Why Britain Cannot Make Good Policy Last

TWIS Deep Research Paper No. 1 — citation-pinned manuscript
Draft status: v0.4, unpublished and not editorially accepted
Evidence cut-off: 31 July 2026¹
Research record: issues #166, #168 and the DRP-01 evidence dossiers
Adversarial review: docs/TWIS_DRP_01_MANUSCRIPT_ADVERSARIAL_REVIEW_V0.1.md
Claim audit: docs/TWIS_DRP_01_CLAIM_AUDIT_V0.1.md
Source pinpoints: docs/TWIS_DRP_01_SOURCE_PINPOINTS_V0.1.md

Scope note: the title names an institutional vulnerability, not a measured national failure rate. The five cases in this paper were selected to test mechanisms. They cannot establish how often British governments preserve good policy or how often they destroy it.²

Executive summary

British governments must be able to change policy. Elections would be hollow if a later government could not repeal legislation, redirect spending, end a contract or replace an institution. Durability cannot mean permanent rule by an earlier Parliament, and continuity is not a virtue when a system has failed, become dangerous or ceased to serve its purpose.

The narrower problem examined here is the quality of major reversal.³ British central government can materially weaken, abolish, fragment or replace ordinary public systems without a consistent public requirement to show four things before capacity withdraws:

1. what function must continue;
2. what evidence shows that repair is inferior to replacement;
3. whether the successor is legally, financially and operationally ready;
4. what money, knowledge, relationships and public protection will be lost, transferred or rebuilt.

Government records expenditure more reliably than it records delivery capacity. Redundancy payments, contract totals and grant reductions may be visible. The value of an experienced team, a trusted local route into help, a functioning data series, a technical standard or a network of professional relationships is often scattered across documents or left unmeasured. When a later government returns to the same public problem, the rebuilding cost appears under a different programme and a different political story.

Five cases test this claim.

Sure Start and children’s centres were not abolished nationally through one decision. Statutory duties remained, and local authorities reorganised provision differently. The removal of protected funding during severe local fiscal pressure nevertheless exposed preventive services, staff and referral relationships to uneven weakening before the longest-term outcomes had matured. Later research found meaningful health and educational benefits. Those findings were not fully available in 2010, but they show why preventive systems need an evidence horizon long enough to reveal their effects. The final judgement in this paper is substantial institutional weakening and partial abandonment with uneven local effects, not one national abolition.

Regional Development Agencies were expensive, uneven and indirectly accountable. Reform was defensible. The stronger criticism concerns sequence. Statutory regional bodies closed before equivalent successor capacity was ready; independent audit found a major dip in local-growth spending and activity; Local Enterprise Partnerships later carried large responsibilities with weak staff and skills; and their core functions moved again in 2024. The best-supported judgement is institutional churn, qualified by fiscal consolidation and real differences between the old and new structures.

Zero Carbon Homes was cancelled in July 2015 after roughly nine years of development and only months after a final impact assessment reaffirmed implementation from 2016. The allowable-solutions model was complex, and government had a legitimate interest in regulatory burden and housing delivery. The public record recovered for this research does not contain an equivalent cancellation appraisal comparing implementation, simplification, stronger on-site requirements, a fixed delay and full withdrawal. A related objective returned through the different Future Homes Standard about a decade later. The judgement is destructive cancellation followed by delayed functional restoration. The realised national cost remains unknown.

Transforming Rehabilitation is the necessary corrective case. The reform addressed a real gap by extending statutory supervision to people released after short prison sentences. Its divided operating model, commercial assumptions and information interfaces failed badly enough that independent audit, inspection and parliamentary evidence supported ending the contracts and reuniting offender management in a public Probation Service. The judgement is evidence-led reasonable correction after failed and damaging institutional fragmentation. Reunification did not erase transition costs or restore an effective service automatically.

The Low Pay Commission is the durable comparison. Rates, age thresholds, targets and political ambitions have changed repeatedly while the institution has retained a recurring evidence process involving employers, workers and independent members. Ministers remain able to reject recommendations with reasons. Its survival may owe much to popularity, adaptation and path dependence as well as design. It still demonstrates that democratic change can occur inside a continuing institution. The judgement is durable adaptive institution.

The broader constitutional and administrative system helps explain why these failures are possible, but not why each decision occurred. A government with Commons support normally has a direct route to ordinary legislative and administrative change. Ministers often hold relevant posts for less time than policies need to mature. Treasury guidance already requires whole-life appraisal, alternatives, risk and evaluation, yet application is uneven and transition accounts remain fragmented. Influence records show advocacy and access more often than causation. None of those facts proves a hidden machine or one national cause.

The paper proposes a Durable Policy Standard for major public systems. Its purpose is not to block elected government. It would require a responsible minister to publish a decision record before material withdrawal, and a departmental accounting officer to state whether the necessary evidence and readiness information exist. The record would define the continuing public function; present decision-time evidence and the scheduled review position; compare repair, staged change and replacement; account for whole-life cost and distribution; inventory capacity; test successor readiness; disclose material influence and independent challenge; and set a post-change evaluation date. High-risk transitions would receive independent assurance. Urgent correction would remain available, with shortened stages and a retrospective review made public.

Britain does not need permanent policy. It needs major change to carry its evidence, capacity and cost from one decision to the next.

1. The system that had to be put back together

On 26 June 2021, offender management in England and Wales was reunited in one public Probation Service. Roughly 7,000 professionals moved into the new structure or its commissioned-service arrangements. Contracts ended, staff transferred and direct public responsibility returned for managing every probation case.⁴

The date looks like a clean repair. It was not.

Seven years earlier, Transforming Rehabilitation had divided probation between a public National Probation Service and 21 Community Rehabilitation Companies. High-risk cases remained public; low- and medium-risk cases went to contractors. Risk, however, is not fixed. A person’s circumstances can change. When assessed risk rose, responsibility and information had to move safely between organisations whose staff, systems, incentives and management chains had been separated.

That interface was not a technical detail. Probation depends on continuing knowledge of a person’s behaviour, housing, treatment, relationships, compliance and risk to others. The reform created a delayed information gateway that generated £23.1 million in compensation and was ultimately little used by most contractors. Commercial assumptions also failed. The Ministry expected at least £467 million in additional payments above original contract terms, including minimum termination costs of £171 million. The termination amount sits inside the wider figure and must not be added again.⁵⁶

By 2019, independent evidence supported ending the model. Continuing it merely to avoid another reorganisation would have confused consistency with responsibility.

Reunification still could not restore what an organisation chart had divided. Staff had lived through two major transitions. Information systems and local services required repair. Workloads remained high. The HMPPS workforce quarterly for March 2026 recorded 5,554 full-time-equivalent probation officers, 1,556 below the target of 7,110. Performance had worsened after unification amid wider criminal-justice pressure.⁷⁸

That later weakness does not prove that reunification was wrong. The relevant counterfactual was not a healthy contracted system. It was a failing model with unstable providers, fragmented responsibility and mounting public intervention. The lesson is harder: a justified correction can still be costly because the state cannot instantly recreate experience, trust, staffing and working relationships once they have been broken.

This paper asks when such disruption is necessary, when it is avoidable and what evidence should exist before government takes another system apart.

2. Change, correction, abandonment and churn

Democratic change is lawful alteration by an accountable government or Parliament. It may be prudent, reckless or something between. The term describes authority, not quality.

A reasonable correction is substantial change supported by credible evidence that a policy has failed materially, caused serious harm, become impracticable or unaffordable, or been overtaken by a better approach. Correction must still be judged on transition, cost and preservation of useful capacity.

Abandonment occurs when a public function is materially weakened while the underlying need remains and no credible replacement can carry it. It can happen through formal abolition, funding withdrawal, fragmentation or slow decay.

Institutional churn is a repeated lifecycle of creation, major reorganisation, abolition, fragmentation or recreation that causes avoidable loss of money, knowledge, staff, delivery capacity or public benefit without adequate justification. One change of name is not churn. One reversal is not automatically churn. A later programme addressing a familiar problem is not necessarily the old policy in disguise.

A durable policy lasts long enough to be implemented, evaluated and improved while retaining a practical route for democratic reconsideration. Durability is not permanence. A system earns continuity by remaining useful, inspectable and correctable.

These distinctions prevent two opposite errors. The first is to treat survival as proof of success. The second is to treat the announcement of change as proof of improvement.

3. Method: what five cases can establish

The research began with 15 candidates across public services, regulation, regional policy, employment and institutional reform. A formal matrix tested evidence quality, lifecycle clarity, measurable consequences, financial and capacity evidence, mechanism relevance and contribution to the paper. Extreme scores received an adversarial challenge. The final five were selected as a portfolio rather than a league table.

Three cases test suspected destructive change: Sure Start, Regional Development Agencies and Zero Carbon Homes. Transforming Rehabilitation tests justified reversal. The Low Pay Commission demonstrates continuity with adjustment.

The jurisdictions differ and are stated throughout. Sure Start, RDAs and Zero Carbon Homes concern England. Probation concerns England and Wales. The Low Pay Commission concerns the United Kingdom.

The evidence base prioritises legislation, official policy records, parliamentary material, National Audit Office reports, statutory advisers and inspectable research. Advocacy and testimony help identify questions but do not carry major conclusions alone where stronger evidence should exist.

This design can establish whether particular mechanisms occurred in these cases. It cannot calculate a national churn rate, prove that every British government behaves alike or rank the United Kingdom against all comparable democracies.²

The fair evidence test has three stages:

1. what was reasonably knowable when the decision was made;
2. whether the decision preserved the ability to learn what was not yet known;
3. what later evidence now shows about benefit, loss or correction.

Hindsight can reveal consequence. It cannot be passed backwards as contemporary knowledge.

4. Sure Start: prevention exposed before its evidence matured

Sure Start began in 1998 and expanded from targeted local programmes into a national network of children’s centres. Its purpose was larger than childcare. Centres were intended to make early education, health, parenting support, outreach and family services easier to reach through local places and relationships.⁹

The system was never uniform. Local authorities and individual centres differed in quality, services and reach. Early evaluations found improvements in some family and parenting outcomes but did not establish every expected cognitive or child-development benefit. Families used different combinations of services, local models changed and attribution was difficult.

By 2010, however, the state had built more than a set of buildings. It had created multidisciplinary teams, referral routes, local knowledge, recognisable access points and a national evaluation infrastructure. The outcomes sought—school readiness, health, educational attainment and reduced later intervention—could take years to appear.

The coalition government did not abolish Sure Start nationally. Local authorities retained duties under the Childcare Act framework to secure sufficient children’s-centre provision so far as reasonably practicable. The central change was financial and administrative. The 2010 Spending Review placed early-intervention funding into an unringfenced grant, giving councils more discretion while they faced wider fiscal pressure. Ministers argued that the network should remain but focus more effectively on families in greatest need.¹⁰

That defence matters. Ring-fenced national funding can protect weak provision as well as strong provision. Councils may understand local need better than Whitehall. A merged building is not necessarily a lost service, and a changed management structure can improve coordination.

The difficulty is that discretion and risk moved together. The Education Committee recorded a fall in registered centres from 3,631 in April 2010 to 3,116 in April 2013, while also documenting disagreement over what counted as closure, merger or redesignation. It found large local variation. Evidence cited by the Committee indicated that 72 per cent of centres had changed services because of cuts and 80 per cent expected further change. Short contracts and uncertain funding weakened the stability on which outreach and early intervention depend.¹¹

The financial record is incomplete. Reported children’s-centre expenditure was about £1.1 billion in 2011–12. The Early Intervention Grant total of £2.365 billion in 2012–13 covered more than children’s centres and cannot be treated as their budget. An estimate for 2013–14 placed children’s-centre spending at £854 million, 28 per cent below the 2010 level. There is no consolidated national account of reorganisation costs, lost staff capacity or the cost of rebuilding local relationships.¹²¹³

At the time, the public finances were under severe pressure and the long-term evidence was mixed. It would be false to claim that ministers already possessed the later results. It would be equally false to treat immature evidence as evidence of no value.

Later linked-data studies by the Institute for Fiscal Studies found reduced hospital admissions through childhood, with larger effects in disadvantaged areas. Further educational research found that children living near a centre throughout their first five years achieved, on average, roughly 0.8 additional GCSE grades across subjects, with larger effects among pupils eligible for free school meals. The studies have limits, including proximity-based measures of exposure, but they are substantial evidence that important benefits took time to appear.¹⁴¹⁵

The later development of Family Hubs and Start for Life, followed by Best Start Family Hubs from 2026, returns to overlapping functions: local access to integrated family, health and education support. The newer model covers a broader age range, includes digital provision and uses different funding and accountability. It is not Sure Start renamed. It is evidence that the public problem persisted.

The current programme includes more than £500 million over the 2026–2029 spending-review period. That commitment is not an outcome. It does show that rebuilding or strengthening a related national network requires new money, guidance and institutional effort.¹⁶

The final manuscript judgement is substantial institutional weakening and partial abandonment with uneven local effects. Some local redesign was legitimate and may have improved targeting. The evidence is not sufficient for one national churn label. The central failure was allowing the funding horizon to become shorter than the outcome horizon without a clear national account of which capacity had to survive.¹⁷

5. Regional Development Agencies: closure succeeded before continuity did

Regional Development Agencies were created by the Regional Development Agencies Act 1998 and became operational in 1999. Nine statutory bodies were given broad purposes covering economic development, business competitiveness, employment, skills, regeneration and sustainable development.¹⁸

Their record was mixed. Administrative regions did not always match functional economies. Accountability was indirect. Missions expanded. Appraisal and evaluation varied. Their planned Single Budget funding was roughly £2.2 billion a year around 2009–10 while regional inequality persisted.¹⁹ A government could reasonably prefer structures that were cheaper, more local or more directly accountable.

The agencies also held real capacity. They employed regional and sector specialists, managed land and property, administered grants and investment funds, and maintained relationships among councils, universities, firms and central departments. National Audit Office work found evidence of returns from some regeneration activity while concluding that the system had not demonstrated maximum value.

The coalition government announced abolition in 2010 and promoted Local Enterprise Partnerships organised around functional economic areas and business leadership. Eight non-London agencies ceased operating on 31 March 2012. The London Development Agency followed a different route. Assets, programmes and responsibilities moved among departments, national bodies and local institutions.

The closure operation itself was effective. Bodies closed, staff left or transferred, assets moved and liabilities were handled. That administrative success did not prove that the replacement system could perform the same functions.

The National Audit Office’s 2013 report found that new local-growth programmes were not established in time to avoid a significant dip in spending and jobs created. Direct central-government spending through the initiatives examined fell from £1.461 billion in 2010–11 to £273 million in 2012–13 before a planned recovery. Fiscal consolidation independently reduced spending, so the entire fall cannot be attributed to abolition. The finding still demonstrates a gap between dismantling the old machinery and making the new machinery able to act.²⁰

Direct transition figures show scale without supporting one dramatic total. Government made a four-year closure settlement of £464 million covering salaries, redundancies, transition, closure and pensions. A parliamentary answer placed estimated redundancy costs at £56.4 million at one point. Approximately £300 million of land and property assets were transferred. The periods and categories differ and may overlap.²¹²²

Local Enterprise Partnerships were deliberately lighter institutions. They brought business and councils together without reproducing RDA statutory powers, staff or asset portfolios. That could reduce bureaucracy and align decisions more closely with local economies. It also left delivery dependent on small teams, partner organisations and accountable bodies.

By 2016, LEPs had access to a projected £12 billion Local Growth Fund, yet the National Audit Office found serious capacity and accountability weaknesses. Only 5 per cent considered their resources sufficient for government expectations. Sixty-nine per cent reported insufficient staff, and 28 per cent did not believe their staff had sufficient skills. Government had not set sufficiently clear and quantifiable objectives for Growth Deals.²³²⁴

Those findings do not show that RDAs would necessarily have performed better. They show that large responsibilities were placed on successor structures without equivalent permanent capacity.

Government later ended LEP core funding and moved principal functions into upper-tier local authorities and combined authorities from April 2024. The integration guidance retained broad functions such as business representation, strategic economic planning and programme delivery.²⁵

This sequence is not simple rebranding. RDAs, LEPs and elected local institutions differ in law, geography, powers, funding and accountability. It is repeated redesign around a continuing territorial function.

The strongest defence is evolutionary. RDAs were replaced by structures closer to functional economies; LEP weaknesses then prompted transfer into elected institutions. A sequence of reforms may eventually create better local accountability.

The criticism is about the route. A government intending to abolish regional machinery should know which functions must continue, where staff and data will go, how cross-boundary projects will be managed and whether successors have authority, funding and people on the first operational day.

Scotland provides a bounded contrast. Scottish Enterprise and Highlands and Islands Enterprise survived the Scottish Government’s 2016 Enterprise and Skills Review while coordination and governance changed. Scotland is not an outcome-identical counterfactual. It shows that reform can preserve a named institutional home while altering priorities.²⁶

The final manuscript judgement is institutional churn, with medium–strong confidence. The legitimate case for reform does not remove the documented delivery dip, dispersed capacity, LEP under-capacity and later functional relocation. Fiscal consolidation and institutional differences prevent a claim that government merely abolished and recreated the same bodies.¹⁷

6. Zero Carbon Homes: cancellation built into long-lived assets

In 2006, government announced that new homes in England should become zero carbon from 2016. Over the following years, departments, housebuilders and technical bodies developed a model combining stronger on-site performance with “allowable solutions” for carbon reductions that could not reasonably be achieved on the site itself.²⁷

The design was contestable. Off-site abatement introduced complexity. Price caps, administration and the boundary between building performance and offsetting remained disputed. Housebuilding was constrained by land, planning, finance, skills and demand. Government had a legitimate interest in avoiding a costly requirement whose benefit was uncertain.

By early 2015, however, the policy was near operation. Parliament had enacted an enabling power in section 37 of the Infrastructure Act 2015. On 27 March, the housing department published a final impact assessment stating that government remained committed to implementation from 2016. Firms and technical organisations had planned around a timetable developed over roughly nine years.²⁸

On 10 July 2015, the government’s productivity plan cancelled the policy. A later parliamentary answer cited regulatory burden and the need to let recent energy standards bed in. Those were real public reasons. Time already spent does not make a policy correct.

The evidential weakness is the missing decision comparison. This research did not recover an equivalent published cancellation assessment weighing the main options: proceed as designed; simplify allowable solutions; retain stronger on-site requirements while dropping offsets; delay implementation to a fixed review; or cancel the package. The enabling power remained in law, but the regulations and timetable disappeared before operational evidence could be produced.²⁹

Housing standards matter because the asset survives the political decision. A weak administrative programme can be altered next year. A home built to a lower standard may consume more energy for decades and require physical correction later.

In 2020, the Climate Change Committee estimated that constructing a genuinely zero-carbon home from the outset was around five times cheaper than retrofitting it later and would generally reduce energy bills. This is modelled counterfactual evidence, not a realised national cancellation cost.³⁰ It should not be multiplied across an assumed housing stock without a reproducible method.

The later Future Homes Standard confirms that the problem persisted but does not prove the original design was right. Regulations published in 2026 are intended to make new homes zero-carbon-ready in use once the electricity grid is fully decarbonised. The model relies on strengthened Building Regulations and low-carbon systems rather than recreating allowable solutions. Main implementation is scheduled for March 2027, with different arrangements for high-rise buildings.³¹

The current policy also requires scrutiny. Its impact assessment estimated an equivalent annual net direct cost to business of £709 million and a net present social value of £11.3 billion. The Regulatory Policy Committee rated the assessment “not fit for purpose” because key option and small-business comparisons were insufficiently evidenced. That judgement concerns the assessment, not the objective.³²

The distribution of risk is clear even where the total is not. Developers avoided or deferred some compliance costs. Occupants of homes built under weaker standards may face higher energy use and later retrofit requirements. Tenants and lower-income buyers have less power to correct the building they occupy. Supply chains lost a stable regulatory signal. Future governments inherited a harder carbon and efficiency task.³³

The exact value of those effects remains unknown. A 2026 government statement linked the cancellation to more than one million homes with higher bills, but this research has not independently reproduced that figure and does not treat it as settled fact. Historic preparation costs are also unconsolidated.³⁴³⁵

The strongest defence remains credible: the original offsetting mechanism may have been too complex, technology changed and a later standard may be better. That defence increases the need for a published option analysis. If offsets were the problem, stronger fabric and on-site standards could have been assessed separately. If timing was the problem, government could have preserved technical preparation through a fixed delay and review.

The final manuscript judgement is destructive cancellation followed by delayed functional restoration. Confidence is strong on the reversal and the long delay, but only medium on realised national consequences. The case does not establish that the original package should have proceeded unchanged.¹⁷

7. Transforming Rehabilitation: correction after fragmentation

Transforming Rehabilitation began with a public problem that government was right to confront. People released after prison sentences of less than 12 months did not receive the same statutory supervision as those serving longer sentences. Reoffending was high, and many people left prison facing housing, treatment, employment and family problems.

The Offender Rehabilitation Act 2014 extended supervision. The operating model then split probation. A public National Probation Service retained high-risk cases and court advice; 21 Community Rehabilitation Companies managed low- and medium-risk cases. Seven public regions and 21 contract areas replaced 35 probation trusts. Payment mechanisms were intended to reward reduced reoffending and encourage innovation.³⁶

The extension of supervision was a real gain. It should not disappear from the judgement because the delivery model failed.

The model created high-risk interfaces. Risk can change. Courts, prisons, public probation staff, contractors and voluntary services needed reliable information and working relationships. The market needed viable providers and accurate assumptions about caseload, cost and payment.

Government implemented the national model without a full-scale pilot capable of testing those relationships and commercial assumptions. This does not prove that failure was inevitable. It shows that a high-risk design was scaled before the system had generated operational evidence.³⁷

The National Audit Office found that Community Rehabilitation Companies forecast collective losses of £294 million, compared with £269 million of profits expected at bid stage. The Ministry expected at least £467 million in additional payments above original terms between 2016–17 and termination, including minimum termination costs of £171 million. Underinvestment persisted and expected innovation was limited. A delayed ICT gateway produced £23.1 million in compensation while most companies did not ultimately use it.⁵⁶

Outcome data were not one clean failure measure. The proportion of proven reoffenders fell over the period cited by the NAO, while the number of reoffences per reoffender rose by 22 per cent. Sentencing, prisons, housing, treatment and wider justice conditions affect both. The case against the model rested on converging evidence of commercial instability, fragmented responsibility, weak delivery and independent criticism.³⁸

By 2018–2019, continuing the contracts was not neutral. Provider viability and public service were at risk. Government announced early termination and confirmed that offender management would return to a unified public service.

The correction created another transition. Around 7,000 professionals moved in 2021. Core offender management returned to public responsibility, while commissioned rehabilitative services continued in defined areas. The new service used 12 regions rather than recreating the old trusts. Systems, staffing and local delivery still required repair.⁴

The National Audit Office’s 2025 assessment found that performance had worsened since unification and staffing remained below target. This prevents a triumphal ending. It does not make the failed contracted model preferable.⁸

The financial comparison cannot support a simple saving claim. Original maximum CRC payments had been expected to reach up to £3.7 billion; revised expected payments were up to £2.3 billion through December 2020. The lower figure reflects early termination, changed scope and stabilisation payments. The full cost of unification and long-term correction has not been consolidated.⁵³⁹

The final manuscript judgement is evidence-led reasonable correction after failed and damaging institutional fragmentation. Confidence is strong. The case does not show that all private or voluntary provision is illegitimate. It shows the danger of splitting core responsibility for a relationship-based service whose risk classification changes over time.¹⁷

8. The Low Pay Commission: adjustment without institutional amnesia

The Low Pay Commission was created in 1997 and placed on a statutory basis through the National Minimum Wage Act 1998. The first main adult minimum wage of £3.60 took effect in April 1999. By April 2026, the National Living Wage for workers aged 21 and over was £12.71.⁴⁰

The policy values changed dramatically. The institution remained recognisable.

The Commission has nine members drawn from employer, employee and independent or academic backgrounds. It receives a government remit, gathers written and oral evidence, commissions research, visits workplaces, analyses labour-market data and publishes recommendations. Ministers and Parliament retain the final decision. If government rejects a recommendation, it must explain why.⁴¹

The arrangement does not remove politics. Government sets the remit and can alter ambition. The introduction of the National Living Wage in 2016 changed the framework. Age thresholds and targets have moved. In 2015, government rejected the Commission’s recommended apprentice rate and published a statutory explanation.⁴² Durability did not transfer sovereignty to experts.

The process gives disagreement a continuing home. Employers present evidence about costs, prices, investment and employment. Workers and unions present evidence about pay and insecurity. Independent members test those claims against data and research. The next cycle can compare forecasts with outcomes.

The 2025 report showed that rate-setting was not treated as a formula. Commissioners had to judge uncertain future median wages, wider economic conditions and data quality. They declined to move 20-year-olds immediately onto the full National Living Wage because the required increase was considered too risky for the youth labour market. An institution associated with higher pay retained the capacity to recommend waiting.⁴³

Its survival across Labour, coalition and Conservative governments, recession, Brexit, the pandemic and inflation is directly observable. The cause is not isolated. The wage floor became familiar and politically popular; employers adapted; abolition acquired a high political cost; no obvious replacement emerged. The Commission may give governments political cover for decisions they already favour.⁴⁴

Those explanations do not remove the observable mechanism. The Commission preserves:

• time-series and earlier forecasts;
• institutional memory of recommendations;
• recurring relationships with employers and workers;
• a public record of government acceptance and departure;
• a scheduled point for correction;
• continuity across ministerial change.

Durability is not completeness. In 2024–25, HM Revenue & Customs identified around £5.8 million in arrears owed to 25,200 workers and issued around 750 penalties totalling £4.2 million. Those are detected cases, not all underpayment. Age-based rates permit some younger workers to receive less for the same job.⁴⁵

Germany’s Minimum Wage Commission offers a related mechanism: employer and employee representation, academic advice, recurring recommendations and written reasons. Germany’s legislature has intervened directly in the wage level, showing that a standing institution does not eliminate political override.

The final manuscript judgement is durable adaptive institution. Confidence is strong on continuity and process, but only medium on why the institution survived. The Low Pay Commission is a demonstration case, not a statistical control or a template for every service.¹⁷

9. What the cases show together

9.1 The public function survives the institution

Integrated family support, regional coordination, efficient new homes, probation and wage-setting remained necessary throughout the changes examined here. That persistence does not vindicate the original design. It changes the meaning of abolition.

When an organisation ends, its function may move to a department, council, contractor or later programme. The cost has not vanished because the budget line changed. A responsible decision must distinguish the function that is genuinely ending from the function that another institution must inherit.

The Low Pay Commission keeps the function, evidence and correction process together. The other cases show different ways those elements can be separated.

9.2 Reform merit and transition readiness are independent

Government can make a defensible case against an old institution and still mishandle the transition.

RDA closure was administered effectively while successor local-growth delivery was not ready enough to avoid a major dip. Transforming Rehabilitation addressed a genuine supervision gap while using an operating model whose interfaces and commercial assumptions were insufficiently tested. Zero Carbon Homes contained a contestable offsetting design, but cancellation removed the timetable without a ready alternative. Sure Start contained uneven provision, but local discretion was expanded while fiscal pressure threatened the ability to preserve core capacity.

The question is not only whether change is justified. It is whether the public function can continue safely on the day the old system withdraws.

9.3 Capacity disappears quietly

Direct spending leaves figures. Capacity often does not.

A children’s centre can retain its building while losing outreach staff and referral relationships. A regional programme can transfer land while dispersing the team that knows how to use it. A housing standard can disappear while technical preparation and supply-chain expectations are stranded. Probation can reunite under one employer while workloads and experience remain damaged. The Low Pay Commission shows the inverse: data, people and relationships accumulate because the institution continues.

Capacity is not sacred. Some expertise supports obsolete practice. A public account should still state what will be retained, transferred, archived, lost and rebuilt.

9.4 Evidence horizons and political horizons diverge

Preventive policy may need years before health or education outcomes mature. Housing standards shape assets lasting decades. Regional institutions build relationships over long periods. Ministers, budgets and programmes often operate on shorter cycles.

Later evidence cannot prove that an earlier decision was irrational. It can show that the decision destroyed or preserved the route by which uncertainty would have been resolved.

The critical question is whether government protects learning during uncertainty. Sure Start was exposed while long-term evidence was incomplete. Zero Carbon Homes was cancelled before operational evidence could be generated. Transforming Rehabilitation shows the opposite condition: once independent evidence converged on systemic failure, uncertainty was no longer a reason to wait.

9.5 Cost moves across institutions and time

The cases do not support one aggregate churn cost. Figures differ in scope, period and price basis; some sit inside others; opportunity costs are often modelled rather than observed.

A pattern is still visible. Savings can appear in one department while costs move to councils, households, contractors or a later government. Preventive spending may benefit health or education years later. Weaker building standards may lower immediate developer costs while increasing household and retrofit risk. Contract termination can be necessary even while taxpayers fund the correction.

A credible account should separate direct cancellation and redundancy, asset transfer, systems and legal transition, interim delivery, replacement and rebuilding, claimed savings, realised savings, supported opportunity costs and unknown values. The purpose is not to monetise everything. It is to stop treating missing cost as zero.

10. Conditions that make starting again easier

The cases do not reveal one cause. They are consistent with an interaction between constitutional permission, short political ownership, fragmented budgeting, weak evaluation practice and the political visibility of announcing a new beginning.

10.1 A direct route to ordinary reversal

The United Kingdom has no general constitutional test requiring a government to prove that a major public system has failed before it changes it.

From 1997 to 2024, the largest governing or government-forming party repeatedly won a larger share of Commons seats than of votes. Labour won 63.4 per cent of seats in 1997 on 43.2 per cent of the vote, 55 per cent of seats in 2005 on 35.2 per cent, and 63.2 per cent in 2024 on 33.7 per cent. The Conservatives won 56.2 per cent of seats in 2019 on 43.6 per cent of the vote. Coalition and minority periods complicate the pattern, and these results do not make individual decisions illegitimate. They show how strong parliamentary control can arise without majority vote support.⁴⁶

Standing Order No. 14 normally gives government business precedence. Party discipline usually supports the programme, although MPs remain free to rebel and governments can lose. Parliamentary sovereignty allows later legislation to amend or repeal earlier ordinary statutes.⁴⁷

Several cases required no new primary Act. Sure Start changed through spending and grant design under continuing duties. Zero Carbon Homes was cancelled by withdrawing intended implementation. Transforming Rehabilitation combined legislation, contracts and administration.

The system contains constraints: the Lords can revise and delay; courts can review executive action; devolution divides competence; select committees and audit bodies can expose failure; coalition and minority governments must negotiate. The narrower finding is that no general preservation, capacity or readiness test applies across major reversals.

10.2 Ministers change faster than many systems mature

A reproducible internal appointments analysis covering 1997 to July 2026 found median tenures of about 1.51 years in education and children’s policy, 1.46 years in communities, local growth and housing, 1.25 years in business and employment, and 1.19 years in justice and probation. Chancellors were more stable at about 2.04 years. Role boundaries and machinery-of-government changes affect the calculation.⁴⁸

Short tenure does not prove poor policy. Several consequential reforms were led by ministers who stayed longer than the median. Long tenure can sustain a risky programme as easily as a good one.

The continuity risk is practical. Policies outlast the minister who launches them. Responsibilities move between departments. Each move requires briefing, transfer and reconstruction of ownership. Civil servants and arm’s-length bodies preserve some memory, but a policy whose evidence and review date live only inside one ministerial programme is vulnerable to relaunch.

The Low Pay Commission reduces that dependence by keeping a recurring timetable and public evidence record through ministerial and departmental change.

10.3 Formal appraisal is stronger than its use

The Green Book 2026 requires options to be assessed through social costs and benefits, distribution, risk, uncertainty, change management and benefits realisation. The Magenta Book says evaluation should be designed early and linked to appraisal.⁴⁹⁵⁰

The formal method does not instruct government to ignore prevention or the future. Discounting reduces the present weight of later effects but does not delete them. Sunk-cost discipline correctly prevents continuation merely because money has already been spent. It should not be used to describe productive staff, assets, data and relationships as valueless.

The problem lies in practice. Departments operate within spending settlements, annual controls and capital–resource boundaries. The organisation paying now may not receive the later benefit. A preventive programme may require local resource spending while benefits appear in health, education or justice. A capital building can survive while the staff needed to make it useful are cut.

A Public Accounts Committee report cited a 2019 review finding robust impact-evaluation plans for only 8 per cent of £432 billion of major-project spending, while 64 per cent had no evaluation arrangements. That snapshot does not describe every programme. It shows that good guidance does not ensure good use.⁵¹

The reform opportunity is therefore not primarily another appraisal manual. It is a public point at which government must show whether the existing appraisal, transition and evaluation work has actually been done.

11. What this research did not find

Policy options are shaped before Parliament through consultations, meetings, market engagement, professional advice, party development, think tanks, unions, trade bodies, charities and the media. Participation is necessary. Government cannot design a housing standard without builders and engineers or a wage floor without employers and workers.

The evidence distinguishes presence, access, advocacy, adoption, documented reliance, causal influence and improper influence. Most public records stop before the top of that ladder.

The selected cases show adopted ideas more clearly than actor-specific causation: local flexibility in Sure Start; business leadership in LEPs; regulatory burden in Zero Carbon Homes; competition and payment by results in probation. The record does not establish that a donor, lobbyist, consultancy, think tank or media proprietor secretly controlled any selected decision.

Fragmented transparency is an accountability problem. It is not proof of conspiracy.

A major decision record should therefore publish consultations, advisory membership, commissioned analysis, material meetings and submissions, and the reasons important arguments were accepted or rejected.

12. Mechanisms used elsewhere

International and devolved comparison is useful only when it compares mechanisms rather than national character.

Denmark frequently operates with minority or coalition governments. Political agreements can give several parties ownership before legislation. A later government remains free to change course, but repudiation may carry a wider political cost. The trade-off is slower or less transparent bargaining.

Germany’s Bundesrat brings territorial governments into federal legislation. Some laws require consent; others can trigger objection or mediation. This makes implementation interests part of the legislative route. It can also create stalemate. Germany’s Minimum Wage Commission shows another form of continuity: recurring employer, worker and academic participation alongside political override.

Scotland retained Scottish Enterprise and Highlands and Islands Enterprise while reforming coordination after the 2016 Enterprise and Skills Review. The contrast with England does not establish superior economic outcomes. It shows that priorities and governance can change without abolishing the institutional home. Scottish community justice also uses statutory duties and review cycles, while continuing to face uneven delivery and resource pressure.

The lesson is not that Britain needs a veto at every stage. It needs proportionate friction before high-risk, hard-to-reverse change: enough to expose evidence, alternatives, readiness and cost without making urgent correction impossible.⁵²

13. The Durable Policy Standard

The Durable Policy Standard is a proposed democratic quality test for major public-system change. It is not a claim that every earlier institution deserves preservation. It should use existing appraisal, evaluation, audit and parliamentary machinery wherever possible.

13.1 When the full standard applies

The full process should be triggered by a material weakening, abolition, fragmentation or replacement of a major public system where one or more of the following make transition consequential:

• statutory rights or duties;
• large public reliance or restricted access to alternatives;
• a substantial workforce, asset base, contract portfolio or liability;
• national, multi-regional or cross-boundary delivery;
• serious public-safety or service-continuity risk;
• effects that are difficult or expensive to reverse;
• evidence or capability that takes years to rebuild.

Minor reporting-line changes and routine contract renewals should not receive the same process. Departments should publish how they apply the trigger rather than decide silently that a change is too small.

13.2 Who is responsible

The responsible minister should publish the decision record before material withdrawal begins or before Parliament is asked to authorise the decisive step.

The departmental accounting officer should state whether the record contains the required evidence, cost and readiness information. This is a completeness and stewardship statement, not a veto over the elected policy objective.

The highest-risk transitions should receive independent assurance under published terms. The assurer should be able to state that evidence is missing or a successor is not ready. Ministers and Parliament remain able to proceed, but the departure becomes visible.

Implementing authorities should have a formal route to report that funding, staff, systems or legal powers are insufficient.

Duty 1: define the public function and jurisdiction

The record should identify:

• the problem;
• the people and places affected;
• legal rights and duties;
• the service, standard or coordination function currently provided;
• what need will remain if the institution ends;
• responsibility before and after change.

This prevents the end of a programme from being mistaken for the end of the problem.

Duty 2: publish decision-time evidence and the review position

The record should distinguish delivered outcomes from promises, national averages from local variation, direct evidence from forecasts, and current knowledge from later evidence still expected.

At creation, a major policy should specify success measures, failure conditions, the evaluator, the minimum evidence period and a review date. Before reversal, government should publish the scheduled evaluation or explain why it cannot safely wait.

The strongest case against the preferred option should appear in the same record.

Duty 3: compare repair, staged change and replacement

The options should extend beyond a preferred reform and a weak “do nothing” baseline. They should consider retention, improvement, narrower remit, stronger accountability, pilot, staged transition, fixed delay, merger, abolition and replacement where relevant.

Government should explain why correction inside the existing structure is inferior. A repair option is not presumed better; it is simply not allowed to disappear because abolition creates a clearer announcement.

Duty 4: produce a whole-life cost and distribution account

The public record should separate:

1. cancellation, redundancy and contract costs;
2. asset disposal or transfer;
3. legal, information and systems transition;
4. interim delivery;
5. replacement and rebuilding;
6. claimed savings;
7. realised savings;
8. future liabilities;
9. supported opportunity costs;
10. unknown values.

Periods, price bases and embedded amounts must be stated. Projected benefits must not be presented as realised. Effects across councils, households, public bodies and later governments should be shown where evidence permits.

The distribution account should explain who carries service gaps, delayed help, higher bills, job disruption, regional loss, reduced practical access and future correction costs. It should consider the different ability of groups to protect themselves.⁵³

Duty 5: inventory capacity

Before withdrawal, government should record:

• staff and specialist skills;
• professional and local relationships;
• data and longitudinal evidence;
• systems, standards and processes;
• physical assets;
• contracts and supply chains;
• institutional memory;
• public reliance and recognisable access points.

For each category, the transition plan should state what will be retained, transferred, archived, lost or rebuilt.

The inventory creates a rebuttable record. It does not make every employee, process or asset permanent.

Duty 6: pass a successor-readiness gate

A major system should not materially withdraw until the successor has, or the minister explicitly proceeds without:

• legal authority;
• an accountable owner;
• confirmed funding;
• sufficient staff and skills;
• working information systems;
• arrangements for assets and contracts;
• territorial or local delivery agreements;
• an operational commencement date;
• a contingency if readiness fails.

Closure readiness and service readiness are different. The decisive question is whether the public function can continue when responsibility changes.⁵⁴

Duty 7: publish influence, challenge and democratic reasons

The decision record should include formal consultations, advisory membership, commissioned work, material submissions, relevant meetings and the reasons significant arguments were accepted or rejected.

Independent evaluators, auditors or statutory advisers should be able to challenge evidence and readiness. Government remains free to disagree. When it does, it should state the democratic objective, the evidence on which it differs, the risks it accepts and the safeguards it will preserve.

This protects democratic override by making it explicit rather than allowing expert advice or ministerial departure to disappear into administrative correspondence.

Duty 8: set a post-change evaluation date

Every major reversal should have a fixed review comparing:

• realised costs and savings;
• service and policy outcomes;
• capacity retained, lost and rebuilt;
• successor performance;
• effects on affected groups;
• the chosen option against credible rejected alternatives;
• whether further correction is required.

The result should update a durable public lifecycle record so that later governments inherit more than the latest announcement.

13.3 Emergency correction

Urgent change may be necessary where delay creates serious harm, public-safety risk, legal failure, insolvency or collapse of essential delivery. The standard must not require a failing service to continue while a lengthy process runs.

An emergency decision should still publish the evidence of urgency, the stages shortened, the capacity being preserved, interim safeguards, the temporary or successor owner and a dated retrospective review.

Transforming Rehabilitation is the direct test. The evidence supported early termination. The standard would not have required the original contracts to run to expiry. It would have strengthened the public account of workforce preservation, transition cost and post-unification evaluation.⁵⁵

13.4 What the standard changes

The standard does not determine the political objective. It changes the record that must exist around the choice.

A government can still say that an institution is unaffordable, a standard is technically wrong, a service has failed or the electorate chose another course. It should also state what happens when the old arrangement ends: who holds responsibility, where staff and data go, whether systems work, who bears the risk and when the decision will be tested.

14. Final case judgements

Case	Final manuscript judgement	Confidence	Main restraint
Sure Start and children’s centres	Substantial institutional weakening and partial abandonment with uneven local effects	Medium	No single national abolition or complete local map
Regional Development Agencies	Institutional churn	Medium–strong	Fiscal consolidation and materially different successors
Zero Carbon Homes	Destructive cancellation followed by delayed functional restoration	Strong on sequence; medium on consequence	No precise national realised cost; original design was contestable
Transforming Rehabilitation	Evidence-led reasonable correction after failed and damaging institutional fragmentation	Strong	Reunification did not restore performance automatically
Low Pay Commission	Durable adaptive institution	Strong on continuity; medium on causal design claim	Popularity and path dependence may explain survival

These classifications describe the evidence now available. They are recorded in schema v0.2 of the Institutional Churn Ledger and should be revisited if the stated evidence gaps are filled.

15. The strongest objections

Fiscal consolidation required hard choices

The financial crisis and subsequent fiscal decisions form the background to Sure Start and RDA reform. Retaining one programme imposed an opportunity cost elsewhere. Future benefits cannot exempt a system from affordability.

This objection defeats any claim that reduction alone proves churn. It does not remove the need to account for transition, unequal local exposure, lost capacity and replacement cost. Fiscal constraint should make prioritisation more explicit, not less.

The predecessor institutions had real weaknesses

Sure Start varied in quality. RDAs were costly and indirectly accountable. Zero Carbon Homes contained a complex offsetting mechanism. The earlier probation system did not supervise short-sentence prisoners. The minimum wage system still contains age and enforcement problems.

These are reasons for reform. They are not evidence that the chosen successor was ready or that destruction was cheaper than repair. A government can diagnose a genuine weakness and still choose a damaging transition.

Later resemblance can mislead

Best Start Family Hubs are not Sure Start restored unchanged. Local and combined authorities are not RDAs. The Future Homes Standard is not the 2016 design. The unified Probation Service is not the old trust system.

The later return of a function proves that the public problem persisted. It does not prove the original institution was optimal. This paper distinguishes functional return from recreation.

External shocks confound outcomes

The cases crossed recession, austerity, Brexit, the pandemic, inflation, changing energy technology, housing pressure and criminal-justice demand. Regional investment would have changed even if RDAs remained. Sure Start outcomes depend on schools, health and local economies. Probation performance depends on prisons, sentencing, housing and treatment.

For that reason, the strongest conclusions concern decision sequence, directly observed capacity and readiness. Claims about long-term national outcomes remain more cautious.

Local variation can be legitimate

Local government exists partly because places differ. Ring-fencing can reduce adaptation. Functional economic areas may be better than administrative regions. A council may integrate services more effectively by merging sites.

Variation becomes a durability problem when central government transfers risk without measuring whether the public function remains accessible. The safeguard is not uniformity. It is a clear minimum function, transparent local evidence and comparable post-change outcomes.

The Low Pay Commission may survive because it is popular

This is the strongest challenge to treating the Commission as evidence of institutional design. Its process may be less important than political acceptance and path dependence.

The paper accepts that limitation. The Commission is used to demonstrate compatibility: policy can change repeatedly while an institution retains evidence, relationships and a review cycle. It is not proof that copying its structure will make every policy durable.

More procedure can become another failure

A Durable Policy Standard could generate long documents, ritual consultation and defensive modelling. Departments may satisfy the form without changing the decision. Independent reviewers may lack time or information. Urgent action may be delayed.

The response is proportionality, visible ownership and use of existing appraisal machinery. The standard focuses on decision facts rather than document length. It permits override and emergency correction. Its purpose is not to create another file; it is to make missing evidence and unready delivery visible.

The case selection is not neutral

The cases were deliberately chosen for explanatory contrast. Three were suspected destructive-change cases. This makes the portfolio useful for mechanism analysis and unsuitable for estimating prevalence.

The title must therefore be read as a challenge to an institutional vulnerability. The paper does not show that most British policy fails, that other countries routinely perform better or that every government prefers destruction.

16. Conclusion

Britain can make policy last. The Low Pay Commission shows that a continuing institution can absorb changes of government, economic shocks and political ambition while retaining evidence and a route for correction.

Britain can also reverse failure responsibly. Probation reunification was supported by independent evidence even though the correction imposed further disruption and did not instantly restore performance.

The failures examined through Sure Start, Regional Development Agencies and Zero Carbon Homes were not identical. Preventive capacity weakened unevenly while long-term evidence was incomplete. Regional bodies closed before equivalent successor delivery was ready. A near-operational housing standard was cancelled without an equivalent public option appraisal or ready replacement timetable.

The five cases do not prove a national law. They expose a recurring procedural weakness: major change can proceed without one public record joining the continuing function, decision-time evidence, repair options, whole-life cost, capacity at risk, successor readiness and future evaluation.

A government should remain free to say that an institution has failed, a programme is unaffordable, a standard is technically wrong or the electorate has chosen another path.

It should also have to explain what happens when responsibility changes hands.

Policy should not last because change is forbidden. It should last long enough for government to learn what it is changing, preserve what remains useful and replace what has genuinely failed without pretending that lost capacity was free.

---

Research and confidence note

This v0.4 manuscript is based on the research record current to 31 July 2026. It incorporates the full adversarial review, claim audit and source-pinpoint pass but remains an unpublished research manuscript.

Material limits remain:

• no individual affected-life example has been authenticated; the opening uses a documented institutional event;
• Sure Start local transitions have not been mapped comprehensively;
• RDA staff, functions and assets have not been compared across a representative regional sample;
• the realised national household and retrofit cost of cancelling Zero Carbon Homes has not been reproduced;
• the complete cost of probation correction and unification is not consolidated;
• the causal contribution of Low Pay Commission design cannot be isolated from popularity and path dependence;
• ministerial, constitutional, Treasury and influence evidence supports enabling conditions rather than a measured case-level causal share;
• the case portfolio cannot estimate national frequency;
• final classifications were transferred into schema v0.2 of the Institutional Churn Ledger through PR #175;
• the claim audit is recorded in docs/TWIS_DRP_01_CLAIM_AUDIT_V0.1.md and source pinpoints in docs/TWIS_DRP_01_SOURCE_PINPOINTS_V0.1.md; a final rendered citation check, legal and reputational review, and the unresolved evidence gaps recorded there remain before publication.

Projected values, official claims and later evidence are identified as such. Missing evidence is not treated as proof of either success or failure.

Sources

Footnotes

1. Manuscript front matter; all five case dossiers; evidence cut-off field Exact audit rule: no evidence after 31 July 2026 is silently imported. Inspect source. ↩

2. docs/TWIS_DRP_01_CASE_SELECTION_MATRIX_V0.1.md; docs/TWIS_DRP_01_CASE_SELECTION_SCORES_V0.3.md; manuscript §3 Internal method boundary; five selected cases cannot estimate national prevalence. Inspect source. ↩ ↩²

3. House of Commons Library, Parliamentary sovereignty, CBP-10377, sections on the legislative supremacy of Parliament; Standing Orders; case decision routes; Current institutional record and bounded absence finding Qualified cross-source institutional inference; no single general statutory preservation gate identified. No general cross-government gate identified; proposed mechanism is not current law. Inspect source. ↩

4. Official Probation Service account of 26 June 2021 transition; approximately 7,000 professionals Approximate movement, not an exact workforce census. Inspect source. ↩ ↩²

5. NAO, Transforming Rehabilitation: progress review, key facts and summary; Same NAO report, key facts and summary pp.9–11; Same NAO report, conclusions and key facts £294m forecast losses versus £269m forecast bid-stage profit. £171m termination floor sits inside £467m additional projected payments. Original/revised maximum payment figures are not a saving calculation. Inspect source. ↩ ↩² ↩³

6. PAC, Transforming rehabilitation: progress review, chapter 2, paras 20–21; earlier PAC report, chapter 1, para 17 Resolves prior pinpoint gap: delayed gateway, £23.1m compensation, two users by January 2019 and twelve opting out. Inspect source. ↩ ↩²

7. HMPPS workforce quarterly, March 2026, section 3 paragraph after Figure 4; Recruitment Annex Exact 5,554 / 1,556 / 7,110 snapshot. Inspect source. ↩

8. NAO, Building an effective and resilient probation service, summary and performance chapters Later weakness restrains triumphal claims; does not validate failed contracts. Inspect source. ↩ ↩²

9. House of Commons Education Committee, Sure Start children’s centres, chapter 1, paras 6–10 Distinguish original local programmes from later children’s-centre phases. Inspect source. ↩

10. Childcare Act 2006, s.5A framework; Education Committee, chapter 5, para 107; DfE, “Funding for Sure Start children’s centres”, headings on EIG and 2011–12 outturn No national abolition claim; funding became unringfenced from April 2011. Inspect source. ↩

11. Education Committee, chapter 1, para 7; Education Committee, chapter 5, para 108, citing ECCE Strand 3 Exact registered-centre counts; closure/merger dispute travels with the figures. Percentages apply to the cited study population, not every centre. Inspect source. ↩

12. DfE, “Funding for Sure Start children’s centres”, EIG and Section 251 headings; Education Committee, chapter 5, para 107 and footnote 237 (Policy Exchange 2013, p.20) Resolves prior pinpoint gap. £2.365bn EIG covers multiple services; £854m is an estimate, not audited outturn. Inspect source. ↩

13. Sure Start dossier cost/gap sections; ICL-2026-001 reopening conditions National transition, lost-capacity and rebuilding total remains unknown. Inspect source. ↩

14. IFS, The short- and medium-term effects of Sure Start on children’s outcomes, “Key findings” and methodology Later quasi-experimental evidence; not mature decision-time evidence in 2010. Inspect source. ↩

15. IFS, The short- and medium-term impacts of Sure Start on educational outcomes, “Key findings” Retain proximity-based exposure and “across subjects” wording. Inspect source. ↩

16. Department for Education/DHSC press release, 6 July 2025, “Best Start Family Hubs in every local authority by April 2026”; funding and rollout paragraphs; Giving every child the best start in life Resolves prior pinpoint gap. Commitment and timetable, not outcome or identity with Sure Start. Inspect source. ↩

17. data/institutional-churn-ledger/ledger.json, ICL-2026-001 to ICL-2026-005; PR #175; merge ff92aeaf7a1b6cc7eb2710af7da118be72910718; ICL-2026-001 final classification and authority; ICL-2026-002 final classification and authority; ICL-2026-003 final classification and authority; ICL-2026-004 final classification and authority; ICL-2026-005 final classification and authority; ICL-2026-005 confidence summary and restraints Repository-status correction only. Exact accepted judgement; no national abolition or definitive churn label. Exact accepted judgement with fiscal and successor restraints. Exact accepted judgement; original design need not have proceeded unchanged. Exact accepted judgement. Exact accepted wording: durable adaptive institution. Strong on continuity/process; medium on survival cause. Inspect source. ↩ ↩² ↩³ ↩⁴ ↩⁵

18. Regional Development Agencies Act 1998; NAO closure material; RDA dossier lifecycle table Nine bodies from 1999; eight non-London agencies ceased 31 March 2012. Inspect source. ↩

19. House of Lords Science and Technology Committee written evidence, paras 31–32 and single-budget table: £2.219bn (2008–09), £2.190bn (2009–10), indicative £2.139bn (2010–11) Resolves prior pinpoint gap by narrowing wording to planned single-budget funding around 2009–10; not realised net cost. Inspect source. ↩

20. NAO, Funding and structures for local economic growth, full report pp.9–11 Spending dip is not attributed wholly to abolition; fiscal consolidation retained. Inspect source. ↩

21. Parliamentary answer cited as RDA-10 in dossier; four-year £464m closure provision and scope; Parliamentary answer RDA-11 and asset-transfer statement RDA-05 in dossier Provision, not separately audited realised net cost. £56.4m redundancy estimate and about £300m assets must not be added without reconciliation. Inspect source. ↩

22. RDA dossier gaps; ICL-2026-002 reopening conditions Net transition cost and function/staff/asset reconciliation remain unknown. Inspect source. ↩

23. NAO, Local Enterprise Partnerships, report summary and key facts Projected £12bn envelope, not LEP administrative capacity or realised outcome. Inspect source. ↩

24. Same NAO report, paras 16–17 / report pp.8–9 Survey percentages apply to responding LEPs at the 2016 snapshot. Inspect source. ↩

25. Government LEP integration guidance, headings “Commencement” and “LEP functions” Decision status supported; post-transfer outcomes immature at cut-off. Inspect source. ↩

26. Scottish Government, Enterprise and Skills Review: report on phase 1, part 3 Institutional-continuity mechanism only. Inspect source. ↩

27. House of Commons Library, Zero Carbon Homes, Briefing Paper 6678, document pp.3–4 and p.10 Resolves prior pinpoint gap: 2006 origin, developed route, 2015 cancellation; “roughly nine years” retained. Inspect source. ↩

28. Infrastructure Act 2015, s.37; final allowable-solutions IA, 27 March 2015; HM Treasury/BIS, Fixing the foundations, chapter 9, para 9.17, document p.47 (PDF p.50) Resolves prior pinpoint gap: cancellation covers allowable solutions and proposed 2016 on-site uplift. Inspect source. ↩

29. Dossier search record and decision section; Commons Library timeline; recovered public cancellation record Documented absence in the public record examined, not proof no internal analysis existed. Inspect source. ↩

30. Climate Change Committee, 2020 Future Homes Standard letter, HTML heading “2. Key recommendations”, recommendation 1 Resolves prior pinpoint gap. Modelled comparison; not realised national cancellation cost. Inspect source. ↩

31. Future Homes and Buildings Standards Building Circular 01/2026 and Building Regulations etc. (Amendment) (England) Regulations 2026, commencement provisions Current legal/timetable status; high-rise transition differs. Inspect source. ↩

32. MHCLG, Future Homes Standard final impact assessment, document pp.55–56, Table 12 and paras 14.1–14.4; RPC opinion, “RPC opinion” and “Summary of proposal” Resolves prior pinpoint gap: modelled £11.251bn central net benefit (rounded £11.3bn), £709m annual business cost; red rating applies to IA, not policy objective. Inspect source. ↩

33. FHS IA, strategic case and results/distribution sections; CCC evidence Directional inference; realised totals unknown. Inspect source. ↩

34. Government Warm Homes statement, 2026, paragraph making the one-million-homes claim Official claim only; not independently reproduced. Inspect source. ↩

35. Zero Carbon Homes dossier gaps; ICL-2026-003 reopening conditions No reproducible national stock counterfactual or consolidated preparation-cost account. Inspect source. ↩

36. Offender Rehabilitation Act 2014; Justice Committee, Transforming Rehabilitation, chapter 2; dossier design section Expanded supervision retained as a gain distinct from operating-model failure. Inspect source. ↩

37. Justice Committee, Transforming Rehabilitation, Conclusions and recommendations para 1 (referring to report para 18) and para 37 (report para 200); PAC progress review, conclusion 1 Resolves prior pinpoint gap: national reform introduced without thorough/adequate piloting. Inspect source. ↩

38. NAO progress review, key facts and performance section Binary and frequency measures remain separate; causation qualified. Inspect source. ↩

39. Probation dossier cost table; ICL-2026-004 reopening conditions Consolidated correction and unification cost remains unknown. Inspect source. ↩

40. National Minimum Wage Act 1998; LPC institutional history; The National Minimum Wage in 2026 Nominal rate history only; no real-wage calculation. Inspect source. ↩

41. Low Pay Commission, “About us” and terms-of-reference/process headings Formal representation does not prove equal influence. Inspect source. ↩

42. BIS, National Minimum Wage: report on the 2015 apprentice rate, GOV.UK Details and Cm 9061; National Minimum Wage Act reporting requirement Resolves prior pinpoint gap: government rejected £2.80 recommendation, set £3.30 and laid reasons before Parliament. Inspect source. ↩

43. Low Pay Commission Report 2025, HTML age-rates chapter and recommendations Evidence of judgement under uncertainty, not proof of optimality. Inspect source. ↩

44. Continuity/process record and stated competing explanations Causal contribution of design remains unresolved. Inspect source. ↩

45. National Minimum Wage: 2024 to 2025 Enforcement and Compliance Report, document p.2 Box 1 and p.5 “Enforcement in 2024/25”; Same report, same pages Exact current-at-cut-off figures. Detected cases are not total underpayment. Inspect source. ↩

46. House of Commons Library election-result briefings/datasets for 1997, 2005, 2019 and 2024; arithmetic from seats divided by total seats Calculations retained with denominators; no inference of illegitimacy. Inspect source. ↩

47. House of Commons Standing Orders, 27 March 2026, Standing Order No. 14(1) Exact primary parliamentary rule; retain “normally”. Inspect source. ↩

48. docs/TWIS_DRP_01_MINISTERIAL_TURNOVER_EVIDENCE_V0.1.md, §§1–2 and methodology Reproducible curated-role analysis, not government-wide tenure averages. Inspect source. ↩

49. HM Treasury, Green Book 2026, chapters on the five-case model, options, risk/uncertainty, implementation and benefits management Official guidance; does not prove consistent application. Inspect source. ↩

50. HM Treasury, Magenta Book, HTML headings “The role of evaluation in policy making” and evaluation-design principles Keep “should”; guidance is not a universal statutory duty. Inspect source. ↩

51. House of Commons Committee of Public Accounts, Use of evaluation and modelling in government, Fourth Report 2022–23, “Conclusions and recommendations”, para 1; Summary Resolves prior pinpoint gap: dated 2019 review, not a 2026 government-wide rate. Inspect source. ↩

52. Danish Parliament, “The Government”; German Bundestag, “Adoption of legislation”; German Minimum Wage Commission, “Information on the work”; Scottish Government, Enterprise and Skills Review, phase 1, part 3 Mechanism comparison only; no national outcome ranking. Inspect source. ↩

53. Green Book principles; five dossier cost tables and numerical-audit rules Normative synthesis, not existing law. Inspect source. ↩

54. RDA transition evidence and probation correction evidence Cross-case mechanism; not a measured national prevalence claim. Inspect source. ↩

55. Transforming Rehabilitation evidence and accepted correction judgement Proposal preserves urgent correction. Inspect source. ↩