What is a mine’s total financial liability beyond its closure bond?

A mine’s total financial liability cannot be read from its reclamation bond or the annual premium for that bond. The real picture combines the current closure obligation, the discounted accounting provision, regulatory security, collateral locked behind that security and separate commitments made to communities, governments and licence holders.

Peabody Energy’s 2025 accounts contain a number pairing that looks wrong until you ask the right question.

The United States coal producer reported US$878.6 million of reclamation bonding requirements. Its United States asset-retirement obligations, the accounting estimate of future retirement work, were US$476.4 million. Its total group asset-retirement obligations were US$754.9 million.

The bond was larger than the provision.

That is not a typo. It is a warning against one of mining finance’s most persistent shortcuts: treating a closure bond as though it were the mine’s total liability.

Peabody explained the difference plainly in its 2025 Form 10-K. The bonds assume reclamation starts now. The accounting liability is discounted from expected future closure dates. One figure asks what an outside party may need to do today. The other asks what future expenditure is worth in today’s money.

Now take Ranger in Australia, where the implications are harsher. At 31 December 2025, Energy Resources of Australia, the company responsible for rehabilitating the former Ranger uranium mine, held a A$2.301 billion rehabilitation provision. Government-held security was A$561 million in the Ranger Rehabilitation Trust Fund and A$125 million in bank guarantees.

There is no single number called “the answer”.

ERA spent A$221 million during 2025 and said further funding would probably be needed by the third quarter of 2027. Activities after 2027 remained uncertain. Rio Tinto funded ERA’s A$766.5 million equity raising in 2024 and assumed management of the rehabilitation programme. Put those facts beside the security and provision figures, and the operating decision becomes clearer: the historic security balance did not eliminate the need for new capital or active oversight.

That is the concise comparison investors need. Peabody’s reconciliation explains why a bond can exceed a provision without implying a larger economic burden. Ranger’s reconciliation shows why security far below a provision can sit alongside an urgent need for funding.

Ranger demonstrates that government-held security may cover only part of a rehabilitation estimate that changes with work scope and monitoring needs. Photo: Kgbo / Wikimedia Commons, CC BY-SA 4.0.

The simple question is this: when someone says a mine has a A$100 million bond, what exactly do they mean?

Usually, less than they think.

GI Network’s view: A closure bond is an instrument of security, not a verdict on total liability. The serious question is whether the full obligation has been costed, refreshed, funded and reconciled against the available security.

The number everyone reaches for

A reclamation bond, bank guarantee, trust or letter of credit gives a regulator financial recourse if an operator fails to perform. United States federal guidance says an operator must indemnify the surety, the company standing behind a bond, for any amount it pays. Some sureties also require partial or full collateral.

Think of it as a guarantor on a tenancy. The guarantor gives the landlord somewhere to turn if the tenant defaults. That does not reveal the full repair bill. It does not reveal the tenant’s other promises. And it does not show how much of the tenant’s cash has been tied up to persuade the guarantor to participate.

The premium is further removed still. It is the annual price of obtaining security, normally calculated as a percentage of the bond’s maximum payable amount. It is not insurance against the operator’s own closure cost.

This is where most people stop looking. A bond amount feels concrete, which makes it tempting to treat it as the answer. It is only one line of the answer.

Under IAS 37, a provision is a liability with uncertain timing or amount, measured with risks and uncertainties in mind and discounted to present value when time matters. IFRIC 1 requires remeasurement when expected costs, timing or discount rates change. The accounting number can therefore be lower than today’s cash cost because the work is expected later. A regulator’s security requirement may instead assume a third party has to close the site now.

That is not a contradiction. It is a timing problem wearing a finance hat.

Marikana. Photo: JMK / Wikimedia Commons, CC BY-SA 3.0.

Why investors do not see one number

Imagine you are deciding what a mine is worth. A large closure provision can reduce the value you place on its future cash generation. Restricted cash behind a guarantee is cash that cannot freely fund operations, distributions or expansion. A requirement to replace or increase security can reduce financing capacity just when the business needs it most.

That is how closure risk moves through an investment decision.

It affects valuation because future closure spending reduces the cash an owner expects to receive. It affects returns because a larger or earlier payment leaves less cash for investors. It affects covenants, the promises made to lenders, because lenders care whether cash is restricted and whether a new obligation could strain available funding. It affects debt capacity because a lender cannot sensibly treat cash pledged behind security as freely available.

Peabody reported approximately US$740 million of restricted cash and other balances supporting reclamation requirements. That figure matters as much as the bond itself. It shows that financial assurance is not merely an environmental compliance issue. It can be a capital-allocation issue.

The US Government Accountability Office found that three major US coal companies entered bankruptcy in 2015 and 2016 carrying about US$2 billion of self-bonds, unsecured promises from the companies themselves. A corporate promise can appear sufficient until the company’s distress is exactly the reason the regulator needs to call it.

For operators, this is the distinction behind funding the proof rather than the fantasy. Start with the work that must be done, not the friendliest headline number.

The costs that hide in plain sight

Sibanye-Stillwater, the mining company reporting on its Marikana operations in South Africa, offers a useful corrective to the idea that closure means earthmoving and demolition.

Its December 2025 estimate for an unscheduled closure at Marikana was R2.8 billion, undiscounted and updated annually. Mining areas accounted for R1.252 billion. Infrastructure accounted for R675.9 million.

Then come the lines that are often waved away as detail. Preliminaries, general costs and contingencies were R331.6 million, 12% of the total. Monitoring and aftercare were R231.8 million, 8.2%.

More than one-fifth of the estimate sat outside the obvious physical works.

South Africa’s financial-provision framework requires annual rehabilitation plans to update both 12-month and final-closure costs. This is not bureaucratic fussiness. It recognises that an immediate closure and an orderly end-of-mine-life closure can require different assumptions, prices and plans.

What surprised us was how consistently this point reappears across the cases. The apparent disagreement between bond and provision often has less to do with bad arithmetic than with different definitions of the event being funded.

Ranger’s moving target

ERA’s A$2.301 billion Ranger provision used a 2.5% real discount rate. A one-percentage-point decrease in that rate would have increased the provision by A$111 million.

That sensitivity gives the reader a useful mental test. The site may not physically change overnight, but the measured liability can move materially when the assumed timing of the work changes.

Rio Tinto’s Ranger information says rehabilitation continues to evolve with closure design, water inventory, Traditional Owner requirements and the monitoring period. The historical trust balance records security already held. It does not cap what remains to be done.

A company can therefore have a large provision, a large bond and a funding gap at the same time. Each figure may be accurate. None is sufficient alone.

The obligations not called closure costs

At Didipio in the Philippines, OceanaGold Philippines, the operator holding a renewed Financial or Technical Assistance Agreement, accepted commitments that sit alongside physical rehabilitation.

The renewal added annual funds equal to 1% and 0.5% of preceding-year gross mining revenue for community and provincial development. Philippine mining law separately requires at least 1.5% of operating costs for social-development programmes. The environmental approval requires a final rehabilitation and decommissioning plan, while national rules require financial requirements covering a ten-year post-decommissioning monitoring period.

Didipio shows how community and development commitments can add mine-related obligations outside a rehabilitation fund. Photo: European Space Agency / Wikimedia Commons, CC BY-SA 3.0 igo.

These obligations should not be casually added to a rehabilitation provision and called one number. But neither should they vanish because they sit in another agreement. They are real calls on cash and part of the wider cost of retaining and closing a mine responsibly.

Put Ranger, Marikana and Didipio side by side and a pattern appears that none of their disclosures states outright: the biggest error is not underestimating one item. It is assuming there is one item to estimate.

A deal does not make the obligation disappear

Elk Valley Resources, or EVR, is the Canadian operating entity whose provincial authorisations remained with it when Glencore acquired a 77% interest in July 2024 for approximately US$6.93 billion. Glencore’s transaction announcement records Canadian undertakings including at least C$200 million of rehabilitation and closure expenditure.

The same transaction information says British Columbia reviewed and updated EVR’s financial-security and environmental-protection requirements following the ownership transfer. Federal approval was required, while the provincial authorisations remained with the operating entity.

That is the evidence behind a point often lost in deal excitement. A share purchase does not make the permit holder disappear. Nor does it remove the practical need to identify consent requirements, security replacement, parent guarantees, Indigenous undertakings and seller indemnities.

This is why signed documents do not automatically settle operational risk, a principle explored in what can still stop funds after investment documents are signed.

Talvivaara, the Finnish mining company that disclosed €15.3 million of environmental security in 2009, offers the darker historical warning. Its operating subsidiary failed in 2014. Finland initially allocated €50 million for environmental stabilisation, and in 2015 the government estimated orderly active closure at approximately €300 million over several years. The state continued operations through Terrafame rather than execute immediate closure.

The figures and subsequent state-backed operations show what a stale security figure cannot tell you: whether the scope still reflects technological performance, operating conditions or the operator’s ability to survive long enough to deliver the work.

When the bond really is useful

A bond is not meaningless. A credible, current and properly supported instrument can protect the public from an operator’s failure. In Peabody’s case, the difference between bond and provision was explained by timing. At Marikana, financial guarantees fund the closure estimate.

The evidence does not support a universal rule that bonds must always exceed provisions, or vice versa. Nor does it say every commitment outside a rehabilitation fund belongs inside it.

The more useful rule is narrower: every number needs a label. What work does it cover? When is it assumed to happen? Who can draw on it? What assets have been restricted to support it? What remains outside it?

What operators should put in the room

For management teams, the task is not to make closure look small. It is to make it legible.

Build one register of obligations drawn from permits, environmental approvals, stakeholder agreements, financial-assurance requirements and transaction undertakings. Maintain a current technical estimate showing work scope, unit costs, contingency, escalation, water assumptions and monitoring requirements. Then reconcile that estimate to the provision, each security instrument and every restricted-cash or collateral arrangement.

Do this before fundraising. A lender or equity investor will eventually ask whether cash pledged behind a guarantee is actually available for operations, distributions or growth. Better to answer before the question acquires a slightly hostile tone.

GI Network would turn the technical closure estimate, permits, assurance instruments, restricted-cash position, stakeholder commitments and change-of-control conditions into one capital-risk reconciliation. That work identifies where the liability is underwritten, where it is merely disclosed and which questions an investment committee will raise before considering debt, equity or blended capital.

What experienced investors look for

First-time investors often ask whether the bond is “enough”. Experienced investors ask enough for what.

They test an immediate-closure case, a lower discount rate, extended monitoring and a need to replace security following a change in control. They distinguish the accounting provision from cash genuinely available to the business. They look for obligations outside the environmental model. Then they ask who pays if the estimate changes, and whether that payment reduces returns, breaches financing promises or requires fresh capital.

That is not pessimism. It is ownership.

The useful way to remember it is the Closure Five-Line Reconciliation:

1. 1.Work: What is the current undiscounted cost of every legally and contractually required closure activity?
2. 2.Provision: What does the accounting provision assume about timing, scope and discount rate?
3. 3.Security: What bonds, guarantees and trust balances can a regulator actually call?
4. 4.Capital lock-up: What cash, collateral and guarantee capacity are tied behind that security?
5. 5.Outside commitments: What community, licence, ownership and transaction obligations sit beyond the environmental closure model?

If a company cannot walk an investor through all five lines, the bond amount is not an answer. It is the beginning of the investigation.