Optimism Bias in Capital Projects: a Governance Risk, Not a Personality Flaw
Updated: Oct 11
The most damaging project forecasts are rarely produced by reckless people. Capable teams often produce them within systems that reward confidence, speed, and approval.
That is why optimism bias deserves more serious treatment in capital project governance. In oil and gas, EPC, construction, energy, infrastructure, mining, utilities, marine, and major industrial work, early forecasts do more than inform a business case. They shape funding decisions, contracting strategy, delivery expectations, stakeholder commitments, and sometimes public promises.
Once a project secures approval on a low-cost estimate, a short schedule, a narrow contingency, or an incomplete scope definition, the delivery team inherits a problem created upstream. By then, the language changes. The issue becomes “cost overrun,” “delay,” “scope creep,” or “poor execution.” Yet the root cause may sit in concept selection, feasibility, FEED, estimating, risk assessment, procurement strategy, or sanction.
Optimism Bias in Capital Projects: A Governance Risk, Not a Personality Flaw is the right framing because the issue is structural. Individual judgment matters, but organizational pressure, competitive capital allocation, sponsor expectations, weak challenge, and approval incentives often shape the forecast before the estimator or planner ever opens the model.
Optimism Bias Starts Before the Project Looks Like A Project
Optimism bias is often discussed as a psychological tendency to expect better outcomes than evidence supports. That definition is useful, but incomplete for capital projects. In major projects, optimism bias is also produced by process design.
A business case is rarely a neutral document. It is a proposal competing for scarce capital, executive attention, board confidence, regulator acceptance, or government funding. A development team may know, openly or quietly, that a more complete cost view could weaken the case. A sponsor may not ask anyone to “understate the estimate,” but the message can be clear enough: make the project viable.
This is where optimism bias becomes a project risk.
In early development, the team is usually trying to answer several questions at once:
Is the opportunity technically possible?
Is the commercial case attractive?
Can it beat other projects competing for capital?
Can it meet a market, regulatory, shutdown, or production window?
Can the organization tolerate the risk?
Those questions are valid. The danger appears when viability becomes the test that the estimate must pass.
Concept Selection Can Narrow The Evidence Too Early
Concept selection should compare options. In practice, it can become a search for the option that best satisfies a preferred strategic answer.
Consider an offshore tieback where the business wants first production within a tight market window. A high-level concept may assume reuse of existing host facilities, limited topsides modification, ready marine vessel availability, and manageable subsea interface risk. Each assumption may be defensible in isolation. Together, they can create a forecast that treats difficult integration as routine.
The selected concept then gains momentum. Alternative concepts look slower, more expensive, or less elegant. The team may keep refining the preferred case while treating evidence against it as a risk to manage later.
That pattern appears across sectors:
A utility upgrade assumes outage windows will be available because they were available on a smaller prior job.
A brownfield refinery project assumes tie-ins can be completed during a planned turnaround without fully testing discovery risk.
A mining expansion assumes logistics capacity on regional roads, ports, or camps will scale smoothly.
A rail or road project assumes land access, utility diversions, permitting, and stakeholder interfaces will resolve in sequence.
The problem is not that early assumptions exist. Early-stage work needs assumptions. The problem is when assumptions become commitments before stress-testing.
Feasibility and FEED Can Create False Protection
Feasibility and FEED should reduce uncertainty. They often do. Yet they can also create a misleading sense of precision if the scope's maturity doesn't match the number's precision.
AACE International’s estimate classification guidance is widely used because it recognizes a basic fact: estimates have different expected accuracy ranges depending on scope maturity, engineering definition, and project phase. The exact range depends on the class and context, but the principle is what matters here. A concept estimate and a control estimate are not the same kind of forecast.
Problems develop when governance forums receive single-point numbers that look precise but still rest on immature inputs. A figure presented to one decimal place can carry the visual authority of an engineered fact, even when key drivers remain uncertain.
Examples include:
Route length known, but ground conditions not yet proven.
Major equipment priced, but installation productivity assumed.
Process capacity defined, but brownfield access constraints not fully surveyed.
Procurement packages listed, but market capacity not tested.
Modularization strategy selected, but transport, lifting, and yard interface risks still open.
Sanction dates can also pressure FEED. When the investment approval calendar is fixed, the project may be pushed toward a decision before enough definition exists. The result is a business case that looks complete because the required templates are complete.
That is not assurance. It is documentation.
Why Intelligent Organizations Still Approve Unrealistic Forecasts
Experienced organizations know that projects overrun. Most have governance processes, stage gates, risk registers, assurance reviews, cost procedures, and schedule standards. Many use PMI, APM, ISO 31000, AACE, CII, RICS, ICE, OECD, World Bank, or government infrastructure guidance in some form.
So why does optimism bias still survive?
Because the formal process may not challenge the strongest incentives in the room.
Approval Processes Often Reward Advocacy
Capital allocation is competitive. Projects fight for funding against other projects, other regions, and other business units. Project teams must advocate for it. They must show strategic fit, technical feasibility, attractive returns, and confidence in delivery.
That advocacy role is legitimate. Without it, good projects would never move.
The weakness appears when the same system asks the advocacy team to independently assure its own forecast. Even when people act with integrity, role conflict can shape judgment. If raising a stronger risk view delays sanction, weakens economics, or gives another project funding priority, the system has created a reason to soften the message.
This is where strategic misrepresentation can enter. The term does not always mean fraud. It can mean selective emphasis, postponed risk recognition, or presenting the most favorable plausible case as the base case. Peer-reviewed research on major projects, including work on behavioral economics and megaproject scholarship, has long distinguished genuine optimism from strategic behavior under approval pressure.
Both matter. They can also reinforce each other.
Governance May Test Compliance Rather Than Realism
A stage gate can confirm that required deliverables exist without confirming that the forecast is realistic. The gate asks whether the risk register is present, whether the cost estimate has been reviewed, whether the schedule has a critical path, and whether the procurement strategy has been approved.
Those checks are necessary. They are not enough.
A project can pass a gate while still carrying major underestimation in areas such as:
Interface complexity between contractors, operations, vendors, regulators, and third parties.
Construction productivity in congested, remote, offshore, winter, or high-humidity environments.
Access limitations in live plants, ports, rail corridors, utilities networks, and marine spreads.
Late engineering growth after “frozen” design.
Vendor data delays.
Owner-supplied equipment risk.
Commissioning and start-up complexity.
Capability gaps in the delivery organization.
Professional bodies have repeatedly emphasized the value of risk management, front-end definition, control processes, and assurance. CII’s work on front-end planning, including tools such as the Project Definition Rating Index, reflects the well-established link between scope definition and project performance. ISO 31000 frames risk management as a decision-support discipline, not a paperwork exercise. PMI and APM both place governance, sponsorship, risk, and stakeholder management at the center of project success.
The lesson is clear. Governance must ask whether the evidence supports the forecast, not just whether the file is complete.
Incentives Can Make Bad News Expensive
Optimism bias often survives because bad news costs money.
A planner who adds time for commissioning uncertainty may be seen as conservative. An estimator who increases contingency may be asked to “sharpen the pencil.” A package manager who challenges a fabrication duration may be told that market engagement already confirms capability. A risk lead who challenges the base case may be accused of blocking progress.
No one needs to say this explicitly. Teams learn what gets rewarded.
If the organization celebrates sanction more than forecast quality, optimism becomes rational behavior. If internal credibility depends on getting capital approved, people will produce approval-friendly forecasts. If bonuses, promotion, or business-unit status depend on a positive investment decision, the base case will tend to improve as it moves upward.
This is why treating optimism bias as a personality issue is too easy. It allows the organization to blame “overconfidence” while leaving the decision system untouched.
Where Underestimation Hides in the Project Lifecycle
Optimism bias does not arrive in one dramatic moment. It accumulates through small decisions, each of which may appear reasonable.
Estimating Under Pressure
The estimate is often the visible battleground. Early estimates may understate cost because they exclude or soften items that are hard to price:
Temporary works.
Construction indirects.
Logistics and laydown.
Escalation.
Owner’s costs.
Commissioning support.
Spare parts and first fills.
Systems integration.
Weather standby.
Marine spread inefficiency.
Permitting and compliance costs.
Interface management.
Brownfield discovery.
Contractor risk pricing.
In EPC and construction, a common pattern is to estimate installed quantities with reasonable unit rates, then under-allow for the conditions in which the work must happen. The rate may be sound for a greenfield plot with clear access. It may be weak for a live plant with permit-to-work restrictions, simultaneous operations, scaffold congestion, shutdown windows, radiography controls, and limited lifting zones.
The estimate is not wrong because someone is cheerful. It is wrong because the basis does not reflect execution reality.
Scheduling That Assumes a Clean Sequence
Schedules often carry optimism in logic, calendars, restraints, and productivity assumptions. Early plans may show design, procurement, fabrication, construction, commissioning, and handover in a clean chain. Real projects rarely behave that way.
An offshore module schedule may assume vendor data arrives in time to complete detailed design. A tunneling project may assume ground treatment progresses as planned. A transmission line project may assume access agreements and environmental approvals do not fragment the workfront. A shipyard or marine conversion may assume steel renewal, equipment delivery, inspection, and class approvals align without rework.
The project schedule becomes fragile when it lacks:
Realistic approval durations.
Procurement lead-time uncertainty.
Design iteration cycles.
Constructability review periods.
Mobilization and ramp-up time.
Testing, commissioning, and performance verification logic.
Recovery limits once a critical path slips.
A deterministic schedule can make uncertainty disappear. Probabilistic schedule analysis brings some of it back into view, but only if the model includes credible logic and honest ranges.
Risk Assessment That Dilutes the Difficult Risks
Risk workshops can fail for social reasons. The room may contain senior sponsors, future delivery leaders, commercial staff, engineers, operations representatives, and consultants. People know which risks are politically sensitive.
As a result, the risk register fills with manageable events, while bigger structural risks remain underdeveloped. People may record, score, and assign a risk such as “late vendor data.” A larger risk such as “the contracting model does not match the owner’s interface management capability” may be avoided because it challenges the whole strategy.
Risk analysis also weakens when contingency is built from a list of discrete threats while ignoring systemic uncertainty. Interface density, immature design, market capacity, organizational capability, and novelty do not always fit neatly into a traditional risk register. Yet they can drive the highest cost and schedule movements.
APM and PMI guidance both recognize uncertainty beyond simple event lists. In capital projects, that distinction is commercial. A team that treats uncertainty as a collection of small risks may underfund contingency and escalation exposure.
Procurement Strategy That Transfers Risk on Paper
Procurement strategy can hide optimism when the project assumes risk has been transferred simply because the contract says so.
A lump-sum EPC contract may appear to cap owner exposure. In reality, incomplete scope, late owner decisions, poor site data, interface ambiguity, extreme market conditions, or change-heavy execution can return risk through claims, disputes, schedule relief, quality issues, or contractor distress.
The same applies to alliance, reimbursable, design-build, EPCM, and framework models. No contracting strategy removes risk. It changes who manages it, who prices it, and how visible it is.
A procurement strategy developed under optimism bias may assume:
The market will accept the preferred risk allocation.
Tenderers will price aggressively.
Contractor capability is available when needed.
Long-lead items can be secured within the schedule.
Interface risk can be managed through coordination meetings.
Local content, customs, import, or labor rules will not affect productivity.
Claims risk is a commercial matter for later.
These assumptions may pass sanction. They do not survive delivery for long.
The Commercial Consequences are Wider Than Overruns
Cost growth and schedule delay are the headline impacts, but optimism bias reaches further into commercial, technical, and operational performance.
Contingency Erodes Before Risk is Retired
When contingency is too low at sanction, it stops being a risk allowance and becomes a political reserve. The project enters execution knowing that normal uncertainty can consume the available headroom.
That creates poor behaviors. Teams delay drawdown requests. They reclassify forecast growth as “opportunities to recover.” They push unresolved scope into later packages. They pressure contractors for savings that may not be real. They defer needed work until change becomes unavoidable and more expensive.
AACE and other cost engineering bodies have long treated contingency as a legitimate allowance for uncertainty within the project scope, not as padding. That distinction matters. If governance treats all contingency as negotiable fat, the project loses one of its basic defenses against known uncertainty.
Claims and Disputes Become More Likely
Optimistic baselines create fertile ground for claims. If the schedule assumes ideal access, contractors will seek relief when access is fragmented. If scope is underdefined, change orders will follow. If interface responsibilities are unclear, parties will protect their positions.
In infrastructure and industrial construction, disputes often arise less from one dramatic breach and more from accumulated misalignment:
The owner expected a fixed outcome.
The contractor priced a defined scope.
The project reality required adaptation.
The contract did not reflect the uncertainty at award.
RICS and ICE guidance on cost, commercial management, and infrastructure delivery often emphasizes clear scope, realistic risk allocation, and sound procurement. Those are not administrative niceties. They reduce the chance that optimism at sanction becomes conflict in construction.
Stakeholder Confidence Weakens
Stakeholders can tolerate bad news better when the project has been candid from the start. Confidence falls when public commitments, board papers, lender models, regulator submissions, or community promises rest on forecasts that later prove unrealistic.
OECD and World Bank materials on government-funded infrastructure have often highlighted the need for sound appraisal, transparent decision processes, and disciplined investment planning. The same principle applies in private capital portfolios. Credibility is an asset.
Once stakeholders believe the project team is always revising upward, they discount every new forecast. That damages trust even when the revised forecast is technically better.
Technical and Operational Readiness Risks Grow
Optimism bias can also push technical risk into operations. A project under cost and schedule pressure may compress commissioning, reduce spares, defer training, limit integrated systems testing, or hand over incomplete documentation.
In utilities, this can affect network reliability. In mining, it can reduce ramp-up performance. In oil and gas, it can destabilize start-up. In industrial facilities, it can shift unresolved design or quality issues into early operations.
The business case may have assumed a production ramp, availability target, throughput rate, or maintenance profile. If the project team does not protect operational readiness, the asset may enter service but fail to deliver the value used to justify approval.
Practical Controls That Treat Optimism Bias as a Governance Risk
There is no single fix. The goal is to build a decision system that makes unrealistic forecasts harder to approve and easier to challenge.
Use the Outside View Before the Inside View Wins
Reference-class forecasting, associated with the work of Daniel Kahneman, Amos Tversky, and later major-project researchers such as Bent Flyvbjerg, starts with a simple discipline: compare the proposed project with outcomes from similar completed projects before relying on the team’s detailed internal story.
The inside view asks, “What do we think will happen on this project?”
The outside view asks, “What happened on comparable projects, and why should this one be different?”
That shift is powerful. It helps challenge beliefs such as:
“This contractor knows the site.”
“Our engineering is more mature than usual.”
“The regulator will support the schedule.”
“This technology is proven elsewhere.”
“The market will be competitive.”
“The brownfield scope is limited.”
Reference classes must be chosen carefully. Don't casually benchmark a remote mining project against a city-center building project. Don't treat a brownfield offshore modification like a clean greenfield module yard job. The value lies in honest comparability, not in finding a flattering peer group.
Separate Advocacy From Assurance
Project sponsors should advocate. Delivery teams should develop the case. But assurance must have enough independence, authority, and timing to challenge the forecast before approval momentum becomes irreversible.
Stronger models include:
Independent cost and schedule review.
External technical peer review.
Independent risk and contingency assessment.
Gate reviews chaired outside the sponsoring business unit.
Commercial challenge of procurement and contracting assumptions.
Operations readiness review before sanction.
Clear escalation routes for dissenting views.
The key is separation. Assurance should not report only to the team whose approval depends on a positive finding. It should report to the investment decision body, portfolio board, or governance function with authority to pause, rework, or condition approval.
Present Ranges, Confidence Levels, and Scenarios
Single-point forecasts invite false certainty. Experienced decision-makers need to see ranges and confidence.
Probabilistic cost and schedule analysis can support this, but only when inputs are credible. A Monte Carlo model built on optimistic ranges will still produce optimistic outputs. The discipline lies in setting ranges, correlations, schedule logic, and how you treat systemic risks.
Good governance should ask:
What is the expected range, not just the base number?
What confidence level is being used for sanction?
What risks are excluded from the model?
How has contingency been formed?
What assumptions would break the business case?
What is the downside case if market, productivity, or permitting assumptions move against the project?
What management actions are available if the downside occurs?
Confidence ranges do not weaken decision-making. They make it more honest.
Benchmark The Estimate and The Execution Strategy
Benchmarking should cover both cost and deliverability. When data exists, you can compare cost per unit, installed quantities, engineering hours, indirects, procurement lead times, craft productivity, module weights, commissioning durations, and owner’s costs against prior projects.
Benchmarking also needs context. A low-cost benchmark from a simple site may not be relevant to a congested live facility. A strong procurement outcome in a soft market may not apply when supply chains are tight. A productivity factor from one region may not translate to another.
Useful benchmarking asks why a project differs, then tests whether evidence supports the difference.
Challenge Assumptions as Decisions, Not Footnotes
Every business case has assumptions. Too often they sit in an appendix and receive less attention than the headline economics.
Assumptions should be treated as decision inputs. For major capital projects, governance should maintain a live assumption register that identifies:
The assumption.
The owner.
The evidence behind it.
The sensitivity to cost, schedule, safety, quality, or value.
The date by which it must be validated.
The decision affected if it proves false.
This is especially valuable in FEED and sanction. If a project still depends on unverified access, uncertain ground conditions, untested technology integration, supplier capacity, or unresolved stakeholder approvals, the investment body should see that clearly.
Make Challenge Safe and Commercially Useful
A healthy challenge process is direct, evidence-based, and early. It does not punish teams for surfacing uncertainty. It also does not reward vague pessimism.
Good challenge sounds like this:
“Show the comparable projects used to support that duration.”
“What discovery allowance is included for brownfield tie-ins?”
“Which interfaces are owned by the contractor and which remain with the owner?”
“What is the basis for construction productivity in this location?”
“How much contingency is linked to defined risk, and how much reflects general uncertainty?”
“Which assumptions must hold for the sanction estimate to remain valid?”
“What would an independent contractor price differently?”
This kind of challenge improves commercial decisions. It gives sponsors choices, such as funding more definition, changing the procurement route, adjusting scope, delaying sanction, increasing contingency, phasing delivery, or accepting the risk with open eyes.
A Better Question for Sanction
The usual sanction question is, “Do we approve this project?”
A better question is, “Do we understand the uncertainty well enough to approve this project on these terms?”
That small change matters. It shifts the discussion from belief to evidence. It also separates three different decisions that are often blurred:
Whether the project is strategically worth doing.
Whether the current scope and execution strategy are ready for approval.
Whether the forecast reflects a realistic range of outcomes.
A project can be strategically important and still not ready for sanction. A project can be commercially attractive and still need more definition. A project can have a strong sponsor and still require an independent reset of cost, schedule, and risk.
Experienced teams know this. The challenge is building governance that lets people state the truth before delivery makes it unavoidable.
Optimism bias is not a defect in cheerful personalities. It is a predictable outcome when capital systems reward approval, compress definition, dilute challenge, and treat uncertainty as a threat to the business case. Manage it as a governance and decision-making risk, and projects are more likely to be funded, contracted, delivered, and operated on terms that reflect reality.
