Why Britain Can't Build: The State Capacity Crisis That Has Nothing to Do With the Size of the State
There is a single figure that contains everything essential about how Britain manages major public investment. When HS2's first trains eventually run between London and Birmingham - probably sometime between 2036 and 2043, two to three decades after the project was first announced - the line will have cost somewhere between £87.7 billion and £102.7 billion at 2025 prices. France builds comparable high-speed railway for around £46 million per mile. HS2 Phase One will cost approximately £396 million per mile. The difference cannot be explained by higher British wages, more hostile British geography, or stricter British safety standards. It is explained by something more structural: Britain has spent forty years hollowing out the institutional capacity to deliver major public undertakings at the cost and on the timeline that comparable countries treat as routine. It spends more, takes longer, delivers less, changes its mind mid-project, and walks away from programmes that have already consumed tens of billions because the numbers have become too politically embarrassing to defend. This is not an occasional failure. It is a pattern. And the deeply inconvenient thing about the pattern is that it predates and runs deeper than any argument about whether the British state should be larger or smaller.
The political argument about public services and public investment is conducted almost entirely on the axis of size. The right says the state is too large, taxes too much, and crowds out private initiative. The left says public services are underfunded, the state has been shrunk beyond the point of effectiveness, and what looks like failure is really the consequence of deliberate under-investment. Both arguments contain partial truths. Neither addresses the central question, which is not how much the state spends but whether it is capable of spending it well. A state that is very large and cannot deliver is not fixed by making it larger. A state that has been reduced to the point of dysfunction is not fixed by reducing it further. What both positions evade is the possibility that Britain's public institutions have, through a combination of deliberate policy, institutional fashion, and accumulated neglect, lost the capability to do the things governments ask them to do - and that rebuilding that capability is a different problem from any argument about size.
The Pattern
The evidence for this proposition is not thin. It is forensic, consistent, and spans every government since Thatcher.
The National Programme for IT in the NHS was launched in 2002 by Tony Blair's government. It was supposed to give every patient in England a centralised electronic care record, linking GP systems, hospitals and pharmacies in a single national digital infrastructure. It was abandoned in 2011. By the time it was wound up, the state had spent an estimated £9.8 billion - against benefits that the Department of Health's own eventual accounting found fell far short of costs. The National Audit Office, reviewing the wreckage, found poor contract negotiation, weak programme management, failure to understand complexity, and a catastrophic mismatch between what the system was supposed to do and the clinical reality of the NHS.
Universal Credit, announced by Iain Duncan Smith's Department for Work and Pensions in 2010, was supposed to consolidate six benefits into one and move eight million households onto the new system by 2017. An early "reset" in February 2013 followed Major Projects Authority warnings about the absence of a basic blueprint. Of £425 million spent by April 2013, £303 million had gone on IT, of which £34 million was subsequently written off. The project had six senior responsible owners. Completion slipped from 2017 to somewhere between 2024 and 2028 - a delay of seven to eleven years depending on which end of the current projection you use. The National Audit Office recorded £1.4 billion in extra costs from a two-year delay alone; lifetime administrative cost estimates rose from £2.2 billion to £15.8 billion.
The Emergency Services Network - the programme to replace the Airwave radio system used by police, fire and ambulance services - began in 2015 with a planned switch-off of the old system in September 2017. By March 2023 the Home Office had spent approximately £2 billion on the new network and £2.9 billion maintaining the old one it was supposed to have replaced six years earlier. The NAO found that "the Home Office does not currently know when ESN will be ready or how much it will cost." The programme's 2019 cost estimate was £9.3 billion - £3.1 billion above the original forecast. The Airwave switch-off has since slipped past 2026 toward 2029.
The Ajax armoured vehicle programme - 589 vehicles for the British Army, contracted in 2014 for £5.5 billion with General Dynamics Land Systems UK - set an initial operating capability date of July 2020. Trials were suspended because the vehicles produced noise and vibration severe enough to injure crew members. The National Audit Office found that "the approach to the programme was flawed from the start." Initial operating capability was eventually declared in November 2025 - eight years late - only to be paused again in January 2026 after soldiers fell ill on exercise. By December 2021 the Ministry of Defence had paid £3.167 billion for 26 delivered vehicles. The contracted price for the whole fleet was £5.5 billion.
These are not the occasional failures that any large organisation produces. They are the same failure, repeated across different departments, different governments, different contractors and different policy domains. The Ministry of Defence under both Labour and Conservative governments. The NHS under both Labour and Conservative governments. The Home Office under both Labour and Conservative governments. The DWP under Conservative governments. The pattern does not discriminate by party. It is systemic, and the question of what is systemic about it is worth addressing with some precision.
The Railway That Ate Itself
HS2 is worth examining in detail because it is the most legible case - the most documented, the most expensive, and the one where the governance failure is most clearly on the public record.
The project was first conceived under Gordon Brown's government, with an initial full-network cost estimate of £37.5 billion covering the Y-shaped route from London to Birmingham, Manchester and Leeds. When the project was formally confirmed in 2012, the full Y-network was costed at £32.7 billion in 2011 prices. The benefit-cost ratio promised £2.40 of value for every £1 spent. These figures were arrived at with design work at approximately 10 per cent completion - a point worth holding in mind, because the consequence of publishing headline costs before design is substantially complete is that the costs are wrong, and the political commitment made on the basis of those costs becomes the first obstacle to managing what the project actually turns out to require.
Costs rose immediately and consistently. By 2013 the estimate was approximately £46 billion; by 2015 it was around £55 billion. In 2019, HS2 Ltd's own chairman Allan Cook produced a review putting the full network cost at between £72.1 and £78.4 billion at 2015 prices, citing ground conditions that were "more challenging than predicted" - though exactly how unpredicted ground conditions in an island that has been industrialised for two centuries can be is a question worth asking. The Infrastructure and Projects Authority rated the northern leg "unachievable." The Oakervee Review, commissioned later that year, recommended proceeding with the full network. That advice was then ignored in stages: the eastern leg to Leeds was cut in the Integrated Rail Plan of November 2021; the Birmingham-to-Manchester leg was cancelled by Rishi Sunak at the Conservative conference on 4 October 2023. The benefit-cost ratio had by that point fallen to 0.8 - officially "poor" value for money on post-pandemic demand assumptions.
By February 2023, £24.7 billion had already been spent. By March 2026, Department for Transport figures showed £44.2 billion spent on the programme overall, including £548 million on Euston works that included £105.6 million of design work subsequently written off, and approximately £2.4 billion on the Phase 2 northern legs before their cancellation. On 20 May 2026, Transport Secretary Heidi Alexander gave the Commons the latest estimate: between £87.7 billion and £102.7 billion at 2025 prices to complete the London-to-Birmingham line only - without the northern extensions that were the original justification for the whole enterprise. First trains between Old Oak Common and Birmingham are now expected between May 2036 and October 2039. Full Euston connectivity is projected for somewhere between 2040 and 2043. The original opening target was 2026.
Alexander told the Commons that two thirds of the cost increase was "down to past misunderstanding of the work required, underestimation and inefficiency." One third was inflation. The Public Accounts Committee's assessment was more direct: HS2 is "a cautionary tale that should be studied by future governments in how not to run a major project."
The governance record is worth itemising because it is not incidental to the cost - it is the explanation for the cost. The project had seven transport secretaries during its lifetime before construction was properly under way. It had six chairmen of HS2 Ltd. Its longest-serving CEO, Mark Thurston, who held the post from 2017 to September 2023, was at one point the highest-paid employee in the British public sector. Construction began in 2020 with design at approximately 10 per cent completion. The Institution of Civil Engineers found a "fatal flaw" in the programme: HS2 Ltd had "delegated too many crucial responsibilities, including design and assurance, to external consultants and contractors," creating a situation in which the client body did not really know what it was buying or managing. A 2025 review by Sir James Stewart - which produced nine recommendations, all accepted by the government - found that "multiple layers of assurance cause duplication and delays, driven by a lack of trust." The state had outsourced so much of the project's substance that it could not trust its own contractors without hiring additional consultants to audit the first set of consultants.
The international comparison is not a minor academic footnote. Britain Remade's analysis found that HS2 London-to-Birmingham works out at approximately £396 million per mile. France builds comparable high-speed railway at around £46 million per mile. Even at HS2's 2013 budget - before the worst of the overruns - independent analysts found it was still 3.7 times more expensive than France's Tours-to-Bordeaux high-speed line. The government's own High Speed Rail International Benchmarking Study found that European high-speed lines average around £32 million per route kilometre, with a range of £11 million to £79 million. HS2 was already at the high end of that range in its early estimates; it has since left that range entirely. The tenfold premium that Britain pays over France is not the result of Britain being Britain. It is the result of Britain lacking what France has always maintained: the institutional machinery to commission and deliver complex infrastructure at realistic cost.
The Digital Graveyard
What happened to the NHS National Programme for IT is worth setting out because it is not merely a story of bad execution. It is a story about the specific institutional incapacity that produces the same outcome in domain after domain, across technology programmes spanning three decades.
The programme was launched in October 2002 with a cost estimate of £2.3 billion over three years. A senior responsible team was assembled; major contracts were let to Accenture, Fujitsu, BT, and the CSC consortium, which subsequently acquired iSOFT - then under Financial Services Authority investigation for accounting irregularities. By June 2006 the NAO was putting the ten-year cost at £12.4 billion.
Accenture withdrew from approximately £2 billion of contracts in September 2006. It faced contractual liability of up to £1 billion. The then Director General of NHS Connecting for Health, Richard Granger, charged it £63 million - roughly six per cent of what it owed. Fujitsu's contract for the southern cluster was terminated in 2008. When the NAO reviewed the abandoned programme, it found that the state had demonstrated "poor negotiating capability" and "weak programme management and oversight," had failed to understand the complexity of the task, and had catastrophically failed to engage the clinical workforce whose buy-in was essential to the entire enterprise. The Department of Health's own benefits accounting, reviewed by the NAO in 2013, showed £3.7 billion of benefits realised against £7.3 billion of costs to March 2012 - and that was after a decade of effort and after write-offs already occurred. The Public Accounts Committee called it one of the "worst and most expensive contracting fiascos" in the history of British public administration.
What happened next is the more revealing part. NHSX was created in 2019 to lead NHS digital transformation. The NAO found in 2020 that "national governance arrangements for digital transformation remain confused." The lessons of NPfIT - engage clinicians early, understand the operational reality before specifying the technology, maintain in-house commercial expertise, do not let contractors design the thing they are then paid to build - had not been institutionally absorbed, because there is no institution through which such lessons are absorbed and retained.
Meanwhile the Home Office was developing Atlas, a casework system to replace the Immigration Directorate's core database CID - a system written in Visual Basic 6 that had been operational since 2000. The Home Office committed in June 2019, in the wake of the Windrush scandal, to moving off CID by March 2020. That deadline was missed. Then a September 2023 deadline was missed. The Infrastructure and Projects Authority gave the Immigration Platform Technologies programme, of which Atlas is the centrepiece, a red rating for 2023 and 2024, meaning "successful delivery appears to be unachievable." As of the NAO's December 2025 report, caseworkers were still referring to the twenty-five-year-old CID database. Published contracts showed £71 million spent on Atlas since 2019, with suppliers including Accenture, Mastek and PA Consulting; a September 2022 accounting assessment put the wider Immigration Platform Technologies programme at £406 million since 2014, with a further £66 million to come. Former Independent Chief Inspector of Borders and Immigration David Neal - who was subsequently dismissed after criticising the Home Office publicly - told a parliamentary committee that Atlas "served to slow things down" and that "poor data is everywhere."
The consequences were not abstract. The asylum backlog stood at more than 175,000 people at end-June 2023, up 44 per cent year on year. Hotel accommodation for asylum seekers exceeded £6 million per day at the peak. The Rwanda scheme - the political response to the processing failure - cost a total of approximately £700 million, relocated four volunteers, and was abandoned by the incoming Labour government. The NAO calculated per-person costs of up to £150,874 over five years. The relationship between the IT failure and the policy failure is direct: the Home Office cannot process asylum claims efficiently because the casework system it has been trying to replace for six years still does not work, running on infrastructure that should have been retired a decade ago.
The Hollowing Out
To understand why these failures keep happening, it is necessary to understand what happened to the state's internal capability over four decades.
The Thatcher government's Next Steps initiative, from 1988 onward, separated policy from delivery by creating arm's-length executive agencies. The immediate logic was coherent enough: concentrate ministers on strategy, and let professional managers run operations. What it also did, over time, was begin separating the policy-making civil service from the experience of actually implementing things. If you never run a service, you lose - across cohorts and across generations - the institutional feel for what making promises in policy documents actually requires of the people who have to carry them out. The policy says "an electronic care record for every patient" without anyone in the room who has spent a decade working in NHS IT understanding why that is not a four-year technology project.
The Blair government's contribution was outsourcing on an unprecedented scale. Large swathes of IT capability, operational management and specialist professional functions moved to the private sector - to Capita, to Fujitsu, to Serco, to the management consultancies. This transfer of function had a consequence that was not fully reckoned with at the time: the state became, progressively, an organisation that could commission things but not do them. The gap between commissioning and doing is precisely where the NHS IT programme failed, where the Home Office digital programme failed, and where HS2 failed. You need in-house technical expertise to write specifications that work, to negotiate contracts that protect the public interest, and to identify early when a contractor is telling you what you want to hear rather than what is true. Britain spent twenty years reducing its supply of exactly that expertise, and the bill has been mounting ever since.
The data on civil service churn reflects the structural problem. In 2023 and 2024, around 12.7 per cent of civil servants moved department or left entirely - the second-highest turnover rate since 2010. The Treasury, which is supposed to provide the financial oversight and challenge that keeps major programmes honest, consistently loses around 20 per cent of its staff per year. The Institute for Government's research found that in countries with stronger delivery records - Australia, Canada, Germany - it is common for senior officials to spend the majority of a career in one department, accumulating the institutional knowledge and professional relationships that keep large programmes on track. In Britain, the average tenure of a Senior Responsible Owner on a major government project has been under two years. The average major project runs for a decade or more.
Pay makes the problem structural rather than incidental. The Institute for Government's analysis found that the median Director General in the British civil service earns around £138,000, against £185,000 to £395,000 for comparable roles in local government and the NHS. Permanent Secretary Cat Little acknowledged in public that the civil service "was never going to be able to compete on salary." When the private sector pays three times as much for the same level of expertise, the civil service gets what it can afford - which is systematically less experienced, more junior, and more likely to move on than the expertise on the other side of the table during contract negotiations with Deloitte or General Dynamics. This is the structural heart of the problem: the people writing the contracts are outgunned by the people reading them, and have been for decades.
The Rented State
The consequence of this hollowing out is a dependency that has become so embedded it has ceased to attract serious comment: Britain has rented its own missing capability back from the private sector, at premium prices, and the firms providing it are in large part the same firms that built the dependency in the first place.
Government management consultancy spend reached a record £3.7 billion in the financial year 2024 to 2025, up approximately 6 per cent on the prior year's figure of roughly £3.4 billion, according to Tussell's analysis published in May 2026. This figure is almost certainly an undercount: the National Audit Office warned in November 2025 that the government "lacks a clear, consistent picture" of its consultancy spending because departments use incompatible definitions and inconsistent tracking methods. Central government management consulting grew 79 per cent over five years to £1.9 billion; defence consulting grew fastest, rising 196 per cent from £70 million to £207 million. Ministerial departments spent £459 million on the eight largest consulting firms in 2019 and 2020 alone - a 45 per cent rise over two years, based on Tussell's analysis of more than 11,000 invoices.
The firms at the top of the public earnings table are consistent across years and departments. Deloitte has been the largest public sector earner five consecutive years running. On HS2 specifically, PwC received approximately £102 million in fees, Deloitte £86 million, EY £25 million and KPMG £9 million - at least £222 million across those four firms on a single infrastructure programme that was subsequently found to have been chronically under-designed, poorly specified and inadequately managed. The consultants were not a solution to HS2's governance problem. They were, in significant part, its mechanism.
The NHS Test and Trace programme is the case that most vividly illustrates what this model produces in practice. The programme was given a budget of £37 billion over two years. The Public Accounts Committee found in October 2021 that its aims were "overstated or not achieved," and its chair Dame Meg Hillier specifically attributed responsibility to "the over-priced consultants who 'delivered' this state of affairs." Deloitte had approximately 900 consultants working on the programme at day rates of £1,000 to £1,100; some individual daily rates reached £6,624. A total of 2,300 consultants from 73 firms were involved at various stages. McKinsey was paid £563,400 for a roughly seven-week review to define the "vision, purpose and narrative" of NHS Test and Trace - a function that, in a state with sufficient in-house capability, would have been settled by the officials responsible for running the programme. It was not settled by those officials because those officials did not have the expertise, the status or, in some cases, the seniority to do so without external validation. McKinsey later won a £1 million contract to advise on the structure of the post-pandemic NHS England. The structural failure it had helped to create became the subject of advice it was paid to provide.
The argument sometimes made in defence of consultant dependency is that government brings consultants in for specific, bounded tasks and retains ownership of the strategy. The evidence does not support this framing. The dependency has grown consistently across every administration that has promised to reduce it. Jacob Rees-Mogg's 2022 review found that only 8 per cent of major project spending in 2019 - £35 billion of £432 billion - had robust evaluation plans, making it structurally impossible to determine whether consultant advice was adding value or not. Rachel Reeves pledged on taking office in 2024 to halve the consultancy bill, identifying £550 million of savings in 2024 to 2025 and £1.2 billion by 2026. The bill proceeded to reach a record high of £3.7 billion.
The revolving door between the civil service and the firms that serve it is a structural feature rather than a series of individual choices. Dido Harding, who led Test and Trace, began her career at McKinsey. The Advisory Committee on Business Appointments, which is supposed to manage such transitions, is not a statutory body and has only advisory powers. It cannot prevent the movement; it can only, occasionally, comment on it. What the revolving door produces is a consultancy market that understands Whitehall intimately - its processes, its anxieties, its preference for elaborate frameworks over direct accountability - and a civil service that has increasingly come to depend on external validation for decisions that should be within its own competence.
The Parliament Problem
Beneath all of the institutional pathologies described above is a structural mismatch that no reform programme has yet addressed with any seriousness: the time horizon of democratic politics and the time horizon of major infrastructure are fundamentally incompatible, and Britain has made no institutional provision for managing that incompatibility.
The parliamentary cycle runs to five years. A general election typically changes ministers, often changes government, and consistently produces pressure to differentiate the new administration from what preceded it - which in practice means revisiting, redesigning or abandoning commitments made by predecessors. Major infrastructure projects, meanwhile, run for ten, twenty or thirty years. HS2 was conceived under Labour, approved under Conservative, redesigned under Conservative, partly cancelled under Conservative, and its remaining section is now being completed under Labour. Seven transport secretaries presided over it before construction was properly under way. The Oakervee Review observed with some understatement that there would "almost certainly be dozens more" over the project's remaining life.
Universal Credit had six senior responsible owners over roughly a decade. The National Programme for IT had multiple health secretaries across its nine-year life. Each change of minister brings a new political instinct about what the programme should look like, which officials are trusted, and which contractors are favoured. Each redesign adds cost. Each addition of a new ministerial priority to a programme designed around a different set of priorities adds complexity that the underlying systems were not built to handle. The result is the pattern visible in every case study above: programmes that are consistently more expensive and more delayed than their initiating estimates suggested, because those estimates were produced for the programme as originally specified, not the programme as repeatedly amended by successive ministers trying to make it their own.
The interaction between political churn and project timelines also shapes how cost estimates are produced in the first place. A minister who announces a major programme at an honest cost estimate that the electorate finds alarming will lose the announcement. A minister who announces at an optimistic cost estimate that later proves wrong will have moved on by the time the reckoning arrives. The incentive structure rewards optimism bias at the point of announcement and punishes honesty that might delay approval. The result, documented by Oxford academic Bent Flyvbjerg in research covering 258 transport projects across 20 nations, is what he calls "survival of the unfittest": the projects with the most unrealistic estimates tend to be the ones that clear the approval threshold, because realistic ones look too expensive to proceed. Britain's tendency to begin construction with design barely started - HS2 at 10 per cent design completion - is the institutional expression of this dynamic. If you wait for full design, the real cost becomes visible. Better to break ground while the optimistic figure is still defensible.
The Obstacle Course
There is a further pathology that is structurally distinct from those described above, though it interacts with all of them: the legal and planning environment for major infrastructure in Britain is, by international comparison, extraordinarily permissive of delay - even for projects that have already been approved and are considered nationally necessary.
The Nationally Significant Infrastructure Projects regime, introduced in 2008, was supposed to create a faster and more certain route to consent for major projects. It has, in practice, generated its own complications. Legal challenges against Development Consent Orders spiked to 58 per cent of decisions in recent years - meaning more than half of all approved major projects faced a court challenge. Of 30 such challenges examined, only four approvals were actually overturned. The legal system mostly delays rather than corrects: the projects that were challenged but not overturned simply lost time, at considerable public cost. Each legal challenge takes an average of 1.4 years to resolve. Major road projects have paid up to £121 million per scheme in costs attributable to legal delay alone. The time to prepare an NSIP application has doubled from around one year to around two years since the regime's inception, driven in part by the anticipation of legal challenge and the pre-emptive documentation required to resist it.
HS2's Phase One Environmental Impact Assessment ran to approximately 50,000 pages. The government attempted to insulate the project from judicial review by authorising it through a hybrid Act of Parliament - the HS2 Act received Royal Assent in 2017 - but legal challenges came anyway. Heathrow's third runway has been through multiple rounds of legal challenge across more than a decade. Every contested wind farm consent generates its own round of litigation. Planning is not simply the process of deciding whether a project should proceed; it has become the first battlefield in a war that continues regardless of the planning outcome.
The Planning and Infrastructure Act 2025 introduced reforms targeted at some of this: removing the paper permission stage, limiting "totally without merit" challenges to a single attempt rather than three, and establishing target timescales for decisions. The government estimates the reforms will speed major projects by an average of twelve months and unlock up to £7.5 billion over a decade. Whether that estimate is more reliable than HS2's original benefit-cost ratio is a question that will be answered over time rather than in advance.
How Other Countries Do It
The contrast with France is not, in the final analysis, merely about cost per mile. It is about what kind of institutional infrastructure a state maintains in order to be capable of delivering major public investments at all.
France has, since the eighteenth century, maintained a permanent cadre of technically expert civil servants - the Corps des Ponts et Chaussées, founded in 1716, whose associated engineering school was founded in 1747 and is the oldest civil engineering school in the world. The Corps - now the Corps des Ingénieurs des Ponts, des Eaux et des Forêts - is recruited substantially from the École polytechnique and has supplied France's state with engineering expertise that is institutional rather than personal. Individual engineers retire; the corps persists. The knowledge of how to write a workable specification, manage a contractor who is gaming the process, and identify the point at which a design is sufficiently complete to start attaching costs to it - these things are embedded in a permanent institution rather than held by individuals who will leave for McKinsey in two years' time.
The Société du Grand Paris, the dedicated delivery body for the 200-kilometre Grand Paris Express metro project, exemplifies what this makes possible. Established by statute in 2010 with a defined remit and professional staff drawn substantially from the permanent engineering corps, it employed over 820 people by 2021, the majority of them infrastructure specialists. Infrastructure Australia's research arm assessed the model as achieving delivery through "concentration of resources and expertise in a delivery agency," contrasting it explicitly with the "potential inefficiencies arising from government agency-led delivery" - by which it meant the British model of dispersing accountability across departments, arm's-length bodies and rotating consultants. France's approach is not failure-proof: the Cour des Comptes found in 2017 that the programme's supervisory board lacked the technical and financial expertise to oversee decisions effectively, and costs grew from an initial €19 billion to somewhere around €35 to 42 billion. But the programme has not been abandoned or redesigned mid-construction. It is being built. When it is finished, Paris will have a 200-kilometre automated metro built in roughly fifteen years. Britain has been trying to build 225 kilometres of high-speed rail for fifteen years and has not yet opened a station.
Germany offers an honest counterexample that this article should not evade. Berlin Brandenburg Airport opened in October 2020, nine years late, at roughly €6 to 7 billion - approximately three times its initial estimate. The failure concentrated in fire-safety system engineering and coordination breakdowns that the project's governance structure was unable to resolve. It is a genuine case of a developed state with strong technical traditions producing a catastrophically managed project, and it argues against any simple story about institutional culture producing reliable outcomes. But BER is one airport, a single failure in a country with an otherwise strong infrastructure delivery record. Britain's failures are systemic across rail, IT, defence procurement and welfare technology, spread across four decades and two parties. One airport against a pattern of institutional incapacity is not a symmetrical comparison.
Bent Flyvbjerg's research at Oxford's Saïd Business School, covering 258 transport projects across 20 countries, found that nine in ten produce cost overruns, with rail showing an average overrun of 44.7 per cent and demand overestimated on average by 51.4 per cent. The phenomenon is genuinely international. But it is not evenly distributed. Countries that maintain a permanent technical civil service, that require designs to be substantially complete before costs are published and commitments made, and that sustain project leadership through the natural life of the programme, consistently show lower overruns than countries that do none of these things. Britain shows up at the expensive end of nearly every international comparison for which data exists. That is not geography. It is policy.
What This Is Actually About
It is worth being precise about what the analysis above is and is not arguing.
It is not arguing that the British state should be larger. A larger version of an institution that cannot manage contracts, cannot sustain senior responsible ownership through a programme's natural life, and cannot maintain in-house technical expertise would simply spend more money achieving the same outcomes. The failure of NPfIT was not a failure of budget; it had £12.4 billion. The failure of Test and Trace was not a failure of resource; it had £37 billion. The failure is institutional, not financial, and no amount of additional funding directed at institutions that lack the capability to use it well will produce different results.
It is not arguing that it should be smaller. The state capacity problem predates the Thatcher-era reductions and has worsened through administrations that both expanded and contracted public spending. Austerity made some of the problems worse - reducing civil service headcount without rebuilding in-house capability simply increased consultant dependency - but the dependency was already growing before 2010. The Conservative argument that private sector discipline is the answer is not supported by the evidence of thirty years of outsourcing, which produced £3.7 billion in annual consultant fees and a succession of failed IT programmes that private contractors delivered, late and over budget, on behalf of a client state that lacked the expertise to manage them.
What the evidence argues is something that both sides of the conventional debate systematically avoid: Britain's public institutions, as currently constituted, do not have the in-house expertise to commission complex IT systems at realistic cost, to manage major infrastructure programmes through their natural lifespan, or to negotiate contracts with large private sector organisations on anything approaching equal terms. This is a capability problem. The answer to a capability problem is not spending more or spending less. It is rebuilding the capability. That is politically inconvenient because it requires admitting that decades of policy across both parties have been wrong in ways that cannot be corrected quickly or cheaply, and that the remedies will produce no visible dividend within a single parliament.
The current Labour government has named some of this correctly. Pat McFadden's "test and learn" reform agenda, the £100 million innovation fund supporting delivery pilots from January 2025, and Heidi Alexander's HS2 reset under CEO Mark Wild - who cut the corporate centre by around 300 roles, aligned the railway to European technical standards and reformed the governance structure - are genuine attempts rather than purely rhetorical ones. Wild's appointment carries its own implicit message: he was brought in from the Crossrail delivery team, an acknowledgement that the individual who knows how to deliver is more valuable than the institution that does not. The Planning and Infrastructure Act 2025 makes a genuine start on the legal challenge environment. These are not nothing.
But the Institute for Government's verdict on McFadden's broader agenda is sobering. His lines, the IfG noted, "could have been said by Francis Maude fifteen years ago." Maude, as Cabinet Office minister in the coalition government, wanted public services that were "agile, flexible and digital by default" and made the same promise about reducing consultant dependency. The consultancy bill in Labour's first year in office reached a record high. The diagnosis has been made before, repeatedly, by ministers of both parties. The diagnosis is not the problem. The problem is that the cure requires sustaining a programme of institutional rebuilding across multiple parliaments, in the face of the short-term political pressures that produced the original failure - which is precisely the thing that Britain's political system has consistently proved unable to do.
What Would Have to Change
The conditions under which the verdict above would change are specific rather than rhetorical.
For Britain to deliver infrastructure at anything approaching the cost that comparable countries manage, it would need to maintain, over generations, a permanent cadre of technically expert civil servants - engineers, commercial specialists, digital professionals - with protected career paths as specialists rather than generalists, and pay that acknowledges the market rather than ignoring it. The recurring observation that delivery capability resides in specific individuals rather than institutions - that it was the Elizabeth Line team, brought in wholesale, that was tasked with rescuing HS2 - would have to be reversed. That means accepting that the civil service will never fully compete with the private sector on total compensation, while building a status and career structure around technical expertise that makes public service an attractive alternative for specialists who want to work at scale on consequential things. It means something analogous to, though not a copy of, what the French Corps des Ponts has provided for three centuries. It would take a generation to build and would require political will to sustain through multiple governments - which is precisely what the current political system has consistently failed to provide.
For consultant dependency to reduce in any meaningful sense, cutting the bill and rebuilding the capability would have to happen simultaneously and in that order. A blunt cut to consultant spend without the capability to replace it recreates the dependency at the next crisis, as the pandemic demonstrated with perfect clarity: the civil service could not stand up a national contact-tracing operation without importing thousands of Deloitte consultants because it had spent the previous decade systematically reducing the operational staff who might otherwise have done it. Contracts with remaining consultants would need genuine knowledge-transfer obligations - not as a box-ticking clause but as a condition of payment, audited by the National Audit Office - so that expertise is internalised rather than rented indefinitely.
For project costs to reflect reality from the outset, business cases would need to be built from the actual outcomes of comparable completed projects - reference-class forecasting, as Flyvbjerg argues - rather than from the optimistic bottom-up estimates produced at 10 per cent design completion. HS2's costs were locked in politically before anyone had a realistic grasp of what they would actually be; the political commitment to an unreal figure was then the primary obstacle to honest management of the real one. This is not a technical accounting problem. It is a governance problem created by a political system that rewards the announcement of affordable-sounding projects and punishes the honesty that might delay their approval.
For the planning environment to stop adding years to projects that have already been democratically approved, the 2025 reforms would need to produce measurable reductions in application times and legal challenge costs - and where they do not, further reform would follow rather than a retreat to the status quo. The test is whether application times actually fall from the current average of two years, and whether the rate of legal challenge declines in response to the new single-attempt limit.
And for any of this to happen, a government would need to say, in public and in specific terms, that Britain has spent four decades systematically degrading its own ability to build and deliver; that this happened under governments of both parties for reasons that made short-term political sense; that the ideological argument about state size has been a distraction from the operational question of state capability; and that the remedies are long-term, unglamorous, and will produce no visible political dividend within a single parliament. No current party is saying this. The closest thing to an honest account is scattered across National Audit Office reports, Infrastructure and Projects Authority red ratings, and Public Accounts Committee hearings where permanent secretaries are pressed to explain why the same things keep going wrong - hearings that attract minimal press coverage and generate reports whose recommendations are routinely accepted and then quietly shelved.
The tenfold premium that Britain pays over France for comparable railway infrastructure is not really a story about railways. It is the price of four decades of institutional choices - to outsource capability rather than maintain it, to rotate ministers and officials through programmes rather than sustain them across the full project life, to prioritise the announcement over the preconditions for delivery - that have made Britain one of the most expensive and least reliable places in the developed world to commission major public infrastructure. The choices were made by governments of both parties. The costs are being paid, and will continue to be paid, by everyone.