WTA, because workers voluntarily accept risk in exchange for added compensation. However, since risk is just one attribute of the labor market, we expect WTP and WTA to be the same for these market-determined levels of risk compensation. Similar studies on the purchase of consumer products designed to reduce the risk of death (e.g., smoke detectors) are WTP estimates. Following conventional terminology, throughout the remainder of this paper, we refer to value of life studies as WTP estimates. Reviews of these studies can be found in Fisher et al. (1989) and Miller (1990).
It is important to realize that most of these indirect estimates are attempts to measure the value of a small risk of injury or death—not the value of certain injury or death. The loss in utility due to certain injury or death is more than 1,000 times the utility loss due to a 1/1,000 risk of injury or death. Thus, for example, we would expect WTA estimates made by jury awards to be higher than WTA estimates using hedonic pricing estimation. This distinction is most stark when considering the risk of death. Although people often voluntarily accept increased risks of death in exchange for lower prices or higher wages, few individuals would accept any amount of money in exchange for certain death. Thus, the hedonic pricing approach used to estimate WTA and WTP is really only valid for valuing risks—not any one particular injury.
Finally, it should be noted that there is no definitive literature on how these value of life estimates vary by occupation or income (Miller, 1990:18). This is particularly important for purposes of estimating crime costs, because there is some evidence that the average crime victim is not representative of the average consumer or worker in the population. For example, unemployed persons have a three times higher risk of being injured by crime than employed persons, and those with annual incomes of less than $10,000 have more than double the risk of injury due to victimization than those with annual incomes of more than $20,000 (Harlow, 1989).
In addition, if victims are themselves involved in criminal activity (e.g., drug dealers or gang members), these individuals might display preferences toward risk that suggest they have a lower value of life than the average population. Indeed, there is some evidence that victims of assaults are more likely than the average population to have committed assaults themselves (Singer, 1981) and that homicide victims are themselves very likely to have a previous arrest record for crimes against the person (Wolfgang, 1958:178). However, not only do we lack adequate evidence on