The Monetary Cost of Criminal Trajectories for an Ontario Sample of Offenders

List of Tables

Table 1: Per Crime Victim Cost Estimates Used in this Study

Table 2: Per-Diem Correctional Cost Estimates Used in this Study

Table 3: Multipliers Used to Calculate Undetected Crimes

Table 4: Conviction Frequency and Victim Costs by Offence Type and Age Category

Table 5: Correctional Costs Accrued by Study Participants by Age and Disposition Type

Table 6: Victim Costs Associated with Undetected Crime by Offence Type and Age

Table 7: Aggregate Cost of Crime by Trajectory Group and Cost Category

Table 8: Average Convictions and Costs by Trajectory Group

Table 9: The Total Costs of Crime by Type and Age Interval

Table 10: Average Costs by Offender Disposition Profile

Table B1: Shortlisted Studies Reviewed to Generate Per-Crime Victim Cost Estimates

Table B2: Studies Measuring Victim Costs by Cost Type and Stakeholder who Pays the Cost

Table B3: Offence Type Groups used to Calculate Per-Crime Estimates for Reviewed Studies

List of Figures

Figure 1: Estimated Criminal Trajectories for Seven-Group Model

Figure 2: Victim Costs Associated with Convictions by Age and Broad Offence Category

Figure 3: Total Victim, Correctional, and other Criminal Justice System Costs (N=386)

Figure 4: Average Total Victim, Correctional, and other Criminal Justice System Costs by Trajectory Group (N=386)

Figure 5: Total Cumulative Costs of Crime by Cost Type and Age (N=386)

Figure 6: Distribution of Total Costs for the Entire Sample (N=386)

List of Appendices

A: Offence Codes and Categories

B: Victim Cost Methodological Notes

C: Currency Conversion and Inflation Adjustments

D: Victim Costs and Conviction Counts by Five Age Intervals

E: Victim Costs by Age and Offence Frequency

F: Victim Costs by Broad Offence Type

G: Average Annualized Correctional Costs by Type and Age

H: Correctional Costs for Youth, Provincial and Federal Systems by Age

I: Correctional Costs Comparing Custody to Community Supervision by Age

J: Total Costs of Crime by Cost Category and Five Age Intervals

K: Victim Costs Associated with Undetected Crime by Offence Type and Age

L: Cost of Crime by Age and Trajectory Group

M: Cost of Crime Estimates Assuming Discounting of Undetected Crimes

N: The Total Costs of Crime by Type and Age Interval

Executive Summary

Since the mid-1990s, there has been a growing interest in efforts to estimate the monetary costs of crime. This increase is likely the result of at least two factors: First, significant methodological advances have enabled researchers to generate more precise cost figures for the various domains affected by crime, such as victim costs, offender costs, and criminal justice costs. Second, there is increasing recognition that such information could be used to conduct cost-benefit analyses of crime reduction programs, so that policy makers charged with influencing and making difficult decisions about the allocation of scarce resources can do so in an evidence-based manner. What is most exciting about these developments is the steady accumulation of cost esti­mates that have become available to researchers, making it easier and more efficient to derive high quality estimates (within the limits of the methods and underlying data sources) across jurisdictions and with different populations. As a result, there is now a growing body of published and unpublished material from around the world, making it increasingly possible to assess the global economic burden of crime, as well as quantify the potential return on investment that can be achieved through early intervention.

In keeping with these developments, the present report aims to contribute a Canadian perspective to this literature by estimating the longitudinal costs of crime incurred by a sample of 386 high-risk male criminal offenders in Ontario. By doing so, we are replicating three studies of offence trajectories that have recently been conducted in the United States, the United Kingdom, and Australia.  Because our offence data are longitudinal, we were able to track the pattern of costs by age, as the offenders developed across adolescence and into adulthood. This is significant in that it allowed us to assess whether the monetary costs of crime mirror the well-known age-crime curve, whereby involvement in criminal activity rises sharply around 15 to 17 years of age, followed by a steady decline into adulthood. Examining the costs of crime from a longitudinal perspective is useful to potentially highlight the age periods that could be targeted for early intervention to reduce the costs of criminal offending in Canada.

The offence histories of the present sample were tracked for an average of 16.4 years, and offence costs were tabulated for a 15-year period, between the ages of 12 and 26 years. Costs included both tangible and intangible costs, as well as victim-related costs and disposition-related correctional costs. Cost estimates were scaled up to include other criminal justice costs (e.g., police, courts, legal aid), as well as costs associated with undetected crimes. All dollar values were converted to 2013 Canadian dollars. The main research questions addressed in this report were:

1. What are the monetary costs of crime imposed on society by a sample of 386 high-risk offenders in Ontario over the length of the 15-year follow-up period?
2. What are the costs imposed by offenders across various offending trajectories?
3. How do the costs differ for offences committed in adolescence versus adulthood?
4. What are the costs associated with provincial sentences versus federal sentences?

The Sample

The study sample comprised 386 high-risk male offenders who had been sentenced between 1986 and 1997 as juvenile offenders to one of two open custody facilities in Toronto, operating within the scope of the children's mental health system. Offenders were, on average, 17.7 years at the time of admission into the facility, and the average sentence length was 122.6 days.

Offence Data

For the entire sample, criminal offence records were accumulated across a 29-year period, from as early as June 29, 1978, to September 26, 2007, the end of the follow-up period. The average follow-up period was 16.4 years, from late childhood/early adolescence into adulthood. Their average age at the end of follow-up was 32.0 years. For the purposes of this study, the follow-up period was standardized to 15 years to ensure that all participants had a uniform interval of follow-up.

The measure of criminal activity for the study was based on a count (tabulated by age) of all their unique court contacts arising from a new set of charges (Day et al., 2008). Over the follow-up period, the sample amassed a total of 7,257 convictions across 4,657 unique court contacts. These offences spanned 313 discrete offence codes, which were subsequently collapsed into 29 offence categories to facilitate data analysis and reporting.

Trajectory Analysis

One objective of this study was to estimate the costs of crime associated with subgroups of offenders, based on their longitudinal pattern of offending. For these analyses, we used the seven criminal trajectory groups generated from a previous analysis of the study sample (Day et al., 2012). The following heuristic labels were applied (with the percentage of the total sample) based on the rate and shape of their offence trajectories, to identify the seven trajectory groups:

• (T1) moderate late persister (3.6%);
• (T2) high late (3.9%);
• (T3) high early (4.4%);
• (T4) moderate adolescence-peaked (11.7%);
• (T5) moderate early persister (14.2%);
• (T6) low desister (29.8%); and
• (T7) low persister (32.4%).

Cost Data

Cost estimates for the study were obtained for four components pertaining to the societal impact of offending: 1) victim costs; 2) correctional costs; 3) other criminal justice system (CJS) costs, for example, police, court, prosecution, and legal aid expenditures; and 4) costs associated with undetected crimes. In order to obtain cost estimates for each of these four components, information was drawn from a variety of sources, including the published literature (for victim costs, other CJS costs, and costs for undetected crimes), as well as the grey literature and other government publications and sources (for correctional costs). For this study, there were various ways to calculate costs but, where possible, we erred in the direction of using conservative figures, so as not to overestimate the costs of crime.

Analyses

For the purpose of this report, cost estimates are provided for offences occurring over a 15-year period, from ages 12 to 26 years. Because the criminal offence data were longitudinal, the results (for each of the four components and the total, aggregate costs) are presented by age. The 15-year age period from 12 to 26 years was divided into two sets of intervals. First, in the body of the text, the findings for two broad age categories, 12 to17 years (youth) and 18 to 26 years (adult) are presented. Second, as Appendices to the report, the costs are further desegregated into five age categories, 12 to14 years, 15 to17 years, 18 to 20 years, 21 to 23 years, and 24 to 26 years. Additionally, the results are also presented for the seven criminal trajectory groups, in order to differentiate the costs for different subgroups of offenders.

Results

First, considering only offences that resulted in official convictions in court (i.e., excluding undetected crimes), the aggregate cost of offending for the sample was $671 million or $1,739,176 per person over the 15-year follow-up period. Including estimates for undetected crimes contributed an additional $1.6 billion in aggregate costs, bringing the total aggregate, longitudinal cost of offending to $2.26 billion, an average of $5.86 million per person. The aggregate costs for each of the four components were as follows:

• aggregate victim costs - $182,602,699;
• aggregate correctional costs - $122,179,765;
• aggregate other CJS costs - $366,539,294; and
• aggregate undetected crime costs - $1,588,830,864

As indicated by these numbers, including costs for undetected crime substan­tially increased the overall cost estimates. These estimates were derived using the best available scaling-up procedures and monetary per-crime estimates from published empirical studies. However, even when we discounted the costs of undetected crimes to temper these costs, the aggregate undetected crime costs were roughly $800 million. This resulted in a decrease in the overall aggregate costs of crime from $2.26 billion to $1.5 billion over the same timeframe, or  approximately $3.8 million per person.

Second, in addition to calculating aggregate costs of crime for the sample, we also ex­am­ined how costs accumulated over time. Consistent with Piquero et al. (2013), the most costly developmental period was mid- to late adolescence, between the ages of 15 and 17 years, which ac­count­ed for 40% of the total costs. This was a time when the average correctional, other criminal justice system, and victim costs were at their highest, relative to other age intervals. This was also the time when the costs associated with both property and violent crimes were at their highest.

Third, as expected, there was a considerable difference in the aggregate costs (excluding undetected crimes) for the least expensive ($66,509) and most expensive ($9.9 million) offenders in our sample. These differences are further reflected in our estimates of the average cost of crime per individual across seven trajectory groups. In these analyses, we found that high-rate offenders (T2 and T3) cost between $12 and $17 million each; medium-rate individuals (T1, T4, and T5) cost approximately $8 million each; and low-rate offenders (T6 and T7) cost about $3.5 million each. These findings reinforce the idea that even within a high-risk offender sample, there is heterogeneity in the frequency, severity, and timing of offending, and among the dispositions received in response to it.

Conclusions

Based on the results of this study, a number of conclusions may be drawn:

• The exercise of estimating the costs of crime in Canada is important, as the results may inform policy decisions about resource allocation for crime pre­vention and reduction;
• The aggregate cost of crime for this sample was substantial;
• The costs were highest for the developmental period of mid- to late-adolescence, between the ages of 15 to 17 years;
• Across the criminal trajectories, costs were disproportionally higher for the small group of high-rate offenders, and disproportionally lower for the large group of low-rate offenders.

The study has some important strengths that contribute to the validity of the results and advance our knowledge and methods of calculating the costs associated with crime. First, we capitalized on a very detailed criminal history dataset in calculating the number of days served for each of eight separate disposition types. With these numbers, we used the best available per-diem correctional estimates in Ontario, the province in which the sentences were overwhelmingly served. Therefore, we arrived at highly accurate estimates of correctional costs for the study participants. Second, the relatively long follow-up period of 15 years provided a longitudinal look of the costs of crime during the peak adolescent years and into early adulthood, intervals when crime prevention approaches should be applied with vigour if cost savings are to be maximized.

Third, the list of per-crime victim costs drawn from the literature also extends the work of Farrington and Koegl (2014) to include a longer list of crime types. We were also able to develop new estimates based on a careful matching of offence-specific codes with per-crime estimates. Fourth, we are confident that we have calculated cost estimates that reasonably correspond to the degree of victim harm, yet take into account the unique characteristics of the separate youth and adult criminal justice systems in Canada. Lastly, the detailed coding of the data also allowed us to track costs as they occurred for each offender on an annual basis. In fact, the findings reproduce the well-recognized age-crime curve (albeit, with a peak two years later compared to curves based on offending and not adjudication costs per se).

The limitations of the study also need to be acknowledged. First, our results may be biased by cohort effects. The present study mitigates this problem somewhat, since participants were admitted to the referring institution over a ten-year period, between 1986 and 1997. Therefore, they represent offenders of different ages and exposures to the criminal justice system at different times. Notwithstanding this issue, it should be noted that the offenders in this sample were drawn from an era in which youth incarceration was used more frequently than is currently the case. As such, we might expect youth custody costs to be lower with a sample of offenders fol­lowed-up prospectively, given the system's current preference to use extra-judicial, non-custodial measures allowable under the Youth Criminal Justice Act (YCJA).

Second, our results are specific to a particular sample, which may be a higher risk group than one generated by randomly selecting a sample from the general offender population. Two factors contribute to this possibility. First, the current sample had all been sentenced to a custodial facility (i.e., open cus­tody) in their late teens. As such, they had already penetrated the juvenile justice system more deeply than, for example, a youth who had committed a low severity offence (e.g., shop­lifting) at the age of 15 and never offended again. Second, as the open custody facility from which the sample came was operated by a children's mental health centre, with a greater availability of services and access to mental health specialists, the youth sentenced to this facility may have been considered to be more “at-risk,” with more mental health problems and criminogenic needs than the general population of juvenile offenders. This, of course, would limit the generalizability of the study findings to all offenders. This conclusion may be offset, however, by the fact that there was variability in the frequency of offending within the sample. For example, our sample included offenders who committed only one or a few offences in their criminal careers (9.3% had three or less convictions to age 26). Therefore, although they may have been higher risk, they were not all high-rate offenders.

Third, as noted earlier, our estimates of victim costs were based on non-Canadian figures, and therefore, their applicability to the Canadian context could be called into question. Fourth, due to data limitations, we relied on crude calculations to estimate other criminal justice system costs, such as policing, courts, prosecution and legal aid. Unfortunately there are no good data in Canada to be able to apply estimates by type of offence. Therefore, we universally scaled-up correctional costs by a factor of three to create an averaged index. A related limitation is that it was not possible to code the number of disposition days served by offence type. Doing so would have allowed us to calculate criminal justice system costs for each offence.

A final limitation is that we did not explicitly disaggregate costs to the offender from those accrued by victims of crime (e.g., loss of productivity). Indeed, it could be rightfully argued that our costs underestimate the total economic burden on Canadian society. At present, there is no gold standard method for esti­mating the costs of crime, and every study is subject to its own limitations. In other ways, some social costs, such as the loss of social cohesion in a low-income neighbourhood, per­sonal costs of a child growing up with an incarcerated parent, and family costs of grieving the loss of a child to a life of crime are impossible to quantify, making the true cost of crime incalculable.

In summary, we found that the cost of criminal offending incurred by this sample of 386 high-risk male offenders in Ontario was substantial. As expected, we also found considerable differences in all the component costs as the sample aged from adolescence to adulthood, and as we parsed the sample into different trajectory groups based on their rate of offending over time. Although the issues are complex and potentially politically charged, based on the results of this study, we would argue that the best option to reduce the costs of crime, that is, to get the “biggest bang for the buck,” is through strategic prevention and early intervention programming. We know from studies that have calculated cost-benefit ratios, that early intervention and prevention programs have the poten­tial to yield substantial cost savings with a high degree of confidence on the return on investment. For example, in a recent analysis of the Stop Now and Plan (SNAP^®) program, a Canadian-based early intervention program, Farrington and Koegl (2014) found that for every $1.00 invested in the program, between $3.07 and $5.64 was saved in costs of crime expenditures over a nine-year follow-up period. Accordingly, we believe that the lion's share of investment should be made in early intervention and prevention programs, taking into account the mandate of the criminal justice system to provide public safety and treatment to offenders.

1.0 Literature Review

1.1 Why is it Important to Estimate the Cost of Offending?

A safer society is a worthy investment. But how much of an investment is required to bring about a safer society? How should the investment be distributed to yield the biggest impact? How should funds be allocated to effectively reach certain “high-risk” groups, both within the criminal justice system (e.g., through high-intensity rehabilitative services for high-rate, persistent offenders), and outside the criminal justice system (e.g., through strategic early intervention programs for pre-delinquent children with multiple risk factors), as well as more moderate- and low-risk individuals within and outside the justice system? These are important questions for the development of an effective and efficient national crime reduction and prevention strategy. Decisions about the allocation of resources should be informed by the best available evidence about the costs of crime, and by the potential cost savings to be gained by different crime reduction programs.

Since the mid-1990s, there has been an increased interest in the economics of crime. There is a growing recognition that such information may have practical utility for both researchers interested in evaluating the cost-benefit ratios of crime reduction programs, and policy makers charged with influencing and making difficult decisions about the allocation of scarce resources. Putting a dollar figure on the costs of crime provides an easily understood metric to compare the financial burden of different types of crimes. It also provides a standardized means to convey information that could aid decisions about allocating scarce resources among a set of crime prevention and crime control programs and policies (Miller, Cohen, & Rossman, 1993). In these regards, crime cost estimates may be used for a number of policy-relevant purposes. Cohen (2005, p. 4) lists three: “1) comparison of the relative harm caused by type of crime; 2) comparison of the aggregate harm from crime with other social ills; and 3) benefit-cost analysis of alternative crime control policies.”

With respect to the latter, and in reference to crime prevention efforts specifically, rather than crime control strategies such as community policing (Welsh & Farrington, 2012), there is a growing recognition that early intervention and prevention programs may be effective to prevent or forestall the onset of antisocial and delinquent behaviour in at-risk children and youth (Far­rington & Welsh, 2006; Welsh & Farrington, 2000). The argument in favour of pro­active crime prevention strategies versus more reactive offender rehabilitation approaches has been bolstered by the growing literature on both the staggering costs imposed by crime on so­ciety (Anderson, 1990, 2012; Zhang, 2011), and the cost savings gained by reducing the number of par­tici­pants involved in criminal activity through prevention (Aos, Phipps, Barnoski, & Lieb, 2001).

The prioritization of crime-related programs, for both prevention and rehabilitation, is of interest to criminal justice policy makers, given that a small number of offenders account for a disproportionate amount of crime (Cohen, Piquero, & Jennings, 2010a). For example, Wolfgang, Figlio, and Sellin (1972) reported that 6% of their Philadelphia birth cohort accounted for 50% of the criminal acts of the cohort to age 17. More recently, Welsh, Loeber, Stevens, Stouthamer-Loeber, Cohen, and Farrington (2008) reported that 10% of their sample from the Pittsburgh Youth Study (PYS) accounted for 50% of all self-reported offences. These high-rate and chronic offenders begin their criminal activity at an early age and persist into adulthood, commit both serious and violent crimes, and pose a considerable challenge to the criminal justice system (Piquero, Farrington, & Blumstein, 2003). Using data from the Second Philadelphia Birth Cohort (Tracy & Kempf-Leonard, 1996; Tracy, Wolfgang, & Figlio, 1990), Cohen et al. (2010a) estimated that the financial burden of the small group of high-rate chronic offenders identified in their study (3% of the sample) was nearly half the total cost of offending for the entire sample. There is also mounting evidence to suggest that based on analyses of criminal trajectories, some offender groups impose greater economic costs than others. This may mean that the allocation of resources to certain subsets of offenders may lead to marked reductions not only in recidivism, but also in costs (Cohen et al., 2010a). Implementing sound targeted prevention and early intervention programs makes good economic sense and holds the greatest promise for crime reduction (Tolan & Gorman-Smith, 2002).

Indeed, studies have found that even a modest decrease in crime through effective crime reduction efforts can yield considerable savings (Aos et al., 2001; McCollister, French, & Fang, 2010). Miller, Levy, Cohen, and Cox (2006) reported that a 10% reduction in alcohol- and drug- involved crime could save USD$4.25 billion. McCollister, French, and Fang (2010) found that compared to a similarly-matched sample of parolees who did not attend a program, a residential aftercare program for substance-abusing parolees netted a cost savings of $185,250, based on just twelve fewer incidents of vandalisms and three fewer robberies for an average parolee. As another example, for prevention programs aimed at high-risk children, for every dollar spent by the High/Scope Perry Preschool Program, an early childhood education program in Ypsilanti, Michigan, the program repaid USD$16.14, which included the cost savings for criminal activity (Nores, Belfield, Barnett, & Schweinhart, 2005; Schweinhart 2007). In other words, participation in this program yielded participants higher lifetime earnings, a lower burden on welfare support, and lower rate of criminal activity, compared to participants in a control group. In a recent analysis of the Stop Now and Plan (SNAP^®) program, a Canadian-based early intervention program, Farrington and Koegl (2014) found that for every $1.00 invested in the program, between $3.07 and $5.64 was saved in costs of crime expenditures over a nine-year follow-up period.

Emerging evidence also suggests that the greatest gains from crime prevention efforts come from targeting those individuals with the highest risk factors (Augimeri, Jiang, Koegl, & Carey, 2006; Dodge & McCourt, 2010; Welsh & Farrington, 2007). This was the conclusion reached by Foster, Jones, and the Conduct Problems Prevention Research Group (2006) when they evaluated the cost-effectiveness of the Fast Track intervention, a high intensity, multi-component prevention program for at-risk children in first through tenth grade. Using an index of the incremental costs of program alternatives relative to the incremental difference in outcomes (i.e., in average effect sizes), referred to as the incremental cost-effectiveness ratio (ICER), they determined that, for the group at lowest risk at program intake, based on measures of conduct problems at home and school, the program was neither cost-effective nor likely to be effective. By contrast, for the group at highest risk at intake, the program was both cost-effective and had a high likelihood of being effective.

In addition to the focus on crime prevention, the burgeoning interest in the economics of crime over the past 15 years can also be attributed to significant methodological advances that have enabled researchers to generate more precise cost figures for the various domains and components affected by crime. These include victim costs, offender costs, and criminal justice costs, as well as both tangible (e.g., those associated with law enforcement, the youth and adult criminal justice systems, and loss of productivity and wages) and intangible costs (e.g., those associated with loss of life, trauma, pain, and the suffering of victims). These advances include developments in the “bottom-up” and “top-down” methodologies, the two main approaches for determining costs estimates (Cohen, 2005; Cohen & Bowles, 2011).

The bottom-up approach derives total costs of crime by tallying up the component costs, including those incurred by the criminal justice system (e.g., for police, public defenders, courts, corrections), the victim (e.g., property loss and damage, medical and victim support services costs, financial loss), as well as criminal career costs, such as the loss of productivity of offenders resulting from incarceration (Cohen 1998; McCollister, French, & Fang, 2010; McIntosh & Li, 2012). The top-down approach uses the “willingness-to-pay” (WTP) method, which estimates the cost of crime based on the “contingent valuation” (CV) methodology employed in other fields such as environmental economics. CV is used “to place dollar values on nonmarket goods such as improvements in air quality, saving endangered species, and reducing the risk of early death” (Cohen, Rust, Steen, & Tidd, 2004, p. 91). Note that the bottom-up approach is an estimate of costs incurred after a crime has occurred (ex post), whereas the top-down approach is an estimate of costs in anticipation of crime, before a crime has occurred (ex ante), such as a person's actions in response to a fear of crime.

What is most exciting about the recent developments in the cost of crime research literature is the steady accumulation of detailed cost estimates that have become available to researchers, making it easier to derive high quality estimates (within the limits of the methods and the data sources underlying them) across juris­dic­tions and with different populations. As a result, there is now a growing body of published and unpublished material from around the world, making the endeavour to assess the economic burden of crime on society seem like a global venture. In keeping with these developments, the present study aims to contribute a Canadian presence to this literature by estimating the cumulative costs of crime incurred by a sample of high-risk male criminal offenders in Ontario, based on offence trajectories. The next section reviews findings from the literature in three areas: 1) aggregate costs of crime; 2) costs of high-risk offenders; and, most relevant to the present study, 3) costs based on criminal trajectories.

1.2 Results of Previous Studies

1.21 Aggregate Costs of Crime

Estimates of the costs of crime in North America, including both tangible and intangible costs, have been as high as USD$1.7 trillion per year when considering the aggregate or collective costs on society (Anderson, 2012; see also Leung, 2004). Zhang (2011) reported that the annual social and economic cost of crime in Canada in 2008 was estimated to be CAD$31.4 billion. Remarkably, this figure is an underrepresentation of the total costs associated with crime once intangible costs are factored into the equation; Inclusion of intangible costs (CAD$68.2 billion) brought the total cost of crime closer to CAD$100 billion (Zhang, 2011). Given the substantial economic burden crime imposes on society, in addition to the personal and social costs, the rigorous measurement of costs associated with crime is a valuable and worthwhile endeavour for a number of reasons, including the development of evidence-based crime prevention policies.

Miller et al. (2006) estimated the costs associated with alcohol and drug-involved crimes and determined that, when factoring in victimization costs of pain, suffering, and diminished quality of life, alongside the costs of justice services, public services (e.g., police services), and medical and mental health care expenses, the total financial burden on society was USD$205 billion. Using the WTP approach, Cohen, Rust, Steen, and Tidd (2004) conducted a large-scale survey of 1,300 residents in the United States to determine the public's willingness to pay for crime reduction programs for specific types of offences, for example, to reduce the commission of a particular crime by 10%. Using these estimates and aggregating across five offence types (burglary, armed robbery, serious assaults, rape and sexual assaults, and murder), Cohen et al. estimated the cost of crime to be USD$625 billion. This figure is considerably higher than previous estimates (e.g., Miller, Cohen, & Wiersema, 1996), but is thought to be more representative of the social costs of crime because it takes into consideration such factors as the public's fear of crime.

Lastly, Welsh, Loeber, Stevens, Stouthamer-Loeber, Cohen, and Farrington (2008) used self-report offence data from 500 youth in the Pittsburgh Youth Study (PYS) to estimate the victim-related costs (e.g., financial loss to replace stolen items, and pain and suffering) associated with juvenile crime. They determined that the aggregate costs for the crimes committed by their sample between the ages of 7 and 17 years ranged from USD$89 million to USD$110 million. The costs for violent crime were estimated to be between USD$82 million and USD$103 million and the costs for property crime were estimated to be USD$6.5 million. Welsh et al. also estimated the costs associated with the chronic offenders in the PYS sample. This small group of offenders (10% of the sample), who accounted for 50% of the self-reported crimes, imposed an average victim cost per offender of between USD$793,000 and USD$861,000, approximately five to eight times the average victim cost imposed by all the other offenders (USD$101,000 to USD$147,000).

1.22 Costs of High-Risk Offenders

Taking a different approach, Cohen (1998) calculated the total costs incurred by a typical criminal career across adolescence to adulthood. Based on the work of Blumstein, Roth, Cohen, and Visher (1986), Cohen estimated the offence characteristics (e.g., offence frequency, offence types, criminal career length) of a typical “chronic juvenile offender,” and determined that the marginal costs imposed by a typical high-risk youth was between USD$1.3 and USD$1.5 million (in 1997 dollars). This included costs for prosecution, incarceration, attempts at treatment and rehabilita­tion, loss of productivity for victims and offenders, as well as victims' costs of pain, suffering, and diminished quality of life.

Nearly a decade later, Cohen and Piquero (2009) extended this work in several important ways: First, to capture real rather than “typical” criminal careers, they used offence data from the Second Philadelphia Birth Cohort Study (Tracy & Kempf-Leonard, 1996; Tracy et al, 1990), a longitudinal investigation of 27,160 male and female participants born in Philadelphia in 1958. Second, they included a larger number of different offence types in their calculation, up to 14 from 8. Third, they provided results using both bottom-up and top-down estimates. Armed with these data, Cohen and Piquero (2009) determined that the lifetime cost (to age 32) of a single career criminal was between USD$2.6 and USD$4.4 million (in 2007 dollars).

Likewise, Koegl (2011) found that some offender groups imposed more costs on society than others. In particular, persistent and prolific offenders (i.e., who were in the top 10% of all offenders in terms of offending frequency and who had at least one conviction during each of three age intervals of 12 to14 years, 15 to 17 years, and 18 to 20 years) accumulated CAD$6,325,085 per person in costs to society. This was in comparison to the low- and moderate-risk offenders who cost much less over the same timeframe, averaging CAD$609,042 and CAD$1,294,888 per person, respectively, be­tween the ages of 12 and 21. Taken together, the findings from these studies suggest that preventative interventions targeting juvenile offenders may be valuable and can lead to high payoffs if effective.

1.23 Costs Based on Criminal Trajectories

Some researchers have suggested that the distribution of resources should be based on an individual's offending trajectory, given that some trajectory groups impose greater costs on society than others (Cohen et al., 2010a). As a result, decisions about targeted preventative initiatives may not be as straightforward as allocating the bulk of resources to “high-risk” young offenders, more generally. Rather, resources may be allocated, for example, to certain groups identified from a trajec­tory analysis as “persistent” and/or “high-rate” offenders.

To our knowledge, only three major studies have been conducted on the monetary costs of crimes associated with offending trajectories. These studies were conducted with data from the United States (Cohen, Piquero, & Jennings, 2012a, 2012b), England (Piquero, Jennings, & Farrington, 2013), and Australia (Allard, Stewart, Smith, Dennison, Chrzanowski, & Thompson, 2014). Trajectory analysis uses longitudinal data to cluster individuals with similar develop­mental growth curves into distinct classes or groups based on their longitudinal rate of offending. Because the trajectories are not directly observed or measured by the researcher, they are referred to as “latent” classes. A criminal trajectory represents the course or progression in the rate of criminal behaviour over age or time at the level of the individual. The objective of studying criminal trajectories is to understand the longitudinal course of offending at the level of the individual, and the factors that influence each trajectory's progression.

In addition, trajectory analysis is based on the premise that offenders comprise a hetero­geneous population and studies typically identify between two and eight trajectory groups. These groups include some combination of high-rate persisters, moderate-rate offenders, low-rate desisters, low-rate persisters, adolescent-peaked offenders, and adult-peaked offenders. Trajectory groups may be differentiated by the shape of the growth curve, the age at which the trajectory peaks, its maximum value, and the ages at which the trajectory starts and stops (based on the available data). As well, researchers may be interested in identifying characteristics that differentiate the groups, such as crime mix (Blokland, Nagin, & Nieuwbeerta, 2005), personality traits or characteristics like psychopathy (McCuish, Corrado, Lussier, & Hart, 2014), and developmental risk factors (Day et al., 2012). To this list we could add the costs imposed on society as a result of their criminal activity.

In the first study to estimate costs associated with trajectory groups, Cohen et al. (2010) extended the findings of the Cohen and Piquero (2009) study by using a 25% randomly selected subsample of 6,750 members of the Second Cohort of the Philadelphia Study (Tracy & Kempf-Leonard, 1996; Tracy et al., 1990). The cohort had been followed from ages 10 to 18 years, tracking offence data in the form of official police contacts during adolescence and police arrests from ages 18 to 26 years. The researchers used both bottom-up and top-down (i.e., WTP) methods to estimate the costs of the police contacts/arrests during the follow-up period. Based on the figures from Cohen and Piquero (2009), cost estimates (in U.S. dollars) were generated for 14 different crimes (for the bottom-up and top down methods, respectively), including murder ($5 million and $11.8 million), rape ($150,000 and $290,000), simple assault ($11,000 and $19,000), burglary ($5,000 and $35,000), arson ($60,000 and $115,000), fraud ($3,500 and $5,5,00), and vandalism ($1,000 and $2,000).

The results of their trajectory analyses yielded four trajectory groups, identified as nonoffenders comprising 76.3% of the sample; low-rate chronics comprising 18.6% of the sample; adolescence-peaked comprising 2.0% of the sample; and high-rate chronics comprising 3.1% of the sample. The aggregate costs of the offending trajectories of the subsample were USD$265 million and USD$529 million, based on the bottom-up and top-down methods, respectively. For the high-rate chronic offence trajectory group, the average total costs were USD$108 million and USD$226 million, based on the bottom-up and top-down methods, respectively, and this group accounted for about 42% of the total costs incurred by the sample. Unlike the other three groups, the costs for the high-rate chronic group were higher in adulthood than adolescence. Lastly, the cost for the most frequent offender (53 crimes across the offence trajectory, mostly committed as an adolescent) was USD$1,696,000 and the cost for the most expensive individual (9 crimes across the offence trajectory, mostly committed as an adult) was an astonishing USD$35,406,000. Both these individuals were in the high-rate chronic trajectory group. What is noteworthy about these results is that the most expensive offender was not the most frequent offender. What this suggests is that the cost of crime is a function of both the frequency and severity or seriousness of the offending, as well as the timing of the offending in the life course (i.e., adolescence or adulthood). Indeed, Cohen et al. (2010) found that the frequency of offending was driving the costs in adolescence, but the seriousness of offending was driving the costs in adulthood.

Cohen, Piquero, and Jennings (2010b) took these analyses one step further and examined the costs of crime for offence trajectories, from ages 8 to 26, disaggregated across gender and ethnicity. Briefly, using the WTP method, their results indicated that male high-rate chronic offenders imposed the greatest costs (USD$1,528,759 on average) and that chronic female offenders, although rare, also imposed considerable costs (USD$754,440 on average). African-Americans in the chronic trajectory group imposed the greatest costs of all racial/ethnic groups (USD$1,641,700 on average), compared to Hispanics (USD$209,212 on average) and Whites (USD$111,881 on average).

In the second trajectory study, Piquero, Jennings, and Farrington (2013) calculated the costs associated with the 411 individuals in the Cambridge Study in Delinquent Development (CSDD) sample. The CSDD is a long-standing, prospective longitudinal study of youth born in 1953 in London, England. They have been followed through interviews and archival data from ages 10 to 50 years, with their life events and criminal activity being documented. The measure of criminal activity for this study was official convictions. Cost estimates for a range of offences (e.g., theft, sex offence, assault, robbery, vandalism) were generated from analyses by Brand and Price (2000), and updated to 2003 dollars by Dubourg and Hamed (2005). Costs included those associated with the anticipation of crime, as a consequence of crime, and in response to crime.

Piquero et al.'s (2013) trajectory analyses yielded five groups, identified as nonoffenders comprising 62.3% of the sample; low adolescence-peaked comprising 18.6% of the sample; very low-rate chronics comprising 11.3% of the sample; high adolescence-peaked comprising 5.4% of the sample; and high-rate chronics comprising 2.5% of the sample. The aggregate cost (expressed in 2003 U.S. dollars) of the offending trajectories of the subsample was $3,254,935. With respect to the costs for each of the trajectory groups, the high-rate chronics amassed an average of USD$95,241 per offender in costs, compared to USD$40,933, USD$21,576, and USD$10,210 for the high-adolescence-peaked, very low-rate, and low-adolescence-peaked groups, respectively. Lastly, the cost for the most expensive individual in the sample was $153,318. These figures are considerably lower than the Cohen et al. (2010) study, which was attributed by Piquero et al. (2013) to the fact that the Dubourg and Hamed (2005) crime cost estimates were conservative and did not incorporate WTP costs, which tend to yield higher estimates than bottom-up approaches.

In the third trajectory study, Allard et al (2014) examined the longitudinal costs for a sample of 41,377 individuals from Queensland, Australia, whose offence trajectories were followed from ages 10 to 25 years. The offence data was based on six types of “finalised criminal justice events” (p. 5), such as formal cautions, police-referred youth justice conferences, and court appearances. Cost estimates were generated for 15 types of offences (e.g., homicide, assault, robbery, theft, fraud, arson) from Rollings (2008), and updated to 2012 dollars. The cost estimates were based on the bottom-up calculation method, and included criminal justice costs as well as broader social and economic costs (e.g., medical costs and costs of property loss).

Their trajectory analyses yielded five groups, identified as adolescent peaking-low rate comprising 30.6% of the sample; adult onset-low rate comprising 53.3% of the sample; adolescent onset–chronic comprising 1.8% of the sample; adolescent onset – moderate rate comprising 11.3% of the sample; and early onset–chronic comprising 3.0% of the sample. The results indicated that the aggregate costs of the offending trajectories (expressed in 2012 Australian dollars) were $1.14 billion. With respect to the average costs of offenders in each trajectory group, the early onset-chronic group cost an average of AUD$262,799 per offender, which accounted for 28.7% of the total cost incurred by the sample. This is compared with the average costs of AUD$8,559, AUD$10,740, and AUD$69,611, for the adolescent peaking-low rate, adult onset-low rate, and adolescent onset–moderate rate groups, respectively. The adolescent onset–chronic group was the second most expensive at AUD$186,366 per individual, on average.

A fourth study, by Craig, Schumann, Petrunka, Khan, and Peters (2011), is discussed here because it stands out as the only other available Canadian study. However, their investigation only estimated the costs associated with the delinquent behaviour of children and youth over a very limited six-year period, from ages 8 to 14 years. As a result of the short time period during which to track criminal behaviour (i.e., between ages 12 and 14 years), this study was only able to estimate the costs of crime over a limited amount of time. Nonetheless, it is included here because it reported costs for some criminal activity based on a trajectory analysis.

The sample of 842 included young people involved in the Better Beginnings Better Futures (BBBF) program, a universally available prevention program in high-risk communities in Ontario, and a control group from the same communities not involved in the program. A single delinquency scale was created by principal components analysis (PCA) from items completed by parents, teachers, and the youth at three time periods, Grades 3 (age 8), 6 (age 11), and 9 (age 14). The items covered a range of beha­viours, including: vandalism, stealing, destruction of property, lying and cheating, and gang member­ship. The cost data included those associated with utilizing government resources pertaining to: social services (e.g., family involvement with child welfare); social assistance (e.g., Ontario Disability Support Program payments); health care (e.g., hospital stays, visits to a physician and nurse); remedial education (grade repetition, use of specialized services); and the criminal justice system (arrests, court appearances).

The trajectory analyses yielded six delinquent trajectories as follows: lowest (rate) delinquency comprising 6.7% and 10.6% of the boys and girls in the sample, respectively; second lowest (rate) delinquency comprising 70.4% and 76.5% of the boys and girls in the sample, respectively; moderate (rate) desisters comprising 13.5% and 8.1% of the boys and girls in the sample, respectively; highest (rate) desisters comprising 3.4% and 1.0% of the boys and girls in the sample, respectively; escalators comprising 4.0% and 2.8% of the boys and girls in the sample, respectively; and high (rate) delinquency comprising 2.0% and 1.0% of the boys and girls in the sample, respectively. Interestingly, the aggregate costs of the offending trajectories (in 2001 Canadian dollars) were slightly higher for girls than boys at $244,056 and $229,236, respectively. The highest costs were associated with remedial education (64% of the costs) and lowest for those associated with the criminal justice system (1% of the costs). The latter finding is not surprising given the age of the sample. Lastly, 80% of the costs were accounted for by 18% of the sample (on the moderate desister, highest desister, escalators, and high delinquency trajectories) and 80% of the criminal justice costs were accounted for by the individuals on two trajectories: high delin­quency and escalators.

A number of conclusions may be drawn from these studies. First, there is considerable variability in the cost estimates across the studies. Some of the variability may be attributed to the different calculation approaches used (e.g., top-down versus bottom-up), various methodological issues such as the measure of criminal activity (e.g., police contacts, convictions), the length of the follow-up (particularly for persistent offenders), the number and type of crimes included in the analyses, the number of different components that were entered into the bottom-up calculations (e.g., criminal justice costs, victim costs), and jurisdictional differences in terms of how costs for each of the components are tabulated.

Second, all three studies involving criminal offenders identified a small group of high-rate offenders who, although comprising a small proportion of the sample, were respon­si­ble for a disproportionate amount of both the crimes committed and costs incurred. For example, the two chronic groups identified in the Allard et al. (2014) study made up only 4.8% of the sample, but accounted for 41.1% of the total costs. Similarly, Piquero et al. (2013) found that the average cost of the high-rate chronic group, which had only 8 members, was more than double the average cost of the high adolescence-peaked group, which had 21 members. In light of the finding reported earlier about the impact of the Fast Track early intervention program on the highest risk children, such programs have the potential to save a considerable amount of money. The accum­ulating evidence is clear in pointing to the high costs incurred by the small group of persis­tent offenders. and these types of studies are useful for policy makers to consider the most effec­tive and cost-beneficial responses to crime. However, there is a need for such a study of offenders in Canada to inform domestic policies and practices.

1.3 The Present Study

The objective of the present study was to estimate the monetary cost of the crimes committed by a sample of 386 high-risk male offenders in Ontario, based on trajectory groups. The cost estimates generated in this report provide a valuable comparison with the monetary costs associated with crimes by the offence trajectories generated for samples in other countries, including the second Philadelphia Birth Cohort in the U.S. (Cohen et al., 2010a, 2010b), the CSDD in the U.K. (Piquero et al., 2011), and for an Australian sample of offenders (Allard et al., 2014). Calculating the costs of crime in Canada is important from a policy perspective, both for understanding the monetary impact of crime on society and, more importantly, for estimating the cost-savings of crime prevention strategies. Decisions about the allocation of resources to support criminal justice programs and policies should be informed by the best available data. For example, drawing on the example by Raffan-Gower and Farrington (in press), if two programs were being considered, one designed to reduce car thefts by 25% and one designed to reduce break and enters by 15%, but only one can be funded, which should the government choose to fund? Decisions about the potential savings to be gained, relative to a program's costs, are informed by research on the monetary cost for specific types of crimes. Moreover, knowledge of the differential costs imposed by different subgroups of offender, such as based on trajectory analysis, also may be factored into policy decisions. As McIntosh and Li (2012) stated, “In order to arrive at a balanced, impartial, and equitable decision, [decision-makers] must identify policies and practices that are not only outcome effective but also economically efficient” (p. 4).

Lastly, because our offence data are longitudinal, we were able to track the pattern of monetary costs by age, as the offenders developed across adolescence and into adulthood. This is significant in that it allowed us to assess whether the costs of crime mirror the well-known age-crime curve, whereby involvement in criminal activity rises sharply to around age 15 to 17 years, followed by a steady decline into adulthood. Examining the costs of crime from a longitudinal perspective is useful to potentially highlight the age periods that could be targeted for early intervention efforts in order to reduce the costs of offending in Canada.

The main research questions addressed by this report were:

1. What are the monetary costs of crime imposed on society by a sample of offenders in Ontario for a 15-year follow-up period?
2. What are the costs imposed by offenders across seven trajectory groups derived for this sample, as reported in Day et al. (2012)?
3. How do the costs differ for offences committed in adolescence versus adulthood?
4. What are the costs associated with provincial sentences versus federal sentences?

2.0 Method

2.1 Overview

2.11 Sample Characteristics

The sample used in this study comprised 386 male offenders in Ontario whose offence trajectories were followed for an average of 16.4 years (range: 9.8 to 28.7 years). These 386 offenders represent a 50% randomly selected sample of the population of juvenile offenders who had been sentenced between January 1986 and December 1997 to one of two open custody facilities in Toronto. These facilities, operated by a children's mental health centre, were intended for youth aged 16 to 17 years. Therefore, our results are specific to a particular sample. Moreover, the study sample may be higher risk than one generated by randomly selecting a sample from the general offender population. For example, the offenders had all been sentenced to a custodial facility (i.e., open cus­tody) in their late teens, and so had already penetrated the juvenile justice system more deeply than, for example, a youth who had committed a low severity offence (e.g., shop­lifting) at the age of 15 and never offended again. Second, the open custody facility from which the sample came was operated by a children's mental health centre, with a greater availability of services and access to mental health specialists. As a result, youth sentenced to this facility may have been considered to be more “at-risk,” with more mental health problems and criminogenic needs than the general population of juvenile offenders. This conclusion may be tempered, however, by the fact that there was variability in the frequency of offending within the sample. For example, our sample included offenders who committed only one or a few offences in their criminal careers (9.3% had three or less convictions to age 26). Therefore, although they may have been higher risk, they were not all high-rate offenders.

The study sample was, on average, 17.7 years of age at the time of admission into the facility (range: 16 to 25 years) and the average sentence length was 122.6 days (median: 93 days). Youth could have been older than 17 years at the time of admission if they were under 18 at the time of committing their index offence, but 18 years or older at the time of sentencing. They were, on average, 15.6 years at the time of their first court contact (range: 9.6 to 19.4 years).^{Footnote 1} Their average age at the end of the follow-up was 32.0 years (range: 26.3 to 40.2 years). For the entire sample, records were accumulated across a 29-year period, from as early as June 29, 1978 to September 26, 2007, the end of the follow-up period. Further details about the sample, including the trajectory analysis, may be found in Day, Nielsen, Ward, Sun, Rosenthal, Duchesne, Bevc, and Rossman (2012), and Day, Nielsen, Ward, Rosenthal, Sun, Bevc, Duchesne, Rossman, and Samuels (2010).

2.12 Offence Data

Official criminal records for juvenile and adult offences were obtained from four sources:

• The (Ontario) Ministry of Community and Social Services (MCSS);
• The (Ontario) Ministry of Community Safety and Correctional Services (MCSCS);
• The Canadian Police Information Centre (CPIC); and
• Predisposition Reports (PDR) from the client files maintained by the children's mental health centre that operated the open custody facilities.

These four data sources were used to ensure a high degree of completeness and accuracy for the sequenced, longitudinal conviction data which is essential for research that requires an accurate tem­poral sequencing of criminal convictions (Smith, Smith, & Norma, 1984). Official records were appropriate for our purposes because they provided the requisite precision with regard to the timing and sequence of offending, as well as offenders' movement data into and out of custodial settings.

From these sources, counts by age of all their unique court contacts arising from a new set of charges were recorded to September 26, 2007, the end of the follow-up period. During the follow-up period, the sample accumulated a total of 4,657 unique court contacts arising from a new set of charges, for an average of 12.1 new court contacts for each offender. Unique court contacts included those that resulted in a conviction and disposition (e.g., secure or open custody, fine, etc.), including a suspended sentence; those that resulted in a finding of guilt but not a conviction (e.g., absolute or conditional discharge); and those that resulted in either a withdrawal of charges, stay of proceedings, or determination that the person was unfit to stand trial (e.g., due to cognitive competence). These latter types of court contacts, which involved neither a finding of guilt nor a conviction, only accounted for 6.5% of the total number of court contacts. Lastly, for 8.0% of the court contacts, the final status in the official records was “remand,” and as such, no specific outcomes were available. For the present study, only offences that resulted in a conviction or finding of guilt was included in generating victim cost estimates.

For each court contact, the offender may have been convicted of a number of offences, including property, violent, drug, sex, and breach, and, perhaps even multiple counts of these different offences. All these offences and counts were considered in calculating the costs associated with their criminal activity. Given that the offender could have been charged with multiple offences at each court contact, the number of offences incurred by the sample exceeded the number of court contacts, for a total of 7,257 offences.

2.13 Age

Because all of the study participants were followed up to age 26, this age was used as the upper cut-off for counting criminal convictions for the entire sample. Therefore, for all results, we present costs incurred between the ages of 12 and 26 years (inclusive). In the main text of this report, we break down this 15-year period into two age intervals: 12 to 17 years (youth), and 18 to 26 years (adult). The follow-up period was further broken into five three-year intervals, 12 to 14 years, 15 to 17 years, 18 to 20 years, 21 to 23 years, 24 to 26 years, which permitted a more nuanced look at the timing of expenditures. These results are presented as Appendices to this report.

2.14 Trajectory Analysis

Consistent with previous studies, offence costs were also examined in relation to the seven trajectories identified by Day et al. (2012) (see Figure 1). As reported in the Day et al., article, prior to conducting the trajectory analysis, the criminal count data were adjusted for two factors: 1) time-at-risk, which takes into account the time spent not incarcerated, and so at risk to offend (Eggleston, Laub, & Sampson, 2004); and 2) an estimated age of the offender at the time of offence rather than at court contact, as only the latter was available from the official criminal record (see e.g., Farrington, Coid, Harnett, Jolliffe, Soteriou, Turner, & West, 2006). Additional details on these adjustments may be found in Day, Bevc, Duchesne, Rosenthal, Rossman, and Theodor (2007), while technical information about the trajectory analysis may be found in Nielsen, Rosenthal, Sun, Day, Bevc, and Duchesne (in press).

The following heuristic labels were applied to the seven trajectory groups, based on the rate and shape of their offence trajectories. The moderate late persister group was the least prevalent trajectory (3.6%) and had the longest average trajectory length (M = 16.6 years) and the latest peak age of offending of all the trajectory groups. The high late group had the highest average rate of court contacts and a peak age of offending in their mid-20s. Moreover, this group comprised only 3.9% of the sample but accounted for 15.8% of the total number of court contacts incurred by the sample. The high early group (4.4%) had the second highest rate of offending and a peak age of offending in their early 20s. The fourth trajectory, labeled the moderate adolescence-peaked group (11.7%), showed a peak age of offending in early adolescence and an average trajectory length of 10.2 years. The moderate early persister group (14.2%) had the second longest average trajectory length, up to age 30.1 years (M = 14.8 years), and a peak age of offending in their early 20s, earlier than the moderate late persister group. The low desister group, comprising 29.8% of the sample, had the shortest average trajectory length (M = 3.1 years) and the fewest average number of court contacts, accounting for only 7.4% of the total court contacts incurred by the sample. Lastly, the low persister group comprised 32.4% of the sample and had an average trajectory length of 11.1 years, to age 27 years, on average.

Figure 1: Estimated Criminal Trajectories for Seven-Group Model

Image Description

For this sample, seven criminal trajectory groups were estimated, based on the rate of offending (shown on the vertical axis) over age (shown on the horizontal axis). Each trajectory is depicted as a function that varies in terms of its overall shape, whereby the higher amplitude represents more frequent offending at a particular age. The figure shows these trajectories over an age range of 15 to 35 years. The average frequency of offending at each age ranges between zero to eight offences. For example, low risk trajectory curves are flatter whereas high risk trajectories have a comparatively more steep peak in their shape.

2.2 Calculating Costs

In general terms, we used conviction and disposition-based correctional data to generate cost of crime estimates for study participants. For the current report, the estimated costs associated with these offences in­cluded both tangible costs (i.e., those associated with law enforcement, the youth and adult criminal justice systems, and loss of victims' productivity and wages) and intangible costs (i.e., those associated with loss of life and trauma, pain, and suffering of victims). We calculated the total cost of crime for each participant by adding together the following components: victim costs (based on offence/conviction data), correctional costs (based on disposition data), other criminal justice system costs (based on Canadian estimates of criminal justice system expenditures, relative to correctional costs), and the costs of undetected crime (based on previous empirical research).

2.21 Offence-Based Victim Costs

First, we determined the number of convictions for each person. As noted earlier, the sample of participants had a total of 7,257 offences across 4,657 unique court contacts. These offences spanned 313 discrete offence codes, which were subsequently collapsed into 29 categories for this report. Cate­gories were developed in reference to per-crime estimates available in the published literature. Where such estimates were available, we grouped offence codes into conceptually equivalent categories for analysis. For other categories, we further sub­divided offence categories in order to better fit monetary estimates with actual conviction data, based on the 313 offence codes available to us. For example, the costs associated with the offence of “theft” was further subdivided into “minor theft” and “serious theft.” There was variation across published studies in how offence categories were tabulated, and therefore, part of the early work of this study involved a content analysis of what types of criminal activities were included (and not included) in each estimate. Appendix A shows the mapping of these codes into the 29 offence categories used in this study, and the frequency (count) at which each offence occurred in the sample.

Next, to determine crime-specific victim cost estimates, we drew from four published studies to generate average cost-per-crime estimates for these offence categories: Cohen and Piquero (2009); Dubourg, Hamed, and Thorns (2005), McCollister, French, and Fang (2010); and Roman (2011). The cost of an offence was calculated by taking the average of all available estimates within each category. For example, the estimate for homicide was based on figures from all four studies, whereas the estimate for armed robbery was based only on one study (Cohen and Piquero, 2009). To arrive at final estimates for each category, criminal justice system costs were first subtracted from each study estimate. These figures were then converted into Canadian currency for the base year(s) cited in each of the studies. We used the midpoint of the year, July 1^st, as the date to compare exchange rates using the calculator provided at http://www.oanda.com/currency/converter/. Lastly, these dollar figures were converted to 2013 dollars using the inflation adjust­ment calculator provided by the Bank of Canada at http://www.bankofcanada.ca/rates/related/inflation-calculator. Appendix B provides additional information about the methodology and costing components used to generate these estimates, and the calculations for the currency conversions and inflation adjustments are shown in Appendix C.

Table 1 lists these estimates by offence type. The estimates include direct and indirect victim losses: death or risk of death; property loss and damage; medical treatment; psychological harm; quality of life; pain and suffering; and victim and offender productivity (i.e., work, school, household). They do not include pre- and post-adjudi­ca­tion criminal justice system costs. When applicable, and to ensure comparability of the data, we used mean and not median monetary estimates of victim costs (e.g., Roman, 2011). In the final analysis, offence-specific costs were generated for 24 of the 29 offence categories (the remaining five categories were assumed to be victimless crimes, or in practical terms, have victim costs close to, or equal to zero). Offence categories were further grouped into broad categories (property, violence, other) for analysis (Table 1).

Lastly, to calculate the victim costs that each offender in the sample incurred, we multiplied the frequency of convictions for each offender for each of the criminal offence categories by the monetary cost of crime estimates in Table 1. For example, an average petty theft might be estimated to cost society around $2,000 per incident. Therefore, a person with five theft convictions would accrue $10,000 in costs for this offence category.

Sources: A=Dubourg et al. (2005); B=McCollister et al. (2010); C=Cohen & Piquero (2009); D=Roman (2011); E=Estimated. Crime Type: V=violence, P=property, O=other.

2.22 Disposition-Based Correctional Costs

A strength of this study is that we were able to generate highly accurate estimates of correctional costs for study participants. We capitalized on detailed provincial and federal datasets from MCSCS and CPIC to document the number of non-overlapping days served for each of eight categories of dispositions. These figures were multiplied by per-diem incarceration cost estimates based on Canadian published reports, and figures obtained directly from MCSCS and MCYS. To the extent that they were not publicly available or published in the grey literature, the project researchers worked collaboratively with the Interde­partmental Working Group on Costing and Justice (IWGCJ) and other governmental officials to obtain per-diem correctional cost estimates. As it was not known what level of prison security (i.e., minimum, medium, maximum) and/or programming had been offered or received by incar­cer­ates, only gross estimates were applied (i.e., there was one per-diem factor applied to each of the youth open, youth closed, adult provincial and adult federal custodial categories). These per-diem cost estimates and the sources are shown in Table 2.

To enumerate the number of days served for each of the eight disposition types, we capitalized on a “timeline” dataset developed for this project that tracked each participant's move­ment within the criminal justice system. These data were focused at the provincial and federal disposition level, and as such, did not include police or court-level information. As these data were event-based, there were multiple records for each individual representing transitions between disposition types (e.g., from incarceration to probation). We started with 17,247 records, of which 3,214 represented days spent in the community without supervision by a probation or parole officer. The remaining 14,033 records were coded into the following eight mutually exclusive categories: 1) youth open custody; 2) youth secure custody; 3) youth probation; 4) adult provincial incarceration; 5) adult provincial probation and parole; 6) adult federal incarceration (based on sentences > 2 years); 7) adult federal supervision in the community (i.e., following federal incarceration); and 8) conditional sentences. Days were allocated to these categories if they resulted from a finding of guilt. However we also included provincial remand/pretrial adult and youth custodial days because they represent tangible costs to the criminal justice system.

For each person, we calculated the number of days served at each age (i.e., from 12 to 26 years) for each of the eight disposition types. These figures were then multiplied by per-diem incarceration cost estimates shown in Table 2. We relied on both published and non-published figures when calculating per-diem estimates for the present study. In cases where there were multiple estimates for a disposition category, we calculated an average based on available numbers and standardized these to 2013 Canadian dollars (CAD), taking into account inflation (see Calculation Notes in Table 2). For youth open and secure custody estimates, we combined per-diems based on bed capacity and those based on actual utilization^{Footnote 2} to arrive at an averaged figure. It would have been preferable to multiply historical per-diem values to days served (in the original timeframes they were served), and then adjust these values for inflation. However, these data were not available. Accordingly, we used more recent per-diem estimates to calculate costs as they would be accrued today, given existing criminal justice system resources and infrastructure.

Sources: A: Correctional Services Division, Ministry of Community Safety and Correctional Services (2014; email correspondence). B: Public Safety Canada (December 2012). 2012 Annual Report: Corrections and Conditional Release Statistical Overview; C: MCYS via the Auditor. D: Zhang (2013). The Justice System Costs of Administration of Justice Offences in Canada, 2009. Department of Justice Canada; E: Ministry of Child and Youth Services (2014; email correspondence).

2.23 Other Criminal Justice System Costs

Disposition-based correctional costs represent only one component of the costs accrued in the investigation and pros­ecution of criminal offences. Other costs include police costs (e.g., pre- and post- inves­tigation, charging and conviction), court costs (e.g., judges, clerks, court services), prosecution costs (e.g., court prosecutor and investigations), and legal aid costs. Unfortunately, there are no published, averaged estimates for these costs in Canada or elsewhere, broken down by offence category. Therefore, we developed estimates of these costs based on published reports of governmental criminal justice spending. This approach is consistent with Koegl (2011, pp. 159-168), who used provincial global expenditures to calculate other criminal justice system costs.

Based on annualized spending between 1994 and 1998, as reported by Taylor-Butts (2002), Koegl (2011) calculated that corrections-related costs constituted 24.8% of criminal justice spending, and on average, police, courts, and legal aid and prosecutions accounted for the remaining 75.2% of annualized budgets. This means that other criminal justice system costs were approximately three times larger (75.2/24.8 = 3.03) than correctional costs. As well, Zhang (2011, p. 7) reported that in Canada, corrections constituted 32.2% of total criminal justice spending in 2008, which means that other criminal justice system expenditures are more than double that (67.8/32.2 = 2.11). More recent figures reported by the Parliamentary Budget Office (Story & Yalkin, 2013, p. 18) revealed that 23% of criminal justice system expenditures in 2012 were allocated to corrections, which means that other criminal justice system costs constituted more than three quarters of the total costs (77/23 = 3.35). For this study, we chose the “middle ground” scaling up factor of 3.03 to calculate other criminal justice system costs. For ease of computation and interpretation, this figure was rounded down to an even value of 3.00.

2.24 Undetected Crime Costs

Because only a fraction of criminal offences ever comes to the attention of legal authorities, it is important to apply multipliers to criminal convictions to include undetected crime. Putting aside the fact that some offenders are never caught for their crimes, Frank and Carrington (2007) suggest that upwards of 90% of criminal incidents by offenders are not captured in official reports of offending. Farrington, Auty, Coid, and Turner (2013) estimated that there are 25 offences for every conviction for persons convicted of at least one offence (as compared to their calculated ra­tio of 38:1 which includes both convicted and non-convicted individuals). For this reason, it is im­por­tant to factor in some sort of estimate of the “dark figure of crime” (Frank & Carrington, 2007).

There is limited empirical research that calculates multipliers for crimes by comparing self-reported and official offending in samples of adolescents (i.e., Farrington et al., 2007; 2003). Farrington and Koegl (2014) reviewed these findings using multipliers reported by Cohen and Piquero (2009) for adults, and recalculated multipliers for 16 different offence categories. Other sources also provide multipliers (e.g., Dubourg & Hamed, 2005; Farrington, Auty, Coid, & Turner, 2013; Mahew, 2003). However, these sources cover only a limited range of broadly-defined offence categories, and, for this reason, we relied on the more specific estimates provided by Farrington and Koegl (2014), as shown in Table 3.

Notes: Table based on Farrington, D.P. & Koegl (2014). Monetary benefits and costs of the Stop Now And Plan Program for boys aged 6-11, based on the prevention of later offending. Manuscript under review; C=Cohen & Piquero (2009); E=Estimated; F1= Farrington et al. (2003); F2=Farrington et al. (2013).

We estimated multipliers for some offence categories that did not have empirical estimates by substituting values for comparable offence categories: abduction/kidnapping was given the same value as homicide (1.0) because it was assumed that reporting for this crime would be near-perfect; serious assaults were given the same value as aggravated assault (7.9); intimida­tion and extortion were given the same value as robbery (12.2); serious mischief was given the same value as arson (7.9). We applied the empirical estimate for weapons offences (7.9) from Cohen and Piquero (2009) to both categories of weapons use and weapons possession. The estimate for larceny/theft (15.0) was applied to the serious theft category for this study. To be conservative, we used the smallest (29.6) of three multipliers from Farrington et al. (2013) to estimate undetected crimes for minor thefts: theft from vending machine (29.6), theft from vehicle (32.9), shoplifting (70.1). By definition, undetected crimes do not come to the attention of the criminal justice system. As such, these multipliers were only applied to the victim costs calculated for each individual based on conviction data.

2.3 Analytic Strategy and Data Presentation

Based on the monetary estimates derived for each individual in the sample, we calculated the aggregate cost of crime by summing these figures for all 386 study participants. As noted earlier, costs were calculated up to the end of age 26, because this was the upper age for which 100% of the study participants were followed up. For all analyses reported in the body of the text, we reported costs for two broad age ranges: adoles­cence (i.e., between the ages of 12 and 17 years) and adulthood (i.e., ages 18 to 26 years). We further broke the follow-up period into five age intervals, each spanning three years: 12 to 14 years, 15 to 17 years, 18 to 20 years, 21 to 23 years, and 24 to 26 years. This permitted a more nuanced analysis of how costs added up over time (reported as Appendices).

In addition to calculating aggregate costs, we also calculated average per-person costs by dividing cost estimates by the number of study participants. Annual, averaged per-person costs were also calculated by dividing per-person estimates by the number of relevant follow-up years. This latter calculation assumes that the costs are equally distributed over the number of years in the follow-up period. As will be seen, however, crime expenditures and costs do vary by age, and, accordingly, the reader should keep this in mind when interpreting these results.

In this report, we focus on showing the costs of crime by age and trajectory group (broken down by cost type: victim; correctional; other criminal justice system; undetected crime). Using correctional data, it was further possible to distinguish between costs to the federal justice system versus the provin­cial adult and provincial youth justice systems. Aggregate and average costs were also calculated for property, person and other offences. Lastly, we calculated the aggregate and average cost for specific groups of offenders (e.g., adolescent-only offenders versus adult-only offenders) to assist policy makers in developing strategies to respond to these discrete populations of offenders.

3.0 Results

3.1 Victim Costs by Age and Offence Type

Table 4 shows the total number of convictions for the sample by offence category, and their associated victim costs between the ages of 12 to 17 years and 18 to 26 years. These costs are broken down into five age intervals in Appendix D, whereas Appendix E shows these costs rank-ordered by the frequency of offence categories. Due to their high per-crime victim costs, crimes involving violence account for the largest proportion of costs for the sample of 386 participants (see Figure 2). Together, homicide, rape, serious assault, robbery and common assault account for nearly three quarters (72%) of the total costs of crime ($182,602,699) between the ages of 12 and 26 years. Because the follow-up age intervals are different (6 years for the group aged 12 to17 years, and 9 years for the group aged 18 to 26 years), we calculated the annualized average cost of crime per person, comparing the youth versus adult intervals. The average cost per person, per year, between the ages of 12 to 17 years was $29,981, compared to $33,242 between the ages of 18 and 26; (subtracting the six homicides in the 18 to 26 years interval had the effect of dropping the second average down to $22,944). For the entire time period, from 12 to 26 years of age, the average victim costs per offender were $473,063, approximately $31,538 per offender, per year. Interestingly, the highest annualized per person costs of $50,145, were accumulated between the ages of 15 and 17 (see Appendix D). So the peak at which victim costs are incurred (see Figure 2) corresponds to the peak of the well-established age-crime in offending, at around 17 years of age (Piquero, Farrington, & Blumstein, 2007).

Additionally, we created three categories of offences (violent, property, and other) to examine patterns in the cost of crime in relation to age, by the nature of the offence type. The sample of 386 offenders accumulated $165,441,175 in victim costs ($428,604 per person) for violent crimes, $16,784,524 costs ($43,483 per person) in property crimes, and $377,000 ($977 per person) for other crimes. Although convictions for violent offences only accounted for 19% of all convictions, the victim costs related to these account for 91% of the total. Conversely, property crimes are the most frequent type of crime being committed (44%), but incur only 9% of victim costs.

In further analyses by age and nature of offence, Figure 2 shows that victim costs keep increasing up to age 17, at which point they decrease between the ages of 17 and 19 years, increase slightly (by about $5 million) up to age 20, then continue to decrease to the end of the follow-up period. Further details and per person, per year breakdowns of these figures are available in Appendix F, where it is shown that, across the five age intervals, victim costs associ­ated with property crime are proportionally highest in the mid-to-late adolescent period (ages 15 to 17 years). This is due to the high rate of property crimes committed in this age period relative to other types of offences (Dauvergne, 2013).

Note: “Type” refers to crime type, where: V=Violent; P=Property; and O=Other.

Figure 2: Victim Costs Associated with Convictions by Age and Broad Offence Category

Image Description

Victim offences account for the greatest proportion of the costs of crime in this sample and that the highest costs are incurred between the ages of 16 and 18 years of age. The figure shows that the vast majority of victim costs are associated with crimes involving violence rather than property crime.

Note: The costs of "other offences" is quite small in relation to "violent" and "property" offences, and therefore, is not visible in this figure.

3.2 Disposition-based Correctional Costs by Age

Table 5 shows the number of days served for each of the eight disposition types as youth, between the ages of 12 and 17, and as adults, aged 18 to 26 years. The total cost of dispositions for the sample of 386 offenders for the 15-year follow-up period was $122,179,765, at $10,441,501 per year during the adolescent years, and $6,614,528 per year during the adult years. Further details and annualized per person estimates of correctional costs by disposition type are provided in Appendix G.

From Table 5, one can see that approximately $85 million (70.1% of the total) was spent on youth open and secure custody alone. Most of the expenditures were made in the adolescent years, despite the fact that the follow-up interval was shorter (6 years versus 9 years). Youth secure custody costs, calculated over a 6-year period, were the highest of all correctional cost categories, and they constituted the largest proportion of spending over the entire follow-up period, due to the high per-diems and the large number of secure custody days served by study participants. In examining all disposition types, youth dispositions were more expensive than adult dispositions (see Appendices G and H).

Not surprisingly, incarceration was much more expensive than community supervision, and this was the case for both the adolescent years and the adult years (see Appendix I). Youth open and secure custody, as well as adult provincial and federal incarceration cost more than 18 times as much as community-based options such as probation, parole and conditional sentences ($115,910,650 and $6,269,115 respectively).

Note: P=provincial, F=federal.

3.3 Other Criminal Justice System Costs

As noted earlier, we multiplied correctional cost estimates by a factor of 3.00 to estimate other criminal justice system expenditures such as police, courts, prosecution and legal aid spending. Figure 3 shows these costs, with the addition of victim costs, in relation to age. For the entire sample, other criminal justice system costs were $366,539,294 over the 15-year follow-up period, or an average of approximately $949,584 per person ($63,306 per person per year). Because these costs were calculated as a factor of correctional costs, they have a similar relationship to age (see Figure 3). As such, they were also highest between the ages of 15 and 17 years, totalling $174,928,152 or $453,182 per person (for three years), or $151,061 per person per year (see Appendix J for details).

Figure 3: Total Victim, Correctional, and other Criminal Justice System Costs (N=386)

Image Description

This figure is a stacked line graph that plots aggregate victim, correctional and other criminal justice system (CJS) costs (on the vertical axis) against age (on the horizontal axis). The maximum total monetary burden of crime, combining these three cost types, increases from age 12 onward and then begins to increase sharply at age 15, peaking at age 17 at approximately $160 million dollars, and then declining sharply until about age 19 where it levels off to about $30 million dollars per year, on average. Other CJS costs accounted for the greatest proportion of monetary costs incurred by this sample, followed by victim costs and correctional costs.

3.4 The Victim-related Costs of Undetected Crime

Using the multipliers provided in Table 3, we estimated the frequency and victim costs of undetected crimes committed by sample participants between the ages of 12 and 26 years. Table 6 shows the results of these calculations. Of the 24 offence categories that have victim costs assigned to them (see Table 1), common assault, robbery, serious assault, sexual assault, and intimidation constituted the largest proportion of undetected crime costs (78.3% combined). Between the ages of 12 and 26 years, the estimated victim costs of all undetected crime incurred by the 386 study males was nearly $1.6 billion ($1,588,830,864) in 2013 Canadian dollars, or approximately $4,116,142 per person. Comparable analyses subdividing these costs by the five age intervals are shown in Appendix K.

3.5 The Costs of Offending by Trajectory Group

Using trajectory analysis, offenders in the sample were assigned to seven discrete groups based on their longitudinal rate of offending (see Figure 1). For additional comparison purposes, we examined their cost of crime estimates by age. Tables 7 and 8 sum­marize the findings for the seven trajectory groups, adding together all costs accrued between the ages of 12 and 26 years. Detailed analyses for each trajectory group can be found in Appendix L. Table 7 shows the aggregate numbers and number of individuals in each trajectory (T) group. As one example, the low desister (T6) group amassed approximately $408 million in total costs. The moderate early persister (T5) group produced a comparable number: $445 million. However, there were twice as many individuals in the T6 group contributing to the total cost figure, as compared to the T5 group. Table 8 takesthe number of individuals in each group into account and shows calculated per-person averages for the seven trajectory groups which, for ease of discussion, could be classified into two low-rate groups (T6 and T7), three moderate-rate groups (T4, T1 and T5), and two high-rate groups (T3 and T2).

Looking at the average cost per person, individuals in the low-rate groups cost between $3.5 and $3.8 million each; moderate-rate individuals cost between $7.9 and $8.1 million each; and individuals in the high-rate groups cost between $12.4 and $17.0 million each. The average frequency of convictions follows the same pattern for low-, moderate-, and high-rate groups: 6.5 to 8.5, 18.1 to 19.7, and 27.1 to 28.5, respectively. It is worth noting that although high-rate offenders only constituted 8.3% of the sample, they contributed 20.6% of the costs. Moderate-rate offenders also had disproportionate costs, but this did not hold true for the low-rate offenders.

Figure 4 presents these results in graphical form for the collapsed trajectory groups. These results show the differential monetary impact of subgroups of offenders based on their longitudinal rate of offending. Across the three groups, the costs incurred rose sharply to age 17 then generally fell off into adulthood. The degree of separation of the lines suggests that low rate offenders can be most clearly differentiated from the moderate- and high-rate groups. These results reinforce the notion that early intervention programs for children and youth with multiple risk factors have the potential to yield considerable savings over the long-term. As well, although for high rate offenders the average cost per person follows the same pattern over time as moderate- and low-rate offenders, between the ages of 20 and 24 years, the costs rise to levels between $400,000 and $500,000 per person (solid line).  This increase was attributable to five homicides committed by high rate offenders between the ages of 20 and 24 (one homicide was registered each year by five separate individuals). Removing these expensive crimes from the calculations had the effect of lowering the average costs per person to amounts between $200,000 and $300,000 over the same time interval (dotted line). Nonetheless, the costs remained higher than for the moderate- and low-rate offence trajectory groups.

Lastly, with regard to the trajectory group analysis, there was one important consideration that may have impacted estimates of the costs of offending, particularly with regard to how undetected crime costs were calculated. At first glance, it may be argued that the multipliers for undetected crime are too high because offences that do not result in investigations and charges being laid are arguably less serious than those that result in a formal conviction in criminal court. Although we applied the multipliers (in Table 3) universally across all sample participants, it could be argued that, taking into account the results of the trajectory group analysis, some groups by definition engage in less criminal activity (e.g., low-rate offenders) relative to other groups (i.e., moderate- and high-rate offenders), and accordingly, some “discounting” of multipliers might be warranted.

To explore this possibility, we conducted an additional analysis, whereby we discounted victim cost values for undetected offences listed in Table 1 by 25%. Therefore, whereas a common assault resulting in a conviction would cost $54,737, the victim cost associated with an equivalent, undetected assault would be valued at $41,053. Second, we discounted the multipliers based on trajectory group mem­ber­ship. Undetected offences for high-rate offenders (T2 and T3) remained the same, as they were calculated using the original multipliers in Table 3. However, undetected offences were calculated at 75% of these values for mod­erate-rate offenders (T1, T4, and T5), and 50% of the original values for low-rate offenders (T6, T7).

Applying these discounts had the overall effect of halving the estimate for victim costs related to undetected crime from $1.6 billion to approximately $800 million ($801,326,964) or $2,075,977 per person ($138,398 per person annually). Despite this decrease, the estimated total cost of crime (victim, correctional, other criminal justice system and undetected crime) was still considerable, at nearly $1.5 billion for all offenders or $3,815,152 per person ($254,343 per person annually). Appendix M shows the result of applying these discounting factors to the calculation of undetected crimes.

Figure 4: Average Total Victim, Correctional, and other Criminal Justice System Costs by Trajectory Group (N=386)

Image Description

This figure presents results in graphical form for the collapsed trajectory groups. These results show the differential monetary impact of subgroups of offenders based on their longitudinal rate of offending.

3.6 The Cumulative Costs of Crime

Adding up the individual cost components yields a staggeringly large number representing the monetary costs of criminal behaviour for the 386 individuals in this study. In this section, we combine victim, correctional, other criminal justice system and (undiscounted) undetected crime costs to arrive at a total estimate of the cost of offending for the 15-year follow-up period. These numbers are shown in Table 9 for offending that occurred in the adolescent and adult years (see Appendix N for a breakdown of these costs using the five age interval categories). Figure 5 shows how these costs accumulate in one-year intervals between the ages of 12 and 26 years.

Adding all the estimates together, we found that this sample of offenders amassed more than $2.26 billion in costs associated with crime over the 15-year period. This trans­lates into $5,855,317 per offender or $390,355 per offender, per year. Even if only considering costs (victim harm, correctional, and other CJS costs) associated with actual convictions and not undetected crime, the average was still quite high at $1,739,176 per person, or $115,945 per person annually. The most costly three-year interval was between the ages of 15 and 17, during which time the sample of 386 offenders cost society roughly $900 million, or 40.0% of the total costs for the entire follow-up period.

Figure 5: Total Cumulative Costs of Crime by Cost Type and Age (N=386)

Image Description

The cumulative costs of crime are enormous. This figure shows the accumulating costs of crime by age, calculated as the sum of victim, correctional, other criminal justice system and undetected victim crime costs. For this sample, the costs were calculated as zero at age 12 where they increase steadily to about age 15 where they increase sharply between the ages of 15 and 17, and continue to rise up to the end of the follow-up period, at age 26. The costs for undetected crime account for the majority of the costs incurred by the sample, followed by other criminal justice system costs, victim costs and correctional costs.

3.7 Serious and Frequent Offender Costs

Figure 6 shows the distribution of costs for the 386 individuals in the sample. Based on the entire sample, the mean and median total costs were $1,739,176 and $1,249,512, respectively. The values for the lower and upper 25% of the sample were $648,638 and $2,318,271, respectively. We did not include costs associated with undetected crime for these analyses because it would be misleading to apply multipliers to individual cases (since they are based on group averages).

Figure 6: Distribution of Total Costs for the Entire Sample (N=386)

Image Description

This chart illustrates the distribution of total costs for the entire sample.

By definition, offenders in the high-rate groups (T2 and T3) commit more frequent and more serious crimes than individuals in other trajectories (see Table 7). In the current sample of 386 males, we identified the offender who, based on actual convictions (i.e., excluding undetected crime cost estimates), had the highest victim, correctional, and other criminal justice system costs. This person (in the high late trajectory group, T2) had 18 convictions between the ages of 12 and 26 years, which resulted in a total cost of $11,305,897, broken down as follows:

• Victim costs: $6,307,605 (one homicide, serious assaults, various property offences);
• Correctional costs: $1,249,498 (mostly federal incarceration costs); and
• Other CJS costs: $3,748,494 (scaled up costs).

The next most costly individual (also in group T2) accrued a total of 28 convictions and incurred $9,929,836 in victim, correctional, and other criminal justice system costs as follows:

• Victim costs: $6,862,236 (one homicide, many serious assaults, robbery, intimidation charges);
• Correctional costs: $766,900 (mostly federal but substantial provincial incarceration costs); and
• Other CJS costs: $2,300,700 (scaled up costs).

It is perhaps not surprising that the top five individuals in terms of total combined victim, correctional, and other criminal justice system costs all had one conviction for homicide. Each of these individuals began their offending in their adolescent years, and four out of five had at least one conviction at or before the age of 15. Combined, these five individuals (three of whom were from T2, and two were from high early trajectory group, T3) had 103 convictions that cost society $48,256,907 between the ages of 12 and 26, or approximately $9.6 million per person.

The most frequent offender (from the moderate adolescence-peaked trajectory group, T4) had a total of 104 convictions, 84 of which were for break and enter offences. This person accrued $784,509 in victim costs, $761,395 in correctional costs and $2,284,185 in other criminal justice system costs for a total of $3,830,090. Looking at the most frequent offenders, the top five had a total of 443 convictions, which translated into a total of $17,126,346 in costs, $5,067,104 of which were victim costs, $3,014,811 were correctional, and $9,044,432 were other criminal justice system costs.

As a point of contrast, the least expensive offender (from the low deisister trajectory group, T6) accumulated $66,509 in victim, correctional and other CJS costs over the 15-year follow-up. The 10 least costly offenders (equally represented in the low rate, T6 and T7 groups) accumulated $1,131,137 in equivalent costs over the same timeframe.

3.8 Specific Offender Profiles

We used correctional data to compare the total monetary costs of crime between the ages of 12 and 26 for four groups, defined as individuals who:

• G1: Served time in federal custody (N=35, 9.1%);
• G2: Served time in provincial (but not federal) adult custody (N=257, 66.6%);
• G3: Had an adult conviction but served no adult custody (N=49, 12.7%); and
• G4: Had a youth conviction, but no adult convictions (N=45, 11.7%).

Costs for G1 included costs associated with both federal and provincial adult and youth convictions. Table 10 shows the average costs by cost type for the four groups. Not surprisingly, individuals who had served federal time had the highest average costs across all cost categories.

4.0 Discussion, Limitations and Conclusions

4.1 Discussion

The aim of this study was to estimate the longitudinal costs of offending for a sample of 386 high-risk male offenders in Ontario, whose offence histories were tabulated over 15 years, from ages 12 to 26 years. Costs included both tangible and intangible costs, victim-related costs and disposition-related correctional costs. Estimates were scaled up to include other criminal justice costs, as well as costs associated with undetected crimes. All dollar values were converted to 2013 CAD.

The aggregate cost of crime for the sample was $2.26 billion, an average of $5.86 million per person in our sample, over the 15-year follow-up. Although this figure may seem high, it makes sense in the following context: 1) in terms of victim costs, crimes involving violence incur the most costs, particularly in terms of victim-related harms; 2) costs related to the psycho­log­ical harm, pain and suffering of victims constituted a substantial proportion of monetary estimates; and 3) including costs for undetected crime contributed considerably to increasing the overall cost estimates.

Considering only offences that resulted in official convictions in court (versus undetected crimes), we found that the sample of 386 males created an aggregate burden of crime in the amount in excess of $671 million or $1,739,176 per person over the 15-year follow-up period ($115,945 per person annually). Adding in estimates for undetected offences contributed an additional $1.6 billion in aggregate costs or $4,116,142 per person ($274,409 per person annually) to our estimates, for a total of $5.86 million per person. These estimates were derived using the best available scaling up estimates from published empirical studies. However, even when discounting the costs of undetected crimes, we still found the costs to be extremely large, at approximately $3.8 million per person ($254,343 per person annually).

In addition to calculating the aggregate cost of crime for the sample, we also examined how costs accumulated over time. Consistent with other research (Piquero et al., 2013), the most costly developmental period was mid- to late-adolescence, between the ages of 15 and 17 years, which accounted for 40% of the total costs. This was a time when the average correctional, other criminal justice system, and victim costs were at their highest relative to other age intervals. This was also the time when the costs associated with both property and violent crimes were at their highest.

Using the trajectory analysis results from Day et al. (2012), we examined the costs of the different trajectory groups and found substantial differences in terms of the average cost of crime per individual: high-rate offenders cost between $12 to $17 million each; medium-rate individuals cost approximately $8 million each; and low-rate offenders cost about $3.5 million each, although the timing of these costs varied among trajectory groups. These findings reinforce the idea that even within a high-risk offender sample, there is heterogeneity in the frequency, severity, and timing of offending, and among the dispositions received in response to it. Looking within trajectory groups, low-rate offenders had the highest per-person total costs in the 15 to 17 year age interval; this was mostly true for moderate-rate offenders (trajectory group T1 had the highest average, per person costs between the ages of 18 and 20); and high-rate offenders had the highest costs between the ages of 18 and 20.

4.2 Strengths and Limitations of the Present Study

This study has a number of strengths that increase the validity of the results and advance our know­ledge and methods of calculating the costs associated with crime. First, we capitalized on a very detailed criminal history dataset in calculating the number of days served for each of eight separate disposition types. Thus, it was possible to calculate with great certainty the number of days actually spent serving these dispositions (i.e., using “days served” as opposed to “days sentenced” when calculating estimates). With these numbers, we used the best available per-diem correctional estimates in Ontario, the province in which the sentences were overwhelmingly served. This combination of an accurate count of “days served” and reliable per-diem costs allowed us to arrive at very good estimates of correctional costs for the 386 individuals in the study. In addition, the relatively long follow-up period of 15 years provides a longitudinal look of the costs of crime during the peak adolescent years and into early adulthood, intervals when crime prevention approaches should be applied with vigour if cost savings are to be realized.

Second, the list of per-crime victim costs drawn from the literature also extends the work of Farrington and Koegl (2014) to include a longer list of crime types. We were also able to develop new estimates based on a careful matching of offence-specific codes with per-crime estimates. Although this is a strength of the study, research is desperately needed to fine tune the per-crime estimates, for example, by expanding the list of included offences further, and to more carefully cost out components within each estimate. As no per-crime cost estimates currently exist within the Canadian literature, this is clearly an area for future research. As well, because we used a hybrid approach to calculating costs for each individual using disposition and offence informa­tion, we are confident that we have calculated cost estimates that reasonably estimate victim harm, yet take into account the unique characteristics of the separate youth and adult criminal justice systems in Canada.

Lastly, the detailed coding of the data also allowed us to track costs as they occurred for each offender on an annual basis. In fact, the findings reproduce the well-recognized age-crime curve (albeit, with a peak two years later compared to curves based on offending and not adjudication/costs per se).

Despite the many strengths of this study, some limitations require explanation. First, could our results be biased due to cohort effects? For example, might we expect different results if the sample was drawn from different years (e.g., O'Brien, 2010)? The present study mitigates this problem somewhat since participants were admitted to the referring institution over a ten-year period, between 1986 and 1997; therefore, they represent offenders of different ages and exposures to the criminal justice system at different times. Notwithstanding this issue, it should be noted that the offenders in this sample were drawn from an era when youth incarceration was used more frequently than is currently the case (Bala, 1997; Bala, Carrington, & Roberts, 2009; Dauvergne, 2013). As such, we might expect youth custody costs to be lower with a contemporary sample of offenders fol­lowed-up prospectively, given the system's current preference to use extra-judicial, non-custodial measures allowable under the Youth Criminal Justice Act (YCJA).

Second, our results are specific to a particular sample, which may be a higher risk group than one generated by randomly selecting a sample from the general offender population. Two factors contribute to this possibility. First, the current sample had all been sentenced to a custodial facility (i.e., open cus­tody) in their late teens. As such, they had already penetrated the juvenile justice system more deeply than, for example, a youth who had committed a low severity offence (e.g., shop­lifting) at the age of 15 and never offended again. Second, as the open custody facility from which the sample came was operated by a children's mental health centre, with a greater availability of services and access to mental health specialists, the youth sentenced to this facility may have been considered to be more “at-risk,” with more mental health problems and criminogenic needs than the general population of juvenile offenders. This, of course, would limit the generalizability of the study findings to all offenders. This conclusion may be offset, however, by the fact that there was variability in the frequency of offending within the sample. For example, our sample included offenders who committed only one or a few offences in their criminal careers (9.3% had three or less convictions to age 26). Therefore, although they may have been higher risk, they were not all high-rate offenders. We hope the estimates for the four cost components (victim, corrections, other criminal justice system, and undetected crimes) provided in this report may be used by other researchers to generate cost estimates with other samples of offenders to examine this possibility.

Third, as with any cost of crime study, it was not possible to disaggregate the cost of a criminal offence versus a criminal episode. For example, a person might break into ten cars on one day to steal property and this might be considered one criminal episode, but for this analysis, it would be treated as ten thefts. This is a particular problem with respect to property crime, where many minor crimes can appear larger “on paper” because of charging practices.

Fourth, as noted earlier, our estimates of victim costs were based on non-Canadian figures, and therefore, their applicability to the Canadian context can be called into question. This is an area where we would encourage future Canadian researchers to focus on, specifically, to develop estimate categories based on Criminal Code of Canada terminology, for example, distinguishing between summary and indictable offences within relevant offence categories.

Fifth, we relied on elementary calculations to estimate other criminal justice system costs, such as policing, courts, prosecution and legal aid. Unfortunately there are no good data in Canada to be able to apply these estimates by type of offence. Therefore, we universally scaled-up correctional costs by a factor of three to create an averaged index. It would seem obvious that some offences (e.g., homicide, sexual assault) require more personnel time and resources within the criminal justice system compared to others (e.g., petty theft). A related limitation is that it was not possible to code the number of disposition days served by offence type within the time constraints of the project. Doing so would have allowed us to calculate criminal justice system costs for each offence. Unfortunately this was not possible because: 1) the timeline and conviction data were in different datasets and could not be meaningfully merged together; and more importantly, 2) it would have been extremely difficult to disaggregate sentence length and time served in those cases where individuals were convicted of multiple offences (e.g., applying severity weightings).

A final limitation is that we did not explicitly disaggregate costs to the offender from those accrued by victims of crime, such as the loss of productivity of offenders as a result of incarceration (Cohen, 2005). To our knowledge, and to this point, there are currently no published studies that calculate the indirectcosts of incarceration. Given increasing support in the research literature that incarceration without treatment contributes to increased rates of recidivism among offenders (e.g., Cullen, Jonson, & Nagin, 2011), it would be valuable for future research to calculate the monetary costs of exposing offenders to other highly antisocial individuals during incarcera­tion. Such a focus would highlight the need to provide evidence-based programmatic approaches with positive benefit-to-cost ratios to persons in conflict with the law.

Indeed, our review of the literature revealed a wide range of approaches in calculating crime costs. For example, Cohen et al. (2010) incorporated additional, more intangible costs such as fear of crime to generate their cost estimates (using the top-down method). Indeed, it could be rightfully argued that our costs underestimate the total economic burden on Canadian society. At present, there is no gold standard method for estimating the costs of crime, and every study is subject to their own limitations. Indeed, in another sense, some social costs, such as the loss of social cohesion in a low-income neighbourhood, personal costs of a child growing up with a parent incarcerated, and family costs of grieving the loss of a child to a life of crime are impossible to quantify, making the true cost of crime incalculable.

4.3 Conclusion

The above limitations notwithstanding, we found that the costs of criminal offending, incurred by this sample of 386 high-risk male offenders in Ontario, was substantial. As expected, we found considerable differences in all the component costs as the sample moved from adolescence to adulthood, and as we parsed the sample into different trajectory groups based on their rate of offending over time. Although the issues are complex and potentially politically charged (see Raffan-Gowar & Farrington, in press), based on the results of this study, we would argue that the best option to reduce the costs of crime is to invest in strategic crime prevention and early intervention programming. We know from studies that have calculated the cost-benefit ratios (see http://www.wsipp.wa.gov/BenefitCost) that early intervention and prevention programs have demonstrated poten­tial to yield substantial cost savings with a high degree of confidence on the return on investment. Accordingly, we believe this is where the largest investment needs to be made. Given that even the ten least costly offenders still accumulated over $1.1 million dollars each, on average, bolsters the argument that even the most modest investments in developmental crime prevention programs can pay enormous dividends to society.

References

Allard, T., Stewart, A., Smith, C., Dennison, S., Chrzanowski, A., & Thompson, C. (2014). The monetary cost of offender trajectories: Findings from Queensland (Australia). Australian & New Zealand Journal of Criminology, 47(1), 81–101.

Anderson, D. A. (1999). The aggregate burden of crime. Journal of Law and Economics, 42, 611-637.

Anderson, D.A. (2012). The costs of crime. Foundations and Trends in Microeconomics, 7(3), 209–265.

Aos, S., Miller, M., & Drake, E. (2007). Evidence-based public policy options to reduce future prison construction, criminal justice costs and crime rates. Federal Sentencing Reporter, 19, 275-290.

Aos, S. Phipps, P., Barnoski, R., & Lieb, R. (2001). The comparative costs and benefits of programs to reduce crime, Version 4.0. Washington State Institute for Public Policy. Available from http://www.wsipp.wa.gov/rptfiles/costbenefit.pdf.

Augimeri, L. K., Jiang, D., Koegl, C. J., & Carey, J. (2006). Differential Effects of the Under 12 Outreach Project (ORP) Associated with Client Risk and Treatment Intensity. Report to the Provincial Centre of Excellence for Child and Youth Mental Health at CHEO, Program Evaluation Grant # PEG162606-101. Available from http://www.excellenceforchildandyouth.ca/sites/default/files/gai_attach/peg-631_final_outcomes_report.pdf.

Bala, N. (1997). Youth criminal justice law. Concord, Ontario: Irwin Law.

Bala, N., Carrington, P. J., & Roberts, J. V. (2009). Evaluating the Youth Criminal Justice act after five years: A qualified success. Canadian Journal of Criminology and Criminal Justice, 51, 131-167.

Blokland, A. A. J., Nagin, D., & Nieuwbeerta, 2005). Life span offending trajectories of a Dutch conviction sample. Criminology, 43, 919-954.

Brand, S., & Price, R. (2000). The economic and social costs of crime. London: Home Office, Economics and Resource Analysis Unit.

Byrnes, J. M., Doran, C. M., & Shakeshaft, A. P. (2012). Cost per incident of alcohol‐related crime in New South Wales. Drug and alcohol review, 31(7), 854-860.

Cohen, M. A. (1988). Pain, suffering, and jury awards: A study of the costs of crime to victims. Law and Society Review, 22, 537-555.

Cohen, M. A. (1998). The monetary value of saving a high risk youth. Journal of Quantitative Criminology, 14, 5-33.

Cohen, M. A. (2005). The cost of crime and justice. NY: Routledge.

Cohen, M. A. & Bowles, R. (2011). Estimating costs of crime. In A. R. Piquero & D. Weisburd (Eds.), Handbook of Quantitative Criminology (pp. 143-162). New York: Springer.

Cohen, M. A., Miller, T., & Rossman, S. (1994). .The costs and consequences of violent behavior in the United States. In A.J. Reiss Jr., & J. A. Roth (Eds.), Understanding and preventing violence: Consequences and control (pp. 67-166). Washington DC: National Academy Press.

Cohen, M. A., & Piquero, A. R. (2009). New evidence on the monetary value of saving a high risk youth. Journal of Quantitative Criminology, 25, 25-49.

Cohen, M. A., Piquero, A. R., & Jennings, W. G. (2010a). Estimating the costs of bad outcomes for at-risk youth and the benefits of early childhood interventions to reduce them. Criminal Justice Policy Review, 21, 391-434.

Cohen, M., A. Piquero, A. R., & Jennings, W. G. (2010b). Monetary costs of gender and ethnicity: Disaggregated group-based offending. American Journal of Criminal Justice, 35, 159-172.

Cohen, M. A., Rust, R. T., Steen, S., & Tidd, S. (2004). Willingness to pay for crime control programs. Criminology, 42, 89-109.

Craig, W., Schumann, L., Petrunka, K., Khan, S, and Peters, R. (2011). Government costs associated with delinquent trajectories. International Journal of Child, Youth, and Family Studies, 2, 263-293.

Cullen, F. T., Jonson, C. L., & Nagin, D. S. (2011). Prisons do not reduce recidivism: The high cost of ignoring science. The Prison Journal, 91(3 suppl), 48S-65S.

Dauvergne, M. (2013). Youth court statistics in Canada, 2011/1012. Canadian Centre for Justice Statistics, Statistics Canada, Catalogue no. 85-002-X

Day, D. M., Bevc, I., Duchesne, T., Rosenthal, J, S., Rossman L., & Theodor, F. (2007). Comparison of adult offense prediction methods based on juvenile offense trajectories using cross-validation. Advances and Applications in Statistics, 7, 1-46.

Day, D. M., Bevc, I., Theodor, F., Rosenthal, J. S., & Duchesne, T. (2008). Change and continuity in criminal offending: Criminal trajectories of the “Toronto” sample. Final research report submitted to the Ministry of Children and Youth Services (MCYS), Toronto, Ontario.

Day, D. M., Nielsen, J. D., Ward, A. K., Rosenthal, J. S., Sun, Y., Bevc, I., Duchesne, T., Rossman, L., & Samuels, S. (2010). Criminal trajectories of two subsamples of adjudicated Ontario youths. Final research report submitted to the National Crime Prevention Centre (NCPC), Ottawa, Ontario. Electronic copy available at http://www.publicsafety.gc.ca/cnt/rsrcs/ pblctns/djdctd-ntr-yth/djdctd-ntr-yth-eng.pdf).

Day, D. M., Nielsen, J. D., Ward, A. K., Sun, Y., Rosenthal, J. S., Duchesne, T., Bevc, I., & Rossman, L. (2012). Long-term follow-up of criminal activity with adjudicated youth in Ontario: Identifying offence trajectories and predictors/correlates of trajectory group membership. Canadian Journal of Criminology and Criminal Justice, 54, 377-413.

Detotto, C., & Vannini, M. (2010). Counting the cost of crime in Italy. Global Crime, 11, 421-435.

Dubourg, R., Hamed, J., & Thorns, J. (2005). The economic and social costs of crime against individuals and households 2003/04: Home Office online report 30/05. London: Home Office, Research, Development and Statistics Directorate.

Dodge, K. A., & McCourt, S. N. (2010). Translating models of antisocial behavioural develop­ment into efficacious intervention policy to prevent adolescent violence. Developmental Psychobiology 52, 277-285.

Eggleston, E. P., Laub, J. H., & Sampson, R. J. (2004). Methodological sensitivities to latent class analysis of long-term criminal trajectories. Journal of Quantitative Criminology, 20, 1-26.

Farrington, D. P., Auty, K. M., Coid, J. W., & Turner, R. E. (2013). Self-reported and official offending from age 10 to age 56. European Journal on Criminal Policy and Research, 19(2), 135-151.

Farrington D. P., Coid, J. W., Harnett L., Jolliffe, D., Soteriou, N., Turner, R., West, D. J. (2006). Criminal careers up to age 50 and life success up to age 48: New findings from the Cambridge Study in Delinquent Development. London: Home Office (Research Study No. 299).

Farrington, D. P. & Koegl, C. J. (2014). Monetary benefits and costs of the Stop Now and Plan Program for boys aged 6-11, based on the prevention of later offending, Manuscript under review.

Farrington, D. P. & Welsh, B. C. (2006). Saving children from a life of crime: Early risk factors for effective interventions. NY: Oxford University Press.

Foster, E. M., Jones, D., & the Conduct Problems Prevention Research Group. (2006).Can a costly intervention be cost-effective? Archives of General Psychiatry 63, 1284-1291.

Frank, O., & Carrington, P. J. (2007). Estimation of offending and co-offending using available data with model support. Journal of Mathematical Sociology, 31, 1-46.

French, M. T., Popovici, I., & Tapsell, L. (2008). The economic costs of substance abuse treatment: Updated estimates and cost bands for program assessment and reimbursement. Journal of Substance Abuse Treatment, 35, 462-469.

Koegl, C. J. (2011). High-risk antisocial children, predicting future criminal and health outcomes (Unpublished doctoral dissertation). University of Cambridge, United Kingdom.

Leung, A. (2004). The cost of pain and suffering from crime in Canada. Ottawa, ON: Research and Statistics Division, Department of Justice Canada. Retrieved from http://www.justice.gc.ca/eng/rp-pr/cj-jp/victim/rr05_4/rr05_4.pdf

Mayhew, P. (2003). Counting the costs of crime in Australia. Australian Institute of Criminology.

McCollister, K. E., French, M. T., & Fang, H. (2010). The cost of crime to society: New crime-specific estimates for policy and program evaluation. Drug and Alcohol Dependence, 108, 98-109.

McIntosh, C., & Li, J. (2012). An introduction to economic analysis in crime prevention: The why, how and so what. National Crime Prevention Centre, Ottawa, Canada. Available from http://www.publicsafety.gc.ca/cnt/rsrcs/pblctns/cnmc-nlss/cnmc-nlss-eng.pdf.

McCuish, E. C., Corrado, R., Lussier, P., & Hart, S. D. (2014). Psychopathic traits and offending trajectories from early adolescence to adulthood. Journal of Criminal Justice, 42, 66-76.

Miller, T. R., Cohen, M. A., & Rossman, S. B. (1993).Victim costs of violent crime and resulting injuries. Health Affairs, 12, 187-197.

Miller, T. R., Cohen, M. A., & Wiersema, B. (1996). Victim costs and consequences: A new look (Document No. NCJ155282). Washington, DC: National Institute of Justice.

Miller, T. R., Levy, D. T., Cohen, M. A., & Cox, K. L. (2006). Costs of alcohol and drug-involved crime. Prevention Science, 7, 333-342.

Moolenaar, D. E. G. (2006). Expenditure on crime in the Netherlands (Memorandum 2006-1): Research and Documentation Centre of the Dutch Ministry of Justice. Retrieved August 22, 2010 from http://english.wodc.nl/onderzoeksdatabase/me2006-1-expenditure-on-crime-in-the-netherlands.aspx?cp=45&cs=6796.

Nielsen, J. D., Rosenthal, J., S., Sun, Y., Day, D. M., Bevc, I., & Duchesne, T. (in press). Group-based criminal trajectory analysis using cross-validation criteria. Communications in Statistics: Theory and Methods.

Nores, M. Belfield, C. R., Barnett, W. S., & Schweinhart, L. (2005). Updating the economic impacts of the High/Scope Perry Preschool Program. Educational Evaluation and Policy Analysis 27, 245-261.

O'Brien, R. M. (2010). Offender trajectories, crime trends, and costs: An invited policy essay on studying the costs of crime across offender trajectories. Criminology & Public Policy, 9, 313-319.

Piquero, A. R., Farrington, D. P., & Blumstein, A. (2003). The criminal careers paradigm: Background and recent developments. In M.Tonry (Ed.), Crime and Justice: A Review of Research (pp. 359–506). Chicago, IL: University of Chicago Press.

Piquero, A. R., Jennings, W. G., & Farrington, D.P. (2013). The monetary costs of crime to middle adulthood: Findings from the Cambridge study in delinquent development. Journal of Research in Crime and Delinquency, 50, 53-74.

Piquero, A. R., Sullivan, C. J., & Farrington, D. P. (2010). Assessing the differences between the short-term, high-rate offenders and long-term, low-rate offenders. Criminal Justice and Behavior, 37, 1309-1329.

Raffan-Gowar, B., & Farrington, D. P. (in press). The monetary costs of criminal careers. In K. Boers, T. Feltes, J. Kinzig, L. W. Sherman, F. Streng, & F. Trueg (Eds.). Criminology, crime, policy, penal law: Festscrift fur Hans-Jurgen Kerner on the occasion of his 70th birthday. Tubingen, Germany: Mohr Siebeck.

Rajkumar, A. S., & French, M. T. (1997). Drug abuse, crime costs, and the economic benefits of treatment. Journal of Quantitative Criminology, 13, 291-323.

Rollings, K. (2008). Counting the costs of crime in Australia: a 2005 update. Canberra: Australian Institute of Criminology.

Roman, J. K. (2011). How do we measure the severity of crimes? New estimates of the cost of criminal victimization. In J. MacDonald (Ed.), Measuring crime, advances in crimino­logical theory (pp. 37-70). New Brunswick, U.S.A.: Transaction Publishers.

Schweinhart, L. (2007). Crime prevention by the High/Scope Perry Preschool Program. Victims and Offenders 2, 141-160.

Smith, D. R. Smith, W. R., & Norma, E. (1984). Delinquent career-lines: A conceptual link between theory and juvenile offenses. The Sociological Quarterly, 25, 155-172.

Story, R. & Yalkin, T. R. (2013). Expenditure Analysis of Criminal Justice in Canada. Ottawa: Office of the Parliamentary Budget Officer.

Taylor-Butts, T. (2002). Justice Spending in Canada, 2000/01. Ottawa: Industry Canada.

Tolan, P. H., & Gorman-Smith, D. (2002). What violence prevention research can tell us about developmental psychopathology. Development and Psychopathology 14, 713-729.

Tracy, P. E. & Kempf-Leonard, K. (1996). Continuity and Discontinuity in Criminal Careers. New York: Plenum.

Tracy, P. E., Wolfgang, M. E., & Figlio, R. M. (1990). Delinquency Careers in Two Birth Cohorts. New York: Plenum.

Welsh, B. C., & Farrington, D. P. (2000). Monetary costs and benefits of crime prevention programs. Crime and Justice, 27, 305-361.

Welsh, B. C., & Farrington, D. P. (2007).Scientific support for early prevention and delinquency and later offending. Victims and Offenders 2, 125-140.

Welsh, B. C., & Farrington, D. P. (2012). Science, politics, and crime prevention: toward a new crime policy. Journal of Criminal Justice, 40, 128-133.

Welsh, B. C., Loeber, R., Stevens, B. R., Stouthamer-Loeber, M., Cohen, M. A., & Farrington, D. P. (2008). Costs of juvenile crime in urban areas: A longitudinal perspective. Youth Violence and Juvenile Justice, 6, 3-27.

Wolfgang, M.E., Figlio, R.M., & Sellin, T. (1972). Delinquency in a birth cohort. Chicago: University of Chicago Press.

Zhang, T. (2011). Costs of crime in Canada, 2008. Department of Justice Canada, Ottawa. Electronic copy available at http://www.justice.gc.ca/eng/rp-pr/csj-sjc/crime/rr10_5/ rr10_5.pdf.

Zhang, T. (2013). The Justice System Costs of Administration of Justice Offences in Canada, 2009. Ottawa: Department of Justice Canada.

Appendices

Appendix A: Offence Codes and Categories

Each table (from left to right) lists the offence label, the observed count in the raw dataset, the percentage of cases observed for the total number of offences, and the label description that references Criminal Code sections.

Appendix B: Victim Cost Methodological Notes

Social science databases were searched to locate studies on the costs associated with crime. We identified research that pro­vided estimates by type of crime, and further subdivided these into tangible and intangible costs, to the extent that it was possible to do so. We conducted a search of the international literature with an emphasis on published Canadian research, limiting the search scope to the last ten years, but prioritizing the last five years of publication.

Participants in the Toronto sample were convicted of a total of 7,257 convictions across 4,657 court contacts, spanning 313 discrete offence codes. These codes were subsequently collapsed into 29 categories. Categories were developed in reference to per-crime estimates available in the published literature. Where such estimates were available, we grouped offence codes into conceptually equivalent categories for analysis. If no monetary estimate was available, offence codes were classified into the best alternative category. It is worth noting that these estimates represent the tangible and intangible costs of crime, and are focused primarily on enumerating victim costs. The following decision tree was used to evaluate empirical studies to derive costing categories and monetary estimates. In deciding which studies to include, of all the available studies, the ones that met the most criteria relative to the others were used in our calculation of victim costs by offence type.

1. Is the source a primary research study?
2. Is the methodology used to generate estimates well-defined and scientifically sound?
3. Was the study published in the past 10 years?
4. Can criminal justice system costs be removed from the estimate?
5. Does the estimate capture all or most of the following harms for relevant offences?
   • Death or risk of death;
   • Property loss and damage;
   • Medical treatment;
   • Psychological harm;
   • Quality of life, pain and suffering; and
   • Victim productivity (e.g., work, wages, school, household)
6. Are the monetary estimates reasonable, given estimates for other types of crime?
7. For international studies, are the estimates generalizable to Canada?

The literature review began with a search for primary research articles using combinations of the following keywords and their variants: cost, crime, offences, monetary, cost-benefit, cost-effectiveness, per-diem, criminal justice system. A previous review by Farrington and Koegl (2014) identified two studies (Cohen & Piquero, 2009; McCollister, French, & Fang, 2010) as the most comprehensive, recently published articles on this topic. For the present study, we therefore performed an additional literature search by searching for articles that cited these two studies. The purpose of this review was to locate additional studies provid­ing estimates of the cost of crime by specific offence types. This search yielded a total of 65 articles. From our review of these studies, a surprisingly small number (N=15) were primary studies with published monetary estimates of the costs of specific crimes. Moreover, many of these studies in this group relied on one another to fill a gap in terms of calculating certain types of cost expenditures. Table B1 lists these studies and shows the methodologies used to generate per-crime cost estimates. Studies could be grouped into three main categories: 1) jury compensation/court awards; 2) willingness to pay; and 3) studies using an empirical approach to calculate costs. Table B1 also lists the currency and base years in which estimates were reported. This information was subsequently used to standardize costs and currencies so that estimates could be combined and averaged for this study.

Because the current study drew upon several different methodologies to estimate the costs of crime, it was important to know what types of costs were included and not included in each of the estimates. To date, the most comprehensive cost of crime typology in the published literature is Cohen (2005; see Table 1.1, pp. 10-11). He organizes costs into three broad categories: 1) costs of crime; 2) cost of society's response to crime; and 3) offender costs, and these are further subdivided into more detailed groupings. Another well-cited example is Brand and Price (2000), which distinguishes between costs incurred: 1) in anticipation of crime; 2) as a consequence of crime; and 3) in response to crime. These typologies were analyzed for their content and reorganized to highlight the groups that suffer the losses associated with criminal behaviour. The typologies have a temporal element to them; they classify costs according to whether they are accrued before, during or after criminal events. Brand and Prince (2000) and Cohen (2005) also identify stakeholders who incur the costs of crime, for example, society, third parties, the criminal justice system, and offenders. For this study, a modified typology was developed, one that builds upon, but fully disaggregates, the temporal and stakeholder elements of the Cohen (2005) and Brand and Price (2000) models (see Table B2).

With respect to the decision tree specified above, it is clear that some of the studies reviewed did not meet all the selection criteria. For example, several studies were published outside the 10-year window (e.g., Miller et al, 1993). However, they were included because they continue to be the most comprehensive and most widely-cited analyses to date. From a review of the literature, tangible and intangible victim costs were broken down into seven categories, shown in Table B2. The first of these, death or risk of death, pertains mainly to homicide offences where the monetary cost of a life/death has been calculated; property loss and damage is applic­able to property offences including mischief; medical costs include physician, hospital and other health related costs resulting from person offences where a victim suffers injury; psychological costs refer to fee for service activities to treat mental health and psychological trauma following victimization; the quality of life, pain and suffering category includes psychological or material person-based damages resulting from victimization. Such costs are typically calculated from analysis of jury awards for specific types of crimes. Productivity costs refer to lost time at work, school, and in the household due to victimization. Lastly, fear of crime and avoidance behaviour costs refer to a more generalized pattern of avoidance and/or fear of victimization resulting from being victimized. Arguably, this is the most subjective of all the categories of costs, and is there­fore the most difficult to estimate. As such, it does not factor into the analyses of this study. Lastly, it is worth noting that most of the monetary estimates cited in the literature include immediate and short-term costs, but typically exclude long-term costs. One exception to this is the monetary cost of a life lost (i.e., the “death or risk of death” category), where research­ers attempt to calculate the effect of lost productivity over the course of a person's lifetime.

Notes: Roman (2011) presents data in 2000, 2003 and 2008 USD dollars; Aos, Phipps, Barnoski, Lieb (2001) and McCollister, French, & Fang (2010) rely on Miller et al (1996) for estimating some victim costs.

Studies were additionally enumerated in terms of who bears each of the seven types of victim costs. For this analysis, six categories were used: the victim him/herself (V); family/friends of the victim or offender (F); any level of government (G); third parties or the members of the private sector (P); the offender (O); and society in general (S). Of course, some of these categories could theoretically overlap with one another for any specific criminal episode or event. Looking at Table B2, it is clear that no single study calculates all seven categories of victim costs. More­over, where most of the studies expectedly enumerate costs from the perspective of the victim, there is substantial variation in terms of whether costs borne by other parties are included.

Table B3 shows the major offence types for which costs were calculated for each study. It should be noted that some studies further broke down some of these categories to differentiate, for example, victim types and the severity of harm inflicted. These data are not shown here because the present study could not capitalize on such distinctions; conviction records used for this study did not have information about victims or the circumstances surrounding each offence. As should be quite obvious from Table B3, there is substantial variation across studies in terms of the specificity and range of criminal categories included. For example, some studies only distinguished between violent and property offences, whereas others made fine distinctions between different types of theft (e.g., theft of vehicle, theft from vehicle, shoplifting). This list was used to narrow down the number of offence categories into a smaller, more manageable set so that they could be applied to our conviction data. The number and type of categories used for this study was determined by considering two factors: 1) the quality of the data underlying the monetary estimates cited in the primary research studies; and 2) the frequency and range of convictions by offence type registered for the Toronto sample.

Notes: V=victim; F=family/friends; G=government; P=private sector, O=the offender, and S=society in general.

Appendix C: Currency Conversion and Inflation Adjustments

Notes: Serious assault was calculated as the average of Aggravated Assault and Common Assault; Use of Weapons was given the same value as Common Assault; Serious Theft was calculated as the average of Minor Theft and Motor Vehicle Theft; Serious Mischief was given a value of $5000 as a midway value between Vandalism/Mischief ($1,209) and Serious Theft ($8,008); Possession of Weapons, Drug Dealing and Other offences were given arbitrary, maximum values of $1,000 each. The last five categories were assumed to have no, or extremely small, victim costs. “RAW” refers to the raw monetary estimate, as provided in the original study; “CAD base” refers to the value of the RAW estimate in Canadian dollars for the base year; and 2013 CAD refers to the monetary estimate adjusted for inflation, in 2013 Canadian dollars.

Appendix D: Victim Costs and Conviction Counts by Five Age Intervals

Appendix E: Victim Costs by Age and Offence Frequency