Choosing between programmes rather than within one

Public appraisal guidance organises spending into a hierarchy of four levels. A policy sets an intent. Portfolios implement it. Programmes coordinate sets of projects. Projects deliver.

Each level takes its objectives from the level above. A project's objectives are set by its programme, a programme's by its portfolio, a portfolio's by the policy. The alignment running down that chain has a name in the guidance: the golden thread.

Then comes the sentence that decides what appraisal can and cannot see.

The appraisal of the project does not need to justify the overarching programme. Similarly, the objectives of a programme are established with reference to its governing portfolio. The appraisal of the programme does not need to justify the overarching portfolio.

Why that is the right rule

It would be easy to read this as a loophole. It is not. It is the only sensible way to appraise anything, and the guidance illustrates why with an example worth keeping.

Take a programme extending a railway to a town with poor connectivity, with three projects underneath it: station refurbishment, track, and signalling. There are several ways to install the signalling, at different costs.

The appraisal of the signalling system should not include the benefits of the railway. Doing so would not help identify which signalling option optimises value for money, which is the only question that appraisal is being asked. Nor should it re-examine the transport needs of the town. That was settled when the railway programme was.

As the guidance puts it, there is no need to imagine that railway signals have some social value in isolation, and it is neither credible nor useful to apportion a share of the programme's benefits to them. The value of the railway is that stations, track and signals function together.

Anyone who has read a business case that solemnly attributes regional economic uplift to a substation will recognise what the rule is preventing.

What the rule leaves behind

Accept that it is right, and a structural consequence follows that the guidance does not dwell on.

The golden thread is a downward alignment check. It confirms that each level serves the one above it. Every mechanism attached to it runs the same way: objectives cascade, benefits are traced down, appraisal optimises within a level.

Nothing runs back up.

FIG. 01Objectives, appraisal and assurance all run down the hierarchy. Each level is tested against the one above it, which means each level takes the one above as given. The portfolio is the only level with nothing above it to be tested against, and it is where the largest sums are committed.

So a project appraisal can be exemplary while serving a programme that should not exist. A programme appraisal can be exemplary while serving a portfolio that was assembled without comparison. And because each appraisal is correct at its own level, the reviewing bodies below the error will all report that things are in order. They are.

The portfolio is the level with nothing above it, in practical terms, and it is the level where the largest sums are committed by the smallest group, earliest, on the least evidence.

The same asymmetry one level down

The question owners actually face

Appraisal as described chooses between options for meeting an objective that has already been set. On a portfolio the harder question is different in kind: which of several defensible things to do, with money that will only stretch to some of them.

That question has no shortlist, because the candidates are not variants of each other. It has no common unit, because a hospital programme, a water programme and a housing programme produce incommensurable things. And it has no natural owner, because each programme has a sponsor who is accountable for delivering it and nobody is accountable for the comparison.

What it does have is a decision, taken every year, usually implicitly.

What happens when the money tightens

The comparison gets made whether or not anybody appraises it. It simply gets made under pressure, quickly, and on the wrong criteria.

At a board meeting in December 2024 the Public Investment Fund approved a minimum twenty per cent spending reduction across its portfolio, and timelines on several giga-projects have been recalibrated since. The reporting of that decision rests on unnamed sources rather than a statement by the fund, so its internal reasoning is not public and should not be guessed at.

What is worth saying is what a reduction of that kind requires of any owner, anywhere. Deciding which programmes slow and which continue is a portfolio appraisal, conducted at speed. In the absence of a pre-existing basis for comparison, the criteria that present themselves are delivery criteria: how far advanced a programme is, how much is contractually committed, how visible it would be to stop, whose sponsor is most senior.

None of those is a measure of value. Every one of them is a measure of momentum. An owner without a portfolio-level basis for comparison does not avoid the decision; it takes the decision using whatever proxies are lying around, in the month it has to be taken.

What a portfolio-level basis requires

The guidance does give one instruction that reaches upward, and it is easily passed over: practitioners should consider the value of projects and programmes in aggregate, because a project may have no measurable social benefit alone while being essential to a programme that does.

That is a rule about aggregation, not about choosing between portfolios, but it names the right unit of analysis. Three things follow from taking it seriously.

Objectives at the portfolio level that are not the sum of the programme objectives. If the portfolio's objective is the list of what its programmes are doing, there is nothing to compare them against and the hierarchy has collapsed into one level wearing four labels.

A common measure, accepted in advance. Not a precise one; a shared one. Programmes that produce incommensurable outputs can still be compared on the contribution each makes to a stated portfolio objective, provided that objective was written before the programmes were.

Somebody entitled to decline. This is the constraint that actually binds. A portfolio that has never declined anything is not allocating. It is recording what was already going to happen, in priority order.

The same test applied to the asset side

What to examine

Ask for the portfolio objectives, and check whether they were written before or after the programmes they are said to govern. The dates settle it.
Ask what the portfolio has declined, and when. Not deferred, declined.
Ask on what basis two programmes in different sectors were compared the last time money was reallocated between them. If the answer describes delivery status rather than value, that is the finding.
Ask which body owns the comparison. Where every programme has a sponsor and the portfolio has a secretariat, the comparison is nobody's and will be made by whoever is loudest at the right meeting.

Sources. HM Treasury, The Green Book, 2026 edition. Paragraphs 3.2 to 3.4 for the policy hierarchy and the definitions of portfolio, programme and project, 3.5 for the golden thread, 3.6 for the statement that appraisal at one level need not justify the level above, 3.7 to 3.8 for the railway signalling example, and 3.9 to 3.10 for considering value in aggregate. The definitions of portfolio, programme and project are drawn by the Green Book from the Government Functional Standard for Project Delivery. The Public Investment Fund spending reduction was reported by Arabian Gulf Business Insight in March 2025, describing a board decision taken in December 2024 and attributed to unnamed sources rather than to a statement by the fund.

Every appraisal can be right and the portfolio still wrong