that making cost estimates is a policy imperative because, in their absence, policy decisions are made on a less informed and arbitrary basis. However, that does not mean that all policy decisions should be based on simple benefit-cost comparisons. Indeed, given the uncertainty in estimating both the benefits and the costs of various policy outcomes, it is incumbent upon the analyst to attempt to account for the uncertainty in these estimates when comparing policy options. Thus, for example, one could construct ranges of costs and benefits based on various plausible assumptions to determine how often the benefit-cost ratio exceeds 1 and to estimate the mean ratio (Hofler and Witte, 1979).
It is interesting to compare the cost estimates developed here to aggregate U.S. expenditures on crime and justice. For example, the total cost of sanctions associated with murder, rape, robbery, and aggravated assault implied by Table 25 is about $12 billion, compared to the $15.75 billion total estimated for the United States in 1986 (Jamieson and Flanagan, 1989:2). This suggests that a good portion of the cost of sanctions in the United States is associated with punishment of violent offenders. This is consistent with the fact that about 55 percent of all prison inmates are serving time for violent offenses.
On the other hand, total criminal justice costs (police and courts) for the crimes estimated here are only about $1.9 billion compared to total U.S. costs of $37.75 billion. There are several reasons why violent offenses account for such a small part of the total police and court costs. First, a good portion of police expenditures are preventive in nature or related to traffic safety and control. (Although police expenditures designed to prevent crimes are not included here, some portion could be allocated to violent offenses—at least in theory.) Second, although violent offenders receive relatively more severe sentences, they are much fewer in number than other criminal offenders. Arrests for violent crimes account for only about 20 percent of FBI crime index arrests. Arrests for violent crimes (including simple assaults) account for only about 12 percent of all arrests nationwide. Thus, violent behavior is a major factor in the cost of corrections, but much less a factor in the cost of the rest of the criminal justice process.
There are several major gaps in the literature on the costs and consequences of violent behavior. In terms of offense types, we have not included kidnappings, bombings, arson, or child abuse. Although few estimates are available for these offenses, the methodology used here could also be applied to them. Cohen (1988a) provides some preliminary estimates of victim costs in the case of