaching back centuries, but neither says a word about who dug the metal up or how much remains. SPDR Gold Shares fund data tracks tonnes held in ETF vaults, which is one demand bucket, not supply.

This dataset is the supply side: tonnage mined, metal refined, scrap recovered, reserves remaining, and who imports what from whom. The practical stack pairs them - annual fundamentals here give the daily and intraday price series a reason for being where it is, and the end-use split tells you which demand pocket to blame when price and supply diverge.

Three things, stated upfront.

First, respect the grain. This is an annual, national-level series - the right tool for structural analysis and citation, the wrong tool for anything needing monthly resolution or mine-level attribution. If you need that depth, say so in the sample request and we will point the sample at the heavier federal sources that carry it.

Second, treat 'e' cells as soft. Every current-year figure revises in the following edition, and world totals arrive rounded, so components need not sum. Build pipelines that re-pull rather than cache if your series touches the latest year.

Third, the verified dictionary covers the seven field groups above; the narrative material - substitutes, world resources, stockpile detail, revision footnotes - gets confirmed against the chapter when we cut your sample, which is also where column naming locks down for your pipeline.

Estimate discipline note - figures marked 'e' are estimates and revise in the next edition; Datadory preserves the flags so downstream models can discount provisional cells automatically.

Rounding note - the world total row is published rounded, and component countries may not sum exactly to it; reconcile at the total, never the parts.

Provenance note - every chapter carries a named author and the National Minerals Information Center's imprint; the current gold chapter credits Kristin N. Sheaffer, giving each statistic a traceable federal chain of custody.

Where to go next:

• Kitco - Gold Price and Charts - the live spot board the chapter's annual average eventually settles into.
• Stooq - XAUUSD Historical Gold Data - daily OHLC price history to pair with the annual averages.
• SPDR Gold Shares (GLD) Fund Data - ETF holdings in tonnes, the demand bucket the end-use split deliberately excludes.
• GoldPrice.org - Gold Price in 160 Currencies - multi-currency spot context around dollar-denominated fundamentals.
• Best gold datasets - the ranked shortlist across the whole vertical.
• Gold data hub - the pooled industry view.