Sixty-seven analyses have estimated the value of a statistical life, generally from estimates of how much people pay for small changes in their survival probabilities. These analyses fall into four classes: (1) Wage-Risk Studies, which analyse compensating wage differentials associated with risky jobs, (2) Market Studies, which analyse the market for products that affect health and safety, (3) Behavioural Studies, which examine risk-avoidance behaviour in inherently risky situations, and (4) Contingent Valuation Surveys, which probe how much people are willing to pay for small changes in risk. Notable reviews of this literature include Smith (1979), Bailey (1980), Blomquist (1981), Violette and Chestnut (1983), Jones-Lee (1984), Miller, Reinert, and Whiting (1984), and Fisher, Chestnut, and Violette (1989). In 1988 dollars, these studies yield values of a statistical life ranging from $0 (e.g., Dorsey, 1983, for non-union workers) to $50,000 (Acton, 1973) to $15 million (Jones-Lee, 1976; Viscusi and Moore, 1989). This wide range hinders credible use of willingness to pay in benefit-cost analysis or courtroom debate over liability damages. This paper narrows the range substantially by weeding out studies with serious flaws and adjusting out some sources of inconsistency. The adjustments include correction for selected systematic biases and reanalysis using uniform values for travel time and the discount rate to convert risk aversion estimates into values. (Author/publisher)
Abstract