Cost-Benefit Analysis

This page explains what cost-benefit analysis asks, how its baseline and perspective determine which effects are included, and how costs and benefits are valued over time. It then shows how net present value and benefit-cost ratios are calculated, why those measures can rank options differently, and how transfers, distribution, uncertainty, unmonetised effects and affordability should be handled.

Cost-benefit analysis is commonly shortened to CBA. Its purpose is to examine whether the social gains created by an option exceed the value of the resources and opportunities it displaces. Expressing consequences in money creates a common analytical unit, but it does not mean that health, life, equity or other social consequences have only monetary value.

What cost-benefit analysis asks

Cost-benefit analysis asks whether an alternative produces benefits that are worth more than its costs under a stated perspective and set of assumptions. Because costs and benefits are both expressed in monetary terms, CBA can compare consequences that would otherwise use different units. It can also compare programmes operating in different sectors.

The result is commonly expressed as net present value, which subtracts discounted costs from discounted benefits. A positive result indicates that the monetised benefits exceed the monetised resource costs under the stated assumptions. In welfare-economic terms, this corresponds to a potential Pareto improvement under the Kaldor-Hicks compensation test: those who gain could in principle compensate those who lose, although no compensation need actually be paid. It does not automatically establish that the option should be adopted.

Which choices define a cost-benefit analysis

A CBA is shaped by decisions made before costs and benefits are calculated. These choices determine whose consequences are counted, what is compared, how effects are valued and how the results should be interpreted. They should be visible in the analysis rather than hidden in a model.

1. Define the decision problem and intended decision-maker.
2. Specify the baseline and mutually compatible alternatives.
3. Select the analytical perspective and affected population.
4. Identify incremental resource costs, benefits, transfers and other consequences.
5. Measure the quantity and timing of each consequence.
6. Apply defensible market, shadow or non-market monetary values.
7. Discount future costs and benefits using the stated convention.
8. Calculate present values, net present value and any supporting ratios.
9. Compare mutually exclusive alternatives incrementally.
10. Examine uncertainty, switching values and scenario results.
11. Report who receives the benefits and who bears the costs.
12. Retain material unmonetised effects explicitly.
13. Interpret social value alongside affordability, equity, feasibility and evidence quality.
14. Report methods, assumptions, limitations and results transparently.

Choosing the baseline and perspective

The baseline describes what is expected to happen without the proposed change. Costs and benefits should be measured relative to a credible baseline, such as current practice, a do-minimum option, no additional intervention or the expected future position without the proposed change.

An unrealistic baseline can create apparent costs or benefits that are not genuinely incremental. The baseline should therefore reflect what is reasonably expected to occur if the evaluated option is not adopted.

The perspective determines whose costs and benefits are included. A healthcare-system perspective may include different consequences from a government, public-sector or societal perspective. The analysis should state whose welfare counts and apply that boundary consistently.

Identifying resource use and consequences

A CBA should begin with a clear inventory of consequences before monetary values are assigned. This helps prevent important effects from being omitted or counted more than once.

For each cost or benefit, the inventory should record:

• The resource, outcome or consequence being measured.
• Whether it is a cost, benefit, transfer or unmonetised effect.
• The affected person, organisation or population.
• The quantity and unit of measurement.
• When the consequence occurs.
• The monetary valuation method.
• The price year and currency.
• The evidence source.
• Any overlap with another recorded consequence.
• Important assumptions and uncertainty.

Relevant resource use might include medicines, staff time, facilities, equipment, hospital care, community services, patient time, caregiver time, implementation, training and administrative activity. Which resources are included depends on the analytical perspective and decision problem.

For a resource item:

Resource cost = Resource quantity × Unit value

Resource quantities and unit values should remain separate so that each can be checked, updated and tested independently.

Valuing market and non-market consequences

Some consequences have observable market prices, while others require an estimated social value. The valuation method should fit the consequence, population, perspective and decision context. No single valuation method is automatically appropriate for every effect.

Possible valuation approaches include:

• Observed market prices.
• Market prices adjusted to reflect social opportunity cost.
• Shadow prices.
• Revealed-preference methods.
• Stated-preference methods.
• Willingness to pay.
• Willingness to accept.
• Values assigned to changes in mortality or morbidity risk.
• Human capital estimates, which value changes in productive output rather than individual welfare.

For each monetary value, the analysis should report its source, population, elicitation method, jurisdiction, currency, price year, uncertainty and transferability limitations.

A value estimated for one population or jurisdiction should not be transferred automatically to another. Differences in income, preferences, risks, services, institutions and valuation methods can materially affect the estimate.

How willingness to pay can reflect income

Willingness to pay measures the maximum amount a person would give up to obtain a benefit or avoid a loss. Willingness to accept measures the minimum compensation a person would require to accept a loss or give up a benefit.

Ability to pay can influence willingness-to-pay estimates. Analysts should therefore identify whose preferences were measured and examine whether differences in income affect the estimated values or the distribution of benefits.

Willingness to pay should not be interpreted as a complete measure of moral worth, need or entitlement.

What a value of statistical life represents

A value of statistical life aggregates individual valuations of small changes in mortality risk across a population. It is not the price of an identified person’s life.

For example, using illustrative figures, if 100,000 people would each pay £50 for a reduction in mortality risk of 1 in 100,000, the aggregate willingness to pay would be £5 million for one expected statistical life.

Applying a value of statistical life requires:

• A defensible baseline risk.
• A clearly measured change in risk.
• An identified affected population.
• An appropriate valuation source.
• A stated price year and currency.
• Consideration of jurisdictional and population relevance.
• Transparent treatment of uncertainty.

Bringing future costs and benefits to present values

Costs and benefits may occur at different times. Discounting converts future monetary amounts into present values so that the timing of alternatives can be compared consistently. The selected discount rate and timing convention can materially affect the results for programmes with long-term consequences.

For an amount occurring in period t:

Present Valueₜ = Amountₜ ÷ (1 + r)ᵗ

In this formula:

• r is the applicable discount rate.
• t represents the timing of the amount under the stated convention.
• An amount occurring at time zero normally has a discount factor of one.

For option j:

PV Benefitsⱼ = Σ[Benefitsⱼₜ ÷ (1 + r)ᵗ]

PV Costsⱼ = Σ[Costsⱼₜ ÷ (1 + r)ᵗ]

NPVⱼ = PV Benefitsⱼ − PV Costsⱼ

When discounted resource costs are positive:

BCRⱼ = PV Benefitsⱼ ÷ PV Costsⱼ

A positive NPV means that monetised benefits exceed monetised resource costs under the stated baseline, perspective, valuation methods, time horizon and assumptions.

A benefit-cost ratio is undefined when discounted resource costs equal zero. Ratios can also be misleading when costs and benefits have been classified inconsistently.

Calculating net present value and the benefit-cost ratio

Net present value measures the absolute difference between discounted monetised benefits and discounted resource costs. It estimates how much net monetised value an option produces.

The benefit-cost ratio expresses discounted monetised benefits relative to discounted resource costs. A ratio greater than one means that monetised benefits exceed resource costs under the stated assumptions.

NPV and BCR answer related but different questions:

• NPV asks: How much net monetised value does the option produce?
• BCR asks: How large are monetised benefits relative to resource costs?

The measures should not be treated as interchangeable.

Comparing mutually exclusive alternatives

Mutually exclusive alternatives cannot all be adopted together. Their comparison should normally focus on incremental social value.

The option with the highest benefit-cost ratio is not necessarily the option that produces the greatest total net value. Differences in programme scale can cause NPV and BCR to rank options differently.

For Option B compared with Option A:

Incremental NPV = NPVᴮ − NPVᴬ

When alternatives are mutually exclusive, the preferred option on monetised net-social-value grounds is normally the feasible alternative with the greatest defensible NPV. It is not automatically the alternative with the highest BCR.

Worked example: when NPV and BCR disagree

Consider two hypothetical mutually exclusive options (illustrative figures).

Option A

• Present value of resource costs: £133,168
• Present value of monetised benefits: £251,505
• Net present value: £251,505 − £133,168 = £118,337
• Benefit-cost ratio: £251,505 ÷ £133,168 = 1.89

Option B

• Present value of resource costs: £504,015
• Present value of monetised benefits: £893,954
• Net present value: £893,954 − £504,015 = £389,939
• Benefit-cost ratio: £893,954 ÷ £504,015 = 1.77

Option A has the higher BCR, but Option B has the higher NPV.

Incremental NPV = £389,939 − £118,337 = £271,602

On monetised net-social-value grounds, Option B produces £271,602 more net value than Option A.

That conclusion remains subject to uncertainty, affordability, distribution, feasibility, evidence quality and material unmonetised effects.

Assumptions behind the worked example

The worked result assumes that:

• Options A and B are genuinely mutually exclusive.
• Both options are compared with a consistent baseline.
• Costs and benefits follow the same analytical perspective.
• Resource quantities and monetary values are measured correctly.
• The same time horizon and timing conventions are used.
• Discounting is applied consistently.
• Transfers have not been misclassified as resource costs or benefits.
• Costs and benefits are not counted more than once.
• Material unmonetised effects remain visible.
• Both options are feasible.

Changing any of these assumptions could change the ranking.

Recognising transfers and real resource effects

A transfer changes who holds money or another claim but does not automatically consume or create social resources. Taxes, subsidies and benefit payments may therefore be treated differently in a social CBA than in a financial analysis.

A transfer can still have important consequences. Administrative costs, behavioural responses, tax distortions and distributional effects may represent real social effects even when the payment itself is treated as a transfer.

The analysis should distinguish:

• Financial payments.
• Transfers between groups.
• Real resource use.
• Behavioural or distortionary effects.
• Distributional consequences.

The treatment of each item should follow the stated perspective and remain consistent throughout the analysis.

Preventing double counting

Double counting occurs when overlapping consequences are valued more than once. This can exaggerate benefits or costs and change the preferred alternative.

Examples include:

• Valuing a health improvement directly and also counting a willingness-to-pay estimate that already includes that health improvement.
• Counting productivity gains separately when they are already included within a broader willingness-to-pay value.
• Counting the same implementation resource under multiple budget headings.
• Treating a transfer payment as both a payer cost and a net social cost.
• Counting caregiver effects in both patient utility estimates and separate caregiver benefits without checking for overlap.

A benefit-and-cost inventory should show each consequence, affected group, valuation method and overlap check.

Showing who gains and who loses

Aggregate NPV can conceal who receives the benefits and who bears the costs. A programme may produce positive total net value while imposing substantial losses on a particular group.

Material differences should be reported by:

• Patient or population group.
• Income.
• Geography.
• Payer.
• Provider.
• Sector.
• Timing.
• Generation.

Distributional analysis complements aggregate social value. Distribution should not be hidden inside an unexplained adjustment to the final result.

Keeping unmonetised effects visible

Some important effects cannot be monetised reliably using the available evidence. Failure to assign a monetary value does not mean that the effect has no value.

For each material unmonetised effect, the analysis should report:

• The affected group.
• Expected direction.
• Likely scale.
• Supporting evidence.
• Reason monetisation was not defensible.
• Relevant uncertainty.
• Whether the effect could change the preferred option.

Unmonetised effects should remain visible to decision-makers rather than being silently assigned a value of zero.

Using switching values to test uncertain effects

A switching value shows how large an uncertain or unmonetised effect would need to be to change the ranking of two alternatives. It can help decision-makers judge whether missing information is likely to matter.

For an omitted effect that falls on only one of two options:

Switching Value = |NPVᴬ − NPVᴮ|

Suppose Option B has an NPV that is £271,602 higher than Option A. An omitted net disadvantage associated with Option B would need a monetary value greater than £271,602 to reverse the monetised ranking, assuming everything else remains unchanged.

A switching value does not prove that an omitted effect has that value. It shows how consequential the effect would need to be to change the result.

How uncertainty affects the conclusion

CBA results depend on estimated quantities, monetary values, discount rates, timing and analytical assumptions. Sensitivity analysis examines whether plausible changes alter NPV, the ranking of alternatives or the interpretation of the result.

Important sources of uncertainty can include:

• Baseline assumptions.
• Resource quantities.
• Unit values.
• Uptake and implementation.
• Market and shadow prices.
• Non-market valuations.
• Mortality and morbidity risk estimates.
• Time horizon.
• Discount rate.
• Treatment of transfers.
• Distributional assumptions.
• Unmonetised effects.
• Model structure.

One-way sensitivity analysis can vary individual inputs. Scenario analysis can examine alternative baselines, perspectives, valuation methods or structures. Probabilistic analysis can explore the combined effect of uncertain parameters when suitable probability distributions are available.

Uncertainty analysis should show which assumptions drive the conclusion rather than merely presenting a different final number.

How CBA relates to budget impact analysis

CBA examines whether monetised social benefits exceed monetised social costs. Budget impact analysis examines how adoption could change expenditure for a specified budget holder over a defined planning period.

Social value and affordability are therefore related but separate questions.

An option can have a positive NPV and still create an unaffordable budget impact. This may occur when:

• Implementation costs are large.
• Expenditure occurs before benefits are realised.
• The eligible population is large.
• Uptake is rapid.
• The budget holder cannot access savings occurring elsewhere.
• Annual budgets are constrained.

Budget impact analysis complements CBA by showing annual cash flows, population size, uptake and financial consequences for the relevant budget holder.

What CBA can and cannot decide

CBA provides structured evidence about monetised social value. It can help compare consequences expressed in different units, but it does not resolve every factual, ethical, financial or implementation issue relevant to a healthcare decision.

• CBA can compare monetised costs and benefits across alternatives and sectors.
• CBA can identify which feasible option produces the greatest estimated monetised net value.
• CBA can show how results change when assumptions or monetary values change.
• CBA cannot prove that every material consequence has been valued correctly.
• CBA cannot assume that an unmonetised effect has no value.
• CBA cannot determine whether gains and losses are distributed fairly.
• CBA cannot establish affordability.
• CBA cannot replace evidence about effectiveness, safety or feasibility.
• CBA cannot make the final adoption decision.

The economic result should inform the wider decision process rather than replace it.

Common mistakes and how to avoid them

A CBA can appear comprehensive because many consequences are expressed in money while still producing a misleading result. Common errors involve the baseline, perspective, valuation methods, transfers, overlapping benefits or mechanical use of a ratio.

Important safeguards include:

• A positive NPV supports efficiency only under the stated assumptions.
• A positive NPV does not establish affordability or require adoption.
• CBA should use incremental consequences relative to a credible baseline.
• Resource quantities and monetary values should remain separately auditable.
• Transfers are not automatically net social costs or benefits.
• NPV and BCR can rank mutually exclusive options differently.
• The highest BCR does not necessarily identify the option with the greatest net value.
• Monetary values may be influenced by income, elicitation methods and institutional conventions.
• Values should not be transferred between populations or jurisdictions without justification.
• Material unmonetised effects must remain visible.
• Aggregate net value does not reveal the distribution of gains and losses.
• Overlapping costs and benefits must not be counted more than once.
• Results depend on the perspective, baseline, horizon, discount rate, valuation methods and evidence.

CBA conclusions should remain tied to the alternatives, baseline, perspective, population, time horizon, price year, monetary values, evidence, assumptions, discount rate, uncertainty and institutional setting used in the analysis.

What should be reported

A complete CBA report should identify:

• The decision problem and intended decision-maker.
• The alternatives and baseline.
• The analytical perspective.
• The affected people, organisations and sectors.
• The time horizon.
• Resource quantities and units.
• Monetary values and valuation methods.
• Price year and currency.
• Inflation or currency-conversion methods.
• Discount rate and timing convention.
• Present-value calculations.
• NPV for each alternative.
• BCR where appropriate.
• Incremental comparisons between mutually exclusive alternatives.
• Treatment of taxes, subsidies and other transfers.
• Distribution of costs and benefits.
• Material unmonetised effects.
• Switching values where useful.
• Sensitivity, scenario and uncertainty analyses.
• Evidence sources and transferability.
• Important assumptions and limitations.
• Affordability, equity, feasibility and wider decision considerations.

Transparent reporting allows readers to examine both the calculations and the analytical choices behind them.

Sources

• Boardman AE, Greenberg DH, Vining AR, Weimer DL. Cost-Benefit Analysis: Concepts and Practice, 5th edition. Cambridge University Press. 2018.
• HM Treasury. The Green Book: UK government guidance on appraisal. 2026 edition. HM Treasury. 2026.
• Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes, 4th edition. Oxford University Press. 2015.
• Kaldor N. Welfare propositions of economics and interpersonal comparisons of utility. The Economic Journal. 1939;49(195):549-552.
• Hicks JR. The foundations of welfare economics. The Economic Journal. 1939;49(196):696-712.
• Husereau D, Drummond M, Augustovski F, de Bekker-Grob E, Briggs AH, Carswell C, et al. Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement: updated reporting guidance for health economic evaluations. BMJ. 2022;376:e067975.