Topic 14 of 64· Governmental economic appraisal
Stated Preference Valuation
Stated preference methods estimate the value of a non-market good by directly asking people what they would be willing to pay for it, or willing to accept in compensation to give it up, typically through a structured survey describing a hypothetical scenario. Contingent valuation is the best-known technique in the family.
Why it matters
Green Book Annex 2 (supplementary guidance on valuing non-market impacts) endorses stated preference methods for goods that have no observable market transaction to infer value from at all — air quality, biodiversity, flood protection, the existence value of a landscape someone may never visit (https://www.gov.uk/government/publications/the-green-book-appraisal-and-evaluation-in-central-government). Defra has published its own stated-preference guidance for environmental appraisal specifically because so much of environmental value (habitat preservation, water quality) has no proxy market at all, unlike, say, noise, which at least correlates with observable house prices (see revealed preference valuation).
Stated preference's core appeal — it can value literally anything, including goods nobody has ever transacted in — is also the source of its credibility problem. Because respondents are not actually spending money, contingent valuation surveys are vulnerable to hypothetical bias (people overstate willingness-to-pay when there is no real budget constraint), embedding effects (the same good is valued differently depending on what else is in the survey), and starting-point bias in bidding-game designs. The 1993 NOAA panel on contingent valuation, convened after the Exxon Valdez oil spill litigation, set out design standards — a binary "would you pay £X, yes/no" referendum format rather than open-ended bidding, and mandatory reminders of the respondent's actual budget constraint — that remain the reference standard for defensible surveys.
The maths
Contingent valuation (referendum format):
Present a binary choice: "would you pay £X per year for outcome Y? yes/no"
Vary X randomly across respondents.
Fit willingness-to-pay as a function of the yes/no response rate at each X.
Mean WTP = area under the estimated demand curve
Aggregate value = Mean WTP × affected population
Choice experiment (discrete choice modelling) variant:
Present respondents with repeated choices between bundles of attributes
(including a cost attribute), estimate implicit prices for each
non-cost attribute from the trade-offs respondents reveal.
The choice-experiment variant is generally preferred in current UK practice over single-question contingent valuation because forcing respondents to trade off several attributes against cost repeatedly produces more internally consistent, harder-to-game estimates than a single yes/no question.
Worked example
National government: Defra commissions a contingent valuation survey to value a river water-quality improvement programme. A referendum-format survey of 2,000 households finds 62% would pay £40/year via a hypothetical water bill supplement, and the estimated demand curve gives a mean willingness-to-pay of £28/year per household.
Mean WTP = £28/household/year
Households in catchment = 340,000
Aggregate annual value = £28 × 340,000 = £9.52m/year
Over a 20-year appraisal period at 3.5% discount rate (annuity factor ≈ 14.2):
PV(benefit) ≈ £9.52m × 14.2 ≈ £135m
This aggregate figure is then compared against the programme's social cost-benefit analysis cost side. The Green Book requires this kind of stated-preference evidence to be reported alongside its confidence interval and survey methodology, not as a bare point estimate, precisely because the underlying number is more fragile than a market price.
Charity: a heritage trust surveys visitors and non-visitors about willingness-to-pay to prevent the closure of a historic building neither group necessarily visits (its existence value). Because non-visitors who will never see the building still report positive WTP, the survey captures existence and bequest value that a simple visitor-fee revenue count (a revealed-preference proxy) would completely miss — showing stated preference's genuine advantage where no market transaction of any kind exists to reveal value.
Software engineering connection
Stated preference methods rarely apply directly to software engineering work, but engineers building citizen consultation platforms, budget-participation tools, or public survey infrastructure are often building the instrument the economics depends on. Getting the survey design details right — randomized bid amounts, binary referendum framing over open-ended questions, explicit budget-constraint reminders — is not a UX nicety, it is what makes the resulting valuation defensible under scrutiny; a poorly designed in-app survey can invalidate months of subsequent economic analysis. See citizen satisfaction metrics for the more general discipline of eliciting public opinion data that will bear analytical weight.
Pitfalls
- Open-ended "how much would you pay?" questions. These are far more prone to strategic and anchoring bias than binary referendum framing; the NOAA panel's recommendation to use a referendum format exists precisely because open-ended elicitation performs badly.
- No reminder of the respondent's actual budget constraint. Without it, stated WTP routinely exceeds what the same people would pay when a real budget trade-off is in play — hypothetical bias.
- Embedding effects ignored. The same good valued alone versus valued as part of a larger bundle produces different WTP estimates; report what else, if anything, was in the survey frame.
- Treating a single survey's point estimate as settled. Green Book practice expects a range and a discussion of known biases, not a bare number carried forward into the cost-benefit table as if it were a market price.
Sources
- HM Treasury. "The Green Book," Annex 2: valuing non-market impacts. https://www.gov.uk/government/publications/the-green-book-appraisal-and-evaluation-in-central-government
- Defra. "Valuing environmental impacts: practical guidelines" (contingent valuation and choice experiment guidance). https://www.gov.uk/government/collections/valuing-environmental-impacts
- Arrow K, et al. "Report of the NOAA Panel on Contingent Valuation." Federal Register, 1993.
- Mitchell RC, Carson RT. "Using Surveys to Value Public Goods: The Contingent Valuation Method." Resources for the Future, 1989.