Key Takeaways
- 83% of major mining and metals projects face cost and scheduling challenges, with average CAPEX overruns exceeding 40% (McKinsey, 2024). IPA's research consistently identifies front-end loading quality as the single strongest predictor of whether those overruns occur.
- Only 16% of capital projects enter authorisation with Best Practical FEL. For every category below Good on IPA's FEL Index, a project loses 2% in internal rate of return - a $30M value destruction on a $200M portfolio before execution has even begun (IPA, 2025).
- Construction input prices rose at a 12.6% annualised rate in early 2026, driven by tariffs on steel, aluminium, and copper (Construction Dive, 2026). Assumptions baked into FID-stage estimates are outdating faster than at any point in recent memory.
- Stress-testing through sensitivity analysis, scenario planning, and Monte Carlo simulation identifies which assumptions carry the greatest cost risk - but only if the underlying FEL definition is sound enough to make those assumptions credible in the first place.
- Owner teams lacking critical functions deliver projects that are 25% more expensive on average, with 20% greater cost growth (IPA, 2025). Stress-testing is only as effective as the governance structures that act on its findings.
When a capital project is sanctioned, the cost estimate reflects assumptions about commodity prices, labour availability, contractor productivity, exchange rates, and regulatory conditions. Within months of Final Investment Decision (FID), many of those assumptions will have shifted. What is less well understood is that the majority were already compromised - not by market conditions, but by the quality of the front-end work that produced them.
In 2025, aluminium climbed over 30% on the back of tariff escalation. The US dollar index fell 9% against major currencies. Construction input prices rose at a 12.6% annualised rate in early 2026. For Energy, Minerals and Resource (EMR) organisations managing multi-year capital programmes, approved budgets increasingly fail to reflect current delivery reality. But the external environment is only half the problem.
This post examines why project cost assumptions fail, what stress-testing actually involves in practice, and how to apply it at each stage - starting with an honest account of why the assumptions themselves are so often built on inadequate foundations.
Why project cost assumptions fail in the Energy, Minerals and Resource (EMR) sector
IPA's database of over 24,000 capital projects shows that while companies expect ±10% accuracy from their authorisation estimates at the end of FEL 3, actual outcomes range from -17% to +42%. McKinsey's 2024 analysis found that 83% of major mining and metals projects experience cost and scheduling challenges, with approximately two-thirds of overruns attributable to poor initial assessments rather than execution failures. IPA Founder Ed Merrow, whose research spans 20,000+ projects, puts the megaproject failure rate at 65% - with the root cause consistently the same: the front end was rushed.
IPA's FEL Index - their quantitative measure of project definition quality - has deteriorated to the Poor range for the first time in the organisation's history. The causes are structural, not incidental.
Schedule pressure. The most destructive driver of FEL dilution is the externally imposed timeline. A board commits to first production by a specific date. A commodity window demands accelerated sanction. The FEL programme is compressed to fit the gap. IPA's data on low-carbon projects illustrates the consequence precisely: projects sanctioned in the 2022 wave were planned to complete execution 14% faster than industry average and were authorised with Poor FEL. The result was a 46% schedule slip. The time saved at the front end was lost many times over in execution.
Optimism bias and strategic misrepresentation. Researcher Bent Flyvbjerg identifies two interlocking forces that corrupt FEL quality: optimism bias (the genuine cognitive tendency to underestimate costs) and strategic misrepresentation (the deliberate shaping of FEL outputs to secure funding). McKinsey documented this in a major capital portfolio review, finding that more than half of potential portfolio value was lost to underperformance - with root causes traced to unrealistic assumptions about input costs, productivity, and work-hour requirements. Every one of those distortions was a FEL failure. None had been independently challenged.
"The energy and resources sector has a persistent habit of sanctioning projects on the basis of a fiction. Optimism bias is often used as a polite shroud for something less forgivable: the deliberate act of misleading Boards, JV partners, and investors to secure a green light. When you hollow out a budget or compress a schedule just to get sanctioned, you are not being ambitious. You are establishing a misaligned baseline that makes failure the only possible outcome. What gets diagnosed at the worksite as an execution failure is, in truth, the downstream casualty of a boardroom decision made years earlier." - Graham Sutherland, Founder, PDAS
Stage-gate theatre. Perhaps the most insidious form of FEL dilution is the project that passes every gate on schedule, with every deliverable checkbox ticked, but where the actual quality of those deliverables is insufficient. IPA distinguishes between organisations that mandate a work process and those that ensure adherence - measuring the actual completeness of deliverables rather than their nominal existence. The gap between those two practices is where capital value quietly disappears.
External cost volatility. Even well-defined FEL assumptions are exposed to market forces. Copper gained nearly 40% on the LME in 2025. US Section 232 tariffs on steel and aluminium reached 50% and were expanded to copper in April 2026. Medium-voltage transformer lead times have grown from 4-6 months to 18-24 months. The EUR/USD exchange rate swung 14% through 2025. Static estimates do not survive these conditions.
"Far too often we attempt to run too fast on the front end, only to be forced to give all the time back several times over in execution and startup." - Ed Merrow, IPA Founder
The Construction Industry Institute estimates that front-end planning typically costs 2-5% of total installed cost. A 3% investment in FEL quality yields approximately a 15% reduction in overall project cost - a $30M saving on a $200M portfolio, $300M on a $2B programme. For every dollar spent on inadequate project definition, organisations spend four to seven dollars correcting the consequences in execution. These are not theoretical figures. They are the empirical result of decades of capital project performance data. For a broader view of how boards should be evaluating these risks, see What Boards Should Be Asking About Their Capital Projects Right Now.
What does stress-testing project assumptions actually involve?
Stress-testing is the systematic process of identifying which assumptions carry the greatest risk to cost outcomes and quantifying the range of possible results. Three complementary methods form the core toolkit.
Sensitivity analysis isolates individual variables - labour productivity, steel price, equipment lead times, exchange rates - and measures how changes in each affect total project cost. The output is typically a tornado chart ranking assumptions by impact magnitude. This identifies the three to five variables that disproportionately drive cost risk, focusing monitoring and mitigation effort where it matters most.
Scenario planning tests coherent combinations of assumptions simultaneously. A base case uses the most likely values. A bear case combines plausible adverse conditions: higher material costs, lower productivity, extended schedule. A bull case reflects favourable outcomes. Each scenario must be internally consistent - mixing optimistic and pessimistic assumptions within the same scenario undermines the analysis.
Monte Carlo simulation assigns probability distributions to each cost element and runs thousands of iterations to produce a cumulative probability curve, generating P50, P80, and P90 estimates. P80 is the most commonly used basis for contingency setting. However, IPA research indicates that what many organisations label as a P80 estimate with 15-20% contingency is often closer to the true P50 - meaning half of all projects will exceed their contingency-inclusive budget.
The most effective approach runs sensitivity analysis first, builds scenarios around the identified key variables, then applies Monte Carlo to quantify the full probability distribution. For more on how structured risk frameworks support capital project cost management, see Understanding Risk Management Strategies in Mega-projects.
How to apply stress-testing at different project stages
FEL 1 (business planning): Broad scenario analysis tests whether the project concept remains viable across a wide range of economic conditions. AACE Class 5 estimates carry accuracy ranges of -50% to +100%, so stress-testing at this stage focuses on directional viability. The primary question is whether the business case holds across base, bear, and bull scenarios - before significant definition spend is committed. Weak FEL 1, as IPA is explicit in noting, creates cascading problems throughout the entire project lifecycle.
FEL 2 (concept selection / pre-FEED): Sensitivity analysis identifies dominant cost drivers. Monte Carlo modelling establishes initial contingency ranges. This is the stage where poor definition most commonly embeds itself in estimates - scope alternatives not fully evaluated, contracting strategy unconfirmed, site conditions incompletely characterised. IPA data shows that independent estimate validation at this stage provides average cost savings of 6% for large projects.
FEL 3 (FEED / authorisation): Full probabilistic cost and schedule risk analysis should be completed before the FID gate closes. P50 and P80 estimates must be explicitly stated and stress-tested against current market conditions - not those that prevailed during FEL 2. Material costs, tariffs, and FX rates should be updated to reflect the latest data. P&IDs, equipment lists, execution plans, and contracting strategy should all be sufficiently complete that no significant unknowns remain.
Execution: Assumptions require continuous re-baselining against actual performance. This is where earned value management becomes the primary detection mechanism. For a structured view of governance approaches across these stages, see Stage-Gate vs Agile in Infrastructure Programmes.
Why owner governance and EVM determine whether stress-testing has any effect
Earned value management provides the quantitative framework for early detection of cost drift during execution. The cumulative Cost Performance Index (CPI) is a particularly reliable forward indicator: established research across hundreds of projects demonstrates that CPI rarely improves by more than 10% once a project is 20% complete. If CPI falls below 0.95 early in execution, the probability of recovering to budget drops sharply. Tracking CPI alongside schedule performance metrics gives owner teams the objective, early warning they need to intervene before drift becomes overrun. For more on which metrics drive better capital decisions, see What Portfolio KPIs Actually Drive Better Decisions.
But EVM only works if the owner team has the capacity to act on what it reveals. IPA reports that 73% of capital project teams are missing critical functions, using inexperienced people in leadership roles, or are generally understaffed. Projects delivered by insufficiently staffed owner teams cost 25% more on average and experience 20% greater cost growth. Poorly functioning teams average 30% budget overruns and over 50% schedule slip.
A Monte Carlo analysis that identifies a P80 significantly above the sanctioned budget is useful only if the governance structure has the authority and capability to respond - through scope adjustment, commercial renegotiation, or contingency release. Stress-testing a poorly defined project does not make it well defined. It makes the degree of risk visible. What happens next is a governance question. For proven approaches to addressing delivery shortfalls when cost drift is already underway, see 11 Effective Strategies for Project Recovery.
Building cost resilience across the project lifecycle
Approved budgets represent assumptions fixed at a point in time. In a cost environment shaped by tariff escalation, commodity price volatility, and compressed front-end timelines, those assumptions require continuous, independent challenge.
Stress-testing project assumptions is not a one-time exercise performed before FID. It is a governance discipline that spans the full project lifecycle. The organisations that manage capital project cost volatility effectively combine rigorous front-end definition with probabilistic analysis, EVM-based early warning, and owner teams with the authority and experience to act on what the data reveals. Stress-testing a well-defined project produces a reliable risk picture. Stress-testing a poorly defined one produces a false sense of control. The distinction starts at FEL 1 - and it runs all the way through to closeout. For more on building operational resilience across capital programmes, see 5 Strategies to Ensure Business Certainty in the Energy Sector.
Is your capital programme stress-tested for current market conditions?
PDAS provides independent project governance and assurance for EMR sector organisations. If your cost assumptions need a structured, objective review, book a discovery call with our team.
