A government deciding whether to fund research faces a problem that cannot always be solved by a more elaborate spreadsheet: the costs may be clear years before the benefits are. An investment can produce knowledge, skills or opportunities without providing a reliable forecast of the money it will eventually generate.
A report published on 4 November proposes a more explicit way of making that judgement. Instead of forcing every expected benefit into a financial estimate, it sets out a framework for assessing research and development proposals using monetary calculations, other quantitative evidence and qualitative reasoning together.
That is not an announcement that the government has adopted a new funding system. Nor does it mean cost-benefit analysis has been abandoned. The proposal retains robust financial comparisons where they are viable. Its more interesting question is what decision-makers should do when they are not.
Why research resists a neat financial return
The study was commissioned by the Department for Science, Innovation and Technology in December 2025, following Lord Willetts’ review of its business-case and approvals process. The publication page says this workstream has since passed to the Department for Business, Innovation, Science and Trade.
The proposed R&D value for money framework, produced by an SQW-led consortium, addresses a familiar difficulty in research spending appraisal: non-monetised benefits can be important without being easy to compare.
The report describes benefits that are intangible, diffuse, uncertain and long-term. Research may also depend on other investments, produce spillovers beyond its original setting, or generate effects that do not follow a simple, predictable path. Establishing how much of a later outcome should be attributed to one programme can be difficult.
Those problems do not make financial analysis pointless. They make its assumptions consequential. A calculation can look precise while depending on uncertain claims about what will happen, when it will happen and how much the intervention will contribute.
The authors explicitly warn that forcing monetisation can produce misleading precision. Their alternative is not to replace evidence with enthusiasm, but to make the basis of a decision easier to inspect.
What the framework would assess
The proposal uses a rubric: a structured set of criteria, performance standards and supporting evidence. It is primarily intended to compare shortlisted options within an individual business case, including doing nothing or doing the minimum. It is not chiefly designed to rank unrelated research programmes against one another.
Six broad criteria organise the assessment:
- Strategic value: whether the proposal addresses the relevant problem or opportunity and fits wider priorities.
- Economy: whether resources and costs are realistic, with appropriate attention to quality and public finances.
- Effectiveness: whether the intervention is likely to achieve its intended outcomes.
- Efficiency: whether the value generated justifies the resources required, distinguishing outputs from outcomes.
- Equity: where relevant, whether targeted groups and places can participate in and benefit from the work.
- Robustness to uncertainty and risks: whether important uncertainties are visible and there are credible plans to manage them.
The categories are meant to be tailored to the policy being assessed. Early-stage research and a programme closer to commercial deployment need not face identical evidential demands. That flexibility is part of the proposal, rather than an exception to it.
Why there is no overall score
For each criterion, an option would be rated poor, adequate, good or excellent. The rating would come with an explanation, supporting evidence and an assessment of that evidence’s strength: high, moderate or low.
The framework deliberately does not add those ratings into a composite score. The authors argue that aggregation would obscure differences between criteria and risk another form of spurious quantification.
A proposal might be inexpensive to deliver but unlikely to achieve much. Another might offer potentially significant outcomes while carrying substantial technical risk. A single number could conceal precisely the trade-off that decision-makers need to understand.
This leaves responsibility with the people making the decision. They must reach a reasoned overall judgement from the separate assessments, rather than treating a calculated total as the answer. Any extra emphasis given to particular criteria would need to be transparent and justified.
Cost-benefit analysis is still part of the picture
The report is clear that sound, robust benefit-cost ratios should still be presented where viable. Such a ratio compares estimated benefits with estimated costs. The proposed rubric can incorporate one, including a partial calculation whose limitations are acknowledged.
Where benefits can be quantified but not sensibly priced, cost-effectiveness analysis may help. Where neither approach captures an important benefit, qualitative evidence can still contribute to the assessment.
The distinction matters. Not attaching a monetary value to something is not the same as refusing to assess it. The report asks for a structured explanation of why an outcome matters, how the intervention could produce it, and what evidence supports that expectation.
It also distinguishes outputs from outcomes. Counting projects, patents or qualifications can help assess delivery costs, but the report warns against treating such measures as sufficient evidence of value. Where robust evidence about outcomes is available, that should take priority.
How strong is the evidence behind a good rating?
One of the proposal’s most useful distinctions is between the assessment of a proposal and confidence in the evidence supporting it. A favourable rating would not automatically mean the underlying case was well established.
The report gives the example of a judgement supported by only one source: performance might be rated good or excellent while evidence strength remained moderate. That separation makes uncertainty harder to hide behind an upbeat summary.
It also creates a connection between approval and later scrutiny. If a decision depends on an uncertain assumption, the business case should identify what will be monitored and when that assumption will be revisited. The rubric is intended to help establish evidence requirements before implementation, not merely describe success afterwards.
The authors suggest that the preferred option should, as a minimum, be adequate on all relevant criteria and good on strategic value and effectiveness. An adequate assessment could warrant changes or funding conditions, such as checkpoint reviews or a pilot before wider rollout. These are recommendations within the proposal, not newly binding funding rules.
Could qualitative judgement become an easier route to approval?
The obvious objection is that words such as good and adequate can be stretched. A rubric could become another document in which an organisation describes its preferred option favourably.
The report recognises that risk. Practitioners may be reluctant to give their preferred proposal a weak rating. They may also concentrate on easier questions about costs and outputs while paying less attention to whether the intervention will achieve meaningful outcomes.
Its proposed safeguards are procedural: define standards before making the assessment, require reasons and evidence, distinguish confidence from performance, and allow decision-makers to challenge the resulting account. Stakeholder involvement and training may also be needed.
Those safeguards do not remove subjectivity. The authors argue that numerical assessments also involve judgement, including choices about assumptions and which outcomes to count. The relevant comparison is therefore not judgement versus no judgement, but judgement that is visible versus judgement embedded in an apparently authoritative calculation.
Still, the framework cannot make weak evidence strong. The report explicitly says its quality will depend on how rigorously evidence is gathered, analysed and interpreted.
What remains unresolved
The framework needs piloting and testing across different R&D policies. The authors envisage refinement through experience and acknowledge that implementation may require changes in organisational culture, as well as training.
There is no basis in these documents for claiming that the proposal has already improved funding decisions, protected a particular research budget or changed who receives support. Publication establishes that a method has been proposed, not that its intended benefits have been demonstrated.
For readers following UK science policy, the significant development is more specific. The report offers a way to describe uncertainty without either inventing precision or treating unpriced benefits as beyond scrutiny.
Research that cannot promise a financial return still has to justify public support. The proposed framework would make the questions behind that justification more explicit: what matters, what is expected to happen, how strong the evidence is, and what would cause the decision to be reconsidered.
Whether it succeeds will depend less on the labels in the table than on whether decision-makers are willing to expose and challenge the assumptions beneath them.