The Risk Outside the Central List
Researched by an agentic pipeline · reviewed and gated by the author
Above all, reconcile every headline improvement for changes in the denominator.
The green dashboard problem
A central portfolio can improve overnight without a single project improving.
Move the hardest work out of the reporting boundary, admit fewer uncertain projects, or retain only those that merit intensive intervention, and the dashboard changes. The centre appears more focused. The proportion of red projects may fall. Review teams have more time. None of those observations tells us whether risk has fallen across the system.
That is the tension behind the United Kingdom’s 2026 reset of the Government Major Projects Portfolio. From 1 April, the centrally supported portfolio reduced from more than 200 projects to about 80. The published list contains 81. Entry now normally reflects government priority, whole-life cost above £1 billion and a judgement that central support will materially improve delivery confidence; exceptional entry remains possible for strategically important, very high-risk or critical national infrastructure work. [S1]
The reform is a serious attempt to allocate scarce assurance expertise rather than spread it thinly. It is also a live test of a problem that reaches well beyond government: how can a small centre concentrate attention without allowing consequential risk to disappear beyond its line of sight?
The answer is not a larger central list. It is a stronger federated assurance contract.
Assurance is an allocation decision
Executives readily accept that capital and specialist talent are scarce. Assurance capacity is rarely treated with the same discipline. Reviews accumulate because a project is large, visible or politically sensitive. The most capable reviewers are assigned by precedent. Every initiative receives roughly the same governance ceremony, regardless of whether another hour of challenge is likely to change a decision.
This produces two failures. A centre that tries to examine everything becomes procedural: it records risks but cannot investigate them deeply enough to alter the course of work. A centre that narrows its remit without redesigning the surrounding system becomes selectively blind: it improves the quality of scrutiny inside the boundary while losing confidence about what sits outside it.
The UK reset recognises the first failure. NISTA says the smaller portfolio allows more targeted support, tiered assurance and clearer escalation, while departments retain responsibility for delivery. Its 2025–26 annual report is careful about the timing: the report covers 189 projects as at 31 March 2026, immediately before the new model took effect. [S2] Those figures are a baseline, not a result.
That distinction matters. The pre-reset portfolio represented £924.2 billion of whole-life cost and included 29 green, 109 amber and 34 red projects, with 17 exempt from published delivery-confidence ratings. [S2] A delivery-confidence assessment is a forecast at a point in time, not a realised outcome, and NISTA itself notes that portfolio composition changes continuously. Neither the £924.2 billion nor the rating distribution describes the performance of the new 81-project portfolio.
Concentrating assurance is valuable only when the organisation separates seeing all material work from intervening intensively in selected work.
The operating contract around the centre
A tiered model works through a chain of decisions, not through the size of the central list.
First, the organisation maintains a complete inventory of material programmes, including work below the threshold for central attention. Second, it selects where central challenge has the greatest expected value: high consequence, unusual complexity, cross-boundary dependency, weak local capability or a decision that is difficult to reverse. Third, departments or business units govern the remaining work within a common standard. Fourth, comparable data preserve a system-wide view. Fifth, deterioration or strategic change triggers escalation and, when necessary, re-entry. Finally, benefits and outcomes remain visible after central attention ends.
The April reforms include several parts of this mechanism. Government Project Delivery described common reporting for both centrally listed and departmental major projects through the Government Reporting Integration Platform, revised integrated assurance and approval planning, and an early-development portfolio. [S3] These are not administrative accessories. They are the means by which a smaller centre can remain connected to a larger delivery system.
Consider an illustrative enterprise group with 140 major initiatives and a central assurance team capable of conducting twelve deep reviews each quarter. Under the old model, every initiative submits a quarterly pack and the centre conducts abbreviated reviews across the whole population. The process produces coverage, but little challenge.
The group then selects 35 initiatives for intensive central attention. A cyber transformation with cross-business dependencies stays in the central tier. A regional warehouse programme moves to the operating company because the local portfolio office has strong commercial and delivery capability. A smaller identity programme stays outside the central tier on cost, yet its dependency on the cyber transformation is recorded in the common portfolio data.
Six months later, the warehouse programme remains locally governed. The identity programme, however, misses a critical integration gate and creates a new dependency for three central initiatives. A pre-agreed trigger moves it into central assurance for a bounded intervention. Once the dependency is stabilised, it can leave again without losing benefits ownership or outcome tracking.
This is not a case for permanent central control. It is a case for an explicit service contract: what the centre can see, when it intervenes, what authority follows intervention and how work returns to local ownership.
The strongest case for a smaller centre
The argument for concentration deserves to be stated at full strength. More central oversight is not automatically better governance. It can duplicate mature departmental controls, slow decisions and weaken local accountability. Review teams that cover too much work learn less about each programme. Project leaders optimise for committee passage rather than expose the uncertainty that serious challenge requires.
A narrower central portfolio can reverse those incentives. Expert reviewers can spend longer on consequential choices. The centre can intervene earlier and develop pattern recognition across the hardest work. Departments with mature portfolio functions can act closer to delivery, using context that central teams lack. Inclusion no longer becomes a substitute for ownership.
Civil Service World’s comparison of the old and new lists found 71 projects retained, ten added and 133 absent from the new central list. [S4] The absences have different explanations: some projects were completed, replaced or no longer met the criteria; absence does not mean cancellation or lower risk. That heterogeneity reinforces the case for differentiated treatment. It also makes raw comparisons between the old and new portfolios analytically unsafe.
The risk outside the denominator
The sceptical case is equally strong. Delegation works only when local capability is real, yet NISTA reports that portfolio-management maturity varies across departments. [S2] Common reporting can preserve visibility only if data are complete, comparable and used in decisions. An escalation route matters only if it can change resources, scope or authority.
History supplies a sharper warning. In 2018, the National Audit Office examined projects that had left the central portfolio. For 22 of 48 sampled projects, it could not determine whether intended outcomes had been achieved. [S5] Processes have changed since then, so this is not evidence that the 2026 design will fail. It is evidence that exit governance and benefits continuity are control problems, not clerical ones.
The evaluation baseline is also weak. A 2025 government review found that 34 per cent of the 2023–24 portfolio had robust evaluation plans under its assessment; 66 per cent did not provide evidence of plans meeting that standard. [S6] If more work moves to departmental ownership without strengthening evaluation, the system can gain autonomy while losing evidence.
There is a further composition problem. Suppose the central portfolio’s red share declines after the reset. That could mean deeper assurance changed decisions. It could also mean difficult projects moved out, newly selected projects entered at different lifecycle stages, or rating practices changed. NISTA explicitly says delivery-confidence distributions do not track a consistent set of projects over time. [S2] The honest test must therefore examine the whole major-project system and compare like with like.
Four controls that make federation credible
The federated assurance contract rests on four controls.
System-wide visibility
The centre needs a complete view of material cost, schedule, dependency, benefit and delivery risk, even where it does not govern intensively. A common data platform is useful only when identifiers and definitions remain stable enough to trace work across entry, exit, merger and replacement.
Entry and re-entry rules
Cost is a crude proxy for the value of attention. Selection should also reflect strategic consequence, reversibility, dependency and the capability of the local owner. Re-entry triggers should be observable: material forecast movement, a dependency crossing business boundaries, loss of sponsor capability or a decision exceeding delegated risk appetite.
Credible local capability
Delegation should follow evidence, not organisational optimism. The relevant test is whether the local portfolio can challenge assumptions independently, allocate scarce resources, resolve dependencies and escalate bad news. The National Audit Office’s 2025 work on mega-projects emphasises clear authority, accountability, sponsor capability and investing-organisation control. [S7] Formal review architecture cannot compensate for their absence.
Benefits after exit
Leaving central assurance must not end outcome accountability. Benefits owners, evaluation dates and evidence standards should survive the transition. Otherwise the centre learns which projects passed its gates but not which investments created value.
Measure the system, not the list
No post-reset evidence yet shows whether the 81-project portfolio improves delivery, realised benefits or decision quality. The appropriate response is neither faith nor cynicism. It is a measurement design that can distinguish concentration from recategorisation.
Track review depth, time to decision and material changes produced by assurance. Compare retained and delegated projects after adjusting for complexity, lifecycle and baseline risk. Record why each project enters, exits or returns. Measure data completeness outside the central tier. Follow realised cost, schedule, scope and benefits rather than treating confidence ratings as outcomes.
Above all, reconcile every headline improvement for changes in the denominator.
A small centre can be a sign of governance maturity: it sees the whole system, intervenes where its attention has value and leaves capable owners free to deliver. It can also be a cleaner dashboard over an unchanged field of risk. The difference is not the number of projects on the central list. It is whether the organisation can still see, challenge and learn from the projects that are not.
Sources
- National Infrastructure and Service Transformation Authority — Government Major Projects Portfolio — 31 March 2026 — https://www.gov.uk/government/publications/government-major-projects-portfolio
- National Infrastructure and Service Transformation Authority — NISTA Major Projects Annual Report 2025–26 — 13 July 2026 — https://www.gov.uk/government/publications/nista-major-projects-annual-report-2025-26/nista-major-projects-annual-report-2025-26
- Government Project Delivery — Government strengthens project delivery and accountability — 1 April 2026 — https://projectdelivery.gov.uk/2026/04/01/government-strengthens-project-delivery-and-accountability/
- Civil Service World — Which schemes are in the new streamlined Government Major Projects Portfolio? — 2 April 2026 — https://www.civilserviceworld.com/in-depth/article/in-out-shake-it-all-about-which-schemes-are-in-the-new-streamlined-government-major-projects-portfolio
- National Audit Office — Projects leaving the Government Major Projects Portfolio — 19 October 2018 — https://www.nao.org.uk/wp-content/uploads/2018/10/Projects-leaving-the-Govenment-Major-Projects-Portfolio.pdf
- Evaluation Task Force / UK Government — Government Major Projects Evaluation Review — 23 April 2025 — https://www.gov.uk/government/publications/government-major-projects-evaluation-review/government-major-projects-evaluation-review-html
- National Audit Office — Lessons learned: Governance and decision-making on mega-projects — 14 March 2025 — https://www.nao.org.uk/wp-content/uploads/2025/03/lessons-learned-governance-and-decision-making-on-mega-projects.pdf