Defining the value of a statistical life: a comment

In a recent paper [J. Hlth. Econom. 20 (2001) 131] Per-Olov Johansson claims to demonstrate that a commonly used intuitive definition of the value of a statistical life (VSL) is wrong, and that empirical estimates of VSLs are biased estimators of what he defines as the theoretically correct concept of the value of saving a life. In this comment I first argue in favor of a theoretically correct concept that is different from Johansson's because it incorporates the risk of death into the individual's lifetime budget constraint. I then show that the common intuitive definition of a VSL in fact is consistent with the theoretically correct concept and thus, provides an appropriate empirical basis for estimating the value to society of mortality-reducing projects.