EITI (Extractive Transparency): Intelligence Source Guide

EITI is the standard under which producer states publish what extractive companies paid them and what they received, reconciled by an independent administrator. For tracing resource revenue, licence allocation and beneficial ownership, it is the only systematic public disclosure regime.

At a glance

Source	EITI (Extractive Transparency)
Category	Conflict, Crime & Human Security › Environmental & Wildlife Crime
Homepage	https://eiti.org/
Machine interface	`https://api.eiti.org/`
Format	`JSON`
Access	Open — no account required
Disciplines	Economic Intelligence, Corporate Intelligence
Mission domains	Mining & Resource Crime

Extractive-industry payment/ownership transparency. — as catalogued in the platform’s own source registry.

The Extractive Industries Transparency Initiative is a governance standard and a body of disclosures produced under it. A country volunteers to implement, forms a multi-stakeholder group of government, industry and civil society representatives, and then discloses a defined set of information about its oil, gas and mining sector. The core of that disclosure is a reconciliation: companies report what they paid to the state, government agencies report what they received, an Independent Administrator compares the two, and the unexplained discrepancies are published rather than resolved privately. Around that core sits a wider set of contextual disclosures required by the Standard – the legal and fiscal regime, licence registers and how licences were allocated, production and export volumes, state-owned enterprise transactions and quasi-fiscal expenditure, revenue allocation to subnational governments, social and environmental payments, and beneficial ownership of the companies holding licences. The material appears as country reports, summary data files in structured formats, and a programmatic interface. Implementation is assessed through Validation, which scores a country's performance and can result in suspension or delisting. The output is therefore two things at once: a dataset of payments and licences, and a documented record of how seriously a given government takes disclosing them.

Corporate registries tell you who owns a company. Sanctions lists tell you who is prohibited. Court records tell you what has already gone wrong. EITI is the only routine source that tells you how much money moved from a named company to a named government agency for a named licence, on the state's own account, with an auditor's note where the two sides disagree. That makes it the primary structured collection route for ECONINT and CORPINT work on resource-linked corruption. The analytical job it does that nothing else does is to establish the expected baseline. When you have an allegation that a mining concession was allocated to a politically exposed shell company at below value, EITI gives you the licence register entry, the allocation process the state says it followed, the signature bonus or fee the state says it received, and – where the country has implemented the beneficial ownership requirement – a declared owner to test against your own research. Discrepancy is the product. A company that appears in the licence register but not in the payment reconciliation, a payment stream that appears on the company side and not the government side, a state-owned enterprise whose transfers to the treasury do not match its declared revenue: these are the findings that justify the source. It also does something less obvious. Because participation is voluntary and Validation is public, the initiative's own membership record is an indicator in its own right. A country that joins, is downgraded over civil society space, and then withdraws has told you something about its governance trajectory that no revenue table contains.

Who publishes it, and why that matters

EITI is governed by a board with equal representation from implementing and supporting countries, companies and investors, and civil society organisations, and is administered by an International Secretariat based in Oslo. Funding comes from supporting governments, participating companies and institutional donors. That structure explains both the strengths and the limits of the product. The strength is that the disclosures are not an NGO's estimate or a journalist's reconstruction; they are numbers the government and the companies have both signed up to, which makes them very hard for either side to disown later. The limit is that a consensus body moves at the speed of its least willing member. Requirements are negotiated, deadlines slip, and the enforcement instruments are reputational – suspension and delisting – rather than legal. Countries have left when the assessment went against them, and at least one major supporting government has withdrawn from implementation for domestic political reasons. The initiative depends on civil society participation being genuinely free, and it has repeatedly had to confront member states where it is not. For an analyst, the practical implications are these: the data exists because a government agreed it should, so the categories reflect what governments were prepared to disclose; coverage is a political map, not a resource map; and the Standard's requirements have been revised more than once, so a report from one cycle is not directly comparable with a report from another without checking which version of the Standard governed it.

Provenance is the first question to ask of any dataset and the one most often skipped. Who collects it, what their incentive is, whether they publish a methodology, and whether they correct the record when they get something wrong all bear directly on how much weight a finding drawn from it can carry.

What a record actually contains

The fields you will be working with, what each one means, and whether it is something you can pivot on. Read the meanings carefully — more analysis is wrecked by misreading a field than by failing to find one, and a field that looks like an observation is often an inference.

Field	Type	What it means	Pivot value
`country`	string	The implementing country whose multi-stakeholder group produced the disclosure. Membership status is a separate attribute and can be implementing, suspended, or withdrawn.	Country risk profile, licence register, national extractive regulator, other producer-state datasets.
`reporting_period`	string	The fiscal year covered by the reconciliation, which frequently is not the calendar year and frequently is not the year of publication. Reports commonly appear one to three years after the period they describe.	Timeline construction; alignment with company annual reports and commodity price series.
`company_name`	string	The reporting entity as named in the national report. This is the local operating or licence-holding entity, often a subsidiary, and often transliterated or abbreviated inconsistently between reporting cycles.	Corporate registry lookup, parent company resolution, group structure, sanctions screening.
`government_entity`	string	The agency that received the payment – treasury, tax authority, mining cadastre, environmental fund, subnational authority or state-owned enterprise. The split between them is where leakage becomes visible.	Public finance analysis, subnational transfer tracing, state-owned enterprise accounts.
`revenue_stream`	enum	The payment type: corporate income tax, royalty, signature bonus, production entitlement, dividend, licence fee, surface rental, withholding tax, infrastructure provision and others. Categories are standardised in principle and locally interpreted in practice.	Fiscal regime analysis; comparison of the same stream across companies in one country.
`payment_value`	int	The amount reported, in the reporting currency. In-kind revenues – the state's physical share of production – are valued using a stated methodology that varies between countries and materially affects the total.	Comparison against company-side disclosure, production volume and prevailing price; identification of implausible unit values.
`currency`	string	The currency of the reported figure. Local-currency reporting in a high-inflation or multiple-exchange-rate economy makes cross-year and cross-country comparison unsafe without an explicit conversion methodology.	Central bank rate series; recalculation of trend claims.
`discrepancy`	int	The unreconciled difference between the company-reported and government-reported figure for a payment. This field, not the payment total, is the investigative signal.	Independent Administrator commentary, follow-up in later cycles, targeted questions to the regulator.
`licence_id`	string	The concession, block or permit identifier as held in the national cadastre. It is the join key between EITI disclosure and the licence register, and it is frequently formatted differently in each.	Cadastre record, coordinates, licence history, transfer chain.
`licence_type`	enum	Exploration, production, artisanal, or a national category with no clean international equivalent. Determines what obligations attach and therefore what payments should exist.	Expected payment profile; detection of a producing block still classified as exploration.
`commodity`	enum	Oil, gas, gold, copper, cobalt, bauxite, diamonds and so on. Coverage is defined nationally and some countries scope out commodities that matter, including artisanal production.	Commodity trade data, export statistics, downstream supply chain sources.
`beneficial_owner`	string	Declared natural person owning or controlling the licence-holding company, where the country has implemented the beneficial ownership requirement. It is a self-declaration by the company, not a verified fact.	PEP screening, corporate registry, leaks datasets, sanctions and adverse media.
`pep_flag`	enum	Whether the declared owner is identified as a politically exposed person. Definitions of PEP status differ between national frameworks, and self-declaration is the norm.	Asset declarations, procurement records, political office holder lists.
`validation_status`	enum	The country's assessment outcome and the date it was determined. Scoring methodology has changed with revisions of the Standard, so a score is only comparable within its own scheme.	Governance trend analysis; weighting of the country's other disclosures.

Coverage — and what is not in it

Coverage follows membership, and membership is voluntary. Implementing countries number in the fifties and include a substantial share of the world's oil, gas and mineral producers across sub-Saharan Africa, Latin America, Central and Southeast Asia, the Caucasus and a smaller number of high-income states. The large gaps are structural rather than accidental: several of the largest hydrocarbon and mineral producers have never implemented, some joined and left, and no country can be compelled to participate. Within an implementing country, coverage is set by a materiality threshold that the national multi-stakeholder group defines, so smaller companies and smaller revenue streams are legitimately excluded from reconciliation and simply do not appear. Time coverage is long by the standards of any comparable dataset – reporting in the earliest implementing countries goes back to the middle of the 2000s – but it is uneven, with gaps where a country was suspended or where a reporting cycle was missed. The rhythm is annual and lagged. A country report covers a fiscal year and is published one to three years afterwards, so this is a source for structural and historical analysis, not for current activity. Two supplementary layers have shorter latency: the Validation record, which updates as assessments complete, and beneficial ownership registers in the countries that maintain them as live systems rather than as periodic report annexes. Entity coverage is companies, government agencies, state-owned enterprises and licences; it is not individuals except where beneficial ownership is disclosed, and it is not transactions at the level of a single shipment or contract payment.

Known blind spots

Absence of evidence here is not evidence of absence. These are the conditions under which EITI (Extractive Transparency) will not show you something that is nevertheless real:

A non-member state discloses nothing here, and the states with the most acute resource-corruption problems are systematically over-represented among non-members and withdrawn members. Absence of a country is a governance signal, not a data gap to be filled by inference.
Materiality thresholds set nationally mean that a company can hold a licence, produce, and pay taxes without ever appearing in a reconciliation, because the multi-stakeholder group decided it was below the line.
Smuggled, artisanal and informal production is by construction outside a system built on reconciling declared payments; for gold in particular, the volumes that never enter the formal chain can exceed the volumes that do, and none of it is here.
Payments made to individuals rather than institutions – the actual mechanism in most grand corruption cases – are not a reportable category. The reconciliation can balance perfectly while the underlying transaction was corrupt.
Transfer mispricing, offshore marketing arrangements and related-party sales reduce the taxable base before any payment is made, so a company that pays exactly what it owes on an understated base looks compliant in this dataset.
Beneficial ownership disclosure is uneven in implementation and self-declared where it exists. A nominee, a trust or a layered offshore structure defeats it, and a missing declaration is rarely enforced.
Publication lag of one to three years means the dataset cannot support any question about current activity, an ongoing transaction, or an asset that is being moved now.
Revisions of the Standard changed what must be disclosed and how it is assessed, so a longitudinal series built by concatenating reports across cycles will contain breaks that look like real changes in behaviour.
Civil society participation is a formal requirement and is constrained in practice in several implementing countries, which means the internal challenge function that is supposed to make the numbers honest may be absent precisely where it is most needed.

Write the blind spot into the product. A statement that something “was not observed in EITI (Extractive Transparency)” is defensible; a statement that it “did not happen” is not, and the difference is what survives cross-examination.

Access, licensing and what you may do with it

Access model: Open — no account required

Everything published under the initiative is public and free. There is no registration, no tier and no gate: country reports, summary data files and the country-level pages are open, and a programmatic interface is offered for structured retrieval. In practice most analysts use three routes in combination. The structured summary data give comparable fields across countries and are the right basis for anything quantitative. The full country reports, usually PDFs produced by the national Independent Administrator, carry the methodology, the scope decisions, the materiality threshold and the auditor's commentary on discrepancies – which is where the analytical value actually sits and which no structured extract preserves. National implementation websites, maintained by the country multi-stakeholder group, sometimes publish more than reaches the international level, particularly licence registers and contract disclosures. Read the API and data documentation on the initiative's own site for the current shape of the interface rather than reconstructing endpoints from an older integration; the data offering has been reorganised more than once. For any finding you intend to publish or act on, keep the source PDF alongside the extracted number, because the number without its scope note is not defensible.

Licence

The initiative's stated purpose is open publication, and its material is intended to be freely used and redistributed with attribution. Confirm the current terms on the site before building a commercial product on it, particularly for bulk redistribution, because the specific licence attached to different components – narrative reports, summary data, and any interface output – has not always been stated identically. A separate and more important constraint applies to the national layer: the underlying country reports are produced under national arrangements, and material republished from a national implementation site may carry that country's terms rather than the initiative's. When you combine EITI beneficial ownership declarations with other personal data, data protection law applies to the combination regardless of the licence on either input.

Rate limits and fair use

This is a low-volume, high-value source and should be treated as such. The whole corpus of structured country data is small enough to mirror locally in a single collection pass, so there is no legitimate reason to query repeatedly. Pull the summary data on a monthly or quarterly schedule, pull the Validation status on a similar cadence, and fetch individual country reports once and cache them permanently. Identify your client with a descriptive user agent and a contact address. The one component that justifies more frequent polling is a live national beneficial ownership register where one exists, and that should be polled against the national system on terms agreed with it rather than through the international aggregation.

Licensing changes, and it changes without warning. A dataset that was free for research this year may not be free for commercial or evidential use next year. Confirm the current terms before you build a dependency on it, and record the terms you relied on alongside the data — the licence in force at the time of collection is part of the provenance.

Collecting it

How EITI (Extractive Transparency) is actually pulled, in the order you would set it up. Prefer the bulk or export interface over per-item lookups wherever one exists: it is kinder to the publisher, faster for you, and gives a reproducible snapshot rather than a series of point-in-time answers you cannot reconstruct later.

Method	Format	Cadence	Notes
Structured summary data	JSON	quarterly; the underlying reports change annually per country	The comparable cross-country layer. Correct starting point for any quantitative work, and small enough to hold in full.
Programmatic interface	JSON	as needed; mirror rather than poll	Use for country, company, revenue stream and licence entities. Read the current documentation before writing a client; the interface has been reorganised.
Country report documents	bulk	once per publication, then never again	The PDFs carry scope, materiality and the Independent Administrator's discrepancy commentary. Archive them with a hash; they are the evidentiary layer.
Validation records	HTML	monthly	Track membership status changes, suspensions and withdrawals as events in their own right, not as metadata.
National implementation sites	HTML	per country, opportunistically	Licence registers, contract disclosures and beneficial ownership registers sometimes exist here and nowhere else. Treat each as a separate source with its own reliability.
Beneficial ownership extracts	CSV	per report or per register update	Where a country publishes a register, collect it as an entity dataset for resolution rather than as a report annex.

Ingesting it into the platform

Every step below is idempotent and cursor-based: interrupt one and it resumes from where it stopped rather than duplicating rows or losing progress. Collection is recorded per source, so a feed that quietly stops publishing shows up as a stale timestamp instead of silently thinning your coverage.

Register the source and its political metadata — Add EITI in sources.php with implementing-country membership status carried as a field, so that every downstream record inherits whether the country was in good standing during the period it describes. A suspended country's report is still data, but it is data with an asterisk that must travel with it.
Schedule the pull — Configure collect.php on a quarterly cadence under cron.php. This source does not reward frequent collection; it rewards never missing a publication. Failure of the job should raise an alert because a silent gap here looks identical to a country that stopped reporting.
Normalise companies before anything else — In ingest.php, treat company_name as a dirty string. The same operator appears with different transliterations, legal-form suffixes and abbreviations across countries and cycles. Resolve to an organisation entity with the raw string preserved as an alias, never overwritten.
Key payments to licences and agencies — Build the payment record as a triple of company, government entity and revenue stream, joined to licence_id where the report supports it. Store discrepancy as a first-class field rather than deriving it later, because the report's own discrepancy figure reflects reconciliation decisions you cannot reproduce.
Resolve beneficial owners as persons — Run resolve-everything.php over declared owners to produce person entities distinct from the companies, then screen them through the sanctions and PEP sources already in the platform. A declared owner is a claim to be tested, and should be stored with that provenance.
Attach geography — Where licence coordinates or a named concession area exist, enrich to a location so that extractive activity can be placed against conflict, deforestation and mining-site datasets on the same map rather than compared in prose.
Correlate against adjacent sources — Use correlate.php to test each licence-holding company against corporate registries, sanctions lists, leaks datasets and adverse media. The correlations that matter are a licence holder with no findable registration, a beneficial owner who is also a procurement counterparty, and a company appearing in one cycle and vanishing from the next.
Publish to the mission views — Surface the resulting entities in financial-crime.php and org-profile.php, and expose country-level aggregates to country-risk.php so that resource revenue transparency contributes to the country picture rather than sitting in a silo.

Registered sources and their last-collected state are listed in sources.php, and the scheduled chain that keeps them current is in automation.php.

How it is wrong, and how to tell

Every dataset is wrong in characteristic ways. Knowing which ways is the difference between using a source and being used by one, and it is the part of source evaluation most often skipped because it is the part that takes work.

Judge this source on process rather than on precision. The numbers are not measurements; they are two parties' assertions, compared. That comparison is the quality mechanism, and where it is done properly – a competent Independent Administrator, a functioning multi-stakeholder group, full scope, discrepancies published rather than buried – the resulting figures are about as reliable as any public financial disclosure from a producer state gets. Where the process is weak, the same publication format wraps a much softer product: a narrow materiality threshold, key state-owned enterprises out of scope, discrepancies described as immaterial without explanation, or a report signed off by a group whose civil society seats are occupied by people who cannot safely dissent. You can tell these apart, and doing so is the analytical work. Read the Independent Administrator's scope section and the discrepancy commentary before you use a single figure. Check the Validation outcome and the reasons given for it. Check whether the same reconciliation in the previous cycle produced comparable totals and, if not, whether the change is explained. Cross-check the largest payers against their own consolidated accounts, which for listed companies are independently audited and use different accounting rules but should be in the same order of magnitude. Where a country's disclosure survives all four tests it is strong evidence. Where it fails any of them it is still useful, but as an indication of what the government was prepared to say rather than as a measurement of what happened.

Characteristic false positives

A zero or a missing row means the item was out of scope, below the materiality threshold, or not reported – not that the payment was zero. Reading absence as evidence of non-payment is the most common misuse of this source and produces confident, wrong allegations.
Company names collide and diverge simultaneously. Subsidiaries with near-identical names in different countries merge under a naive match, while the same subsidiary transliterated differently across two cycles splits into two entities and creates a phantom new market entrant.
In-kind revenue valuation methods differ by country and sometimes by cycle. A change in how the state's production share is priced can move a national total by a large margin with no change whatsoever in the underlying flows.
Local currency reporting in an economy with high inflation or parallel exchange rates makes year-on-year comparison meaningless unless you re-derive the conversion yourself, and headline trend claims built on the published totals are frequently artefacts of the rate used.
A reconciled discrepancy of zero can be manufactured. Where the Independent Administrator adjusted one side to match the other and documented it in a footnote, the structured extract carries the clean number and drops the footnote.
Beneficial ownership declarations naming a plausible local individual are routinely nominee arrangements. Treating a declared owner as the controlling party without independent corroboration is the failure mode this field was designed to expose and frequently perpetuates instead.
Fiscal year misalignment between the state and the company means the same payment can be attributed to different years on the two sides, generating a discrepancy that is a calendar artefact rather than a leak.
Revenue stream categories are standardised in name and interpreted locally in substance. Comparing royalty as a share of revenue across countries without reading each fiscal regime compares different things under the same label.

None of these make the source unusable. They make it a source that requires corroboration before an assertion built on it goes into a product, which is true of every source and admitted by few.

Ageing

The dataset is inherently retrospective, so the question is not when it goes stale but what a stale record misleads you about. Payment figures do not age – a payment made in a stated fiscal year remains a fact about that year forever, and old EITI data is genuinely useful for reconstructing a historical baseline. Everything attached to the payment ages quickly. Licence ownership transfers, and a concession attributed to a company in a report published two years ago may have changed hands twice since; the register entry is the thing to re-check, not the payment. Beneficial ownership is the fastest-decaying field in the source, because restructuring in response to disclosure is a known response to disclosure. Corporate identity ages through mergers, renames and liquidation, so an organisation entity resolved at ingest should be re-resolved at analysis time. Country membership status changes and can invalidate the assumption that a series will continue. A stale record looks like a perfectly plausible payment attributed to a company that no longer holds the licence, owned by a person who divested when the register went live, in a country that has since been suspended – and nothing in the record itself will tell you any of that. The mitigation is to timestamp every enrichment separately from the disclosure period and to re-run resolution rather than trusting the cached result.

What this source feeds

A source is only worth what it lets you conclude. These are the disciplines that collect through it, the mission domains it serves and the data points it yields — every one is a tag, so you can follow any thread from here into the rest of the library.

Collected by these intelligence disciplines

Serves these mission domains

Mining & Resource Crime

Yields these data points

How each sector uses EITI (Extractive Transparency)

The same dataset is worked very differently depending on who you are, what authority you hold, and what you are ultimately producing. A military analyst is supporting a commander’s decision; a journalist is meeting a publication standard; an NGO caseworker is protecting a person. The records are shared — the constraints, thresholds and outputs are not.

🎖 Military and defence

Resource revenue is the financing layer under most protracted conflicts, and this source gives the formal half of that picture in a form that supports planning rather than speculation. For a joint operations or stability planning cell, the useful products are the licence register mapped against the areas of operation, the identity of the companies with concessions in or adjacent to contested terrain, and the size and destination of the revenue those concessions generate. That tells you which economic actors have an interest in a given outcome, which local authorities are funded by extractive transfers and would lose that funding under a change of control, and where infrastructure obligations attached to a concession might already exist. Use it for the formal economy only. Armed-group financing runs through the informal and smuggled channels this source cannot see, so pair it with mining-site mapping and conflict-minerals reporting rather than treating a clean revenue table as evidence that a sector is clean.

🕵 National intelligence

For ECONINT and CORPINT this is baseline collection, not a curiosity. Three products come out of it reliably. First, the state's own account of its resource revenue, which is the benchmark against which any allegation of diversion has to be measured and which is far harder for a government to repudiate than an external estimate. Second, the licence allocation record, which exposes who obtained rights to what, when, through which process, and at what stated cost – the sequence in which state capture is usually visible before the money moves. Third, the beneficial ownership layer, which is a self-declaration and therefore valuable in two opposite ways: as a lead when it names someone unexpected, and as a documented false statement when it does not name someone you can establish independently. The initiative's own membership and Validation record is a governance indicator worth tracking as a series. The collection discipline point is that publication lag makes this a structural source. It answers how a system is arranged; it does not answer what happened last month.

👮 Law enforcement

In corruption, money laundering and asset recovery casework, EITI provides the predicate documentation that turns an intelligence lead into an investigable proposition. A licence register entry establishes that a specific entity held a specific right on a specific date. A reconciliation establishes what the state says it received. A beneficial ownership declaration is a statement made to a public register that can be tested and, where false, is often an offence in itself under national law. The evidential posture matters: this is public official disclosure, generally admissible as a published government record in most jurisdictions but rarely self-proving, so plan to obtain the underlying records through mutual legal assistance or from the national regulator directly. The material is also useful defensively, in vetting – a company bidding for a public contract that holds extractive licences in a country with a poor Validation outcome and no beneficial ownership disclosure is a due diligence finding, not a hunch.

🔍 Private investigation and corporate security

For corporate intelligence, pre-transaction due diligence and litigation support, this is one of the few free sources that produces hard, citable facts about counterparties in producer states. The standard uses are establishing whether a target company actually holds the concessions it claims, identifying the local partners and their declared owners, quantifying the fiscal exposure attached to a licence, and finding the discrepancy or the missing report that justifies asking a harder question in the data room. It is also useful for the negative finding – a company that describes itself as a significant operator in a country and does not appear in that country's reconciliation is either below materiality or misrepresenting its position, and both are worth knowing. The professional constraints are ordinary but real: the data is lagged, so it describes a historical state; beneficial ownership declarations are unverified and should never be reported to a client as established fact; and the licence register in the report may be superseded by the national cadastre.

📰 Journalism and OSINT media

Investigative reporting on resource corruption has been built on this source since it began, and the reporting patterns are well established: the concession awarded to a company incorporated weeks earlier, the payment stream that appears on one side of the reconciliation and not the other, the state-owned enterprise whose transfers to the treasury do not match its declared sales, the beneficial owner who turns out to be the minister's relative. What makes the source good for journalism is that the numbers are the government's own, which shifts the burden in any subsequent dispute. What makes it dangerous is that the numbers require context to mean anything, and a discrepancy is not by itself evidence of theft – reconciliation differences arise from timing, exchange rates and classification far more often than from fraud. Read the Independent Administrator's commentary before writing, put the discrepancy figure next to the total so readers can judge scale, and put the question to the company and the ministry with enough time to answer.

🌍 NGO, humanitarian and human rights

For advocacy, budget monitoring and community-level accountability work, this is the instrument the sector campaigned for and it repays close use. National coalitions use the reconciliation to check whether subnational transfers that legally follow production actually arrived in the districts where extraction happens, which is frequently the most consequential finding in a whole report and the one closest to affected communities. The licence register supports free, prior and informed consent work by establishing who holds rights over which land before a company arrives to claim otherwise. The Validation process is itself an advocacy surface: civil society participation is a formal requirement, and documenting its absence has real consequences for a government's standing in the initiative. Two cautions. Participation on a multi-stakeholder group carries risk in restrictive environments and that risk belongs to the person taking it, not to the organisation citing their work. And the initiative's requirements can absorb energy that would be better spent on national law; disclosure is a means, and a country can be a model implementer and still capture its resource rents.

🎓 University and research

The corpus supports a substantial literature in political economy, public finance and development studies, and it is unusually well suited to it: it is longitudinal, cross-national, publicly documented and generated by a process whose rules are written down. The productive research programmes have been the resource curse and whether transparency mitigates it, the political determinants of joining and leaving the initiative, the effect of disclosure on subsequent revenue collection, and the reliability of the reconciliation itself as a measurement instrument. The methodological traps are severe enough to require explicit handling. Membership is self-selected, so any comparison of implementing and non-implementing countries is confounded by whatever caused the country to join. Materiality thresholds vary, so national totals are not directly comparable. The Standard has been revised, so a panel dataset has structural breaks that must be modelled rather than smoothed. And the summary data drop the scope and methodology notes that determine what each figure means, which makes the reports rather than the extracts the correct unit of analysis for anything careful.

Playbook: working EITI (Extractive Transparency) end to end

A repeatable sequence from first pull to finished product. Each phase states what you are trying to establish, not merely what to click — the objective is a defensible chain of reasoning, not a completed checklist.

Phase 1 — Establish whether the country is in the system at all

Before any analysis, determine the country's status and history: never implemented, implementing, suspended, or withdrawn, and when each transition happened. This single fact governs how much of your question the source can answer and is itself a finding. A country that withdrew after an adverse assessment has given you a dated signal about its willingness to disclose.

Phase 2 — Read the scope before the numbers

Open the Independent Administrator's report for the period you care about and find the materiality threshold, the list of in-scope companies and revenue streams, and the exclusions. Everything you subsequently say about totals is conditional on that scope. Analysts who skip this step routinely publish national figures that exclude the state-owned enterprise or the entire artisanal sector.

Phase 3 — Build the licence picture first

Extract the licence register – identifiers, holders, dates, commodities, coordinates where present – before touching the payments. The register is the entity backbone. Payments attach to it, and a payment you cannot attach to a licence is either a corporate-level tax or a scope gap you need to understand.

Phase 4 — Resolve the companies to real legal persons

Take every licence holder and reporting entity through corporate registry resolution. You are looking for the incorporation date, jurisdiction, directors and any parent. The finding that matters is the mismatch: a company incorporated shortly before the licence award, registered in a secrecy jurisdiction, or with no traceable registration at all.

Phase 5 — Test the beneficial ownership layer

Where declarations exist, treat each as a hypothesis. Screen the named person against PEP and sanctions data, look for them in procurement and asset declaration records, and check whether the declaration changed between cycles. Where declarations do not exist, record their absence explicitly – a country that has not implemented the requirement is a different situation from a company that declined to declare.

Phase 6 — Work the discrepancies

Rank the reconciliation by unexplained difference, in absolute terms and as a proportion of the stream. Read the Administrator's explanation for each large one. Most will be timing, classification or exchange rate. The ones that survive that filter – a payment one side does not recognise at all, a state-owned enterprise that cannot account for a transfer – are the leads.

Phase 7 — Look for the entities that should be there and are not

Compare the licence register against the reporting company list. A licence holder that is not in the reconciliation, an operator active in the country's export statistics that appears nowhere, or a company that reported in one cycle and vanished in the next are all questions. Absence has to be explained by scope or by something else.

Phase 8 — Place the licences on the ground

Geolocate concessions and overlay them against conflict incidents, protected areas, forest loss alerts and mapped artisanal mining sites. Formal and informal extraction frequently occupy the same ground, and a licence area that shows industrial-scale disturbance without corresponding declared production is a specific, checkable finding.

Phase 9 — Follow the money outward, not just downward

Take the largest payers and trace their ownership up to a group parent, then look for the group's marketing, trading and offshore entities. The fiscal outcome in the producer state is determined by transactions that occur outside it, and the reconciliation only sees the last leg.

Phase 10 — Corroborate against independent measurement

Compare declared production and export volumes against customs data from importing countries, commodity trade statistics and, where relevant, satellite observation of the site. Systematic mirror-trade gaps between what a country reports exporting and what its partners report importing are one of the strongest available indicators of undeclared flows.

Phase 11 — Write the caveats into the finding

Every conclusion should carry the scope threshold, the reporting period, the Validation status of the country and the fact that beneficial ownership is self-declared. A finding stated without these will be dismantled by the first competent respondent, and correctly so.

Phase 12 — Set the watch and re-check

Register the licences, companies and owners you care about in watchlist.php and configure alerts for the next publication cycle and for any change in the country's status. The most valuable EITI findings are longitudinal – the owner who changed, the payment that stopped, the company that stopped reporting – and they only appear if you are comparing cycles deliberately.

The platform ships this as a step-checked workflow in playbooks.php, so progress is recorded against a case rather than held in someone’s head.

What to pair it with

No single source carries a finding. These are the datasets that corroborate, extend or contradict this one — and a source that contradicts is worth more than one that agrees, because it is the only thing that will tell you when you are wrong.

Source	Relationship	What it adds
Open Ownership	extends	Aggregates and standardises beneficial ownership data across jurisdictions, giving you a structure and a data standard for testing EITI declarations against other registers.
OpenCorporates	prerequisite	Company registration, directors and jurisdiction. Resolving a licence holder to a real legal entity is a precondition for almost every EITI analysis.
Natural Resource Governance Institute	corroborates	Independent assessment of resource governance quality, contract analysis and fiscal regime research. Provides the interpretive frame that raw reconciliation figures lack.
Publish What You Pay	corroborates	The civil society coalition behind the initiative. National chapters frequently document what the official process omitted and are the fastest route to informed local critique.
Global Witness	extends	Investigative documentation of resource-linked corruption, frequently starting where the official reconciliation stops. Often the source of the allegation that EITI data is then used to test.
IPIS mining site mapping	contradicts	Maps artisanal and armed-group-controlled extraction that formal reconciliation cannot see, and frequently shows activity in areas the official picture treats as inactive.
Global Forest Watch	corroborates	Independent satellite observation of land disturbance within and around concession boundaries, giving a physical check on declared activity levels.
UN Comtrade and national customs statistics	corroborates	Mirror-trade comparison between declared exports and partner-country imports is the standard external test for undeclared extractive flows.
Sanctions and PEP screening sources	prerequisite	Beneficial owner and licence holder screening is the step that turns a payment table into a compliance finding.

Legal, ethical and operational constraints

The disclosures are official public information published with the intent that they be used, so collection and analysis carry no meaningful legal risk in themselves. The constraints attach to what you do next. Beneficial ownership declarations are personal data about identified natural persons, and combining them with PEP screening, adverse media and other datasets creates a profile that in most jurisdictions triggers data protection obligations – a lawful basis, purpose limitation, retention limits and, in some regimes, rights of access and correction that apply even to a person you are investigating. Publication carries defamation exposure that is not evenly distributed: several implementing countries and several jurisdictions where the relevant companies are incorporated have claimant-friendly libel regimes and an established pattern of strategic litigation against public participation, and a discrepancy reported as evidence of theft when it was in fact a timing difference is exactly the error such a claim is built on. Where your work identifies a possible predicate offence, most regulated organisations have reporting obligations that are triggered by suspicion, not by proof, and those obligations are jurisdiction-specific. Finally, the people who make this source possible in restrictive countries – civil society members of national multi-stakeholder groups – are exposed. Citing their published work is legitimate; identifying an individual as your source of a critical interpretation, in a country where that carries risk, is not something to do casually.

Operational security

Requests to the initiative's own site and interface are ordinary web traffic to an international NGO and reveal little beyond your interest in a particular country. The exposure is at the national layer. Retrieving licence registers, contract disclosures and beneficial ownership registers from a producer state's own implementation site means connecting to infrastructure operated by, or on behalf of, the government you may be investigating, and in several implementing countries that infrastructure is monitored. Access from an attributable corporate range, at volume, focused on a small number of concessions, is a legible pattern. Use general-purpose infrastructure for national-layer collection, spread retrieval over time, and take a full copy on first contact rather than returning repeatedly to the same records. Be aware that some national registers log queries and that a search for a specific individual's ownership interests can be visible to the register operator. The safest posture for a sensitive case is to mirror the entire relevant dataset once and do all subsequent analysis locally, so that the pattern of your interest never appears in anyone's logs.

Two rules that hold regardless of jurisdiction. Collection that is lawful is not automatically proportionate, and a dataset assembled for one purpose does not carry consent for another. Where the records concern identifiable people, the question is not only whether you may hold the data but whether holding it serves the purpose you are accountable for.

Is it earning its place?

Sources accumulate. Feeds get added during an incident and are never reviewed again, and a decade later the pipeline is carrying dead weight that nobody dares remove. These are the measures that show whether EITI (Extractive Transparency) is contributing anything, and they are worth baselining now so the answer is available later.

Proportion of licence holders in your area of interest that resolve to a corporate registry entity with an identifiable parent, tracked over time. A falling ratio means your resolution pipeline is degrading or the register is deteriorating.
Number of reconciliation discrepancies above your materiality threshold that survive the Administrator's own explanation, per country per cycle. This is the source's actual yield of investigative leads.
Count of beneficial ownership declarations that changed between consecutive cycles for licences you are watching, which is one of the few forward-looking signals the source produces.
Publication lag in months between the end of a reporting period and the appearance of the report, per country, as an indicator of implementation health independent of the formal assessment.
Coverage ratio of your target countries: how many of the states relevant to your mission are implementing, suspended or absent, and how that has changed year on year.
Number of cases in cases.php where an EITI record supplied the licence, entity or payment fact that anchored the case, as opposed to being cited as background.
Rate at which EITI-derived entities are subsequently corroborated or contradicted by corporate registry, sanctions or leaks data, which tells you whether you are over-trusting self-declaration.
Time from publication of a new country report to its appearance in your platform, which should be measured in days and is the only operational metric this source needs.

Beware of volume. Indicator counts rise easily and say almost nothing. Unique contribution — findings this source produced that no other source in your stack would have — is the measure that matters, and it is usually far lower than anyone expects.

Tradecraft notes

The distinctions that separate a competent analyst from a fast one:

Read the scope section before the numbers, every time, without exception. The materiality threshold and the list of excluded entities determine what every figure in the report means, and they change between countries and between cycles of the same country.
A discrepancy is a question, not an answer. Timing, exchange rates and classification account for most of them. Learn the boring explanations well enough to eliminate them quickly, because the leads are what is left.
Never compare national totals across countries without checking whether both included the state-owned enterprise, both valued in-kind revenue the same way, and both used the same fiscal year. Two of the three usually differ.
Treat the licence register as the spine and the payments as attributes. Analysts who start from the payment tables end up with company-level aggregates that cannot be tied to any specific asset and therefore cannot be checked against the ground.
A beneficial ownership declaration that names a plausible local businessman with no other public footprint is a finding, not a resolution. The absence of a corroborating trace is itself informative.
Watch the transitions. A company entering or leaving the reporting list, a licence changing hands, an owner changing between cycles: the changes carry far more signal than any single-cycle snapshot.
The initiative's own governance record is data. Suspensions, downgrades and withdrawals are dated, public, and directly relevant to how much weight the country's disclosures deserve.
Where a country reports in local currency and you need a series, do the conversion yourself with a documented rate source and state the method. Published trend claims built on unadjusted local-currency totals are frequently wrong by a factor rather than a margin.
Keep the source PDF, its hash and the page reference next to every extracted figure. When a finding is challenged – and resource-corruption findings are challenged – the structured extract will not defend you and the report will.

Questions analysts actually ask

Can I use EITI figures to prove that money was stolen?

No, and attempting to is the classic misuse. The reconciliation shows what two parties reported and where they disagreed. A discrepancy establishes that the accounts do not tie, which is a reason to investigate. Proof of diversion requires the underlying records, which you obtain through legal process, not from a published report.

Why does a company I know operates in the country not appear?

Most likely it falls below the materiality threshold the national multi-stakeholder group set, or the commodity or licence type is out of scope. Check the scope section first. If it is in scope and still absent, that is a genuine question for the national secretariat and a legitimate line of inquiry.

How current is the data?

Not current. Reports typically cover a fiscal year and appear one to three years later. Use it for structure and history. For anything happening now, go to the national cadastre, company announcements, trade data and satellite observation, and use EITI to establish what normal looked like.

Is the beneficial ownership information reliable?

It is a self-declaration by the company, usually unverified, in a system whose enforcement mechanisms are weak. Treat every declaration as a testable claim. Its value is highest when it names someone unexpected and when it changes, and it is not a substitute for independent ownership research.

A country was suspended. Is its historical data still usable?

Yes, with the suspension recorded alongside it. Data published before a suspension was produced under the same process as any other report and does not become false. What the suspension tells you is that the conditions supporting that process subsequently failed, which is relevant to how much weight later or absent reporting deserves.

Can I compare one country's revenue transparency with another's using the published totals?

Only after normalising scope, in-kind valuation method, fiscal year and currency. In practice the direct comparison is almost never valid as published. What is comparable is the process: whether both countries reconciled the same categories, disclosed the same registers and reached the same assessment outcome.

How does this relate to conflict minerals due diligence?

It is the formal-sector complement. EITI documents licensed industrial extraction and the payments it generates. Conflict minerals frameworks address the artisanal and small-scale chains where armed group financing actually occurs, and those chains are largely invisible here. Use both, and never treat a clean reconciliation as supply chain assurance.

What is the single most useful field?

The licence identifier, because it is the join key. It connects a payment to a physical concession, and a concession to coordinates, ownership history, environmental observation and conflict data. Payment totals without a licence anchor cannot be checked against anything on the ground.

Does the platform generate any of these entities or relationships with a model?

No. Companies, licences, payments and declared owners are ingested from the published disclosures as they stand. The only language model involvement anywhere in the platform is the Summarise skill in copilot.php, which writes prose about records that already exist and creates no indicator, relationship or attribution of its own.

Standards, formats and interoperability

What this source speaks natively, and what it has to be translated into before a partner can consume it. Work that arrives in a recognised format is easier to defend, easier to hand over and easier to automate against:

The EITI Standard itself, whose requirements define what must be disclosed and which has been revised more than once – always check which version governed the report you are reading.
Open Contracting Data Standard, the natural target schema when EITI licence and contract disclosure is combined with public procurement data.
The Beneficial Ownership Data Standard, used by Open Ownership and others to make ownership declarations from different registers comparable.
Legal Entity Identifier, where licence holders have one, giving a stable global key that survives renaming and transliteration.
OECD Guidelines for Multinational Enterprises and the associated due diligence guidance, which set the expectations most large operators claim to meet.
UN Convention against Corruption, the framework under which most asset recovery and mutual legal assistance in this space proceeds.
STIX 2.1 and MISP as the platform's export formats, with EITI-derived companies, licences and persons mapped to identity and location objects for sharing.
National extractive cadastre schemas, which are the authoritative licence records and which the EITI register in a report is derived from rather than equal to.

References

Primary documentation and authoritative references for this source. Publishers revise and retire material, so treat the retrieval date as part of the citation and re-check before relying on any of it in a formal product.

Extractive Industries Transparency Initiative — EITI International Secretariat. The primary site: the Standard, country pages, Validation outcomes, reports and data. Start with the current version of the Standard before reading any country report.
EITI data interface — EITI International Secretariat. The programmatic route to structured country, company and revenue data. Read the current documentation rather than assuming an older endpoint shape.
Natural Resource Governance Institute — NRGI. Independent research on fiscal regimes, state-owned enterprises and contract terms. The best available interpretive layer over raw EITI figures.
Publish What You Pay — PWYP. The civil society coalition that campaigned the initiative into existence. National chapters publish the critique that official reports do not contain.
Open Ownership — Open Ownership. Beneficial ownership data standards, register implementation guidance and cross-jurisdiction data. Essential for testing EITI ownership declarations.
OpenCorporates — OpenCorporates. Company registration data across jurisdictions. The first step in resolving any licence holder to a real legal entity.
OpenSanctions — OpenSanctions. Consolidated sanctions and PEP data for screening licence holders and declared beneficial owners.
Global Witness — Global Witness. Investigative reporting on resource corruption, much of it built on or against official transparency disclosures. Useful for seeing how the data is used in practice.
Open Government Partnership — OGP. The wider transparency commitment framework that many implementing countries also participate in, and a source of national commitments on registers and contract disclosure.
UN Comtrade — United Nations Statistics Division. Bilateral trade statistics for mirror-trade analysis, the standard external check on declared extractive export volumes.

Link integrity: every reference above was verified with a live request when this page was generated. Where a publisher had moved or withdrawn a document, the link was repointed at a preserved copy in the Internet Archive and marked as archived. Anything with no reachable copy anywhere had its link removed rather than left to rot — the source is still credited, it simply cannot be linked.

Put it into practice

The Quantus Intel threat intelligence platform operationalises this source: EITI licence registers, payment reconciliations and beneficial ownership declarations are ingested as resolved company, person and location entities, screened against sanctions and PEP data, and placed on the same map as conflict incidents, mining sites and forest loss.. Browse the full source catalogue, or follow any tag above into the rest of the library.