ESG in Mining
Ratings Standards and Best Practices

Credit rating agencies, evaluating comparable complexity in a different domain, correlate above 0.95. A number like 0.38 means that if MSCI gives a mining company a top-quartile score, Sustainalytics may place the same company in the bottom quartile, using the same publicly available data, in the same fiscal year. For mining, where the rated activity involves reshaping land, redistributing water, and generating chemical liabilities that will persist for centuries after the company ceases to exist, this level of inter-rater disagreement has consequences that extend well beyond portfolio analytics.

The disagreement is not random. It clusters. Water management is one cluster. MSCI evaluates it through financial risk. Sustainalytics evaluates it through unmanaged exposure relative to demonstrated management capacity. S&P Global's Corporate Sustainability Assessment evaluates it through a self-reported annual questionnaire scored against a rubric with keyword sensitivities that an entire consulting industry has learned to optimize. Indigenous rights weighting is another cluster. Scope 3 emissions boundary definitions are a third.

A copper mine in northern Chile drawing water from an aquifer shared with indigenous agricultural communities receives three different water scores from three agencies. The scores may not even agree on direction.

MSCI uses industry-specific key issues weighted by expected financial impact. The model serves investors. It was designed for them. Community health outcomes downstream of a mine do not register in the MSCI framework except to the extent they produce litigation, regulatory intervention, or reputational damage that threatens cash flows. If a contaminated aquifer has not yet generated a lawsuit and the regulatory environment makes a lawsuit unlikely, the contamination is not a risk in the model.

MSCI also has a momentum component. Improvement gets rewarded. Stable high performance does not.

Vale's post-Brumadinho rating trajectory is where this becomes concrete in a way that stays with you. January 25, 2019, Dam I at the Córrego do Feijão mine collapsed. 12 million cubic meters of tailings. 270 dead, many in the employee cafeteria, built in the dam's inundation zone, below a structure whose stability TÜV SÜD had certified. MSCI downgraded Vale to B. Then Vale started decommissioning upstream dams, expanding monitoring, hiring safety staff. The rating recovered. By 2022, upgraded.

There is a lot to say about what happened at Brumadinho, and most ESG analysis says the same things about it: the failure of assurance, the gap between disclosure and reality. What gets discussed less, because it is harder to fit into a narrative about fixing the system, is what the MSCI trajectory means for a competitor. Take a company, any company, that never built upstream tailings dams, that spent the extra capital on dry stacking or filtered tailings from the start, that therefore has an impeccable tailings safety record not because it recovered from disaster but because it engineered the disaster out of its operations from the beginning. That company sat at whatever MSCI rating its quiet, undramatic competence earned, year after year, no momentum bonus, no trajectory story, no dramatic improvement arc for analysts to note. Meanwhile Vale, starting from catastrophe and trending upward, moved through the scoring system faster. The methodology cannot see what did not happen. Prevention is invisible to a momentum-based model. Recovery is highly visible. The incentive structure this creates is something that mining ESG professionals discuss among themselves with a frankness that does not appear in published analysis.

Four years before Brumadinho, in 2015, the Samarco dam had failed. Samarco: a joint venture between Vale and BHP. 19 dead. 43.7 million cubic meters of waste into the Rio Doce. BHP held top-tier ESG ratings throughout. Two catastrophic tailings failures linked to the same parent company in four years, and the ESG ratings architecture absorbed both events and kept going. Agencies adjusted the weighting of tailings management indicators. They did not revisit whether a system built on self-reported data, scored by algorithms calibrated to financial materiality, and validated by third-party auditors whose clients are the companies being audited, was structurally capable of detecting the kind of risk that kills 270 people eating lunch.

Sustainalytics works on a different axis. The risk rating measures unmanaged ESG exposure, lower is better. There is a controversy overlay that can override quantitative scores when negative media coverage reaches a certain volume. The controversy mechanism responds to media events, NGO campaigns, legal proceedings. It does not respond to environmental damage that no one covers. A mine in a remote part of Papua New Guinea discharging waste into a river system, with measurable downstream contamination and no international journalist within 500 kilometers, will carry a different controversy profile than a mine in Chile doing less damage with an active Greenpeace campaign documenting it. The controversy score for these two operations reflects the geography of journalism, not the geography of harm.

S&P Global runs the CSA that feeds the Dow Jones Sustainability Index. Mining companies face a detailed annual questionnaire with sector-specific modules. The CSA has generated an optimization industry, and the optimization industry deserves a longer description than it usually gets because it reveals something about how self-reported scoring systems behave in practice when the stakes are high enough.

The consultancies that help mining companies with their CSA submissions are staffed by people who used to work in sustainability departments at mining companies or at S&P Global itself. They know the rubric. They know which keywords trigger higher scores in specific question categories. They know the evidence formatting that evaluators prefer. They know the difference between a response that earns 80% of available points and one that earns 95%, and the difference often lies not in the underlying performance data, which may be identical, but in how the data is structured, contextualized, and presented. Two mines with the same water recycling rate, the same LTIFR, the same community investment spend, the same closure planning maturity, can receive significantly different CSA scores depending on whether the submission was prepared by an internal team working from the questionnaire text or by a specialist consultancy working from the scoring rubric. Anglo American has consistently ranked at the top of the mining sector in DJSI. Anglo American invests in its CSA process at a level commensurate with its ambition to remain at the top. Whether that ranking primarily reflects Anglo American's operational sustainability performance or its disclosure optimization capacity is a question that the CSA methodology cannot answer from within itself, and that DJSI users almost never ask, because asking it would undermine the utility of the index.

Glencore. The company pleaded guilty in 2022 to bribery and market manipulation charges in the United States and paid over $1.1 billion in combined penalties across U.S., U.K., and Brazilian proceedings. Glencore is an ICMM member. Glencore's board independence metrics, committee structure scores, and risk oversight assessments are adequate or favorable under several ratings methodologies. The board structure did not prevent the criminal conduct. The ratings measure the structure. They do not measure what the structure failed to prevent. These are different quantities, and the ratings capture one of them.

GRI, SASB, TCFD, ICMM, IRMA, EITI.

GRI measures disclosure. The Mining and Metals Sector Supplement adds industry-specific indicators. Newmont has won reporting awards for its GRI-aligned sustainability reports. The reports are well-written, comprehensive, data-rich documents produced by a large and skilled sustainability team. They describe Newmont's sustainability approach. They do not and cannot substitute for independent verification of conditions at each Newmont operation.

SASB, now under the ISSB, is where the quantitative metrics live. Tonnes of CO₂, cubic meters of water, lost-time injury frequency rates. The quantitative insistence is the most useful feature of any framework in the mining ESG ecosystem, because it creates comparability and because building the data infrastructure to report these numbers tends to improve operational awareness even when the initial motivation is purely external.

LTIFR needs its own space here because it is the most reported safety number in mining, the most referenced by ESG ratings, and the most susceptible to management in ways that do not involve changing the underlying safety conditions.

A worker in a processing plant fractures two fingers. The site medical officer examines the injury. The medical officer is employed by the mining company or by a medical services contractor whose contract renewal is influenced by the site general manager. The site has a corporate LTIFR target linked to the management team's annual bonus. Two legitimate classification options exist. One: lost-time injury, the worker goes home, the LTIFR goes up, the bonus target gets harder to hit. Two: modified duties, the worker comes to site with a splinted hand and answers phones for two weeks, no time lost from work, the injury is a medical treatment case, the LTIFR stays flat, the bonus remains within reach.

The reporting standards accommodate both classifications. The judgment call about whether modified duties constitute meaningful work sits with the medical officer, in a room, at a remote mine site, under institutional pressure that runs in one direction. This plays out at thousands of sites every year, in every mining jurisdiction, across every commodity. When a company publishes an LTIFR of 0.4 and a competitor publishes 1.6, the numbers may reflect genuinely different safety environments or they may reflect genuinely different classification cultures, and from the outside, from the desk of a ratings analyst inputting the number into a model, there is no way to tell.

Total recordable injury frequency rate is harder to game because it includes medical treatment cases. High-potential incident frequency, which counts near-misses that could have been fatal, tells a different story entirely: a mine with low LTIFR and high HPIF is a mine where the gap between what happens and what almost happened is wide. Neither metric has displaced LTIFR in the ESG ratings ecosystem. LTIFR has decades of benchmarking history. Replacing it would break the time series. The industry and the ratings agencies are locked together into a metric that everyone close to mine-site safety management knows is unreliable as a standalone indicator, and the lock is maintained by institutional inertia on both sides.

Water recycling. A flotation circuit in a copper concentrator recirculates process water. Each pass counts as recycled. A mine can report 85% recycling while drawing heavily from an external freshwater source and discharging water with elevated metals concentrations. The number describes the efficiency of an internal loop. It says nothing about the mine's net effect on the surrounding watershed.

The metric that would describe that effect is net fresh water consumption per tonne of ore processed, benchmarked against the catchment's renewable water budget. Generating this number requires a catchment-level hydrological model built from years of monitoring data across boreholes, rivers, weather stations, and competing water users spanning an entire drainage basin. Most mines have not built these models. Most ESG frameworks do not require them.

ICMM validates at the corporate level. One assessment per company. A diversified miner with operations on six continents gets one validation that covers all of them. Glencore's ICMM membership coexists with the criminal proceedings and the operational controversies because ICMM evaluates corporate management systems and the proceedings address conduct at specific operations in specific countries. For a community living next to a specific operation that is the subject of environmental complaints, the fact that the parent company has satisfied ICMM's corporate-level expectations is not actionable information.

IRMA assesses at the mine-site level. The standard was developed by a governance body including NGOs, indigenous peoples, and labor representatives. Results are published, including findings of non-compliance. The four-tier achievement scale (Transparency, 50%, 75%, 100%) allows incremental progress. The number of fully assessed mines is small. The assessment is expensive, time-consuming, and transparent in its outcomes. A company that is confident in site-level conditions has an incentive to participate. A company that is not confident in site-level conditions has an incentive to avoid the process. The pattern of participation and avoidance, viewed across the industry, traces the boundary between companies whose corporate ESG narratives are backed by site-level performance and companies where a gap may exist between the two.

IRMA's future influence depends on whether downstream purchasers of mined commodities, particularly the automotive and electronics sectors that have begun incorporating responsible sourcing requirements, create enough market demand for IRMA-verified material to push reluctant operators into the assessment. Without that demand-side pull, participation will remain limited to self-selecting companies whose site-level performance is strong enough to withstand independent scrutiny.

EITI promotes payment transparency between extractive companies and governments. The standard matters most in jurisdictions where the gap between what a government receives from a mining company and what the public knows about those receipts creates space for corruption in the licensing and permitting process. Companies that publish disaggregated payment data by project and jurisdiction in non-EITI countries, without being required to, are doing something that costs nothing financially and signals something about governance posture. Most companies do not do it.

This is the section that determines whether the rest of the article matters or is an exercise in rearranging deck chairs, so it is going to take up more space than the others.

Single materiality: ESG factors matter insofar as they affect enterprise value. Double materiality: ESG factors matter insofar as they affect enterprise value, and also insofar as the company's operations affect people and the environment, even when that effect does not circle back to the company's balance sheet within any foreseeable period.

The ISSB uses single materiality. The EU's CSRD and ESRS use double materiality. Most mining ESG ratings currently operate on single materiality because most ratings agencies serve investor audiences and the ISSB framework was designed for capital markets.

For banks, software companies, retailers, the practical difference between the two materiality definitions is often small. Most of the externalities these businesses generate either feed back into financial risk within a few years (regulatory fines, consumer boycotts, employee attrition) or are diffuse enough that attribution to a specific company is difficult. Mining breaks this pattern. Mining creates concentrated, attributable, geographically specific externalities that persist on timescales measured in decades and centuries. The gap between what single materiality captures and what double materiality captures is wider in mining than in any other rated industry.

The Witwatersrand. South Africa. Over 120 years of gold mining. The geology contains pyrite, iron sulfide, distributed through the gold-bearing reefs. Mining exposed the pyrite to water and oxygen. The chemical reaction is straightforward and relentless: iron sulfide plus water plus oxygen produces sulfuric acid. The acid dissolves heavy metals, arsenic, manganese, uranium, from the surrounding rock. The resulting contaminated acidic water, acid mine drainage, flows into the groundwater and surface water of the Gauteng province. Gauteng is the most densely populated, most economically productive province in South Africa. Johannesburg is in Gauteng.

The acid mine drainage does not stop when the mines close. It continues as long as sulfide-bearing rock is exposed to water and air, which, given the scale of excavation in the Witwatersrand, means indefinitely in any time frame relevant to human civilization. The water treatment required to keep the drainage from overwhelming Gauteng's water supply is continuous and must continue in perpetuity. The South African government funds this treatment. The mining companies that created the sulfide exposure over the course of a century are mostly gone. Some failed. Some were acquired and the acquiring companies structured transactions to limit legacy environmental liability exposure. Some restructured in ways that separated continuing profitable operations from depleting, liability-laden ones. The technique is well understood in mining finance: create a subsidiary, park the aging mine and its obligations in it, let the subsidiary fail when the obligations come due.

Single materiality, applied at any point during the operating life of those mines, would have classified the acid drainage as financially immaterial to the operator. The regulatory environment did not compel remediation spending. Enforcement was minimal. The full cost would materialize decades or centuries after the mine closed, long past any investment horizon. The operating companies extracted the gold, paid dividends, and ceased to exist in their original form. The materiality assessment was correct at each moment in time, within its own framework. The liability transferred from private capital to the public.

Double materiality would have captured the acid drainage from the moment the sulfide exposure became foreseeable, because the framework registers outward environmental impact independent of whether the company ever bears the cost. Same mines, same chemistry, same centuries of contamination, opposite materiality conclusions depending on which framework applies.

This is not confined to historical cases. The same dynamic operates today, at operating mines, in jurisdictions where environmental enforcement is under-resourced or captured. A mine operating under a legally valid permit in a country where the environmental regulator lacks the capacity or political independence to enforce remediation requirements can create century-scale contamination that single materiality frameworks classify as a managed regulatory risk. The contamination is managed in the model. It is not managed in the water.

Barrick Gold's Porgera mine in Papua New Guinea discharged mine waste directly into a river system under a government-approved riverine tailings disposal permit. The permit was legally valid. Under single materiality, the permit manages the regulatory risk component. Community health impacts and ecological damage downstream are external to the enterprise value calculation unless they generate litigation that reaches a jurisdiction with effective courts, or unless media coverage reaches the threshold that triggers a Sustainalytics controversy overlay. Under double materiality, the downstream effects are material when they occur, because the framework does not condition materiality on the probability that the company will be held financially accountable.

Over 60,000 abandoned mine features documented in Australian government inventories alone. The global number is poorly catalogued, which tells you something about how the mining industry has historically prioritized post-closure responsibility relative to pre-production investment.

The subsidiary dissolution technique for escaping closure obligations works like this. A parent company holds a mine through a subsidiary. The mine ages. Production declines. Closure obligations grow. The parent sells the subsidiary to a small operator with limited capital, or allows the subsidiary to enter administration. The closure obligations stay with the subsidiary. The subsidiary's assets, which consist of a depleting mine and a growing liability, are insufficient to cover the remediation. The parent retains the accumulated profits from the productive decades. The environment gets whatever the insolvency process leaves, typically nothing meaningful.

Financial assurance mechanisms that work are structured to survive this sequence. A trust fund or escrow account, legally separated from the parent company's balance sheet, protected from creditors in bankruptcy, transferring with the mine site in any corporate transaction rather than with the corporate entity. If the parent is sold, the fund stays with the mine. If the parent fails, the fund remains. This is expensive. Capital locked in an escrow account does not earn returns on exploration or expansion. The expense is the mechanism's credibility. A closure plan, which every significant mine produces because regulators demand it, is a document describing what the company intends to do. A bankruptcy-proof closure fund is capital committed to what will be done regardless of the company's future.

Juukan Gorge.

In May 2020, Rio Tinto detonated charges that destroyed two rock shelters in Western Australia's Pilbara region. The shelters held evidence of 46,000 years of continuous human occupation. The Puutu Kunti Kurrama and Pinikura peoples had communicated opposition to the blast. Rio Tinto held a Section 18 consent under Western Australia's Aboriginal Heritage Act 1972, obtained years earlier, and the Act contained no mechanism for withdrawal of that consent once granted. Iron ore sat beneath the shelters. The ore was in the mine plan. The production schedule called for the blast. Internal decision-making processes did not elevate the traditional owners' opposition to a level that could override the schedule. The blast went ahead.

Jean-Sébastien Jacques, CEO, departed. Simon Thompson, chairman, announced he would not seek re-election. Chris Salisbury, iron ore division head, departed. The Australian Parliament held an inquiry. Western Australia replaced the 1972 Act. The global mining industry's operational understanding of FPIC moved.

The ESG ratings that evaluated Rio Tinto before May 2020 assessed policy frameworks, board oversight structures, and reported compliance with indigenous engagement standards. All of those assessments were based on documents and descriptions that accurately represented Rio Tinto's formal policies and governance architecture. None detected that a specific decision at a specific mine in a specific week would destroy a 46,000-year-old heritage site because the decision was legal, the ore was valuable, and the internal process assigned greater weight to the production schedule than to the traditional owners' stated opposition. Formal policy evaluation and operational decision prediction are different activities, and ESG ratings do the first.

The temporal dimension of consent. A Section 18 authorization obtained in 2013 does not capture the state of the relationship between Rio Tinto and the PKKP peoples in 2020. New archaeological information had emerged in the intervening years. Community leadership had evolved. The significance of the site had become clearer. Consent is not a fixed commodity. It degrades as circumstances change. Treating it as a permanent acquisition, obtained once during the approvals phase and valid for the life of the mine, is the practice that Juukan Gorge discredited across the industry.

Periodic consent renewal, where the company returns to the affected community at defined intervals and reconfirms that the social license remains intact, is the practice that a subset of companies has adopted in response. The resistance to this practice comes from project finance, because renewable consent gives communities ongoing authority over assets with billions of dollars of committed capital, and capital markets price uncertainty negatively. The tension between the financial logic of certainty and the ethical logic of ongoing consent is unresolved and may be irresolvable in any way that satisfies both sides simultaneously.

Biodiversity offsets. Maron et al., Conservation Biology, 2016. Most offset programs fail to achieve no net loss. The reason is ecological non-fungibility. A mature forest has structural complexity developed over decades or centuries: canopy layers, understory composition, soil microbiology, mycorrhizal networks, deadwood habitats, species interactions calibrated by long coevolution. A newly planted forest has saplings in rows. The offset accounting framework counts hectares. Ecology does not work in hectares. A hectare of littoral forest on the coast of Madagascar and a hectare of inland forest in the same country are different ecosystems hosting different species assemblages adapted to different conditions. Rio Tinto's QMM ilmenite mine in southeastern Madagascar cleared littoral forest and established offset areas of a different forest type. The hectare count balances in the offset ledger. Independent ecologists have questioned whether the offset areas protect ecological values comparable to what was destroyed. The question has not been answered to the satisfaction of the critics.

This matters for ESG ratings because agencies that credit biodiversity offsets at face value are accepting an accounting framework that the underlying science does not fully support. The mitigation hierarchy, avoid, minimize, restore, offset, places offsetting last for a reason. It is the least reliable step. Companies that treat it as a routine tool rather than a last resort, and ratings agencies that score it without interrogating the ecological equivalence of the offset, are building ESG assessments on assumptions that the conservation biology literature has substantially weakened.

BHP's scope 3 emissions. Roughly 14 times scope 1 and scope 2 combined. BHP sells iron ore and metallurgical coal. Customers turn the iron ore into steel. Customers burn the coal to make the steel. The carbon emissions from those processes dwarf the emissions from digging the ore and coal out of the ground and shipping it to port. A BHP sustainability report that covers scope 1 and 2 thoroughly and treats scope 3 as an appendix item is presenting approximately 7% of the company's total carbon footprint.

For years, several major miners did exactly this, and the ESG ratings they received did not penalize the omission because scope 3 completeness was not yet a scoring criterion in most agency methodologies. That has changed. Agencies now penalize incomplete scope reporting. The quality of scope 3 disclosures ranges from detailed value chain emission models with methodology notes, uncertainty quantification, and third-party review to a single aggregate estimate with no supporting documentation. Under current rating methodologies, both qualify as having met the scope 3 reporting requirement. The difference between a rigorous scope 3 analysis and a rough estimate is not captured in the binary has-reported-scope-3 / has-not-reported-scope-3 distinction that most scoring rubrics apply.

Contractor safety. At many mine sites, contractor workers outnumber direct employees. The contractors drill, blast, haul, maintain equipment, cook food, provide security. A contractor employee who is killed at a mine site is dead in the same way a direct employee is dead. The family's experience is the same. The community's experience is the same. The investigation, if there is one, covers the same ground. The ESG reporting treatment diverges. Many mining companies report LTIFR and TRIFR for direct employees only, because most ESG reporting frameworks set the boundary at the reporting entity's own workforce, and contractor employees work for a different entity even though they work at the same site, doing the same tasks, facing the same hazards, breathing the same dust.

A company can report an LTIFR of 0.3 for its direct workforce while serious contractor injuries at the same operations go uncounted in the reported metric. From a ratings analyst's desk, the 0.3 looks excellent. From the mine site, the distinction between a direct employee and a contractor is a legal abstraction that has no physical correlate. The companies that report combined safety data for everyone on site, and that require contractors to meet the same safety management standards, training certifications, and equipment specifications applied to direct employees, bear higher costs. Contractor rates increase when contractors must comply with stricter standards. Work may slow when contractor crews require additional training. The cost is the mechanism by which the reporting boundary is aligned with the physical boundary of the operation.

The exploration phase of mining is when the pit location is chosen, the tailings facility sited, the first water abstractions begin, the first community encounters occur. These decisions happen at junior mining companies. TSX Venture Exchange listings. Pre-revenue. Below the coverage threshold of every major ESG ratings agency. Often operating in jurisdictions selected partly for mineral prospectivity and partly for permitting speed, which in practice sometimes means jurisdictions where the environmental and social assessment process is less rigorous.

By the time a project reaches the scale where its operator attracts ESG ratings coverage, the foundational decisions are physically embedded in the landscape. The pit exists. The waste facility exists. The water table has been affected. The community has formed its assessment of the operation based on years of experience with a company that faced no external ESG scrutiny during the period when the most consequential choices were being made.

Streaming and royalty companies provide financing to junior miners in exchange for future production at agreed prices. Franco-Nevada, Wheaton Precious Metals, Royal Gold. These financing agreements could include independently verified ESG conditions. A streaming contract that requires IRMA assessment before first production, or catchment-level water monitoring during exploration, or a bankruptcy-proof closure fund before ground-breaking, would insert ESG requirements at the lifecycle stage where they are most needed and least present. The competitive dynamics of the streaming sector work against unilateral adoption: if one streamer imposes ESG conditions and a competitor does not, the junior miner signs with the competitor. The shift would need to come from downstream demand, from automotive or electronics companies requiring responsibly sourced minerals and tracing their supply chains back through the streaming arrangements to the mine site, or from coordinated action across the streaming sector, neither of which has materialized at scale.

Framework consolidation is underway. ISSB absorbed SASB and TCFD. The EU is implementing CSRD. Mining jurisdictions in Chile, Australia, Brazil, and South Africa are tightening disclosure mandates.

More consequential than any framework revision is what is happening with data. Planet Labs images the entire Earth's land surface daily at 3-to-5-meter resolution. The European Space Agency's Sentinel constellation provides synthetic aperture radar data, free for anyone to download, capable of detecting ground subsidence near underground mines with millimeter precision. Deforestation at a mining concession. Expansion of a tailings storage facility. Discoloration of a water body downstream of a discharge point. Bare ground at a site that a sustainability report describes as undergoing active rehabilitation. All visible without asking the mining company to report anything.

Environmental ESG performance in mining is becoming independently verifiable for the specific dimensions that satellites and radar can observe: land disturbance, water quality indicators, tailings facility condition, rehabilitation progress. The social and governance dimensions remain largely dependent on reported data and qualitative assessment. The transition is partial. It is also irreversible. Once the satellite time-series archive exists for a mine site, it exists permanently, and the record of what the land looked like in each year of operation does not depend on what the sustainability report said about what the land looked like.

For companies that spent the last decade on operational environmental performance, on catchment-level water management and concurrent rehabilitation that shows up as vegetation recovery in multispectral imagery and filtered tailings and dry stacking, that investment will be validated by an evidence base the company does not control and cannot edit. For companies that spent the decade on sustainability report writing and CSA questionnaire optimization and carefully chosen photographs of rehabilitated areas selected from a larger portfolio of un-rehabilitated ones, the satellite record will contain the full portfolio, every year, at a resolution that distinguishes between a restored landscape and a green-washed one.