The Effect of Risk Context on the Value of a Statistical Life: a Bayesian Meta-model

Abstract

This paper presents the first value of statistical life (VSL) meta-analysis that empirically estimates correction factors for ‘out of context’ benefits transfer (BT) purposes. In the field of mortality risk reductions elicited willingness to pay values in one risk context, say road safety, are frequently applied in other risk contexts like air pollution. However, differences in risk perception and the population at risk across contexts are likely to result in diverging VSL estimates. In a meta-analysis of 26 international stated preference studies, a Bayesian model is estimated regressing contingent values for mortality risk reductions, originating from three different risk contexts, on the characteristics of the risk reduction itself and additional variables characterizing the underlying studies. A willingness to pay (WTP) premium for mortality risk reductions in the air pollution and general mortality risk context relative to improving road safety is observed. Evaluated at the mean, road safety VSL estimates should be multiplied by a factor 1.8 before being applicable in the air pollution context. Moreover, in an illustrative BT exercise we find limited overlap in the set of context specific predictive VSL distributions. Consequently, ‘out of context’ BT results in a substantial over- or underestimation of the VSL.