Public Sector Cost-Benefit Analysis: A Practical Guide

In public sector work, a cost-benefit analysis is not just a finance exercise; it is a way to test whether an option genuinely improves public value once implementation costs, benefits, risks, and distributional effects are all counted. The practical sequence is simple on paper but easy to weaken in practice: define the problem, set a credible baseline, compare realistic alternatives, value the impacts, and check whether the result still holds when assumptions move. In the UK, that means thinking like a public servant responsible for outcomes, not just like someone balancing a budget line.

What cost-benefit analysis is trying to prove in public sector work

For government operations, the point of a cost-benefit analysis is to compare the social value of different options, not to maximise profit. That distinction matters because a proposal can be financially expensive and still be worthwhile if it delivers stronger outcomes for citizens, better service reliability, or lower long-term risk. The current Green Book frames appraisal as a way to assess costs, benefits, and risks so officials can advise decision makers on the best way to achieve government objectives.

When I review a public-sector appraisal, I look for one basic question underneath everything else: does this option produce enough public value to justify the resources and trade-offs it needs? That is a better test than “is it cheap?” and a more honest test than “does the ratio look good?”

The real question behind the spreadsheet

A strong appraisal should tell you which option offers the best balance of outcomes, how sensitive that answer is, and what has not been fully monetised. In practice, that means the method supports judgment rather than replacing it. It also means the analysis has to be proportionate: a small operational change does not need the same level of modelling as a national infrastructure programme, but it still needs a disciplined comparison of alternatives. Once that purpose is clear, the next task is to define the problem and the baseline properly.

Define the problem, the baseline, and the options

The first practical step is to state the decision problem in plain language. If the proposal is trying to reduce waiting times, improve compliance, cut fraud, modernise a service, or redesign a process, say so directly. I also like to force the objective into something measurable, because a vague objective creates a vague appraisal.

1. Define the objective. Write down what the intervention must achieve, who it is for, and what success looks like.
2. Set the baseline. Describe the “do nothing” or “do minimum” position accurately, including current costs, service levels, and known pressures.
3. Build a shortlist of real options. Compare only alternatives that are genuinely deliverable and materially different.
4. Fix the appraisal period and geography. Decide how long effects will be tracked and which places or populations are in scope.
5. Document assumptions early. Anything that later becomes “obvious” should already be recorded and defended.

The baseline is where many analyses quietly drift off course. If you compare a new model against an unrealistically bad status quo, the proposal will look better than it is. If you compare it against an unrealistically perfect service, it will look weaker than it is. The same problem appears when teams include only the options they already like; a shortlist should contain at least one do-minimum option and other credible ways forward. The more disciplined the framing is here, the less damage you have to repair later in the numbers.

Value the costs and benefits with enough discipline to trust the result

This is the part most people think of first, but it works properly only if the earlier framing was sound. In government operations, I try to separate three things: direct financial cost, wider social cost, and impacts that are real but not easily monetised. That keeps the appraisal honest without pretending that every effect can be reduced to pounds and pence.

Impact type	How I would treat it	Typical public-sector example
Direct financial cost	Include capital, staffing, procurement, maintenance, training, and transition costs in real terms.	New case management system, extra caseworkers, supplier onboarding.
Monetised benefit	Put a value on measurable gains where evidence supports it, using market data or accepted valuation methods.	Reduced travel time, fewer repeat contacts, lower processing time.
Unmonetised impact	Describe clearly if valuation is weak or impractical; do not hide it just because it is awkward.	Trust in a service, dignity, user experience, staff morale.
Distributional effect	Show who gains and who loses if different groups or places are affected differently.	Urban versus rural access, effects on low-income households, protected groups.
Risk and uncertainty	Record the main uncertainties and make them visible rather than burying them in a single point estimate.	Demand forecast error, delivery delays, supplier failure, policy change.

The Green Book expects unmonetisable impacts to be handled explicitly, not ignored. That is important in public service settings where a proposal may have modest budget costs but significant effects on fairness, access, safety, or wellbeing. I would rather see a cautious qualitative judgement than a fake monetary figure that looks precise but is built on sand. Once the impacts are captured, the next issue is how to compare them across time and across options.

Compare the options using present value, not gut feel

Future costs and benefits cannot be read properly in their raw form, because a pound next year does not carry the same weight as a pound today. That is why appraisal uses discounting and present value. In the current Green Book, the real social time preference rate is 3.50% for years 1 to 30, 3.00% for years 31 to 75, and 2.50% from year 76 onward; health and life effects are discounted at 1.5%. Those rates matter because many government schemes pay back slowly, and the appraisal has to show that delay clearly.

Metric	What it tells you	Where it can mislead you
Net present social value	The net gain to society after discounting all measured costs and benefits.	It can hide distributional issues if you read it alone.
Benefit-cost ratio	How much value is created for each pound spent.	A high ratio can still sit on a small absolute gain.
Sensitivity analysis	How much the result changes when key assumptions move.	It is only useful if you test the assumptions that actually matter.
Switching value	The point at which the preferred option stops being value for money.	It is easy to overstate confidence if the switching value is close to the central estimate.
Appraisal summary table	A compact view of the core economic, financial, distributional, and risk evidence.	It should summarise judgment, not flatten everything into one score.

I never read a benefit-cost ratio in isolation. A recent Green Book update makes that point even more strongly by discouraging crude pass or fail thinking based on a single ratio, especially where schemes affect different places or communities in very different ways. A ratio is useful, but it is not a verdict. What matters is whether the option still looks good when uncertainty is stressed and the wider consequences are visible. That is where the method becomes useful in real government work.

How these steps change in real UK government operations

In a ministry, arm’s length body, or local authority, cost-benefit analysis is rarely a stand-alone spreadsheet. It sits inside a business case, a policy submission, a funding bid, or a delivery decision. That means the analysis has to speak to finance, policy, procurement, and operational teams at the same time. If it only makes sense to economists, it is not ready for decision makers.

There are a few patterns I see over and over in public sector appraisals:

• Budget savings are treated as total benefits. That misses service quality, user outcomes, and long-term risk.
• Implementation costs are underestimated. New systems, training, migration, and temporary duplication often cost more than the initial model suggests.
• Different places are compared too crudely. A rural access scheme and an urban regeneration scheme should not be judged on the same headline ratio alone.
• Distributional effects are left vague. If a proposal helps one group while burdening another, decision makers need to see that clearly.
• Evaluation is forgotten. The Green Book treats appraisal as one stage in the broader ROAMEF cycle, so monitoring and evaluation should be planned from the outset, not added later as an afterthought.

I also pay attention to how the result is presented. The standard appraisal summary table is useful because it forces a balanced view: the monetised result, the non-monetised impacts, the public sector financial effect, the distribution of impacts, and the uncertainty around all of them. In other words, it makes the appraisal harder to game. That is exactly what good government analysis should do. The final step is making sure the appraisal is decision-ready rather than just technically complete.

What a decision-ready UK public sector appraisal should still check before sign-off

• Is the objective specific enough that someone outside the project team would understand it?
• Does the baseline reflect the real status quo, including current constraints and likely behaviour?
• Have the major costs and benefits been counted in real terms over the right time horizon?
• Have unmonetised impacts, equality issues, and place-based effects been made visible?
• Do the sensitivity results show whether the preferred option is robust or fragile?
• Does the narrative explain why the preferred option is better, not just that its ratio is higher?

If those checks are in place, the analysis is usually strong enough to support a serious public-sector decision. If they are not, I would treat the appraisal as provisional, because the problem is usually not the maths but the framing. In government operations, the best cost-benefit work is proportionate, transparent, and honest about uncertainty, and that is what gives decision makers something they can actually use.