PEA, PFS, FS
Three stages through which a mining project matures from an idea into an investment decision. They differ in essentially one respect: how accurate the numbers are. And that is precisely the point almost everyone gets wrong.
What the studies say about themselves
Inside the reports, accuracy is usually stated as a symmetric range. A pre-feasibility study puts its cost estimate at "±25%", a feasibility study at "±15%".
| Stage | Name | Stated in the report | Rough cost | Duration |
|---|---|---|---|---|
| PEA | Preliminary Economic Assessment | ±35–50% | USD 0.3–1m | 3–6 months |
| PFS | Pre-Feasibility Study | ±20–25% | USD 1–5m | 6–12 months |
| FS | Feasibility Study | ±10–15% | USD 5–20m | 12–24 months |
These ranges are industry convention, not a standard. No rulebook prescribes them, neither NI 43-101 nor the CIM standards. They appear in the documents because that is how the practice settled. Costs and durations, in turn, are observed experience; they vary a great deal with project size, jurisdiction and metal.
What the cost standard actually says
The reference the industry invokes is the AACE 47R-11 cost estimate classification for mining and mineral processing. There, the ranges are not symmetric:
| AACE class | Maturity of the engineering | Low deviation | High deviation |
|---|---|---|---|
| Class 5 | Concept study | −20% to −50% | +30% to +100% |
| Class 4 | Pre-feasibility | −15% to −30% | +20% to +50% |
| Class 3 | Feasibility | −10% to −20% | +10% to +30% |
The axis on which AACE plots these ranges is labelled Growth from Estimated Costs — growth against the estimated cost base. The standard does not expect the error to be evenly spread in both directions. It expects it on the upside.
The difference is not a quibble. Read "±40%" for an early study and an 80% cost overrun looks like an outlier. Under AACE it sits inside the range.
Which class belongs to which stage
AACE assigns the feasibility study to class 3 and the pre-feasibility study to class 4. On the PEA the standard deliberately stays open: it is "usually … Class 5, but not always; it may also be used for a more advanced study that simply does not meet the qualifications for a PFS or FS".
That ambiguity cannot be resolved, and it should not be. "PEA" is a term from disclosure law, not from cost engineering. Two studies carrying the same label may well describe two entirely different levels of engineering maturity. What carries weight is therefore not the heading but the chapter on cost estimation: where do the numbers come from — analogue projects, factored methods, or quotations?
What separates the stages in substance
PEA. May include inferred resources — the exception under Section 2.3(3) of NI 43-101. That is exactly why every PEA carries the prescribed statement that it does not demonstrate economic viability. The cost estimate is drawn largely from comparable projects rather than from quotations.
PFS. Measured and indicated resources only. The mining method has been chosen, the processing tested in the laboratory, part of the cost base quoted. From here on a reserve may be declared. See Inferred, indicated, measured.
FS. Engineered through, costs mostly underpinned by quotations, permitting route as well as water, power and transport access resolved. A bank decides on project finance on the strength of this very document. Hence the epithet bankable.
Where it misleads
The deviation is asymmetric — and that is documented, not assumed. Bertisen and Davis examined 63 mining projects in The Engineering Economist in 2008, comparing actual capital costs with the bankable feasibility study in each case. The result: a mean overrun of 14%, with close to half the projects falling outside ±15% — outside, that is, what an FS claims for itself. Haubrich arrived in 2014 at mean overruns of between 20% and 60% for earlier study stages.
The reason is mundane. At an early stage the missing items are the ones nobody yet knows about: infrastructure, dewatering, reclamation, interest during construction. A line item that has never been estimated can only move one way.
NPV reacts disproportionately. A 30% rise in costs can halve the present value, because the capital goes in at the start and the earnings come at the end. That is why the sensitivity table in the report matters more than the headline.
"Robust economics" is not a category. It is press-release language. What holds up is after-tax NPV, IRR, capital requirement, payback period — and the assumed commodity price, which is the first thing to check.
The stages are not a compulsory sequence. Some projects go from PEA straight to FS; some publish several PEAs at increasing throughput. A new PEA two years on is not automatically progress. It may equally mean that the PFS failed to confirm the earlier arithmetic.
Years still separate an FS from first production. Construction decision, financing, permitting, building. Each step carries its own risk, and none of them is settled by the study.
How we handle it
For every project we state the stage and date of the underlying study, and we recalculate NAV with our price and discount rate, not the study's. We carry the accuracy of the cost base asymmetrically, in the logic of AACE 47R-11 — not as a "±" lifted from the report text. For a PEA the graduated treatment of inferred resources set out in Inferred, indicated, measured applies on top.
The full framework is set out under How we value resource companies.
Sources
- Cost Estimate Classification System — As Applied in Engineering, Procurement, and Construction for the Mining and Mineral Processing Industries, AACE International Recommended Practice No. 47R-11 — web.aacei.org
- Bertisen, J. / Davis, G. A.: Bias and Error in Mine Project Capital Cost Estimation, The Engineering Economist 53(2), 2008 — tandfonline.com
- Haubrich, C.: study of cost overruns in mining projects, 2014. No freely available edition; the 20% to 60% range is corroborated by the specialist literature on mining cost estimation.
- National Instrument 43-101 — Standards of Disclosure for Mineral Projects, consolidated British Columbia version, Section 2.3 — bclaws.gov.bc.ca
- CIM Definition Standards for Mineral Resources and Mineral Reserves, 10 May 2014 — bcsc.bc.ca
As of: 11 August 2026 · Version 2