Why Large Projects Fail: The Structural Causes

Why large projects fail examined: optimism bias, strategic misrepresentation, governance failure, and complexity, tested against megaproject research and documented cases.

By Kateule Sydney — Researcher and Writer | Last Verified: 2026-10-03 | Published by Kat-Syd Resources Hub

Why large projects fail: the structural mechanisms behind megaproject underperformance

Table of Contents

• 1. Optimism Bias Makes Forecasts Wrong Before the Project Starts
• 2. Strategic Misrepresentation and Optimism Bias Operate Together
• 3. Governance Weaknesses Make Corrective Action Difficult
• 4. Complexity Turns Small Errors Into Compounding Failures
• 5. What the Evidence Shows About Preventing Failure
• Judgement
• References

1. Optimism Bias Makes Forecasts Wrong Before the Project Starts

The first structural mechanism of large-project failure is not execution. It is estimation. The forecasts that justify a project can be systematically wrong, and they can be wrong in a predictable direction. This is not necessarily a failure of individual managers. It is a cognitive pattern documented across industries, countries, and decades.

This article focuses particularly on megaprojects and other large, complex projects, where the research literature identifies recurring systemic patterns rather than a single universal cause of failure.

Definition. Optimism bias is the tendency for individuals and groups to underestimate the likelihood of negative events and overestimate positive outcomes. In project management, optimism bias can produce underestimates of costs and schedules and overestimates of benefits, which can subsequently appear as cost overruns, schedule delays, and benefit shortfalls. The planning fallacy, described by Kahneman and Tversky, is the related tendency to underestimate the time, costs, and risks involved in completing tasks while overestimating the benefits and overall success.

The evidence on transportation forecasting accuracy is striking. In a large sample of transport projects, nine out of ten rail projects overestimated passenger demand, with actual ridership averaging 51 percent below forecast. For road projects, 50 percent had traffic forecasts differing from actual traffic by more than 20 percent, while 25 percent differed by more than 40 percent. In the same body of research, nine out of ten transportation projects underestimated construction costs. Flyvbjerg's analysis found that forecasting inaccuracy persisted across time and geography: travel-demand forecasts showed no improvement over the 30-year period studied, while comparable cost-estimate data showed no improvement over roughly 70 years.

The pattern is not limited to transportation. The Channel Tunnel exceeded its estimated costs by approximately 80 percent, and the Sydney Opera House cost approximately 1,400 percent more than the original estimate. The Suez Canal figure of 1,900 percent overrun is also reported in the megaproject literature. These are not isolated curiosities. They illustrate a broader pattern documented in larger samples of megaprojects, in which cost overruns are frequent and sometimes very large.

Why this matters. The forecasts that justify large projects can be structurally optimistic. The project can be approved on numbers that have a predictable bias toward underestimating cost and overestimating benefit. By the time the bias becomes visible, the project is already committed. The distinction between optimism bias and deliberate deception is important, and the next section addresses it directly. But the first point is that even with honest forecasters, the numbers can be wrong.

Carry forward: Optimism bias produces underestimates of cost and schedule and overestimates of benefit. The evidence on transportation projects shows the pattern persists across decades and countries.

2. Strategic Misrepresentation and Optimism Bias Operate Together

Optimism bias does not by itself explain every politically favorable forecast. A second mechanism identified in the megaproject literature is strategic misrepresentation: the deliberate presentation of information in ways intended to secure approval or other strategic advantages. The two mechanisms are analytically distinct and can operate together, and the second can make the first worse.

Definition. Strategic misrepresentation is the deliberate and systematic distortion or misstatement of information for strategic purposes. Bent Flyvbjerg, the most cited scholar on megaprojects, identified two main causes of misinformation in management and policy: strategic misrepresentation, which he characterizes as deception, and optimism bias, which he characterizes as delusion. The two are not alternatives. They can reinforce each other.

The logic is political. To gain support and consent for a project, sponsors may deliberately distort or misstate information. Optimism bias can enable what social scientists call strategic ambiguity, a political ploy to leave room for manoeuvre when things do not work out as planned. Under strategic misrepresentation, the information presented to decision-makers may be systematically more favorable than the information that would result from a neutral assessment of costs, benefits, risks, and uncertainty.

Key point. The distinction between optimism bias and strategic misrepresentation matters because the remedies are different. Optimism bias requires better forecasting methods. Strategic misrepresentation requires accountability and incentive alignment. Flyvbjerg's reference class forecasting method primarily addresses optimism and the planning fallacy by anchoring estimates in empirical outcomes from comparable projects. Because it reduces reliance on project-specific promotional forecasts, it can also reduce the scope for strategic misrepresentation. Daniel Kahneman called it the single most important piece of advice regarding how to increase accuracy in forecasting.

Carry forward: The forecasts that justify large projects can be wrong because of optimism bias and strategic misrepresentation working together. Correcting one without the other can leave the structural mechanism in place.

3. Governance Weaknesses Make Corrective Action Difficult

Optimism bias and strategic misrepresentation explain why weak projects get approved. Governance weaknesses help explain why projects may continue after serious problems become visible. The third structural mechanism is the difficulty of challenging assumptions, escalating problems, changing direction, or stopping a project when the evidence supports that decision.

Definition. Governance in project management refers to the decision rights, oversight mechanisms, and accountability structures that determine how decisions are made and reviewed. A systematic literature review funded by the Project Management Institute analyzed 6,007 titles and abstracts and 86 full papers, identifying 18 causes of poor megaproject performance grouped into six themes. The review found that no single factor was sufficient to explain poor megaproject performance; rather, multiple interrelated causes were identified across six themes: decision-making behaviour; strategy, governance, and procurement; risk and uncertainty; leadership and capable teams; stakeholder engagement and management; and supply-chain integration and coordination.

Governance failure is not simply a lack of decision-making. It can occur when decision structures, incentives, or accountability mechanisms make it difficult to challenge assumptions, escalate problems, change direction, or stop a project when circumstances warrant it. Decision rights, escalation mechanisms, sunk costs, contractual commitments, political pressures, and the availability of credible recovery options can all influence whether a troubled project is continued, restructured, or stopped. That is why the gate structure I discussed in my post on the project life cycle exists precisely to make stopping a genuine option.

The governance failure is visible in documented cases. The National Programme for IT in the NHS was launched in 2002 and formally dismantled as a single national programme in 2011, although its component programmes and contracts continued under separate arrangements. The Department's original contracts with CSC totaled £3.1 billion. By 2013, the full cost was still not certain, with a forecast of £9.8 billion that excluded future costs. Benefits to March 2012 were estimated at £3.7 billion, just half of the costs incurred. The Public Accounts Committee reported poor negotiating capability, weak programme management and oversight, failure to understand the complexity of the tasks, and failure to get people to operate the systems even when they were adopted.

The Edinburgh tram project provides a second case. The original budget was £545 million. The revised budget for the truncated line was £776 million, while the eventual reported cost of that line was approximately £852.6 million. The full Edinburgh tram scheme was ultimately completed to Newhaven in 2023 at a total cost of about £1.043 billion. The Edinburgh Tram Inquiry identified serious and systemic failings, particularly in relation to the lack of proper oversight.

Why this matters. Governance weaknesses are not a lack of decision-making. They are conditions in which decision-making cannot effectively challenge, escalate, or stop. When gates become approval stamps, the project continues past the point where the evidence supports continuation.

4. Complexity Turns Small Errors Into Compounding Failures

The fourth structural mechanism is complexity. Large projects are not simply bigger versions of small projects. They have more components, more interfaces, more dependencies, and more ways for small errors to propagate. The failures that result can arise from interactions that are difficult to predict or manage when the system is designed and integrated under tight schedule and coordination constraints.

Definition. Complexity in large projects arises from interconnected technical components, organizational interfaces, dependencies, institutional relationships, supply chains, and the difficulty of predicting how changes in one part of the system will affect others.

A cross-national study of construction-project delays provides another perspective. Across projects and countries, commonly identified delay factors included payment delays, financial problems, inaccurate time estimates, owner-driven requirement changes, design changes, site conditions, late owner decisions, inadequate planning, and contractor-selection problems. These findings illustrate the interaction of financial, design, managerial, and procurement factors rather than establishing a single cause of delay.

The Denver International Airport automated baggage system is the clearest documented example. The system was designed around thousands of radio-controlled carts, approximately 4,000 in total, traveling over roughly 17 miles of track. The system was supposed to identify each bag and send it to its correct destination automatically. Testing exposed the weakness: bags were misrouted and damaged, carts jammed and derailed, and the system struggled under realistic operating conditions. The airport opening was delayed by approximately 16 months, with the baggage-system problems contributing substantially to the delay and adding roughly $560 million to the airport's costs according to commonly cited project analyses. The automated baggage system itself had a much smaller direct system cost. A conventional baggage system had to be installed alongside the automated equipment, and the automation was used only in a limited portion of the airport.

The failure was not a single mistake. It was the interaction of an unproven technology, unprecedented scale, and a construction project that was still in progress while the baggage system was being installed. The system had to track individual bags, avoid collisions, and respond instantly when something went wrong. A problem with one component affected a much larger sequence of operations. Changes elsewhere in the airport required changes to the baggage system that was supposed to serve it.

Carry forward: Complexity is not a reason to avoid large projects. It is a reason to design them differently. Possible design responses include modular architectures, clearer interfaces, staged integration and testing, and deliberate management of dependencies. These approaches are consistent with the broader megaproject literature's emphasis on system architecture, integration, and coordination. That same logic about system design under complexity appears in my post on supply chain design and resilience, where design decisions made before disruption determine whether the system can absorb it.

5. What the Evidence Shows About Preventing Failure

The structural mechanisms of large-project failure are documented. The evidence on prevention is less settled than the evidence on the existence of recurring performance problems. The literature nevertheless identifies several recurring areas for improvement, including better forecasting, stronger governance, system architecture, collaboration, stakeholder engagement, and supply-chain integration.

Definition. Reference class forecasting is a method developed by Flyvbjerg that uses historical data from comparable projects to adjust estimates, rather than relying on the specific project's own forecast. The method addresses optimism bias and the planning fallacy by anchoring estimates in what actually happened on similar projects.

The PMI-funded systematic literature review identified 54 cures across the 18 causes. The review recommends understanding megaprojects as production systems, encompassing all stages from planning through design, manufacturing, construction, integration, and handover to operations. The research agenda includes designing the system architecture, bridging the gap with manufacturing, building and leading collaborations, engaging institutions and communities, and decomposing and integrating the supply chain. That same integration logic appears in my post on operations management, where the design of the operating system determines whether the firm can execute or is limited by its own structure.

Why this matters. The evidence on prevention is less mature than the evidence on causation, but the direction is clear. Better forecasting methods, stronger governance, system architecture, and deliberate management of dependencies address the structural mechanisms directly. The remedies are not purely technical. They span forecasting, governance, organizational behaviour, system architecture, collaboration, stakeholder engagement, and supply-chain integration.

Judgement

The evidence shows that poor performance in megaprojects cannot be explained solely by individual effort or competence. Structural and systemic factors—including decision-making, governance, risk, leadership, stakeholder relationships, and supply-chain integration—play important roles. The four mechanisms discussed in this post—optimism bias, strategic misrepresentation, governance weakness, and complexity—are an analytical framework for understanding recurring patterns in megaproject performance, not the complete empirical taxonomy of megaproject failure.

My position is that the structural causes are best understood as institutional, not individual. The individual project manager should therefore not automatically be treated as the sole cause of an optimistic forecast or continued investment. Individual judgment operates within institutional incentives, governance arrangements, political commitments, contractual constraints, sunk costs, and organizational culture. Changing outcomes therefore requires attention to the system in which project decisions are made, not only to the people making them.

The strongest counterargument is that cost, schedule, and benefit performance do not fully determine whether a large project ultimately creates value. Some projects perform poorly against their original forecasts but nevertheless produce substantial long-term social, economic, technological, or cultural benefits. The evidence therefore supports treating cost overruns and benefit shortfalls as important performance problems without assuming that every project with an overrun is an overall failure.

My recommendations for practitioners are these:

• Use reference class forecasting — begin with empirical evidence from comparable projects and explicitly model the probability and magnitude of cost, schedule, and benefit deviations rather than assuming that the project-specific forecast will be accurate
• Make gates real — a gate that cannot produce a stop decision is unlikely to provide the full governance value of a genuine gate, and the governance weakness that allows failing projects to continue can be expensive
• Design for modularity — reduce the number of interdependencies that can propagate failure, and test components before integrating them
• Separate optimism from accountability — better forecasting methods address bias, and accountability mechanisms address misrepresentation, but neither works alone
• Treat the base rate as the starting point — begin with empirical evidence from comparable projects rather than assuming the project-specific forecast will be accurate

The final argument of this post is that the evidence suggests that large projects can be exposed to institutional incentives that favour optimistic forecasts, continued commitment, and political momentum, while the complexity of delivery creates additional opportunities for problems to compound. Improving outcomes therefore requires changing not only project-level practices but also the systems in which projects are proposed, approved, governed, and delivered.

References

• Denicol, J., Davies, A., & Krystallis, I. (2020). What Are the Causes and Cures of Poor Megaproject Performance? — Project Management Journal, 51(3), 328–345
• Flyvbjerg, B. (2008). Public Planning of Mega-Projects: Overestimating of Demand and Underestimating of Costs — TRID
• Flyvbjerg, B., Bruzelius, N., & Rothengatter, W. (2003). Megaprojects and Risk — Cambridge University Press
• Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk — Econometrica, 47(2), 263–291
• Public Accounts Committee. (2013). The Dismantled National Programme for IT in the NHS — UK Parliament
• Edinburgh Tram Inquiry. (2023). The Edinburgh Tram Inquiry Report — Edinburgh Tram Inquiry
• United States General Accounting Office. (1995). New Denver Airport: Impact of the Delayed Baggage System — GAO
• Project Management Institute. From Nobel Prize to Project Management: Reference Class Forecasting — PMI

Read Also on Kat-Syd Resources Hub

Project Life Cycle: Phases, Gates, and Transitions — Phases, the PMI Process Groups, phase-gate decision models, and how predictive, adaptive, and hybrid structures distribute project risk.

PMBOK Guide: Origins, Structure & Limits — The PMI standard, the Eighth Edition's principles and domains, and the criticisms of rigidity.

Supply Chain Design and Resilience — Network design, risk categories, visibility and resilience strategies tested against Apple and Safaricom.

Written by Kateule Sydney — Researcher and Writer

© 2026 Kat-Syd Resources Hub. All rights reserved.

Kat-Syd Resources Hub — Educational case studies and analytical reference