Closure is decades away, the provision is discounted heavily, and the person who signs the estimate will not be there when the bill arrives. The incentive structure explains most of the variance.
Bonding is the only honest test
A provision is an accounting entry. A bond is cash a regulator can call. Jurisdictions that require full bonding produce very different estimates from those that accept a note in the accounts.
What we look for
The discount rate, the inflation assumption, the date of the last third-party review, and whether the bond covers the current estimate or the one filed at permitting.
The discount rate assumption does most of the work
A closure provision's present value is extraordinarily sensitive to the discount rate applied to the future rehabilitation cost, given that closure often lies decades ahead. A modest change in the assumed discount rate can alter the reported provision by a large margin without any change whatsoever in the physical scope of the rehabilitation work required. Because the discount rate assumption sits within a range that auditors generally accept as reasonable, it offers considerable, entirely legal, latitude for a company to manage the size of the liability it reports in any given year. That latitude is precisely why the assumption deserves more scrutiny than it typically receives from analysts focused on headline earnings.
Comparing the discount rate a company uses for its closure provision against the discount rate implied by its own cost of capital, and against peers operating in similar jurisdictions, is a useful, if imperfect, check on whether the reported figure is conservative or optimistic.
The pattern at actual closure
Case studies of mines that have reached closure in recent years show a consistent pattern: the final rehabilitation cost, once ground disturbance, water treatment obligations and social transition commitments are fully accounted for, tends to exceed the provision carried on the balance sheet in the years leading up to closure, in some cases substantially. Water treatment in particular has proven to be the most persistently underestimated cost category, since many operations require some form of ongoing treatment long after mining itself has ceased, a perpetual obligation that sits awkwardly within a provisioning framework built around a fixed closure date.
