What Is a Thing Worth When Nobody Sells It? Value Without a Market

A government deciding whether to build a road, a railway or a flood defence must compare costs it can measure against benefits it cannot, and refusing to assign numbers does not avoid the comparison but merely conceals it. Building the road implies that the benefits exceeded the costs, and not building it implies the opposite, so a decision assigns an implicit value either way. Making that value explicit allows it to be examined, criticised and applied consistently across decisions, which is the argument for doing so. The alternative, in which each decision embeds an unstated and inconsistent valuation, produces outcomes that nobody would defend if the implied numbers were written down.

Several techniques are used and each has characteristic weaknesses:

• •Revealed preference, inferring value from choices people actually make
• •House price differences between quiet and noisy streets, for noise
• •Wage differences between riskier and safer jobs, for risk to life
• •Travel cost, inferring the value of a site from what people spend reaching it
• •Stated preference, asking people directly in structured surveys
• •Replacement cost, pricing what it would take to supply the thing artificially

The most contested of these numbers is the one attached to mortality risk, and understanding what it actually measures removes some of the objection. It is not the value of a particular person's life and not a price anybody would accept for dying. It is derived from what large numbers of people implicitly accept in exchange for small changes in risk, through the wages they take for dangerous work and the money they spend on safety equipment, and it is expressed as the value of preventing one statistical death. Governments publish these figures and use them to decide which safety measures are worth funding, which is uncomfortable to state plainly and produces more consistent protection than deciding case by case.

These valuations are not academic exercises, since several governments require them by rule for any substantial spending decision. Transport departments publish standard figures for the value of travel time saved, broken down by whether the journey is for work or leisure, and those figures dominate the appraisal of nearly every road and rail scheme, since time savings are usually the largest single benefit claimed. Health services use a measure combining length and quality of life to decide which treatments to fund, with an explicit threshold per unit above which a treatment is refused. Environmental agencies apply figures for carbon, for air quality and for flood damage avoided. Publishing the numbers makes the resulting decisions contestable, which is the argument for doing so.

The criticisms are substantial and are taken seriously by practitioners. Stated preference surveys produce answers sensitive to how the question is asked, and respondents give similar amounts for protecting one lake and for protecting fifty. Revealed preference methods assume people choose with full information and no constraints, which is frequently false, and a wage difference may reflect lack of alternatives rather than a valuation of risk. Valuations derived from what people can pay embed existing inequality, so benefits to the rich count for more. Effects on future generations require discounting, and the rate chosen dominates the answer for anything long term. And several things being valued may not be the kind of thing people hold consistent preferences about at all.

Public decisions assign a value to unpriced benefits whether or not anybody writes the number down, so making it explicit allows it to be examined and applied consistently. Values are inferred from house prices, wage differences and travel costs, or asked directly. The figure for mortality measures what people accept for small changes in risk rather than any individual's worth. Willingness to pay embeds existing inequality, which is the sharpest objection.