Many of the most important consequences of public projects have no market price: a life saved, an hour of leisure, a clean river, a species preserved, a quieter neighbourhood. To include them in a CBA — rather than ignoring them, which implicitly values them at zero — they must be valued. This module covers how economists put money figures on things that are not bought and sold, and how much to trust those figures.
Two families of method
Revealed vs stated preference
Revealed-preference methods infer the value of a non-market good from actual behaviour in related markets — what people really do reveals what they value. Stated-preference methods ask people directly, through surveys, what they would pay for a (often hypothetical) good. Revealed preference is grounded in real choices (more credible) but works only where behaviour leaves a trace; stated preference can value anything (including pure 'existence' values) but relies on what people say they would do (vulnerable to bias). The appraiser's craft is choosing the right method for the good and knowing how far to trust the result.
Revealed-preference methods
- Travel cost — value a non-priced site (a park, a beach) by what visitors spend in time and money to get there; the demand curve traced by visitors from different distances reveals their willingness to pay for the experience.
- Hedonic pricing — infer the value of an attribute (clean air, quiet, a view, proximity to a road) from how it affects the price of a marketed good that embodies it, usually housing: comparing otherwise-similar houses reveals what the market pays for the attribute.
- Averting/defensive behaviour — value a bad (polluted water, noise) by what people spend to avoid it (bottled water, double glazing).
- Wage-risk (the value of a statistical life) — infer the value people place on mortality risk from the extra wage they require to take riskier jobs.
The value of a statistical life
VSL: valuing mortality risk, not a life
Many projects change mortality risk (safer roads, cleaner air, water treatment). To include this, CBA uses the value of a statistical life (VSL) — NOT the value of a specific person's life, but the aggregate willingness to pay for a small reduction in risk across many people. If 100,000 people each pay $50 for a measure that reduces their annual death risk by 1-in-100,000 (so one statistical death is avoided), they collectively pay $5 million for that one avoided statistical death — the VSL is $5 million. It is estimated mainly from wage-risk studies (the wage premium for risky jobs) and stated preference. The VSL is indispensable (without it, life-saving projects are valued at zero) and ethically fraught — especially the finding that estimated VSL rises with income, which implies a lower VSL in poorer countries and the uncomfortable conclusion that a life-saving project 'scores' lower where people are poorer. Handling this honestly (often by using a common VSL within a country, and being explicit about the value judgement) is part of the appraiser's responsibility.
Stated preference and contingent valuation
When no behaviour reveals the value — especially for 'non-use' values like the existence of a species or an ecosystem no one visits — economists turn to stated preference, chiefly contingent valuation (CV): survey respondents are asked their willingness to pay for a described hypothetical change. CV is powerful (it can value anything) and notoriously vulnerable to biases: hypothetical bias (people overstate WTP when no real money changes hands), embedding/scope insensitivity (people state similar WTP for one lake cleaned and for fifty), starting-point and framing effects, and protest responses. After the Exxon Valdez oil spill made CV legally consequential, a NOAA expert panel (chaired by Arrow and Solow, 1993) issued guidelines to make CV more reliable (incentive-compatible formats, reminders of budget constraints and substitutes), but CV remains the most contested valuation method. Modern practice often prefers discrete-choice experiments (respondents choose among bundles with varying attributes and prices, from which WTP is inferred less directly) as somewhat more robust.
Benefit transfer
Original valuation studies are expensive and slow, so in practice appraisers often use benefit transfer — taking a value estimated in one study (a VSL, a value per hectare of wetland) and applying it to the project at hand, with adjustments for income and context. It is cheap and ubiquitous, and it is risky: a value estimated for one population, ecosystem, or income level may not transfer to another, and uncritical transfer (especially of rich-country values to poor-country contexts, or vice versa, without proper adjustment) is a common source of error. Benefit transfer is a practical necessity but a methodological hazard — use the most similar source study, adjust transparently (especially for income), and flag the uncertainty it introduces.
Exercise
A government must value the benefits of a water-treatment project in a poor rural area that would reduce child mortality from waterborne disease, improve a river used for fishing and recreation, and preserve a downstream wetland with rare birds few people ever visit. (1) Choose an appropriate valuation method for each of the three benefits and justify it. (2) Explain how the VSL would be used for the mortality benefit and the ethical problem if VSL is scaled to local income. (3) Why is contingent valuation both necessary and problematic for the wetland's existence value? (4) The team proposes using a wetland value from a European study via benefit transfer — assess the risks.