When does devolution work?

Our view on devolution

Civil Liberties Action supports devolution because people should have a meaningful say in decisions that affect their lives. Local knowledge can improve those decisions, provided the people responsible have the authority, resources and capability to act. Moving a function out of Whitehall should make responsibility clearer and public power easier to challenge.

Within England, central government should protect rights, maintain essential standards and address inequalities between places. Strategic authorities should take decisions that require regional coordination. Councils and communities should retain or gain control over matters they can handle effectively locally. The division should follow the work and its consequences, with a clear explanation of why each responsibility sits where it does.

The evidence gives grounds to pursue this approach with care. It shows opportunities for more coordinated services and longer-term funding, alongside weaknesses in oversight and limited evidence about some outcomes. Each transfer needs a practical account of what will change, how it will be implemented and how its results will be assessed. This paper considers devolution within England; the constitutional settlements of Scotland, Wales and Northern Ireland are outside its scope.

1. What the evidence tells us

The National Audit Office’s July 2026 review found that Integrated Settlements were encouraging longer-term thinking and had the potential to improve value for money. It also identified reporting burdens greater than anticipated, questions about capacity and a need to strengthen local accountability. Arrangements for responding to underperformance remained untested. The report therefore supports greater flexibility alongside work on how that flexibility is governed. Its scope does not establish the wider effects on economic growth or workforce skills. [E1]

Greater Manchester’s health reforms offer an example of the possible benefits of local coordination. Britteon and colleagues examined 98 health-system measures up to February 2020, using a generalised synthetic control and adjustment for multiple testing. Their 2024 study reports improvements in population health, with mixed findings across some services. The authors identify coordination and additional transformation funding as possible explanations, despite limited devolved control over health resources. [E2]

That comparative approach provides stronger evidence than a simple account of improvement after a reform. It remains an observational study of one region and a package of changes. We reviewed the published abstract, which does not allow us to determine how much of the result came from devolved powers, extra funding or other changes. The study is a reason to investigate how local coordination works; its estimates should not be used as a forecast for every future transfer.

The Community Life Survey gives a useful starting point for the democratic argument. In 2024/25, 24% of adults aged 16 and over in England felt able to influence local decisions, while 50% thought that influence was important. The 24% figure was not significantly different from 23% in the preceding year. These findings identify a problem for public institutions. They do not establish that devolution has caused it, or that moving powers will by itself resolve it. [E3]

The Teesworks review shows why the arrangements for challenge matter. On the information supplied to the panel, the independent review found no evidence supporting corruption or illegality. It nevertheless identified deficiencies in governance, transparency and oversight that prevented sufficient demonstration of value for money. The lesson is to make liabilities, conflicts of interest, delegations and the reasons for major decisions visible before commitments become difficult to reverse. That standard should apply to national and local government alike. [E4]

International experience also points towards the conditions under which decentralisation can work. The OECD emphasises clear responsibilities, adequate funding, capable institutions, coordination, participation, transparency and equalisation between places. Its guidance helps organise the design of a settlement; it does not establish that one institutional model will produce better results everywhere. [E5]

2. Responsibility and resources

Begin with the decisions that need to improve. Identify the people who make them now, the information they use, the approvals required and the points at which work passes between bodies. This should expose delays and duplication, as well as the functions that need to remain shared. The case for a regional decision should explain the cross-boundary effects or service relationships that justify that scale.

For each transferred function, publish who decides, which staff and budgets move, which liabilities follow and which existing approvals end. Give the receiving authority enough financial certainty to plan and enough capability to carry out the work. Retained ministerial decisions and the grounds for intervention need equal clarity. A transfer that leaves the authority seeking permission for the same routine choices may do little to improve delivery.

Funding should reflect the responsibilities accepted, the costs of transition and the needs of the population. National redistribution remains necessary where tax bases and service pressures differ. Local discretion should operate within a settlement that protects essential standards and allows places with weaker starting positions to invest.

These arrangements can be assessed through five questions: does the geography fit the function; do resources match the responsibility; are rights and participation protected; can decisions be scrutinised; and are there workable arrangements for learning, continuity and correction? The answers should be supported by a responsibility map, a costed funding agreement, an equality and participation assessment, published scrutiny arrangements and an evaluation and continuity plan.

3. Accountability and implementation

Responsibility should be visible to the person using a service. A resident should be able to identify the decision-maker, understand the reasons for a decision and find a route to challenge it. Consultation needs to take place while there are choices to make. Scrutiny needs timely information and a response from the executive. Audit findings need an owner, a completion date and evidence that the agreed action happened.

The risks of change should be considered alongside the costs and risks of the present system. The Green Book provides the basis for appraising options and their distributional effects. The Magenta Book helps design evaluation before implementation, while the Orange Book provides a discipline for assigning, monitoring and responding to risk. Together with the OECD guidance, these offer a sound basis for preparing a transfer. [E5] [E6] [E7] [E8]

Preparation should be funded, with operational responsibility moving in stages as the necessary capability is demonstrated. Test payments, information flows and service continuity before the handover. Agree in advance how an emerging problem will be investigated, who may intervene and how responsibility will return to local control. The extent of intervention should reflect the problem and the lawful powers available.

The accompanying Devolution Guarantee applies this approach to adult skills. It sets out the work, funding, responsibilities and protections that would need to be agreed before an authority took on or revised a commissioning function. The purpose is to make a settlement understandable and usable by the people affected.

4. Building the evidence

Our research should follow decisions and their consequences. A useful first exercise would ask residents to find the person responsible for a service, locate the reasons for a decision and identify a complaint route. Repeat the exercise after clearer information and responsibilities have been introduced. Record completion, time taken, understanding and accessibility, with the recruitment method and non-response made clear.

Adult skills provides a practical opportunity to examine service outcomes. A study should follow participation, completion, progression and learners’ experience, using pre-transfer trends and credible comparison areas. It should distinguish the effect of changed responsibilities from additional funding, altered provision and wider labour-market conditions. The analysis should be agreed before results are examined, including the main outcomes, timing and treatment of uncertainty. [E7]

A further study should trace scrutiny recommendations through executive responses and subsequent delivery. Select decisions by a rule published in advance, and include recommendations that were rejected or left incomplete. This would show whether challenge reaches the people who can act and whether commitments are carried through.

Administrative costs also need to be measured. Record the staff time spent on bidding, approvals and reporting before and after a settlement changes. Include the cost of new local assurance arrangements. A reduction in the number of forms is useful only if it reduces the overall burden or improves the quality of decisions.

Finally, repeat common questions about influence and examine who participates, whose proposals affect decisions and who remains excluded. Area averages can conceal worse access for rural communities, disabled residents or people with low incomes. Collect only justified data, report missing observations and protect identities, including small groups that could be recognised. Documented experience can explain a finding without being treated as representative of everyone.

These are proposed studies. No fieldwork, recruitment or authority partnership has begun. A pilot’s first six months should establish the baseline, record implementation and test whether residents can use the arrangements. Employment and earnings require longer follow-up, potentially a year or more. Funding for evaluation and access to the necessary data should be agreed before launch.

5. Judging the results

A convincing evaluation needs to consider what would have happened without the change. Comparison areas should be selected using prior characteristics and trends. Analysis should examine changes in funding, other reforms, movement of people or activity between areas and changes in boundaries. Where a credible comparison is unavailable, process and contribution evidence can still be useful, provided the uncertainty about causation is explicit. [E7]

Better access, clearer responsibility and sustained improvements against a credible comparator would strengthen the case for extending a transfer. Worse access, higher net costs without compensating benefits, greater exclusion or ineffective remedies would require correction and could weaken the case for that particular arrangement. If gains depend mainly on additional funding, that should change the argument about institutional design even where the funded programme remains worthwhile.

Publishing unfavourable and inconclusive findings should be part of the research agreement. CLA’s commitment is to decisions that people can influence, services that respond to their needs and public power that remains accountable. Evidence should guide where responsibility sits and how it is exercised.

Scope and sources

This is a targeted review of primary publications checked on 16 October 2026. Searches covered English devolution, outcomes, funding, accountability, health, participation and risk, using government, National Audit Office, OECD and academic sources. We selected material on observed outcomes, documented institutional risks and the design and evaluation of transfers, including adverse and mixed findings. Commentary and promotional case studies were not treated as impact evaluations.

The review was not an exhaustive database search, a meta-analysis or independent peer review. Public audit informs the diagnosis of institutions; the survey measures reported influence; the Teesworks review concerns a particular case; and the guidance offers design principles. The health study was reviewed at abstract level, without a full risk-of-bias assessment. A fuller review should examine its methods and supplementary analysis, search systematically for null findings and use a second reviewer to check selection and interpretation. The references identify the material inspected.

References

Sources checked on 16 October 2026. The notes below identify the material inspected.

• [E1] NAO (2026), Devolution in England: funding and accountability. 1 July 2026; Scope and Conclusions on the report landing page. Independent public audit.
• [E2] Britteon et al. (2024), The impact of devolution on local health systems. Social Science & Medicine 348, 116801; repository abstract and bibliographic record. DOI: 10.1016/j.socscimed.2024.116801. Peer-reviewed observational study; abstract reviewed.
• [E3] Community Life Survey 2024/25: Civic engagement and social action. Civic participation: influencing local decisions; England adults aged 16+. Official survey statistics.
• [E4] Independent Teesworks review (2024): executive summary and recommendations. 29 January 2024; paragraphs 1.6-1.8 and recommendations on oversight, liabilities and conflicts. Independent commissioned case review.
• [E5] OECD (2019), Making Decentralisation Work. Handbook chapter: ten guidelines for effective decentralisation. Comparative policy guidance.
• [E6] HM Treasury, The Green Book (2026). Appraisal of options, costs, benefits, risk and distributional effects. Government appraisal guidance.
• [E7] HM Treasury, Magenta Book: Central Government guidance on evaluation. Theory of change, impact evaluation and counterfactual approaches. Current HTML edition checked. Government evaluation guidance.
• [E8] HM Treasury, The Orange Book: Management of Risk. Risk governance, responsibilities, monitoring and assurance. Current HTML edition checked. Government risk guidance.

Suggested citation: Civil Liberties Action (2026), When does devolution work?, Published briefing | Version 1.1, 16 October 2026.

Corrections and documented evidence: contribute evidence or policy@civillibertiesaction.org.uk. Please identify the passage and supporting source.