The Green Book's fix for megaproject blowouts is a mandatory optimism bias adjustment, made explicit before work starts. High Speed Two got that adjustment, overran it anyway, and the government's own later arithmetic now rates the original approval as poor value for money.
Estimated reading time: 11 minutes
In short. The Green Book is HM Treasury's guidance on how every central government proposal is appraised, and its flagship defence against megaproject cost blowouts is a mandatory, explicit adjustment for optimism bias, made before a project is approved. High Speed Two Phase 1 received exactly that adjustment in April 2020: a benefit-cost ratio of 1.2, already barely inside the Green Book's own "low value for money" band. Six years and roughly 60 billion pounds later, the Department for Transport's own paperwork says that if the true costs had been known at the time, the ratio would have been 0.3 to 0.4, poor value for money by the Treasury's own published bands. The evaluation that is supposed to feed this lesson back into future adjustments is not due until 2050. Four documented mistakes: three stars.
Picked from the registry's queued Tier 3 entries: the next-highest-scoring candidate after this week's Sourcing Playbook teardown, verified live on GOV.UK with no withdrawal notice. It scores on sector spend above a billion pounds, an update to a tracked doctrine, high search salience and no enforcement or review clause of its own, and it names its own rich seam: appraisal optimism bias. This is a Tier 3 deep teardown, the document's own words tested against the fullest evidence the record allows.
The rule the document exists to enforce
HM Treasury published the latest Green Book on 5 February 2026, describing it as "radically shorter and simpler" than its predecessor, with "duplication and wordiness reduced throughout" [1]. It remains live guidance, checked on GOV.UK on 13 September 2026 with no withdrawal notice [1]. Its core mechanism for stopping projects being approved on rosy numbers is set out in Chapter 6, "Shortlist appraisal": "Optimism bias is the demonstrated systematic tendency for practitioners to be over-optimistic about key assumptions in appraisal, such as social costs, social benefits or project duration... Practitioners must account for optimism bias by making explicit adjustments at the outset of an appraisal" (p.44) [2].
High Speed Two is the single largest live test of that rule anywhere in the state. It is also, by the government's own admission, the case in which the rule visibly did not hold.
The claims, tested
| The Green Book's claim | What we found | Verdict |
|---|---|---|
| A shortlisted option is chosen through "a balanced judgement based on monetisable costs and benefits, unmonetisable costs and benefits, public sector financial impact, risks and uncertainties" (p.31) [2] | HS2 Phase 1's April 2020 Notice to Proceed approval carried a benefit-cost ratio of 1.2, already assessed as "low value-for-money" [4]. The Department's own May 2026 paperwork says that on a like-for-like basis, using today's known costs, the ratio would be 0.3 to 0.4 [5], poor value for money under HM Treasury's own published bands [10] | The balanced judgement that approved the project would have failed the Green Book's own floor, on the government's own later arithmetic |
| "Practitioners must account for optimism bias by making explicit adjustments at the outset of an appraisal" (p.44) [2] | The National Audit Office found in January 2020, before Notice to Proceed, that HS2 Ltd's contingency of 7 billion pounds, 37% of forecast Phase One costs, "was not enough to address the significant increases in cost that emerged" [6]. Academic analysis of NAO data later found the main civil construction estimate had nearly doubled "well beyond the highest suggested optimism bias adjustment for civil engineering work" [8] | The adjustment was made, on schedule, by the book. It was not enough, on the record, twice |
| Evaluations should let "practitioners... use data from previous interventions to identify historical forecast errors and derive appropriate optimism bias adjustments" (p.73), and departments "are required to register their planned, live and completed evaluations on the Government Evaluation Registry" (p.74) [2] | HS2 Phase 1's own Evaluation Registry entry lists the evaluation stage as "Planned", no link to an evaluation plan, no published report, and a final report expected in December 2050 [9] | Registered, not done, and not due for another 24 years |
| The Green Book's own publication page states that "the 2003 edition and the 2022 edition of the Green Book can be found on the National Archives website" for "academic study and public information" [1] | The link labelled "2022 edition" resolves to a National Archives capture of an unrelated Planning Inspectorate document, "Norfolk Parishes Movement for an OTN, comments on written representations" [3] | A document about citing evidence correctly does not correctly cite its own back catalogue |
One point two, then nought point three to nought point four
The Five Case Model is the Green Book's spine, and the economic case exists to answer one question: does the preferred option deliver value for money. Chapter 6 says the answer comes from "a balanced judgement based on monetisable costs and benefits, unmonetisable costs and benefits, public sector financial impact, risks and uncertainties" [2]. For HS2 Phase 1, that judgement was made public on 14 April 2020, when the Department for Transport and HS2 Ltd's joint Accounting Officer Assessment approved Notice to Proceed on a "cost estimate of 35 to 45 billion pounds (2019 prices)" and recorded a benefit-cost ratio of "1.2:1, including wider economic impacts", explicitly "assessed as 'low' value-for-money" [4]. Under HM Treasury's own bands, a ratio below 1.5 sits in "low"; below 1.0, it falls into "poor" [10]. At 1.2, the project that would go on to become the single most scrutinised infrastructure programme in the country was approved only just clear of that floor.
It did not stay clear of it. The Department's own Accounting Officer Assessment for the 2026 reset, published 19 May 2026, sets Phase 1's cost at "87.7 billion to 102.7 billion pounds" and states plainly that "the value for money assessment has been conducted in accordance with Green Book methodology" [5]. Against a counterfactual of cancelling the scheme, it judges the ratio "likely exceeds 1.5". Then comes the sentence that matters most: "In the interest of transparency and to enable a comparison of historic BCR trends on a like-for-like basis, if current costs were known at the time of Notice to Proceed (2020), the range would be 0.3 to 0.4" [5]. That is not a rounding error inside "low". It is a fall straight through the floor, into the band HM Treasury's own guidance defines by example as a scheme whose costs exceed its benefits [10].
The adjustment that was made, and was not enough
The Green Book does not merely recommend caution about optimism bias. It requires a specific mechanism: costs and timeframes must be increased, benefits decreased, "to provide a more realistic view", with the size of the adjustment informed by "the evidence base of the originating organisation" and, as risks are retired, "practitioners should then reduce the optimism bias adjustment in proportion to the level of risk prevented" (p.44) [2]. HS2 Ltd used exactly this method before Notice to Proceed. The National Audit Office's January 2020 report on the programme found that HS2 Ltd's cost estimate "resulted in 7 billion pounds of contingency being set, 37% of forecast future Phase One costs. The amount of contingency was not enough to address the significant increases in cost that emerged as the design became more detailed" [6]. That finding predates the 14 April 2020 approval by three months. The adjustment was made, checked by the National Audit Office, found wanting, and the project proceeded on a revised range regardless.
It went wrong the same way a second time. Writing in February 2024, the academics Cahal Moran and Ganga Shreedhar traced how the initial optimism-bias-adjusted estimate for "main civil construction" of 5,751 million pounds had, within a few years of construction starting, already grown to 10,667 million pounds, "a near-doubling of costs only a few years into the commencement of the project", which they describe as "well beyond the highest suggested optimism bias adjustment for civil engineering work" in HM Treasury's own guidance [8]. Their explanation is the mechanism itself: the Green Book's own instruction to "reduce the optimism bias adjustment in proportion to the level of risk prevented" as a project matures gives practitioners a standing invitation to release the buffer early, before the risks it was covering have actually passed [8]. The safeguard is not being ignored. It is being followed exactly as written, on the one project where following it has visibly failed to catch the bias it is named for.
An evaluation due the year the railway is meant to have been open for a decade
Chapter 10 closes the loop the first two chapters open. Evaluation is supposed to correct the next appraisal: "practitioners should use data from previous interventions to identify historical forecast errors and derive appropriate optimism bias adjustments" (p.73), and every department "is required to register their planned, live and completed evaluations on the Government Evaluation Registry", with the expectation that reports are "placed in the public domain" (p.74) [2]. The registry itself, run by the Cabinet Office and HM Treasury's Evaluation Task Force, describes itself as "a single home for evaluations across Government" that "any member of the public can use... to search and browse".
Searching it for HS2 returns one entry: "Evaluation of High Speed 2, Phase 1", led by the Department for Transport, evaluation stage "A planned evaluation", no link to an evaluation plan, no published report, and an events table giving "Publication of final results" a date of December 2050 [9]. The Green Book's own bar for registration has been cleared: the entry exists, and it is honest about its own stage. But the mechanism Chapter 10 describes, using this project's real forecast errors to correct the next one's optimism bias adjustment, cannot function on a report due a quarter of a century from now, for a railway whose own current schedule range puts services running from 2036 at the earliest [5]. Every practitioner appraising every major project between now and then will have made their own optimism bias adjustment, on the historical record available at the time, without the single biggest data point the Green Book says they should be using.
A citation that does not go where it says
Not every failure here is about billions of pounds. HM Treasury's own Green Book publication page carries a short paragraph on its predecessors: "Previous editions of the Green Book are now withdrawn. For purposes of academic study and public information, the 2003 edition and the 2022 edition of the Green Book can be found on the National Archives website" [1]. The 2003 link resolves correctly. The 2022 link, checked live on 13 September 2026, is a National Archives capture, dated 7 December 2024, of a document titled "Norfolk Parishes Movement for an OTN, comments on written representations", a Planning Inspectorate submission with no connection to appraisal guidance of any edition [3]. It is a small thing next to a 60 billion pound cost swing. It is also the kind of error that a document built entirely around the discipline of checking a source before you rely on it should not carry in its own footnotes.
Credit where due
The sentence that makes this piece possible is one HM Treasury and the Department for Transport did not have to publish. Nothing in the Green Book or in Managing Public Money required the Department to calculate what the 2020 approval would score using today's costs; the reset's own value for money case already clears the bar it needed to clear, on the cancellation-counterfactual basis the Department chose to use. Publishing the 0.3 to 0.4 comparison anyway, under a heading that says "in the interest of transparency", is the single most useful sentence in either Accounting Officer Assessment, and it is the government marking its own past work down in public rather than waiting for the National Audit Office to do it first [5]. The National Audit Office's own June 2026 report on the reset credits the Department and HS2 Ltd with "taking a considered approach" this time, aligned with the Office for Value for Money's lessons on megaprojects [7]. And the 2026 Green Book's own editorial reset, cut down and restructured specifically to reduce "duplication and wordiness" after a dedicated review, is a genuine improvement in the document as a document [1]. None of that changes what the numbers say. All of it is real.
Verdict
Three stars, from four documented mistakes in a document whose central safeguard against exactly this outcome, an explicit, mandatory optimism bias adjustment, was applied to the single largest capital programme in the state and did not hold, twice, on a record the government itself now publishes. None of the four is a case of HS2 Ltd or the Department breaking a rule the Green Book sets out. Each is the rule being followed and the outcome it was written to prevent happening regardless. A rulebook whose worst live test is also its most transparent one is not nothing. It is also not the "more realistic view" Chapter 6 promises.
The star score counts four documented mistakes against the Green Book's own text: the balanced-judgement, benefit-cost-ratio principle (p.31) tested against HS2 Phase 1's 1.2 ratio at Notice to Proceed and the Department's own 0.3 to 0.4 like-for-like restatement six years later; the mandatory optimism bias adjustment (p.44) tested against the National Audit Office's finding that HS2 Ltd's pre-approval contingency was insufficient and academic analysis showing the adjustment mechanism itself allowed the buffer to be released early; the evaluation feedback loop and Government Evaluation Registry requirement (p.73 to p.74) tested against HS2 Phase 1's own registry entry, still at the planned stage with a final report due in 2050; and the publication page's own broken citation to the withdrawn 2022 edition, which links to an unrelated Planning Inspectorate document. Four falls in the 4 to 9 band: three stars; the bands are on the ratings page. Analysis is of the Green Book as served from GOV.UK on 13 September 2026, last updated 5 February 2026. The document is tracked in the Daily Register; a revision will get a diff entry.
Sources
- HM Treasury / GOV.UK, "The Green Book" (publication page, checked live 13 September 2026, last updated 5 February 2026, no withdrawal notice). https://www.gov.uk/government/publications/the-green-book-appraisal-and-evaluation-in-central-government
- HM Treasury, "The Green Book" (2026 edition, PDF, 88pp). https://assets.publishing.service.gov.uk/media/698dbcd17da91680ad7f4308/The_Green_Book_2026.pdf
- UK Government Web Archive / National Archives, capture dated 7 December 2024 of the URL linked from source 1 as "the 2022 edition", in fact a Planning Inspectorate document, "EN010109-001912, Norfolk Parishes Movement for an OTN, comments on written representations". https://webarchive.nationalarchives.gov.uk/ukgwa/20241207185311/https:/infrastructure.planninginspectorate.gov.uk/wp-content/ipc/uploads/projects/EN010109/EN010109-001912-Norfolk%20Parishes%20Movement%20for%20an%20OTN%20-%20Comments%20on%20Written%20Representations%201.pdf
- Department for Transport / HS2 Ltd, "High Speed 2 (HS2) Phase One: Notice to Proceed, Accounting Officer Assessment" (dated 14 April 2020, published on GOV.UK). https://www.gov.uk/government/publications/government-major-projects-portfolio-accounting-officer-assessments/high-speed-2-hs2-phase-one-notice-to-proceed-accounting-officer-assessment-april-2020
- Department for Transport, "High Speed 2 (HS2) Phase 1: decision to continue delivery following programme reset, Accounting Officer Assessment summary" (19 May 2026). https://www.gov.uk/government/publications/department-for-transports-accounting-officer-assessment-summaries-for-the-government-major-projects-portfolio/high-speed-2-hs2-phase-1-decision-to-continue-delivery-following-programme-reset-accounting-officer-assessment-summary-may-2026
- National Audit Office, "High Speed Two: A progress update", HC 40, Session 2019-20, 24 January 2020. https://www.nao.org.uk/wp-content/uploads/2020/01/High-Speed-Two-A-progress-update.pdf
- National Audit Office, "High Speed Two reset", HC 52, Session 2026-27, 29 June 2026. https://www.nao.org.uk/reports/high-speed-two-reset/
- Cahal Moran and Ganga Shreedhar, "HS2 reveals the pervasiveness of optimism bias in government decision making", LSE British Politics and Policy blog, 5 February 2024. https://blogs.lse.ac.uk/politicsandpolicy/hs2-reveals-the-pervasiveness-of-optimism-bias-in-government-decision-making/
- Evaluation Registry (Cabinet Office / HM Treasury Evaluation Task Force), "Evaluation of High Speed 2, Phase 1" (checked live 13 September 2026). https://evaluation-registry.cabinetoffice.gov.uk/search/ba119288-a763-47dd-aa35-cd81b0030ca6/
- HM Treasury, "Value for Money: supplementary guidance on categories" (November 2024). https://assets.publishing.service.gov.uk/media/673e1cd02ff787d4e01b08d6/value-for-money-supplementary-guidance-on-categories.pdf