EIA Coal Data (production, consumption, prices, reserves)

It is the official United States coal ledger as typed rows instead of a yearbook. The U.S. Energy Information Administration publishes the collection under one description: reserves, production, prices, employment and productivity, distribution, stocks, imports and exports. That breadth is the point - supply, trade, logistics, burn, inventory and price live in one normalized row shape, so a question about why a generator paid what it paid does not require stitching five publications together.

Three axes organize everything. The geographic ladder descends from national totals through states and counties to individually named mines. The frequency switch runs a subject at annual, quarterly or monthly grain depending on the series. And the facets - consuming sector and coal rank - cut demand and pricing along the lines contracts and procurement actually use. Within Datadory's catalog of 1,744 datasets across 159 viable industries, this record scores 9 out of 10, against a catalog-wide average of 7.81.

The row anatomy, flat as records arrive:

series      : COAL.PRODUCTION
period      : <annual | quarterly | monthly>
value       : <short tons | thousand short tons | dollars per short ton | Btu | employees>
state       : <facet present where the series is state-disaggregated>
sector      : <electric power | industrial | commercial | residential>
coalRank    : <bituminous | subbituminous | lignite | anthracite>

# shape of the slice
grain=one row per series-period-facet   geography=United States, to county and mine
frequency_enum=annual|quarterly|monthly   rank_enum=bituminous|subbituminous|lignite|anthracite

Read the anatomy as a promise about shape, not a claim about any particular print. One identifier names the series; one column carries the number; the facets declare which slice of the market the number describes. Because the units travel with the series rather than the column, nothing in a delivery ever asks you to guess whether a figure is short tons or thousand short tons - the declaration rides beside the value. Live payloads ship with your sample request, keyed to the series you name, so your first join attempt runs against evidence rather than hope.

Six documented fields, grouped into three jobs. Identification: series is the time-series identifier and the join key every downstream merge hangs on - one stable string per subject, geography and frequency combination, with the record's worked example reading COAL.PRODUCTION. Timing and measurement: period stamps the observation at annual, quarterly or monthly grain, and value carries the number in whatever unit the series speaks - short tons, thousand short tons, dollars per short ton, Btu, or employees.

Cutting facets: state appears where a series disaggregates below the national line; sector marks the consuming side as electric power, industrial, commercial or residential; coalRank splits pricing and quality into bituminous, subbituminous, lignite and anthracite. Rank is the facet most copies of coal data lose - without it, metallurgical and thermal economics average into one meaningless number.

The dictionary above is the documented core. Series-family selections, unit normalizations and county-or-mine extracts fold under additional fields on request rather than being promised blind.

Geography: the United States, four rungs deep - national totals, states, counties, individual mines. The lower rungs are the differentiators: basin questions (which seam, which operation, which county) have answers here that state-level matrices structurally cannot give.

Temporal: depth varies by series, stated plainly rather than blended into one recency claim. Mine-level survey detail reaches back to the 1980s, national totals run further, and current periods publish alongside the history - so a forty-year panel and a this-quarter read come from the same collection.

Granularity: one observation per series, period and facet, across eleven series families - production, imports, exports, shipments, receipts at electric power plants, consumption, stocks, prices, reserves, employment and productivity. Annual, quarterly and monthly frequencies coexist, which means the frequency choice belongs to the analysis, not the source.

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

Pick the channel your stack already speaks. The rows arrive identical either way - one record per series, period and facet, with the identifier, the timestamp grain and the unit declaration carried in-schema - extracted so nothing in your pipeline re-assembles a statistical program by hand.

Every delivery ships with the field dictionary above plus a payload keyed to the series you named, so validation happens against your slice, not a demo. The cadence setting governs how fast an observation reaches you, never what its columns mean.

• Utility fuel-cost analysts track receipts, stocks and prices as the input bill behind generation economics; the workflow pattern continues on our market sizing page.
• Traders and procurement teams hold rank-level official prints as the reference line market quotes get discounted or marked up against; see price monitoring.
• Quant researchers backtest on decades of definition-stable panels instead of repairing scraped splices; the panel pattern lives on quant backtesting.
• Forecasters read consumption by sector beside plant stocks as a fuel-switching signal; see demand forecasting.
• Journalists and academics cite the federal statistical source itself rather than somebody's chart of it; see citation-grade research.

Market researchers and consultants get the structural layer - fundamentals from the national line down to a named mine - that turns a headline tonnage into a ranked supplier or customer view; the fuller workflow lives on market researchers in coal consumable fuels. Investors and quant researchers turn receipts and prices into utility-cost and demand factors; see investors and quants in coal consumable fuels. Data scientists and ML engineers get long typed panels keyed on one identifier; see data scientists use cases. Journalists, academics and students get numbers that survive citation checks; see journalists and academics in coal consumable fuels.

Every close alternative trades something away. The EIA State Energy Data System covers every fuel across all 50 states from 1960 but locks to an annual spine and averages rank out of existence - the head-to-head is argued properly in our EIA Coal Data vs SEDS comparison. The Annual Coal Report freezes roughly 46 tables into one yearly edition where this record stays a living series. MSHA's Mine Data Retrieval System holds the operating record of roughly 92,000 mines - inspections, violations, employment - rather than the commodity statistics. And the Global Coal Plant Tracker inventories the 14,674-unit generating fleet that burns what these series count: the demand-side mirror. This record is the one answering how much moved, burned and cost what, over what period.

Because the official coal series are famous and annoying: famous enough that everyone quotes them, scattered enough that assembling them means reconciling frequencies, units and facet vocabularies by hand. Datadory delivers them as one typed row shape, documents the unit-per-series rule in schema rather than tribal knowledge, confirms facet spellings against your named selection before anything ships, and keeps successive deliveries accumulating so history grows on your clock - daily, weekly, or hourly. Browse the rest of the slice on the coal & consumable fuels data hub, the best coal & consumable fuels datasets ranking, or the source profile behind this record.