Global Coal Exit List (GCEL) Data

It is the corporate map of the thermal coal economy - one row per firm instead of one statistic per country. Compiled by urgewald, the German NGO whose exit lists anchor this catalog, the Global Coal Exit List follows the thermal coal value chain end to end: miners, coal-fired power producers, project developers, and service providers across exploration, processing, trading, transport and logistics, equipment manufacturing, maintenance and engineering services, and coal-to-liquids and coal-to-gas conversion. The current edition carries roughly 3,000 companies - over 1,500 parent companies plus their subsidiaries - which urgewald states represent more than 90% of global thermal coal production and coal-fired capacity, headquartered across more than 80 countries.

Membership comes from three published gates rather than a keyword match. Expansion criteria flag firms developing new mines or mine extensions, new coal-fired capacity of at least 100 MW, or coal transport and infrastructure such as coal-to-gas facilities. Relative criteria set a 10% floor on coal share of revenue - or on coal share of power production, for utilities. Absolute criteria take the giants outright: 10 million tons of annual thermal coal production or 5 GW of installed coal-fired capacity. Metallurgical coal is deliberately out of scope; steelmaking coal sits on the separate Metallurgical Coal Exit List.

Against the wider catalog - 1,744 datasets averaging 7.81 - this record scores 8/10, carried by completeness plus attribution: few shelves can say which companies own nine-tenths of a fuel, company by company, and defend every inclusion with published criteria.

Five of the roughly 3,000 companies, chosen because they sit in four different corners of the value chain:

# company-level records as delivered (5 of ~3,000)
company=Coal India Ltd   hq=India
  csr=>70%   cspp=9%
  power_capacity=20 MW   mining_production=721.4 Mt
  expansion_power=>2730 MW

company=CHN ENERGY Investment Group Co Ltd   hq=China
  csr=>69%   cspp=80%
  power_capacity=211000 MW   mining_production=620 Mt
  expansion_power=47806 MW

company=SUEK JSC   hq=Russia
  csr=>70%   cspp=98%
  power_capacity=16913 MW   mining_production=112.8 Mt
  expansion_power=967 MW

company=Glencore PLC   hq=Switzerland
  csr=>4%   cspp=-
  power_capacity=-   mining_production=99.6 Mt
  expansion_power=>577 MW

company=Peabody Energy Corp   hq=USA
  csr=>73%
  mining_production=96.9 Mt
  expansion_mining=Australia

# shape of the slice
grain=one row per company (parents + subsidiaries)   hq_countries=80+
ratios=csr and cspp carried as threshold bands (e.g. >70%)
expansion_flags=power | mining | infrastructure   dash=field not carried by that company

Read the anatomy rather than the digits. CSR arrives as a band, not a decimal - '>70%' is the native unit of disclosure, so treat it categorically instead of averaging bands back into a fake mean. CSPP belongs to utilities: Coal India carries 9% on just 20 MW of coal-fired capacity while CHN ENERGY carries 80% on 211,000 MW, and Glencore's row leaves the field empty because a trader-miner with no coal-fired fleet gives it nothing to measure. Inclusion logic shows straight through the numbers: a 4% revenue share would fail the relative gate entirely, yet 99.6 million tons of annual production clears the absolute gate - and the '>577 MW' expansion flag explains why the row matters regardless. Peabody's row shows the second flag doing different work: expansion_mining names a country of new mine development rather than a megawatt figure. Dashes mark fields a company does not carry; they are not zeros, and parsing them as zeros manufactures phantom utilities.

Nine verified core fields, written against the published methodology rather than guessed. Exposure comes first: the two ratio fields separate merchants from utilities, with coal share of revenue (CSR) reading across the whole value chain and coal share of power production (CSPP) applying to generators. Physical scale follows - installed coal-fired capacity and annual thermal coal production - which is what makes the absolute inclusion gates computable rather than rhetorical. The three expansion flags split development activity into power, mining and infrastructure, so a firm adding mines reads differently from one adding gigawatts. Geography and corporate structure close the row: headquarters country, and the parent-subsidiary link that keeps rollups honest.

The dictionary below is the verified core. Anything adjacent confirmed during sample preparation folds under additional fields on request rather than being promised blind.

Geography: global by construction. Headquarters span more than 80 countries, from Argentina to Zimbabwe, and urgewald states the covered firms together hold over 90% of world thermal coal production and coal-fired capacity - so a screen run against this file generalizes to the market rather than to a regional sample.

Temporal: annual editions since 2017, with GCEL 2025 as the current reference edition. Edition-over-edition continuity is itself analytical: a company entering or leaving the list is a signal, because crossing a published threshold is what causes the move.

Granularity: company level, with the parent-subsidiary link preserving the difference between operating exposure and ultimate ownership. Roughly 1,500 parents carry their subsidiaries with them, so rollups to corporate groups stay lossless instead of counting every listed group twice.

Set against the wider catalog - 1,744 datasets averaging 7.81 - this record scores 8/10.

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

Pick the channel your stack already speaks: flat files sized to the companies, thresholds and columns you actually named, or a direct pipe into Snowflake, BigQuery or Redshift. Rows arrive exactly as shaped above - one record per company with its ratio bands, capacities, production figure, expansion flags and parent link - so your first screen runs against evidence rather than a rebuilt spreadsheet. Cadence gets decided after the sample validates, not before.

• ESG and climate teams screen portfolios against the three expansion gates before exposure turns into headline exposure; worked flows continue on our esg & emissions analysis page.
• Credit and risk analysts read CSR bands and capacity as quantified coal dependence per borrower, with the expansion flags showing direction of travel; see credit risk screening.
• Ownership researchers roll the parent-subsidiary links upward, connecting a subsidiary mine or plant to the listed group that ultimately owns it; fuller methods sit on our supply-chain mapping page.
• Analysts and writers cite named companies under published inclusion criteria instead of an anonymous estimate of 'the firms responsible for most coal'; see citation-grade research.

For contrast inside the same catalog: Global Coal Plant Tracker counts the megawatts asset by asset where this list screens the companies behind them, the EIA Annual Coal Report stays statistical and American, and the SteelHome Database prices the coke and coal these firms trade. None puts 1,500+ parents and their subsidiaries in one screened table.

Investors and quant researchers get a screened universe they can join against holdings - subsidiary exposure rolled to the parent, thresholds already applied; workflows continue on investors and quants in coal consumable fuels. Market researchers and consultants start from a ready-made counterparty universe with quantified dependence per firm; see market researchers in coal consumable fuels. Journalists, academics and students get named companies under published criteria, which makes a divestment claim citable rather than anecdotal; see journalists and academics in coal consumable fuels. Data scientists and ML engineers get typed rows whose enums and numeric scales survive a join; see data scientists in coal consumable fuels. Developers and data-product builders get a stable nine-field schema that ships inside screening tools unchanged across editions; see developers and builders in coal consumable fuels.

Six notes travel with this page - the asset-level tracker underneath these companies, the met-coal sibling, the US statistical mirror, the source profile, and the two routes into the wider catalog. Each earns its place on a build, not a brochure. Skim them below, then request a sample cut to your companies, thresholds and cadence.