Cost–benefit analysis is based on societal willingness to pay for the health improvements and other consequences of social determinants of health interventions. Like cost–effectiveness analysis, cost–benefit analysis relies on well-developed methods for estimating willingness to pay for health, especially mortality risks. One general approach is to use methods based on revealed preferences. Revealed-preference methods analyse market behaviour to infer willingness to pay for non-market outcomes. For example, analysis of workers’ choices about job safety and wages provide the basis for estimating the marginal value of mortality risks, often summarized as the “value of a statistical life”. More precisely, these studies estimate the dollar value people place on a small reduction in the risk of death. An example is useful to explain the terminology. Suppose a social determinants of health intervention in the housing sector improves safety and reduces the risk of accidental death, say by 1 in 10 000. If each of 10 000 people are willing to pay $600 for that risk reduction, on aggregate the net benefits of the risk reduction are valued at $6 million. Because the intervention can be expected (in a statistical sense) to save one life, this product is then called the statistical value of life. The same approach is also used to estimate willingness to pay to reduce the risks of on-the-job injuries (11).9 Instead of studying revealed preferences for health in labour and other markets, an alternative approach to estimate willingness to pay for health is to use stated preferences. Stated-preference methods use contingent valuation surveys that directly elicit willingness to pay for non-market outcomes, including health. The methodology of contingent valuation surveys has been extensively studied and refined. Much of the research on the contingent valuation method concerns the application of the method to value environmental quality. In an important legitimization of the method, a “blue ribbon” panel of social scientists convened by the United States National Oceanic and Atmospheric Administration concluded that the contingent valuation method could provide useful estimates for the assessment of damages to natural resources (12). Standard references such as Boardman et al. (3, chapter 14) provide in-depth discussions of the contingent valuation method and its strengths and weaknesses. An important advantage of cost–benefit analysis over cost–utility analysis is that it is in principle straightforward to apply to the multiple impacts of social determinants of health interventions (13, 14).10 The potential impacts of those interventions on social welfare include improvements in life expectancy, health-related quality of life, cognitive development, behaviour and social competence, educational attainment and earnings, and reductions in delinquency and crime. Economic methods