What must change in Whitehall

1 The decision we are asking government to take

Government should require every substantial transfer of a public function to include a published account of the decisions that will move, the resources needed to exercise them and the approvals that will end. Parliament and residents should then be able to see whether the transfer happened. This is the next practical step in CLA’s Power Closer to People campaign.

The starting point should be adult skills in an authority already responsible for the service. It offers a bounded way to examine whether local discretion is usable, whether funding reaches the intended provision and whether a learner can challenge a decision. Beginning with an existing programme reduces the disruption involved in testing a new accountability arrangement. It does not remove financial or delivery risk.

The national interest remains substantial. People should retain their legal rights wherever they live. Government must provide a fair funding settlement, support services that cross boundaries and intervene lawfully where serious failure requires it. Local discretion needs to operate within those protections. The purpose of the proposed settlement is to make the division of responsibility intelligible and to test whether it works.

2 The legal starting point

The English Devolution and Community Empowerment Act 2026 received Royal Assent on 29 April 2026. Section 32 and Schedule 13, concerning education, came into force on that day under section 108. Their commencement should not be confused with the date on which every authority can exercise every function. The inserted section 120B makes separate provision for authorities already exercising adult education functions and for other authorities. [L1, L2]

Greater Manchester’s earlier adult education settlement applied to academic years beginning on or after 1 August 2019. The new statutory framework continues pre-existing functions through the amended legislation. A current operating agreement must therefore start with the amended 2009 Act, the 2026 Act and applicable instruments, rather than reproducing the original 2018 order as if it were the whole present settlement. [L2, L4]

National influence also continues. The inserted section 114A preserves directions concerning eligibility rules for awards and requires regard to relevant Secretary of State guidance. The adult education provisions contain specific exclusions and limits, including apprenticeship training and particular provision for younger learners with Education, Health and Care plans. These details matter when describing what a mayor or authority can decide. [L2]

Other parts of the 2026 Act follow different commencement arrangements. For example, the second commencement regulations brought specified audit provisions into force on 15 July 2026, with express limitations. CLA’s proposed Devolution Guarantee is an additional policy proposal. We do not describe it as a statutory scheme or assume that the whole Act commenced together. [L3]

3 Publish the decisions that will move

For each function, the sponsoring department and receiving authority should prepare one schedule identifying the decision, its legal basis, the person or body entitled to take it and any necessary national approval. Every retained approval should have an explanation. If the reason is that legislation requires it, the schedule should say so. If the approval is an administrative habit or a condition government has chosen to impose, it should be examined on its merits.

For adult skills, the schedule should distinguish programme priorities, provider allocations, eligibility, individual learner decisions and service complaints. These are different decisions. Giving an authority discretion over the distribution of a budget does not give it unrestricted power to alter an entitlement or a provider’s contractual obligations.

The schedule should also show where further decisions can sit with councils and providers. An authority may set regional priorities while a college determines how to organise teaching and a council addresses local barriers to participation. Residents need an effective opportunity to influence choices that remain open. A learner panel can advise on access and service design, but it cannot acquire a statutory power simply because an agreement calls it a decision-maker.

Before the transfer starts, the department should publish the approvals it will cease to require and the reporting returns it will withdraw. Where an approval cannot yet be removed, it should give the reason and the next review date. This provides a concrete test of whether Whitehall has changed its own work.

4 Make the funding usable

The published functional responsibilities for Integrated Settlements describe the intended policy scope, including non-apprenticeship adult skills. They are not, by themselves, a complete account of an authority’s statutory powers, available budget or grant conditions. Those documents must be read together. The framework also identifies a distinct Greater Manchester employment-support pilot, which should not be treated as a general power held by every area. [F2]

The national adult skills funding rules for 2026/27 explain that policy responsibility moved to DWP in September 2025, while DfE continues implementation on its behalf. Those rules generally exclude residents of devolved areas, subject to specified continuing-learner arrangements. A Greater Manchester pilot must use the applicable local rules and funding agreement. We have not yet obtained the complete current local rules for this exercise. [F1]

The published section 31 adult skills allocation tables exclude authorities whose relevant funding sits within Integrated Settlements. They therefore cannot be used as a substitute for Greater Manchester’s actual skills budget. Academic-year commitments also need to be reconciled with financial-year funding. A budget can appear balanced while leaving an unfunded commitment to learners who continue into the next year. [F3]

CLA proposes a funding schedule that shows the committed learning budget, learner support, administration, scrutiny, evaluation and any permitted reserve separately. It should identify which funding source can lawfully pay for each cost. Transition costs should be explicit. A promise to devolve should not depend on concealing the cost of finance staff, contract management, accessible participation or a functioning complaints service.

Government should maintain a published method for addressing differences in need and resources between places. Further local revenue powers would need a separate distributional assessment and a workable equalisation arrangement. The pilot proposed here assumes no new tax power and makes no claim about savings that have not been measured.

5 Change the work of departments

The department should establish what it currently does for the function: making allocations, negotiating exceptions, collecting returns, checking compliance, resolving problems and maintaining national data. It should record the staff time and cost of that work before a transfer. The authority should make the equivalent record. Otherwise an apparent reduction in central administration may simply be an unfunded increase locally.

Some specialist capability may need to move with the function. Some may be better provided through a shared service that several authorities can use. Staff transfers, access to systems, contract changes and information-sharing arrangements require their own lawful implementation. The relevant agreement should name the organisation providing each capability, when it will be available and who pays. This paper does not assume that a ministerial announcement transfers staff or contracts.

National data should remain comparable where comparison serves a clear public purpose. Departments should agree a small common set of definitions with authorities and providers, explain why each return is required and reuse information already collected where lawful and suitable. An outcomes framework should not become a second set of detailed controls over every local delivery choice.

The NAO’s July 2026 examination identifies both the potential of Integrated Settlements and unresolved questions about funding and accountability. That supports a careful implementation programme with explicit tests. It does not establish that devolution has already improved adult learning or reduced administration. [E1]

6 Give residents a route to influence and remedy

The public account of a programme should identify who made the relevant decision, what evidence they considered, what options were available and where a resident can raise a concern. Supporting documents should remain accessible, with versions and dates. People who need a telephone or paper route should be able to obtain equivalent information.

Consultation should take place while a meaningful choice remains. An authority should explain which parts of the proposal are fixed by law or funding and which it can change. After consultation, it should publish the main issues raised, its response and the changes made. Where it rejects a proposal, it should give its reasons. A count of responses is not evidence that residents controlled the decision.

Scrutiny should be able to follow an issue through to an executive response, a responsible officer and a completed action. Our three-authority review has located published arrangements. It has not established that those arrangements are effective in practice. Its observations remain subject to factual correction, with no rankings attached.

A complaint about the authority, a provider’s decision and a matter within a regulator’s jurisdiction may have different routes. The public guide should explain those distinctions and applicable time limits. Participation in a CLA pilot would neither replace an existing remedy nor extend the jurisdiction of an Ombudsman. The authority’s legal and complaints teams should approve the guide before it is used.

7 Deal with failure without undermining responsibility

A low-risk transfer requires preparation for foreseeable failure. The authority should identify who can authorise corrective action if a provider fails, payments stop, a serious rights issue arises or spending exceeds available resources. Continuing learners need a practical continuity plan, including records, alternative provision and communication with those affected.

The response should match the problem and the lawful powers available. An immediate threat to safety or rights may require urgent action. A weak outcome measure may require investigation before a delivery change. A national intervention should have a stated legal basis, reasons, scope and review arrangements. The pilot should not invent an automatic power for Whitehall to resume a function.

Existing statutory officers and decision-makers retain their responsibilities. An independent reviewer can examine evidence and report concerns, but should not be presented as a substitute for the authority’s governance. Published review dates and an honest account of unresolved action will do more to establish credibility than a general assurance that risk is under control.

8 Test the arrangement before extending it

CLA proposes two connected exercises. The first traces a defined sample of adult-skills decisions and observes how a small, varied group of residents finds and understands the relevant information. The second, if an authority agrees, tests the Devolution Guarantee within one existing programme. Both should publish their methods before collecting the main evidence.

The resident exercise is exploratory. Twenty participants cannot establish how the whole population experiences devolution. It can identify obstacles, misunderstandings and promising changes worth testing more widely. The document review should record an unavailable answer as unknown, rather than assume that a missing web page proves the underlying action did not happen.

A six-month pilot review should examine whether responsibilities became clearer, promised actions were completed, reporting effort changed and serious problems were resolved. Learner progression may require longer follow-up. Before-and-after changes cannot, on their own, show that the Guarantee caused better outcomes. A stronger impact study would need a credible comparison and sufficient observations. [M1]

Within three months, we should be able to produce a corrected account of the three authorities, report the first bounded study if recruitment and review are in place, and reach a documented decision about a pilot. The decision could be to proceed, revise the design or stop. Each would be a useful result if supported by evidence.