For nonfatal injuries, the research team estimated values of pain, suffering, fear, and lost quality of life by analyzing jury awards to crime victims and burn victims (Cohen and Miller, 1994a; Miller, Brigham et al., 1993). Although violent offenders rarely have adequate assets and are thus infrequently sued for damages, some victims are able to sue third parties for inadequate security or other forms of negligence. These suits have become common enough that the researchers were able to obtain data from 1,106 jury awards and settlements to assault survivors and 361 to rape survivors (including 39 assault awards and many rape awards involving psychological injury only), as well as 606 to burn survivors (used to compute arson losses). This study ignored jury awards for punitive damages and instead focused solely on that portion of the jury verdict designed to "compensate" the victim for pain, suffering, and lost quality of life. Since cases brought to trial are not necessarily representative of crime cases, the researchers could not apply the pain and suffering estimates directly. Instead, they estimated the functional relationship between the out-of-pocket costs of crime (lost wages and medical expenses); characteristics of the victim (age, sex, work status, etc.); severity of injury (body part, hospitalization, etc.); and the jury's award for pain and suffering. This functional relationship was then applied to the actual distribution of crime victims in the project's data set. In this manner, the researchers were able to estimate what the average jury award for pain and suffering would be for the typical crime in the project's data set.

Analysis of jury awards was based on victims, not victimizations. The data did not permit disentanglement of the effect of series victimizations on pain and suffering. Thus, the lost quality of life estimates for nonfatal injuries are lower than the estimated jury award to a crime victim. For example, the researchers estimated that the 1.1 million rape victims suffer 1.45 million rape victimizations annually. That means annual rape victimizations average 1.27 per victim. Multiplying 1.27 by the $81,400 quality of life loss per rape victimization yields estimated quality of life losses of $103,400 per rape victim.

Serious errors can occur if policy analysts ignore the intangibles when allocating resources. For example, Cohen (1988) describes a study of an early release program in which the authors compared the out-of-pocket costs of crime committed by early release recidivists to the savings to the community that decided not to build more prison space. That earlier study concluded that the program passed a benefit-cost test. However, when the value of pain, suffering, and lost quality of life were added, the early release program failed the benefit-cost test; more prison space was preferable.

Today, benefit-cost analyses typically include the intangible losses. Miller (1993) cites the extensive theoretical literature supporting their use. The U.S. Office of Management and Budget (1989) requires regulatory analysts to use a method that includes intangible losses whenever a benefit-cost analysis values saving human lives.

Intangible costs--lost quality of life--are clearly the largest cost component for crimes of violence. They are also subject to the most uncertainty, a subject that is discussed in a later section. Perhaps more importantly, however, intangible costs are less meaningful when applied to any one particular crime victim. For example, although the lost quality of life for a murder victim is $1.9 million, that does not mean anyone would voluntarily exchange their life for $1.9 million. Instead, that number is arrived at by estimating the incremental amount that individuals are willing to pay for a reduced risk of death, where the commodity is "risk of death," not death itself. Thus, if 100,000 people would collectively pay $30 each to reduce their risk of dying from 1/100,000 to 0, one would say that the group values the "statistical life" that is likely to be saved by $3 million ($30 x 100,000 people).

Victimization and victim costs per crime

Intangible costs are the largest component for all but burglary, larceny, and motor vehicle theft. Intangible pain, suffering, and lost quality of life costs generally exceed all other tangible categories combined. Within the tangible loss category, productivity losses are generally the largest, although medical costs are also substantial. For example, the average rape victim incurs about $500 in medical costs and $2,200 in productivity losses. Drunk driving victims average $1,400 in medical costs, while productivity losses are $2,700. If only those with injury are included, medical costs increase to $6,400, while productivity losses increase to $15,400.

Table 4 summarizes the costs per crime in table 2 and pro�vides two other methods of calculating total costs. The first column repeats the estimates in table 2. The second column excludes the crime of murder and instead allocates each murder to its underlying crime. This is the approach taken