Intermediatepublic finance
Cost-Benefit Analysis: Putting Numbers on Policy
Overview
Should the dam be built? The regulation imposed? The vaccine program funded? Cost-benefit analysis (CBA) answers by comparing monetized costs and benefits — the applied arm of welfare economics and the standard toolkit of regulatory policy worldwide. But pricing lives, time, and nature is fraught. This lesson covers the CBA mechanics, valuing the non-market (VSL, ecosystem services), discounting's ethics, the Kaldor-Hicks controversy, and how to read a CBA critically.
Learning objectives
- Define cost-benefit analysis and its welfare basis
- Explain valuation: market prices, revealed and stated preference
- Describe the value of a statistical life (VSL)
- Analyze discounting: rates and intergenerational ethics
- Evaluate CBA's critiques and read it critically
The mechanics
CBA's logic: sum all benefits (willingness to pay), sum all costs (opportunity costs), approve if net benefits are positive. Steps: define the policy and counterfactual (vs. what?); enumerate effects; monetize (the hard part); discount future flows to present value; apply decision rule (NPV > 0); run sensitivity analysis. Its welfare basis: Kaldor-Hicks — winners could compensate losers. Governments institutionalize it: the US requires regulatory CBA (OIRA review since Reagan), the EU and UK use it routinely. Done well, it's rationality infrastructure — forcing explicit comparison of consequences. Done badly, it's numerology justifying foregone conclusions.
CBA's most important step is the baseline: compared to what? Biased baselines (rosy no-action scenarios) corrupt everything downstream. Always interrogate the 'do nothing' case.
Valuing the non-market
Many benefits lack prices: lives saved, clean air, quiet, ecosystems. Techniques: revealed preference (wage-risk premiums for dangerous jobs → value of a statistical life; hedonic pricing — house prices revealing clean-air value); stated preference (contingent valuation surveys — 'what would you pay?'; choice experiments); and cost-based proxies (replacement cost for ecosystems). Each has flaws: surveys are hypothetical (hypothetical bias), revealed preferences assume informed rational choice, and some things resist monetization. But 'don't value it' implicitly values it at zero — the honest choice is explicit, uncertain valuation over implicit dismissal.
Refusing to monetize doesn't avoid tradeoffs — it hides them. CBA's critics are right that valuation is fraught; its defenders reply that implicit zero-valuation is worse.
The value of a statistical life
The VSL (~1,000 extra for a 1-in-10,000 annual fatality risk, the implied VSL is $10 million. Misunderstandings abound: it's not 'the price of a life' — it's the marginal rate of substitution between money and small risks, aggregated. It's lower in poor countries (lower willingness/ability to pay — the ethical controversy: does a Bangladeshi life count less? Practically yes in national CBA, no in global ethics). Age-adjustments ('senior discount') and equity weights are live debates. The VSL is where CBA's ethics get rawest.
VSL values small risk changes, not identified lives. We spend ~$10M per statistical life saved via regulation while refusing $10M to save a named person — consistent, once you see it's about risk-money tradeoffs at the margin.
Discounting: the ethics of time
Future costs/benefits get discounted to present value — at what rate? The Ramsey formula: rate = pure time preference + growth × inequality aversion. The US uses ~2–7% (recently lowered toward 2%); the UK ~3.5% declining. For climate (century-scale), the rate decides everything: Stern (~1.4%) demanded radical action; Nordhaus (~4–5%) counseled gradualism — trillions hinge on it. Components debated: pure time preference (should we discount future welfare just for futurity? ethicists say no); growth projections; and whether discounting consumption equals discounting welfare. Declining/hyperbolic schedules try to split the difference. There's no value-free rate — every CBA embeds an ethics of intergenerational justice.
The most expensive parameter disagreement in economics: ~1.4% vs. ~4% discount rates producing radically different climate policies. CBA doesn't resolve it — it exposes exactly where ethics enters.
Critiques and critical reading
Critiques: Kaldor-Hicks doesn't require compensating losers — CBA can bless policies that harm the poor (no equity weights by default); monetization biases toward the measurable (a rich person's willingness to pay counts more); distributional blindness (a dollar is a dollar regardless of recipient); and false precision (point estimates hiding vast uncertainty). Reforms: distributional weights (weighting poor recipients' dollars more), breakeven analysis (how bad would assumptions need to be to flip the verdict?), and qualitative side-constraints (rights, deontology). How to read a CBA: check the counterfactual, the discount rate, the VSL/source of valuations, whether losers are identified, and what sensitivity analysis shows. A CBA you can't interrogate is a press release, not an analysis.
Key points
- CBA compares monetized costs and benefits (NPV > 0); it's applied Kaldor-Hicks and global regulatory standard practice.
- Non-market valuation via revealed/stated preference; refusing to value implicitly prices at zero — worse.
- VSL (~$10–12M US) values marginal risk-money tradeoffs, not lives; its cross-country and age variations are ethically raw.
- Discounting embeds intergenerational ethics — Stern vs. Nordhaus shows trillions hinge on the rate.
- Critiques: no required compensation, distributional blindness, false precision. Read critically: counterfactual, rate, valuations, losers, sensitivity.
Common misconceptions
✓ Truth: VSL measures marginal tradeoffs between money and small risks — not the value of an identified life. It's how much we already spend per unit of risk reduction, made explicit.
✓ Truth: Every CBA embeds ethics: the discount rate, the VSL, whose willingness to pay counts, whether losers matter. Good CBA makes these explicit; bad CBA hides them.
✓ Truth: Proper CBA includes quantified-but-unmonetized effects and breakeven analysis. 'Unquantified' should appear explicitly — invisibility, not imprecision, is the real danger.
Practice questions
Q1. Cost-benefit analysis approves a policy when:
- Benefits exceed costs in physical units
- Net present value of monetized benefits minus costs is positive — the Kaldor-Hicks criterion applied
- Everyone gains
- Costs are zero
CBA's decision rule: NPV > 0, meaning winners could compensate losers. It doesn't require actual compensation or unanimous gains — the source of its power and its ethical controversy.
Q2. The value of a statistical life (VSL) is best understood as:
- The price of a person's life
- The marginal tradeoff between money and small mortality risks, aggregated — e.g., $1,000 for 1-in-10,000 risk implies $10M VSL
- Lifetime earnings
- A fixed universal constant
VSL comes from observed risk-money tradeoffs (wage premiums for dangerous jobs), not from pricing identified lives. It varies with income and context — the ethical debates follow from this.
Q3. In climate CBA, the Stern-Nordhaus disagreement was fundamentally about:
- Whether warming is real
- The discount rate — how much to weight future damages, an ethical choice about intergenerational justice
- Carbon taxes vs. cap-and-trade
- Solar panel costs
Both used CBA; they differed on pure time preference and growth assumptions in the discount rate (~1.4% vs. ~4%). The trillions at stake show CBA exposes ethical choices rather than settling them.
Click a question to reveal the answer and explanation.
Frequently asked questions
What is the difference between CBA and cost-effectiveness analysis?
CBA monetizes everything and asks 'is it worth it?' (NPV > 0). CEA takes the goal as given and asks 'what's the cheapest way?' (cost per life saved). CEA avoids valuing the goal — useful when valuation is ethically fraught.
What is breakeven analysis?
Instead of point-estimating uncertain values, ask: how large would the benefit need to be to justify the cost? If the breakeven threshold is clearly exceeded, decide without false precision.
Should CBA use distributional weights?
Advocates say yes — a dollar to the poor matters more (diminishing marginal utility), and unweighted CBA favors the rich (higher willingness to pay). Practice is moving slowly this way; it's the main equity reform.
What is OIRA?
The US Office of Information and Regulatory Affairs — reviewing major federal regulations' CBAs since 1981. The institutional home of American regulatory cost-benefit analysis, and a political battleground.