Ask a mining company where the actuaries are, and you will get a puzzled look. Many actuaries sit in the corner of an insurer, some distance from the shaft. Mining is grades and tonnes and commodity cycles, feasibility studies and discounted cash flows. The two worlds are not thought to touch.

They should. A mine is, underneath the rock, a structure of long-dated and deeply uncertain obligations: a cash-flow stream that depends on a price nobody can forecast, an orebody that depletes, a closure bill that falls due decades from now, and a set of promises to the people who did the digging. That is not a mining problem dressed up in finance. That is, almost exactly, the shape of problem the actuarial profession was invented to model. The mining model does not need an actuary in the room. It needs actuarial thinking — and most sector models, in mining and well beyond it, are poorer for the lack of it.

It is worth separating the thinking from the credential. Actuarial thinking is not a set of formulae reserved for people who passed the examinations. It is a small number of habits of mind, and each one is a corrective to a way that ordinary financial models lie.

The first habit is to reserve for what has not happened yet — to recognise a future obligation in the present, honestly, before it arrives to collect. The second is to model the whole distribution rather than a single point: to treat the future as a range of outcomes with probabilities attached, not as one confident line. The third is to take the tail seriously — to give real weight to the low-probability, high-severity events that a best-guess model rounds away, because those are the events that end careers and companies. The fourth is to discount the future honestly, treating the discount rate as a decision to be defended rather than a default to be inherited. The fifth is to name your margin: to state, out loud, the buffer you are holding for the things you cannot see, instead of hiding it inside a spuriously precise estimate. And the sixth is to own your assumptions and then watch them — to run a control cycle in which you set an assumption, monitor what actually happens against it, and recalibrate, rather than building a model once and trusting it forever.

Put together, these habits amount to one thing: a disciplined honesty about uncertainty. That is the whole of it. And it is precisely what the standard mining model lacks.

Consider how a mining asset is usually valued. The overwhelming majority of valuations — by most counts, comfortably more than half of them — rest on a discounted cash flow model. A single commodity price deck is chosen. Most-likely grades are assumed. One life-of-mine plan is laid down. The stream is discounted at a single weighted cost of capital, and out comes a net present value, carried to a precision that no honest person believes. The number looks like knowledge. It is a guess in formal dress.

The trouble is not that DCF is wrong; it is that the standard DCF is static. It assumes the future arrives on schedule and that management, having built the mine, will stand still while the world moves. For a thirty- or forty-year base-metal project, this is close to absurd. The risk profile of such a project is not well-behaved: uncertainty does not grow smoothly with time but lurches — with the price, with the grade, with the exhaustion of a tax shield, with a policy change in a mining jurisdiction. A single discount rate applied to a single cash-flow stream cannot hold that shape. It systematically misprices the very thing that makes a long-life mine valuable or fatal: its exposure to a future that will not sit still.

This is where the first tranche of actuarial thinking earns its place, and where the discipline of forecasting stops being decoration. Model the commodity price as what it is — a volatile, often mean-reverting stochastic process — and run the mine through thousands of futures rather than one. The output is no longer a number but a distribution, and a distribution tells you what the point estimate concealed: how much of the value lives in the tails, how often the project drowns, how fat the downside really is. Go one step further into real-options thinking and the mine reveals its true nature. It is not a fixed cash-flow stream at all. It is a portfolio of decisions — to defer, to expand, to mothball, to walk away — each of which management will take as uncertainty resolves. Valuing those embedded choices is exactly the kind of contingent, path-dependent reasoning the actuarial toolkit was built for. The static DCF cannot see this flexibility, and so it undervalues good optionality and overvalues doomed commitment in the same breath.

But the sharper contribution is not on the asset side of the balance sheet. It is on the side the mining model most reliably forgets: the liabilities.

Mining is unusually rich in obligations that are long-tailed, uncertain in both amount and timing, and payable long after the ore is gone — which is to say, obligations that are actuarial in every respect except the department that owns them. Rehabilitation and closure. Environmental damage and acid mine drainage. Post-retirement medical promises. Occupational disease. Each of these is routinely reduced to a single engineering point-estimate and parked in a note to the accounts, when each is in truth a reserving problem of the classic kind.

Take closure. A rehabilitation liability is an obligation of uncertain cost, falling due at an uncertain date thirty or fifty years out, that must be measured today and funded over the life of the mine. That is a reserving calculation. Treated as one — with a distribution of costs, an honest discount rate, and an explicit margin for the near-certainty that the estimate is optimistic — it produces a provision a company can stand behind. Treated as a single hopeful number, it produces the situation South Africa now lives with. The country carries an estimated six thousand-odd derelict and ownerless mines; older assessments put the cost of cleaning up the legacy at around R100 billion and the timeline at centuries, while the rehabilitation funds actually set aside run to only a fraction of the eventual bill. Closure plans, study after study finds, underestimate both the significance of the impacts and their cost. And when the provision is too low, the liability does not disappear. It transfers — to the state, to the fiscus, and most heavily to the community left beside the tailings dam with contaminated water, unstable land, and the zama-zamas who move into the void. An under-reserved closure liability is not a technical footnote. It is a debt quietly assigned to people who never signed for it.

Or take occupational disease, where the industry’s most expensive lesson was learned the hard way. The silicosis and tuberculosis class action that produced the Tshiamiso Trust turned on a liability with every actuarial characteristic there is: a disease that manifests decades after exposure, a claimant population that must be traced and medically assessed, benefits that depend on severity and survival, and payments spread across a twelve-year life. The settlement’s headline figure — roughly R5 billion — was never a fact. It was described from the outset as a reasonable estimate of an obligation that is formally open-ended and the founding companies’ liability for which is unlimited. Six companies reached it using, by their own account, different bases. It requires annual actuarial assessment to keep it honest. By early 2026, around R2.5 billion had been disbursed, roughly the midpoint of the Trust’s life. This is a textbook actuarial reserve — and the industry arrived at it through litigation rather than foresight, because the thinking that would have surfaced it earlier was not in the room.

None of this thinking survives on its own. The last habit — the control cycle — is what keeps a model honest after the transaction closes. Actuaries do not build a reserve once and file it. They set assumptions, compare actual experience against expected, and recalibrate as reality answers back. Most corporate models never get this second life; they are built for a deal and abandoned the day it signs. An independent validation discipline asks the two questions the confident single number is designed to suppress: is this figure genuinely defensible, and does it update when the world contradicts it? The aim of bringing the discipline into a mining finance team is not to turn geologists and accountants into actuaries. It is to install the habit — to make the range, the margin, and the monitored assumption the normal furniture of the model rather than an audit-season afterthought.

And the habit travels. It is not a mining peculiarity. A retailer’s loyalty programme is a deferred liability that behaves like a reserve; customer lifetime value is a survival problem wearing a marketing label; warranty and returns provisions are small insurance books nobody underwrites as such. Wherever a model quietly collapses an uncertain future into one flattering number, actuarial thinking has something to correct.

Which returns us to the deeper point. The truths that matter most in a mine are the ones furthest out in time and hardest to see — the price that cannot be forecast, the closure decades away, the lung that scars slowly and speaks late. A deterministic model is most confident exactly where it is least entitled to be, and its false confidence is not free. It is paid, eventually, by whoever inherits the consequence: the fiscus, the community, the mineworker whose claim was under-reserved before he was ever counted.

Actuarial thinking, stripped of its mystique, is simply the discipline of telling the truth about uncertainty — of provisioning honestly for the futures we would rather round down. That is why it improves a financial model in any sector that will submit to it, and why it matters most in the sectors that never asked for it. It does not flatter the model. It illuminates what the model would prefer to leave in the dark.