New Evidence on the Monetary Value of Saving a High Risk Youth

Notes

1. At the same time, although most adult offenders have a prior history of juvenile offending, not all juvenile offenders become adult offenders.

2. While we acknowledge that placing dollar values on crimes is not without controversy—especially the valuation of nonpecuniary losses associated with victimization—benefit-cost analysis is now being conducted routinely by policy analysts. For example, the State of Washington has mandated that benefit-cost analyses be conducted for “criminal justice policies, violence prevention programs, and other efforts to decrease the criminal recidivism of juvenile and adult offenders, and certain at-risk behaviors of youth” (Aos et al. 2001). Thus, it is important that existing estimates be refined and improved upon over time.

3. To compare these two figures, we obtained annual Census data for residents under age 18 in Pennsylvania and nationwide. We then divided the number of juvenile arrests by the population to arrive at an arrest rate. In the Philadelphia cohort, there were 2.2 arrests per 100 population under age 18. At the time, this was below the U.S. average of 3.4%. In 2006, these figures have switched, so that the rate in Pennsylvania is 3.8% compared to 2.2% nationally, which is consistent with the recent spikes in crime observed in Pennsylvania generally, and Philadelphia in particular.

4. Cohen (1998) used one offense multiple for adults based on the median reported in Blumstein et al. (1986). While the underlying studies are now quite old, we ultimately use the same sources because they are still among the best data available (see Piquero and Blumstein 2007).

5. Note that some of the juvenile police contacts in the Philadelphia Cohort data do not ultimately result in an arrest. While we have counted incidents as “contacts,” we have not increased these incidents by using an offense multiple.

6. The 966 offenders with 6 or more police contacts represent 3.6% of the 27,160 individuals in the cohort. Of the 14,000 females, only 30 (0.2%) had 6 or more police contacts, while 7.1% (930 out of 13,160) of males were in this category.

7. For simple assaults, Miller et al. (1996) included a category of assault with “no injury,” which was valued at $1,860 in 1993 dollars, and a category titled “NCVS with injury” valued at $24,000. Unfortunately, the definition of “simple assault,” which is how the Philadelphia Cohort data are coded, includes minor injuries. Thus, neither estimate is perfect. According to the most recent NCVS data, 76% of simple assaults result in no injury. If we assume that simple assaults with “minor injury” are about 25% as serious as the average assault with injury, those simple assaults would cost $6,000 ($24,000/4). Based on the ratio of simple assaults with injury versus no injury, this implies an average cost of approximately $2,900—or $4,495 in 2007 dollars (Criminal Victimization in the United States, 2005, Table 91—Percent distribution of victimizations, by type of crime and whether or not reported to the police. Available at: http://www.ojp.usdoj.gov/bjs/pub/pdf/cvus/current/cv0591.pdf).

8. To convert 1993–2007 dollars, we have taken the ratio of the average hourly wage rate in the US from US Census data, resulting in a multiple of 1.55.

9. The Philadelphia Cohort data contain 19 frauds committed by juveniles where victim costs were estimated. The estimated average victim cost for those juvenile frauds was $110, in addition to 74 adult frauds with an average cost of $515. Updated to 2007 dollars, the weighted average cost was $1,100. There were also 724 recorded cases of juvenile vandalism with monetary estimates averaging $74, and an additional 8 adult offenses averaging $79. Updated to 2007 dollars, these offenses average $370.

10. Estimates of time served were taken from BJS (2006a, b, 2007). The average daily cost of prison was estimated to be $62.01 in 2001 (BJS 2004a), or $72.65 in 2007 dollars using the consumer price index.

11. Researchers have recently begun to consider both the economic and non-economic costs on a community or neighborhood’s well-being when an individual is incarcerated. While these costs have not been included here, conceptually, one could consider adding them as well.

12. One exception is Lengyel (2006), who attempts to estimate the social costs of imprisoning a drug offender who is the parent of young children in New York state. Lengyel explicitly accounts for lost quality of life to the incarcerated offender and his/her family, in addition to other costs such as child care. He also attempts to control for the offsetting benefits to some spouses and/or children who are made better off by removing the offender from the family situation.

13. As noted elsewhere in the text, these data only include about half of the crimes committed by 26 year olds. Later, in Table 8, we estimate crimes past that age.

14. This represents about twice the number of crimes as estimated in Cohen (1998). Of course, we include many minor crimes that were not included in Cohen. Limiting our crimes to those covered in Cohen, we estimate about 80–86 offenses (very close to the 68–80 offenses estimated in Cohen).

15. Note that unit cost of crime estimates have increased substantially over time (even after adjusting for inflation) - primarily due to improved estimation methodologies—not due to any underlying change in the severity of crime-related injuries. The earliest estimates by Cohen (circa 1988) were based on jury awards to accident victims—adjusted for the distribution of injuries for crime victims. That method understated the intangible costs since, a broken leg, for example, caused by an accident was given equal value to a broken leg caused by mugging. This was clearly a lower bound estimate as we know that there is added fear and anguish associated with an intentional injury that would not be captured by estimates based on accidental injuries. Miller et al. (1996) overcame this limitation by obtaining jury award data on actual crime victims—which was particularly helpful in valuing rape. It also improved upon the victim survey-based medical cost estimates from BJS that Cohen (1988) used by going directly to hospital data to obtain better estimates of the true cost of intentionally caused injuries (Victims are seldom aware of the full hospital costs or long-term costs of their injuries).

16. The survey data are available from ICPSR at: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/04596.xml.

17. We calculated the number of past month drug users for cocaine, crack, and heroin from age 19 onward, who reported their original use of that drug occurred prior to age 19. By comparing the fraction of age 20 users who began their drug use prior to age 19, for example, to the fraction of age 19 users who also began their drug use prior to age 19, we can estimate a retention percentage by each age group. Using these data, we estimated a quadratic regression equation by age category to obtain year-by-year retention estimates. To arrive at annual estimates for heavy drug abusers, we start at age 14 and continue through age 41 (the last year in which the regression model predicts positive usage for those who begin prior to age 19).

18. This figure is obtained by taking the weighted average of 3.325 million hardcore cocaine users at $9,672 per year; 977,000 heroin users at $10,868; and 350,000 methamphetamine users at $4,680.

19. Since heavy drug use can start at various ages, and since we are providing estimates based on age 14 and 18 throughout this paper, the present value of drug abuse cost estimates are identical for ages 14 and age 18. When we discount from age 10 or birth, we assume that drug abuse begins at age 14.

20. Note that Cohen cited previous research suggesting that there was only a small (if not insignificant) demonstrated relationship between drug use and productivity. However, more recent evidence has refined this by showing that there is a significant reduction in productivity among the chronic (as opposed to the casual) drug abuser (see French et al. 2001; Alexandre and French 2004).

21. ONDCP (2004: IV-3) estimated the cost of hospital and ambulatory care for acute drug-related illness was $1.454 billion in 2002. Additional costs for special diseases associated with drug use were estimated to be $19 million for Tuberculosis, $3.755 billion for HIV/AIDS, and $312 million for Hepatitis B and C. Health insurance administration costs were estimated to be an additional 7.6% on top of these expenses. Combined, medical care costs are estimated to be $5.96 billion, or $7.3 billion inflated to 2007 dollars using the medical care cost component of the CPI. Assuming that 75% of these costs are due to heavy drug abusers, this amounts to $1,500 per person annually (75% × $7.3 billion/3.5 million heavy drug users).

22. The largest difference is due to an estimated increase in drug-attributable crime. While we estimate this to be $700,000, Cohen (1998) estimated a range of $220,000–$606,000. The lower end of that range was based primarily on a one year study of offenders who underwent drug treatment—a selected sample.

23. Bureau of Labor Statistics, “Employer Costs for Employee Compensation—June 2007,” September 20, 2007, http://www.bls.gov/news.release/ecec.t01.htm. Total wages, paid leave and supplemental pay was $22.05 per hour for all workers, while benefits totaled $5.70 per hour, or 25.8%.

24. Policy implications and suggestions for how these estimates might be used are contained in Cohen (1998) and will not be repeated here.