A government can do almost anything; it cannot do everything. Cost–benefit analysis (CBA) is the discipline of deciding whether a particular project or policy is worth its cost, by valuing all its consequences in common units and comparing them. It is the appraisal half of the public-investment problem from the Public Budgeting course — and the toolkit a serious policy analyst is expected to wield.
What CBA does
The core idea and the decision rule
CBA identifies every consequence of a project (benefits and costs, to whomever they accrue), values each in money, discounts future values to the present, and sums them into a net present value: NPV = Σ (Bₜ − Cₜ) / (1 + r)ᵗ The decision rule: undertake the project if NPV > 0 (benefits exceed costs in present-value terms), and among competing projects prefer higher NPV. Equivalently, the benefit-cost ratio (BCR = PV of benefits / PV of costs) should exceed 1. The power of CBA is that it forces every consequence — however different in kind (a life saved, an hour of travel time, a tonne of carbon) — onto a single comparable scale, so unlike things can be weighed against each other and against the cost.
Efficiency, not distribution (yet)
Standard CBA measures economic efficiency: does the project increase total net benefits, regardless of who gets them? It rests on the Kaldor-Hicks criterion (a project is worthwhile if the winners could in principle compensate the losers and still be better off — even if no compensation actually occurs, the subject of module 2). This is CBA's great strength and its great limitation: it gives a clear, aggregable answer about efficiency, but it is silent on distribution — a project with a high NPV might funnel benefits to the rich and costs to the poor. Holding efficiency and distribution apart is deliberate; module 7 brings distribution back in with weights and the MVPF. For now, understand that the basic NPV answers 'is this efficient?', not 'is this fair?'
Framing a CBA correctly
- Standing — whose costs and benefits count? A national appraisal counts effects on citizens; should it count effects on foreigners, or only on a region? Standing decisions are consequential and must be explicit — they can flip a result.
- With-versus-without, not before-versus-after — the correct comparison is the world with the project against the world without it (the counterfactual), not the world after against the world before. A road may carry more traffic after it is built (before-vs-after), but the benefit is the difference between traffic-with-road and traffic-that-would-have-existed-without it — and conflating the two is the single most common appraisal error (the counterfactual logic of the Impact Evaluation course).
- Incremental effects — count only the changes the project causes, avoiding double-counting (e.g., counting both the time saved on a road AND the rise in adjacent land values, when the land value largely capitalises the same time saving).
- The full life — costs and benefits over the whole life of the asset, including maintenance and decommissioning, not just the construction cost.
CBA versus the alternatives
CBA is not the only appraisal tool. Cost-effectiveness analysis (CEA) compares the cost of achieving a fixed objective by different means (cost per life saved, cost per child immunised) without valuing the objective in money — useful when valuing the benefit is impossible or contested (much health appraisal uses cost per DALY averted instead of monetising health). Multi-criteria analysis (MCA) scores options against several weighted criteria when not everything can be monetised — more flexible but less disciplined, and prone to hiding value judgements in the weights. CBA is the most complete and disciplined when benefits can be credibly monetised; CEA and MCA are the fallbacks when they cannot. Knowing which tool fits is itself a skill.
The honest framing
CBA is sometimes attacked as a false precision that reduces life and nature to money. The honest defence: CBA does not claim to be an oracle; it is a structured, transparent argument. Its value is that it forces every assumption — what counts, how it is valued, what discount rate, whose welfare — into the open, where it can be debated and stress-tested, rather than leaving the decision to undisclosed intuition or raw politics. A CBA is only as good as its assumptions, and a good appraiser makes those assumptions visible and tests how much the answer depends on them (the sensitivity analysis of module 6). Used this way, CBA disciplines decisions; misused, it launders a predetermined conclusion — and the rest of this course is about telling the two apart.
Exercise
A county proposes a new bypass road. The promoter argues: 'Traffic on the route has grown 40% in five years, so the road is clearly needed; and property values along the planned route will rise, so we should count that as a benefit on top of the time savings.' (1) Identify the with-vs-without error in the traffic argument. (2) Identify the double-counting error in the benefits argument. (3) Set out the correct framing for this CBA, including standing and the counterfactual. (4) Suppose the benefits are mostly hard to value reliably; what alternative appraisal tool might be more honest, and why?