Key Takeaways
- Research spanning 16,000+ projects shows 91.5% of megaprojects miss cost, schedule, or both - and only 0.5% deliver on all three measures of success.
- The UK's Office for Value for Money confirmed in June 2025 that every UK megaproject since 1980 has exceeded its budget.
- Three current case studies - HS2 (£46bn+ spent, no agreed baseline), Hinkley Point C (cost nearly doubled to £35bn in 2015 prices), and California High-Speed Rail ($126bn vs $33bn voter-approved) - illustrate the same structural failures.
- The systemic fix is independent governance - separating those who promote a project from those who assess it, with publicly named assurers empowered to stop work.
- South Korea's PIMAC model screened out enough loss-making schemes to save an estimated USD 101 billion between 1999 and 2017.
Government megaproject failure is not a series of unlucky accidents. It is a statistical pattern, documented across decades and continents, that continues to repeat because the structural incentives that produce it remain largely unchanged.
In 2025 and 2026, three of the world's most prominent infrastructure programmes - the UK's HS2 high-speed rail, Hinkley Point C nuclear power station, and California's High-Speed Rail - have each delivered fresh evidence of the same underlying problem. Three separate UK governance reviews, published within months of each other, reached the same diagnosis. This post examines why government megaprojects fail so consistently, who bears the cost, and what independent governance looks like when it works.
What Does the Data Actually Show About Megaproject Failure?
The most comprehensive evidence base on megaproject performance comes from Professor Bent Flyvbjerg's Oxford database, now covering more than 16,000 projects across 136 countries. As documented in How Big Things Get Done (2023), the findings are stark: 91.5% of megaprojects miss their cost or schedule targets. Only 8.5% land both. Only 0.5% deliver on cost, time, and benefits combined.
Flyvbjerg describes this as the "iron law" of megaprojects: over budget, over time, under benefits, over and over again. Rail projects average 44.7% cost overruns with demand overestimated by 51.4%. A separate McKinsey study of 48 troubled megaprojects found poor execution was responsible for cost and time overruns in 73% of cases, with bridges and tunnels averaging 35% cost overruns.
The causes are well established. Optimism bias - the cognitive tendency to underestimate cost and overestimate benefit - is part of the picture. But the more damaging driver is strategic misrepresentation: the deliberate understating of costs and overstating of benefits by project promoters competing for funding. As Flyvbjerg's research demonstrates, the distribution of overruns is asymmetric in a way that pure cognitive bias cannot explain. Promoters know the real numbers would not survive sanction, so they present numbers that will.
What Do HS2, Hinkley Point C and California HSR Reveal in 2026?
HS2 is the clearest UK example. Originally estimated at £20.5 billion (2019 prices) for Phase 1, HS2 Ltd's own June 2024 estimate placed the cost at £54-66 billion against a £44.6 billion funding envelope. By end-February 2026, £43.6 billion had been spent on the active programme - £46.2 billion including the cancelled Phase 2 - with no agreed baseline cost. CEO Mark Wild told the Public Accounts Committee that construction started before mature design consents were in place, describing the rush to begin as "in retrospect, a mistake".
Hinkley Point C was approved in 2016 at £18 billion. EDF's 2025 annual results raised the cost to £35 billion in 2015 prices - approximately £48-49 billion in 2026 prices - and pushed first power to 2030, five years behind the original target. The project took a €2.5 billion impairment, contributing to a 26% drop in EDF's group net income.
California High-Speed Rail was voter-approved in 2008 at $33 billion for a San Francisco-to-Los Angeles system by 2020. The Authority's Draft 2026 Business Plan puts Phase 1 at $126.3 billion, with an un-optimised scenario reaching $231 billion. No track had been laid by mid-2025. The project's former peer review chair, Lou Thompson, wrote to the Legislature in March 2026 that the project "has reached a dead end".
Why Do Government Projects Specifically Fail Worse Than Private Ones?
Three structural features distinguish public-sector megaprojects from their private-sector equivalents.
Electoral cycles are shorter than project timelines. A UK Parliament sits for five years; HS2's lifespan from sanction to opening exceeds twenty. Ministers who approve a project rarely see it completed. The OVfM's June 2025 study noted that political support waxes and wanes, with scope and objectives changed without full understanding of the impact on cost or schedule.
Strategic misrepresentation is institutional, not individual. Promoting departments are incentivised to understate cost because honest numbers would lose the funding contest against rival schemes. The OVfM's own language is direct: it is "difficult to challenge estimates from organisations who are incentivised to understate costs and timelines".
Internal reporting marks its own homework. The gap between reported status and observed reality - what delivery teams tell sponsoring departments versus what an independent reviewer would find on site - is a structural feature, not a personnel failure. The NAO's March 2025 report on megaproject governance explicitly noted that governance arrangements that look suitable on paper do not always work as intended in practice. For Boards overseeing capital projects, this gap between dashboards and delivery reality is the central governance risk.
The result is escalating commitment. Once a project is large enough, cancelling it becomes politically more costly than continuing - even when the marginal benefit-cost ratio falls below 1.0. The PAC's February 2025 conclusion on HS2 illustrates the trap: cancellation would write off approximately £24.6 billion of sunk costs (2019 prices), so continuation is rationalised even at a benefit-cost ratio that, without the cancellation-avoidance benefit, would not represent value for money.
What Does Effective Independent Governance Look Like?
Three converging UK reviews in 2025 reached the same prescription: separate project promotion from project assessment.
The Stewart Review (June 2025) found the HS2 governance structure "not fit for purpose", with the Department for Transport simultaneously holding the roles of sponsor, funder, policy owner, client, and shareholder. It recommended a three-tier model - Shareholder Board, Programme Board with independent chair, and Delivery Board - modelled on the London 2012 Olympics, Sizewell C, and Crossrail, all of which had an external buffer separating delivery from political pressure.
The OVfM study proposed five changes the Treasury accepted in full: a Strategy and Delivery Plan signed by PM, Chancellor, and Secretary of State; streamlined decision-making; feasibility studies before commitment; a fixed capital envelope in construction with the right to move money between years; and pay flexibility for specialist project leadership.
International models reinforce the principle. South Korea's PIMAC - an independent unit inside the Korea Development Institute, reporting to the Ministry of Economy and Finance rather than line ministries - conducted mandatory feasibility studies on 767 projects between 1999 and 2017. Only 63.3% were assessed as feasible. The remainder were stopped before they consumed taxpayer money, saving an estimated KRW 141 trillion (USD 101 billion). Norway's external quality-assurance regime (KS1/KS2) has been shown in peer-reviewed research to materially reduce cost overruns on major projects.
The common thread across all of these models is that independent assurers are publicly named, sign off the cost estimates, and can stop a project - not merely flag concerns. A review that cannot say "no" is not assurance. It is theatre.
What Should Boards and Project Directors Be Doing Now?
For organisations overseeing capital programmes in the energy, minerals, and resources sector, the implications are concrete.
Demand reference-class cost and schedule ranges for every major capital project, prepared by an independent third party using comparable-project data rather than the in-house cost model. Reference class forecasting - basing estimates on actual outcomes of similar projects rather than the inside view of the promoter - bypasses both optimism bias and strategic misrepresentation.
Separate the sponsor from the delivery function, even within the same organisation. Anyone whose career or bonus depends on sanction cannot also be the assurer. The Stewart Review's diagnosis of the DfT/HS2 relationship is the clearest warning of what happens when these roles are combined.
Insist on named, independent assurers whose estimates are the ones the Board votes on. Watch for the early signals of strategic misrepresentation: suspiciously narrow cost ranges at sanction, benefit-cost ratios that just clear the hurdle, claims that this project is unique and will not behave like others, and pressure to start construction before design maturity reaches approximately 70%. All three case studies above showed every one of these signals before the overruns materialised.
The data is unambiguous. Government megaproject failure is not a management problem. It is a governance problem. And governance problems require structural solutions - not better spreadsheets, but independent oversight with the authority to act.
PDAS provides independent project assurance and governance for capital programmes in the energy, minerals, and resources sector - verifying that reported status reflects observed reality.
Book a discovery call with our team.
