Economic Outcomes of Canadian Federal Offenders

Economic Outcomes of Canadian Federal Offenders PDF Version (1 Mb)

By Kelly M. Babchishin¹, Leslie-Anne Keown², and Kimberly P. Mularczyk¹

¹ Public Safety Canada
² Correctional Service of Canada

Table of contents

• Introduction
  • Current Study
• Method
  • Participants
  • Procedure
  • Measures
  • Data Analyses
• Results
  • Filing Taxes
  • Labour Force Participation, Employment Income, and Social Assistance
• Discussion
• References
• Appendix A
• Appendix B

Introduction

Corrections services promote public safety by administering sentences and facilitating the successful reintegration of offenders into the community. Securing employment after release from a correctional institution is integral to the successful reintegration of offenders as law-abiding, contributing members of the community (Andrews & Bonta, 2010). Indeed, employment following release from correctional institutions is associated with both reduced reoffending rates and lower rates of return to custody (Berg & Huebner, 2011; Gillis & Nafekh, 2005; Webster, Staton-Tindall, Duvall, Garriety, & Leukefeld, 2007). Given the potential mitigating impact that employment may have on reoffending, many correctional institutions have opted to assess employment skills and provide employment and educational programming for offenders (e.g., prison work, vocational training) in an effort to augment the successful employment of its graduates and to, consequently, decrease the number offenders that will commit new offences (Correctional Service of Canada, 2018). Unfortunately, we know surprisingly little about the long-term community adjustment of adults with past criminal records in Canada; it is estimated that 1 in 10 Canadians aged 19 and over in 2005 had a criminal record (2.9 million of 24.8 million; Powell & Winsa, 2008; Statistics Canada, 2005)^{Note 1}. Boyce, Te, and Brennan (2018) found that adults that had any form of contact with police in Saskatchewan in the 2009/2010 year had incomes 50% lower than adults without police contact, as reported in their 2008 tax form. Chronic offenders (who had 5 or more subsequent recontacts with police) had an average income that was 67% lower than those with one contact with police. Unlike Brennna et al. (2018), the current study provides a national perspective on the community adjustment of federal offenders admitted to Correctional Service of Canada institutions and identifies characteristics associated with better economic outcomes among offenders.

Gaining Employment after Incarceration: An Elusive Challenge

Although estimates of post-prison employment rates tend to vary based on the sample, timeframe, and definitions used (AAltonen, 2016), overall post-prison employment rates are approximately 50% (Visher, Debus-Sherrill, & Yahner, 2011) to 75% (Von Bergen & Bressler, 2016). The low post-prison employment rate has often been explained through the multiple barriers that individuals with criminal records tend to face when trying to secure gainful employment following release from correctional institutions (Petersilia, 2001). The mere existence of a criminal record is one of the main reasons why securing employment after incarceration can be elusive for many. Namely, requesting proof of a clean criminal record is a common prerequisite to many job applications in Canada and United States; a phenomenon that has also been increasing in Europe (Pijoan, 2014). In the United Kingdom, for example, about three quarters of a million convictions that are shared with potential employers are more than ten years old, and a great deal of convictions are also not relevant to the qualifications required for a specific position (Doward, 2017).  Currently, between 50% (UK; Pijoan, 2014) to 92% (US; Von Bergen & Bressler, 2016) of employers request criminal record checks for job applicants, and the vast majority of these checks do not result in a criminal record being identified. For example, only 6% of 4.2 million criminal record checks in 2015 in the United Kingdom produced criminal record information (Doward, 2017).

There are a number of employment sectors where a criminal record automatically precludes employment (e.g., government agencies, vulnerable sectors). In Canada, each province and territory has their own unique regulations regarding employment discrimination^{Note 2} based on criminal records. Some provinces and territories do not include criminal records as a status that can be discriminated against (e.g., prairies), whereas others include criminal records as a protected status, but still allow for the requirements of the position to be considered when making the decision as to whether or not the employer will hire an individual with a criminal record (see Appendix A). The main limitation of current Canadian legislation, when present, is that the legislation does not preclude employers from asking for criminal record verification, nor does it define the instances in which a criminal record verification would be irrelevant (e.g., employment without access to vulnerable populations, such as children). For applicants with a criminal record, requesting criminal record verification also results in job application delays, which can preclude employment. When applicants with a criminal record request a criminal record check in Canada and a record is found, fingerprint verification must be completed before the record is released back to the applicant ─ a process that can take 3 months or more to complete (RCMP, 2014). This is a substantial duration for any one struggling to earn income through legal means.

Employer bias is a second barrier that individuals with criminal records commonly encounter when aiming to secure employment. Studies have found that employers are biased towards hiring individuals without a criminal record (Batastini, Bolanos, Morgan, & Mitchell, 2017; Petersilia, 2001). Employers often cite ensuring a safe workplace and reducing liabilities as their main justification for hiring individuals without criminal histories. Interestingly, however, most workplace violence is brought on by non-workplace employees (AAltonen, 2016). Regardless, and despite the suitable qualifications of individuals with criminal records, a criminal record is often used as justification for not hiring a particular candidate (e.g., AAltonen, 2016). Indeed, in a study where employers were asked to judge hypothetical employment scenarios, training programs were found to increase employers' beliefs that individuals with criminal records are acceptable for employment (Batastini et al., 2017). Yet training programs did not have an effect on whether an employer would seriously consider hiring a candidate with a criminal record (Batastini et al., 2017). In short, the presence of bias among employers toward individuals with criminal records and the delays that result from preliminary or extraneous criminal record verification requests negatively impact an individual's ability to obtain post-release employment, regardless of the individual's competencies or the type of employment sought.

Employment Outcomes Vary Following Release

Though the presence of a criminal record negatively impacts employment prospects, the degree to which a criminal record impacts employment varies across individuals. For example, men are more likely to be employed after release from correctional institutions than women (Duwe & Clark, 2017). Studies have shown that the employment rate following incarceration is lower for women (19% employed; Freudenberg, Daniels, Crum, Perkins, & Richie, 2005) compared to what is typically cited for men (50% employed in Visher et al., 2011). Women also take longer to find employment (median 10 months) then men (median 6 months) following release from correctional institutions (Gillis & Nafekh, 2005). Younger individuals are also more likely to be employed than older individuals following release from a correctional institution (Duwe & Clark, 2017; Visher et al., 2011).

In addition to gender and age, another demographic characteristic associated with differential employment outcomes following release from a correctional institution is race. For instance, following release from a correctional institution, Caucasian individuals are more likely to be employed than non-Caucasian individuals, a finding often attributed to racism (Nally, Lockwood, Knutson, & Taiping, 2013; Pager, 2003). A study conducted by Pager and Shepherd (2008) using identical resumes, found that individuals with traditional African American names (e.g., Jamal) were contacted by managers 50% less than individuals with traditional Caucasian names (e.g., Brad); a disadvantage that persisted even when the traditionally-named Caucasian individuals had a criminal record and the traditionally-named African American individuals did not (Bailey et al., 2017). In an observational study on the relationship between race and employment, research assistants who were matched on dress and qualification, but varied on race (Caucasian vs. African American) and on the presence of a hypothetical criminal record (yes vs. no), were asked to pose as hypothetical employment candidates (Page, 2003). In this study, Pager (2003) found that the research assistants who were Caucasian and had a hypothetical criminal record were over three times more likely to be called back and offered employment than research assistants who were African American with the same fictional criminal record and qualifications (17% vs. 5%, respectively). In fact, Caucasian research assistants with a hypothetical criminal record were more likely to secure employment than African American research assistants without a criminal record (17% vs. 14%; Pager, 2003).

Similarly, in another study, Von Bergen and Bressler (2016) concluded that Black individuals who have indicated that they have a criminal record on an employment application had twice as much difficulty being called back for employment than non-Black individuals. This presence of racial biases in the hiring decisions of some employers is reflected in the observed disparities of employment rates observed across racial groups. Consistently across studies, Caucasian offenders are more likely to be employed following release from correctional institutions compared to non-Caucasian offenders (e.g., African American, Hispanic, and Indigenous offenders; Duwe & Clark, 2017; Gillis, 2002; Visher et al., 2011). For example, the unemployment rate of individuals with a criminal record is approximately 50% post-release, yet when separated by racial background, this unemployment rate is higher for non-Caucasian individuals than Caucasian individuals (e.g., 59% unemployment rate for African American offenders vs. 38% unemployment rate for Caucasian offenders; Nally et al., 2013).

Along with demographic variables, a number of other factors impact an offender's ability to obtain employment once released. Researchers have found that offenders with a greater employment history prior to incarceration and connections to a prior or potential employer are more likely to be employed following release than offenders who were not employed prior to sentencing (AAltonen, 2016; Berg & Huebner, 2011). Several studies, for example, have found that individuals who have worked prior to incarceration are more successful at obtaining employment following incarceration than those without a work history and those with a more limited work history (AAltonen, 2016; Berg & Huebner, 2011). Individuals with a prior history of stable employment prior to incarceration who are released from correctional institutions report that their most successful strategy for obtaining post-release employment included returning to a previous employer (Visher, Debus, & Yahner, 2008). Yet, even after accounting for offenders' past employment history, those with greater prosocial ties to family were found to be more successful at obtaining employment then those without prosocial ties to family (Berg & Huebner, 2011). In a Canadian sample of federal offenders, those with greater affective ties to employment at intake and those who have a greater history of specific employment skills (e.g., mechanics) were related to higher employment success post-release (Gillis, 2002).

In addition, the likelihood to reoffend as assessed by risk assessment tools and the length of the criminal record both influence an individual's potential to find employment. For example, Canadian federal offenders with higher scores on two measures of risk at intake were less likely to be employed following release compared to those scoring lower on these risk measures (the Statistical Information on Recidivism score [SIR] and Risk/Need score; Gillis, 2002). Relatedly, in a large sample of released prisoners, a greater number of convictions and prison misconducts (Duwe & Clark, 2017), as well as greater presence of poor physical and mental health conditions (Visher et al., 2011), were associated with a lower likelihood of employment. In short, many individuals released from correctional institutions find the process of seeking and maintaining fulfilling employment to be a volatile one, with only about half reporting being happy with their pay at 2 and 8-months post-release (Visher et al., 2008).

Underemployment and Sources of Income Support

Despite the fact that about 50% of offenders report being employed following release, half of those employed offenders are classified as being marginally employed, meaning that the amount of income received annually is insufficient to support common living, familial, or personal need expenses (Nally et al., 2013). In addition, as previously mentioned, both women and non-Caucasian individuals tend to earn less than men and Caucasians (Bailey et al., 2017). For example, Nally and colleagues (2013) found that 59% of African Americans were marginally employed compared to 39% of Caucasian individuals. Given that sufficient income plays an important role in the successful reintegration of offenders, the aforementioned trends are concerning. Indeed, as the annual income of offenders increase, reoffending rates decrease (Nally et al., 2013; Webster et al., 2007).

Close to half of the individuals released from incarceration are unemployed (Visher et al., 2011) and, of those employed, close to half have reported incomes below the poverty line (Nally et al., 2013); thus, it is not surprising that a number of offenders report using government services to assist their income (AAltonen, 2016; Freudenberg et al., 2005; Harding, Wyse, Dobson, & Morenoff, 2011). Studies have found that public benefits and government supports (e.g., food stamps, Medicaid, Section 8 housing vouchers, Supplemental Security Income for disability), followed by employment and the support of family and friends, are key resources utilized by offenders once they are released (Harding et al., 2011). Public benefits and government supports are especially important for individuals who do not have family and friends to rely on, as well as for lower income families that are unable to adequately support family members newly released from correctional institutions (Harding et al., 2011). With greater time passing post-release, individuals report relying less on family and friends (66% at 2 months to 48% at 8 months), and more on other sources of income, such as legal employment (30% at 2 months to 41% at 8 months) and informal employment (28% at 2 months to 47% at 8 months; Visher et al., 2008). Coinciding with gender differences commonly found for the employment rates of offenders (Gillis & Nafekh, 2005), women report receiving more income from family and friends and government programs (both 56%), than from formal employment (27%; Freudenberg et al., 2005).

Current Study

The current study examined the economic outcomes of a large cohort of Canadian federal offenders 14 years post-release from correctional institutions. We hypothesized that a large portion of released individuals would be unemployed and, of those who were employed, that a large portion would be underemployed. We also hypothesized that women, non-Caucasian individuals, and those scoring higher on measures of risk to reoffend would have disproportionately poorer employment outcomes than men, Caucasians, and those scoring lower on measures of risk to reoffend.

Method

Participants

Participants included a cohort of 11,158 federal offenders admitted to Correctional Service of Canada (CSC) institutions between January 4th, 1999 and December 31st, 2001 (medianadmission year = 2000) who were released in the community and therefore able to file taxes in 2014 (i.e., not dead, deported, or incarcerated). On average, participants had lived 14 years in the community and were 47 years of age in 2014 following release from CSC institutions (see Table 1). Participants were sampled across Canada, with 30.6% of offenders admitted to the Prairie region, 23.5% of offenders admitted to Québec, 24.9% of offenders admitted to Ontario, 10.7% of offenders admitted to the Atlantic region, and 10.4% of offenders admitted to the Pacific region. Filing rates for the various characteristics is provided in the column 'Filers'. The cohort of released CSC Canadian federal offenders had a filing rate of 52.3% (5,835/11,158), with women filing at a higher rate (61.3%; 352/574) than men (51.8%; 5,483/10,584). In 2014, the filing rate among released CSC Canadian federal offenders (51.8%-61.3%) was lower than the filing rate among the general Canadian population of individuals 25 years of age or older (88%; Statistics Canada, CANSIM Table 11-10-004-01). Specifically, in 2014, 33% of Canadians aged 25 to 44 years of age, 35% of Canadians aged 45 to 64 years of age, and 20% of those aged 65 years or older filed taxes (Statistics Canada, CANSIM Table 11-10-004-01). Appendix B provides additional characteristics of tax filers and predictors separately for men and women. On average, tax filers did not declare any children under the age of 12 (M =0.2; SD = 0.7, N = 5,835), with only 10.8% declaring a child under the age of 18 on their taxes. On risk assessment tools measuring likelihood of reoffending, filers scored moderate (37.2%, n = 2,164) to high (50.8%, n = 2,959) on the Dynamic Factor Assessment at admission, had an average of 12.6 criminal convictions (SD = 7.5, n = 5,819), and had an average SIR-R1 score of 0.7 (SD = 10.7; ranged from -25 to 28; note that lower scores indicating higher risk to reoffend).

Procedure

Exclusion Criteria

Participants were excluded if they (1) did not have an Finger Print System (FPS) number or a Sentence ID, (2) if their sentence was overturned, squashed, or pardoned, (3) if they received provincial sentences, (4) if their status in 2014-2015 was deceased, deported, suspended, Temporary Detention (TD), Unlawfully at large (UAL), or incarcerated, and (5) if they were under the age of 18 at the time of admission.

Data Linkage with Statistics Canada

A database of identifying information of the CSC cohort was securely transmitted to Statistics Canada. These data were linked with the 2014 Statistics Canada's tax registry, which was the most recent year available (record linkage #090-2016).

Measures

Demographic Characteristics

Several demographic characteristics (i.e., age at the time of admission, gender, race, marital status at admission, region) were retrieved from CSC databases (for more information see Table 1B). Provinces and territories that denote criminal offence history as a status that should not be discriminated against for employment include (6): British Columbia, Ontario, Newfoundland and Labrador, Prince Edward Island, Quebec, and Yukon. Provinces and territories that denote pardoned and/or suspended criminal offence history only as a status that should not be discriminated against for employment include (2): Northwest Territories and Nunavut^{Note 3}. Provinces and territories that do not include criminal offence history as a status that should not be discriminated against for employment include (5): Alberta, Manitoba, Nova Scotia, New Brunswick, and Saskatchewan (see Appendix A). Demographic information was also retrieved from the 2014 Statistics Canada's tax registry (marital status at filing). In addition to typical marital status definitions (i.e., married, common-law, widowed, divorce, separated and single), Statistics Canada defines other categories of marital status as Spouse families (i.e., married couple living in the same dwelling, with or without children), Common-law families (i.e., common-law couple living in the same dwelling, with or without children), Lone-parent families (i.e., family with only one parent, male or female, and at least one child), and Non-family persons (i.e., not part of a census family, couple family or lone-parent family).

Tax Information
Information obtained on the filing taxes, labour participation rates, employment income, and social assistance payments were retrieved from the 2014 Statistics Canada's tax registry. Additional CANSIM tables were retrieved from Statistics Canada's website to provide comparisons between released CSC Canadian federal offenders and the general Canadian population. Labour participation rates are calculated by including the count of persons with labour income, which is given as a percentage of total tax filers and dependents in the area.

Dynamic Need Tool
The Dynamic Factors Identification and Analysis – Revised (DFIA-R) (Brown & Motiuk, 2005) is completed by parole officers as part of the Offender Intake Assessment process. The tool measures dynamic risk among male and female offenders, with the goal of identifying areas of criminogenic need. It is comprised of 100 indicators (scored as "yes" or "no") across the following seven domains: (1) personal/emotional, (2) substance abuse, (3) associates, (4) attitudes, (5) marital/family, (6) community functioning, and (7) employment/education. In respect to each domain, offenders are assessed in terms of their level of need and the relevance of the items to their criminal offending. In addition to a rating on each indicator, offenders receive an overall dynamic need rating based on the parole officer's professional judgement. Domain rating levels include: (1) no need for improvement, (2) low need for improvement, (3) asset to community adjustment (not applicable for substance abuse or personal/emotional domains), (4) moderate need for improvement, and (5) high need for improvement. For Indigenous offenders, Aboriginal Social History is taken into account in relation to each contributing dynamic factor. Domain ratings have been found to predict revocations of conditional release (Stewart, Wardrop, Wilton, Thompson, Derkzen & Motiuk, 2017).

SIR-R1
The SIR-R1 (Nafekh & Motiuk, 2002) is a 15-item evaluation tool created to predict general reoffending within 3 years after release among male, non-Aboriginal offenders. The SIR-R1 is a slightly modified version of the General Statistical Information on Recidivism (GSIR; Nuffield, 1982). The SIR-R1 combines demographic and criminal history characteristics to produce a total score ranging from −30 to 27 (higher scores are indicative of lower risk), which can then be classified into one of five risk categories (very low to very high). Each risk category is associated with a probability rating for recidivism. Meta-analytic reviews have demonstrated that the earlier version of the SIR (Nuffield, 1982) is moderately predictive of violent (e.g., d = 0.81; Campbell, French, & Gendreau, 2009; Yang, Wong, & Coid, 2010) and sexual recidivism (e.g., d = 0.64; Hanson & Morton-Bourgon, 2009). It is not scored on women or Indigenous offenders in CSC.

Static Risk Factor
The Static Factors Assessment is completed by parole officers as part of the Offender Intake Assessment process. The tool measures static (historical) risk factors among male and female offenders with the goal of assessing an offender's risk of recidivism. The tool is comprised of 137 indicators (scored as "yes" or "no") across three areas: (1) criminal history record, (2) offence severity record, and (3) sex offence history checklist. The parole officer judges overall risk as low, moderate or high, based on some or all of these areas and/or subscales (Helmus & Forrester, 2014). The overall rating of the tool is associated with community outcomes (Helmus & Forrester, 2014).

Data Analyses

In the current study, logistic regression was used to assess predictors in analyses when dichotomous outcomes were included (e.g., labour participation, yes/no), and linear regression was used to assess predictors in analyses when continuous outcomes were included (e.g., employment income). Logistic regression conducted with dichotomous outcomes provides odds ratios, whereas linear regression conducted with continuous outcomes provides exponent betas. Odds ratios below 1 indicate that higher scores on the predictor are related to lower odds of the outcome. Odds ratios above 1 indicate that higher scores on the predictor are related to higher odds of the outcome. If the 95% confidence intervals include 1 in a logistic regression, the variable is considered not significantly associated with the outcome (p > .05). For linear regression, exponent betas can be positive or negative; a negative exponent beta indicates that higher scores on the predictor are related to lower scores on the continuous outcome, whereas a positive exponent beta indicates that higher scores on the predictor are related to lower scores on the continuous outcome. If the 95% confidence intervals include 0 in a linear regression, the variable is considered not significantly associated with the outcome (p > .05). Exponent betas are standardized and, thus, can be compared across analysis across predictors. In contrast, odds ratios are tied to measurement scale of the predictors and are unstandardized, so we are unable to compare odds ratios across predictors and they cannot be compared across variables (i.e., dichotomous predictors will provide larger odds ratio than continuous predictors). Throughout, both bivariate and multivariate analyses were conducted. Bivariate analyses (unadjusted models) provide the association between the predictor and outcome while not controlling for other variables, whereas multivariate analyses (adjusted models) provide the association between the predictors and outcome, controlling for the other variables in the model.

Results

Filing Taxes

Bolded odds ratio and adjusted odds ratio (AOR, adjusted for other predictors in the model) in Table 2 denotes the characteristics that predicted tax filing (p < .05). Gender, Indigenous status, age, years since release, and score on the Static Risk Tool all predicted tax filing. Specifically, men had 28% lower odds of filing their taxes than women (AOR = 0.72), Indigenous had 10% lower odds of filing their taxes than non-Indigenous (AOR = 0.90), each one year increase in age was associated with a 10% increase in the odds of filing taxes (AOR = 1.01), each one year increase in being released was associated with a 3% decrease in filing taxes (AOR = 0.97), and, each one unit increase in the Static Risk tool was associated with 15% decrease in the odds of submitting taxes. Rates of filing ranged from 46% to 59% across the regions and was highest in Quebec (59%) and the Atlantic region (58%) and lowest in Ontario (46%).

Labour Force Participation, Employment Income, and Social Assistance

Table 3 provides the average income, labour force participation rate, and use of social assistance based on different demographic characteristics for offenders admitted to CSC institutions between January 1999 and December 2001. In total, 50.8% of tax filers reported some labour force income (2,962/5,835). In contrast, the labour participation rate was 68.8% for the general Canadian population aged 25 and over. However, those 45 to 49 years of age, which is most representative of our sample, had a higher labour participation rate of 85.3% (CANSIM Table 11-10-0023-01; Table 111-0009). The labour participation rate was lower for women released from CSC institutions (39%; 137/352) compared to men (51.5%; 2,825/5,483). These labour participation rates are between half to two thirds the rate of the general Canadian population of both women (64.2%) and men (73.8%; CANSIM Table 11-10-0023-01).

In 2014, the median employment income for the general Canadian population was $33,180 (CANSIM Table 111-0004), with men claiming a median employment income of $39,580 and women claiming a median employment income of $27,750. In stark contrast, both men and women within this federal cohort of offenders had a median income of $0, and ranged^{Note 4} from $0 to $118,805 (median^{Note 5} = $0, M = $14,049.41, SD = $22,963.12, n = 5,762). Of those who reported employment income (n = 2,962), the average income was a little over $14,000 (M = 14,118.69, SD = 23,011.29).

Approximately 2 out of every 5 individuals released from CSC institutions received at least some form of social assistance payment (40.7%; 2,377/5,835), with women released from CSC institutions receiving more social assistance payment than men released from CSC institutions (54.5% vs. 39.9%), and receiving close to ten times the social assistance payment rate of the general population^{Note 6} (5.9%; 1,559,570/26,618,560). The rate of social assistance payment in the general population was slightly higher for women (6.1%; 842,850/13,729,040) than men (5.6%; 716,720/12,889,520). Individuals released from CSC institutions received social payment ranging from $25 to $37,936 (median = $7,756.00; M = $8,231.84, SD = $4,870.86), which was also similar to the rate of social assistance payment received by individuals in the Canadian general population ($7,271.68).

Predictors of Labour Force Participation, Employment Income, and Social Assistance
Table 4 provides the characteristics associated with labour force participation outcomes for Canadian federal offenders. Bivariate analyses showed that all examined variables, with the exception of the Associates domain of the Dynamic Risk Factor scale, were statistically associated with labour force participation. The multivariate analyses showed that younger age (Adjusted Odds Ratios [AOR] = 0.95), fewer years since release (AOR = 0.94, 95% CI [0.89, 0.998]), being a male (AOR = 2.40), non-Indigenous status (AOR = 0.80), residing in a region without discrimination laws (AOR = 1.27), lower risk scores as assessed by the Dynamic Risk Factor scale (AOR = 0.66) and Static Risk Factor scale (AOR = 0.75), as well as a history of sexual offences (AOR = 1.28) were related to higher odds of participation in the labour force, even after controlling for other variables in the model. Predictors of labour force participation were found to be similar for both males and females, with confidence intervals largely overlapping (see Appendix B, Table 3B for the final adjusted model of labour force participation presentedfor males and females separately).

Similar variables that were found to predict participation in the labour market also predicted employment income. As can be seen in Table 5, standardized betas from the adjusted model show that younger age (Exp[b] = -.20), being a male (Exp[b] = .09), non-Indigenous status (Exp[b] = -.08), residing in a region without discrimination laws (Exp[b] = .11), lower risk scores as assessed by the Dynamic Risk Factor scale (Exp[b] = -.12) and Static Risk Factor scale (Exp[b] = -.11), and having a sexual offence in their official criminal history (Exp[b] = .04) were associated with higher employment income, even after controlling for other variables in the model. Predictors of employment were found to be similar for both males and females (see Appendix B, Table 4B for the final adjusted model of employment income presented separately for males and females).

Note. Bolder values represent OR that were statistically significant (p <.05).  Reference categories of dichotomous predictors are provided in parenthesis. ^bModel adjusted for other significant variables in the model. Excludes the DFA and the Static item score as these are total scores that include the other items in the table (multicollinearity). Also excludes SIR score as this tool was never scored on women or Indigenous so it was removed from the overall model. The R² of the adjusted model was .131 (N = 5,819).

Note. EXP(b):Standardized beta. Bolder values represent predictors that were statistically significant (p <.05). For dichotomous items, reference category is provided in parenthesis. Unadjusted model represents the bivariate regression. In contrast, the adjusted model controls for the other significant variables in the model. This adjusted model excludes the DFA item score and the Static item score as these are subsumed in the total score. SIR score was also removed from the adjusted model as it was not scored on Indigenous and women offenders. Adjusted model: Adjusted R² =.089.

Finally, as can be seen in Table 6, predictors of social assistance were generally similar but in the opposite direction of those predictors that were associated with labour participation and employment income. Women (AOR = 0.39, 95% CI [0.31, 0.49]), residing in a region without discrimination laws (AOR = 0.70, 95% CI [0.63, 0.79]), scoring higher on risk to reoffend as assessed by the Dynamic Factor Scale (AOR = 1.74, 95% CI [1.57, 1.92]) and Static Risk Factor Scale (AOR = 1.32, 95% CI [1.20, 1.45]), as well as having an exclusively non-sexual offending history (AOR = 0.70, 95% CI [0.61, 0.81]) were associated with greater likelihood of receiving social assistance, after controlling for other variables in the model. Predictors of social assistance were similar for both males and females, with the exception of regions. For women, regions without discrimination laws were associated with higher odds of social assistance. On the other hand, for men, regions without discrimination laws were associated with lower odds of social assistance (see Appendix B, Table 5B for the final adjusted model of social assistance payment presented separately for men and women).

Note. N = 5,474 men, R² = .172; N = 345 women, R² = .126. EXP(b):Standardized beta. Bolder values represent predictors that were statistically significant (p <.05). For dichotomous items, reference category is provided in parenthesis.

Discussion

We found that the economic outcomes of Canadian federal offenders are quite poor, even after an average of 14 years following release from a correctional institution. Only half of released offenders from CSC institutions who filed taxes in 2014 were in the labour market and, consequently, many were earning below the poverty line with a median income of $0. The average income of those who reported employment income was $14,000, which was at least half of the median employment income reported by men ($39,580) and women ($27,750) in the general Canadian population that year. As such, it is not surprising that many individuals with a criminal record are almost 10 times more reliant on government support agencies than individuals within the general population. In our sample, 41% received at least some form of social assistance payment; a rate that is almost 10 times higher than the general Canadian population.

We also found several characteristics that were associated with employment income in released federal Canadian offenders. First, women not only received more social assistance payment than men (54.5% vs. 39.9%), but women also made substantially less than their male counterparts, even 14 years after release from a correctional institution, with women earning on average just under $10,000 and men earning on average just over $14,000. In addition, individuals that were Indigenous earned less than their non-Indigenous counterparts, with Indigenous individuals earning on average just over $10,000 and non-Indigenous earning on average just under $15,000. In terms of type of offenders, top earners were those whose index offence^{Note 7} was a drug offence (just over $18,000) or white collar offence (just under $18,000). There were no differences between those who had an aggregate sentence length of under 4 years (just over $14,000) and those with aggregate sentence length of 4 to 10 years (just over $13,000). However, offenders with sentences longer than 10 years or indeterminate sentences made substantially more (just over $22,000). This is likely because indeterminate offenders on supervised release were a relatively small group (n = 67) who were older and had previous employment histories prior to incarceration. As well, it is possible that these individuals had greater support in the community from their community supervision officers. Finally, we found that risk factors that predict reoffending, as well as older age, both predicted poorer economic outcomes among individuals released from Canadian federal institutions. These findings suggest that higher-risk offenders and older offenders (45 and over) may require additional support once released into the community to find gainful employment.

Moving Forward

In the US, it is estimated that the loss of income due to the underemployment of ex-offenders is between 57 to 65 billion annually, an estimate that does not account for the incremental costs of social assistance programs that released offenders may also be utilizing (Schmitt & Warner, 2010). The current study found that released Canadian federal offenders are also underemployed. Underemployment of offenders does not suggest that there is a loss of income for Canadians as a whole, but rather that this loss of employment opportunities for released offenders is likely a contributor to reoffending. Gaining employment and making sufficient income is a key factor that predicts desistance from crime for offenders (e.g., Andrews & Bonta, 2010; Berg & Huebner, 2011; Gillis & Nafekh, 2005; Webster et al., 2007). Unfortunately, many individuals released from Canadian federal institutions find this goal of obtaining employment elusive, which is a commonly reported concern among released offenders that has been supported by research, showing that only about half of released offenders are able to find employment. Further, even when they find employment, about half of released offenders are making less than the poverty line. The underemployment of individuals with criminal records has commonly been explained through factors such as hiring biases that are present against those with criminal records, regardless of qualifications, and the fact that criminal record biases are compounded due to the higher ratio of minorities having criminal records than non-minorities (AAltonen, 2016; Batastini et al., 2017; Department of Justice Canada, 2017; Nally et al., 2013; Pager, 2003; Visher et al., 2011). Yet, as mentioned, Caucasian individuals with a criminal record are more likely to be employed than African American individuals without a criminal record (e.g., a US study from Pager, 2003).

Given that federal offenders vary in their ability to secure gainful employment, identifying offenders who require additional support before release is essential. The current study suggests that older individuals, women, and Indigenous offenders would benefit from greater support in securing employment in the community. In addition, there are scales designed to identify which steps in the job search process (e.g, resume writing, interview, job search efficacy) where individuals may require extra support or training. For example, the Offender Job Search Self-Efficacy Scale (OFJSSE; Varghese, Anderson, Cummings, & Fitzgerald, 2018) identifies areas of concerns that the offender holds (e.g., preparing for an interview), and can be used to identify areas where correctional staff can provide support on prior to release (e.g., mock interviews).

Strengthening the employment discrimination laws in Canada or adding criminal history as a special status that should not be discriminated upon in the Canadian Charter of Rights and Freedoms could also improve employment outcomes of individuals with criminal records. Currently, only eight provinces and territories have criminal offences as a special status that should not be discriminated against. Of these, two provinces and territories place this special designation on pardoned or suspended offences, therefore allowing criminal offences to be considered by employers. The current study found that these anti-discrimination laws in Canada have no positive effect on the employment rate or income of Canadian federal offenders, perhaps due to ambiguity in the legislation. There are two possible reasons for this. First, despite the legislations, employers are continuing to discriminate against individuals with criminal records. Second, provinces and territories without employment discrimination laws (prairies, Nova Scotia, New Brunswick) may have greater access to employment opportunities that do not require criminal record checks (oil sands, farming, and fishing), which are traditionally more often male dominated. The latter is likely why we found that men tend to use less social assistance in these provinces than women. Regardless, the current study suggests that more can be done in terms of protecting the rights of individuals with criminal records to gain employment.

In response to the dismal economic outcome of offenders, many countries have supported 'Ban the Box' legislation, which bans employers from asking applicants about whether or not they have a criminal record, particularly when the position does not involve access to vulnerable populations, such as children, individuals with disabilities, and the elderly. In the United States, there are over 130 cities and 24 states that have removed questions about criminal history from public employment applications, and 9 of these states have also applied this ban to employers within the private sector (Evans, 2016; Von Bergen & Bressler, 2016). Hawaii found that the rate of reoffending decreased following the Ban the Box law (D'Alessio, Stolzenberg, & Flexon, 2015). The current study suggests that an approach that mimics features of the ban the box law would yield fruitful results, as it would provide individuals with a criminal record greater opportunity to reintegrate into the community and to increase self-reliance that stems from legal employment. Requiring criminal record checks uniformly for any type of job application means that individuals with criminal records will have notably increased challenges when seeking employment, which would add to the many other potential hurdles offenders are facing upon release (e.g., securing housing, substance abuse, health; Henry & Jacobs, 2007). Given the simple fact that holding a criminal record biases employers toward rejecting applicants, regardless of competencies and qualifications (Batastini et al., 2017), a ban across preliminary record verification would be required to make a substantial change in employment income of offenders.

In Canada, we currently have a policy in place for all employment that involves vulnerable populations (e.g., Vulnerable Sector Check; RCMP, 2014). Banning criminal record checks for non-government agencies would not mean that those serving vulnerable sectors would be impacted. Instead, organizations within the vulnerable sectors and government agencies would continue to require record checks. Employment is in fact a key protective factor for the desistance of criminal offending (Andrews & Bonta, 2010). Unfortunately, the current study has shown that there are a number of considerable barriers that individuals with criminal records encounter with regards to gaining employment in Canada, barriers that are impacting women, Indigenous, and older individuals differentially. Policies to improve the economic outcomes of individuals with criminal records would not only reduce the substantial cost to Canadians (e.g., decrease social assistance payments, increase of revenue through taxation), but improved policies would also reduce the likelihood of reoffending.

References

AAltonen, M. (2016). Post-release employment of desisting inmates. British Journal of Criminology, 56, 350-369. doi: 10.1093/bjc/azv047

Andrews, D. A., & Bonta, J. (2010). The psychology of criminal conduct. New York, NY: Routledge.

Bailey, Z. D., Krieger, N., Agenor, M., Graves, J., Linos, N., & Bassett, M. T. (2017). Structural racism and health inequalities in the USA: Evidence and interventions. The Lancet, 389, 1453-1463. doi: 10.1016/S0140-6736(17)30569-X

Batastini, A. B., Bolanos, A. D., Morgan, R. D., & Mitchell, S. M. (2017). Bias in hiring applicants with mental illness and criminal justice involvement: A follow-up study with employers. Criminal Justice and Behavior, 44(6), 777-795. doi: 10.1177/0093854817693663

Berg, M. T., & Huebner, B. M. (2011). Reentry and the ties that bind: An examination of social ties, employment, and recidivism. Justice Quarterly, 28(2), 382-410. doi: 10.1080/07418825.2010.498383

Brown, S.L., & Motiuk, L.L. (2005). The Dynamic Factors Identification and Analysis (DFIA) Component of the Offender Intake Assessment (OIA) Process: A meta-analytic, psychometric and consultative review (Research Report R-164). Ottawa, ON: Correctional Service of Canada.

Campbell, M. A., French, S., & Gendreau, P. (2009). The prediction of violence in adult offenders: A meta-analytic comparison of instruments and methods of assessment. Criminal Justice and Behavior, 36(6), 567-590. doi: 10.1177/0093854809333610

Correctional Service of Canada (2018, June). How offenders benefit from CORCAN apprenticeships. Available from http://www.lte-ene.ca/en/how-offenders-benefit-corcan-apprenticeships-1

D'Alessio, S. J., Stolzenberg, L., & Flexon, J. L. (2015). The effect of Hawaii's Ban the Box law on repeat offending. American Journal of Criminal Justice, 40(2), 336-352. doi: 10.1007/s12103-014-9251-9

Department of Justice Canada (2017). JustFacts: Indigenous overrepresentation in the criminal justice system. Author: Ottawa, Ontario. Available from https://www.justice.gc.ca/eng/rp-pr/jr/jf-pf/2017/docs/jan02.pdf 

Doward, J. (2017, November). 'Irrelevant' criminal record checks harm ex-offenders' job hopes. The Guardian. Available from https://www.theguardian.com/uk-news/2017/nov/25/irrelevant-criminal-record-checks-harm-job-hopes

Duwe, G., & Clark, V. A. (2017). Nothing will work unless you did: The predictors of postprison employment. Criminal Justice and Behavior, 44(5), 657-677. doi: 10.1177/0093854816689104

Evans, L. (2016). Ban the Box in employment: A grassroots history. San Francisco, CA: Legal Services for Prisoners with Children. Available from http://www.prisonerswithchildren.org/wp-content/uploads/2016/10/BTB-Employment-History-Report-2016.pdf

Field, A. (2009). Discovering statistics using IBM SPSS statistics. Thousand Oaks, CA: Sage.

Freudenberg, N., Daniels, J., Crum, M., Perkins, T., & Richie, B. E. (2005). Coming home from jail: The social and health consequences of community re-entry for women, male adolescents, and their families and communities. American Journal of Public Health, 95(10), 1725-1736. doi:10.2105/AJPH.2004.056325

Gillis, C. (2002). Understanding employment: A prospective exploration of factors linked to community-based employment among federal offenders. Doctoral dissertation. Available from https://curve.carleton.ca/e72b2626-5f71-4a45-b48f-655f6c681b79

Gillis, C., & Nafekh, M. (2005). The impact of community-based employment on offender reintegration. Forum on Corrections Research, 17(1), 10-14.

Harding, D. J., Wyse, J. J., Dobson, C., & Morenoff, J. D. (2011). Making ends meet after prison: How former prisoners use employment, social support, public benefits, and crime to meet their basic material needs. Ann Arbor, MI: University of Michigan.

Helmus, L., & Forrester, T. K. (2014). Static Factors Assessment (SFA) in the Offender Intake Assessment in the Offender Intake Assessment Process: Relationship to Release and Community Outcomes (Research Report R-339). Ottawa, ON: Correctional Service of Canada.

Henry, J. S., & Jacobs, J. B. (2007). Ban the Box to promote ex‐offender employment. Criminology & Public Policy, 6(4), 755-761.

Nafekh, M., & Motiuk, L. L. (2002). The Statistical Information on Recidivism, Revised 1 (SIR-R1) scale: A psychometric examination. Ottawa, Ontario: Correctional Service of Canada. Research Branch.

Nally, J., Lockwood, S., Knutson, K., & Taiping, Ho. (2013). The marginally employed offender: A unique phenomenon among released offenders. The Journal of Correctional Education, 64(1), 50-68.

Nuffield, J. (1982). Parole decision-making in Canada: Research towards decision guidelines. Ottawa, Canada: Communication Division, Solicitor General of Canada.

Pager, D., & Shepherd, H. (2008). The sociology of discrimination: Racial discrimination in employment, housing, credit, and consumer markets. Annual Review of Sociology, 34, 181-209.

Pager, D. (2003). The mark of a criminal record. American Journal of Sociology, 108(5), 937-975. doi: 10.1086/374403

Petersilia, J. (2001). When prisoners return to communities: Political, economic, and social consequences. Federal Probation, 65(1), 3-8.

Pijoan, E. L. (2014). Legal protections against criminal background checks in Europe. Punishment & Society, 16(1), 50-73. doi: 10.1177/1462474513506031

Powell, B., & Winsa, P. (2008, July 22). Criminal record: The stain that won't go away. The Star. Available from https://www.thestar.com/news/crime/crime/2008/07/22/criminal_record_the_stain_that_wont_go_away.html

RCMP (2014). Dissemination of criminal record information policy. Ottawa, Ontario. Author: RCMP. Available from http://www.rcmp-grc.gc.ca/en/dissemination-criminal-record-information-policy

Schmitt, J., & Warner, K. (2010). The changing face of US labor, 1983–2008. Journal of Labor and Society, 13(2), 263-279. doi: 10.1111/j.1743-4580.2010.00286.x

Statistics Canada (2005). CANSIM Table 17-10-0005-01 Population estimates on July 1st, by age and sex. Available from https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1710000501

Statistics Canada (2014). CANSIM Table 11-10-0004-01 Selected characteristics of tax filers and dependants, income and demographics (final T1 Family File). Available from https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1110000401

Statistics Canada (2014). CANSIM Table 11-10-0007-01 Tax filers and dependants with income by source of income. Available from https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1110000701

Statistics Canada (2014). CANSIM Table 11-10-009-01 Selected income characteristics of census families by family type. Available from https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1110000901

Statistics Canada (2014). CANSIM Table 11-10-0023-01 Tax filers and dependants 15 years of age and over with labour income by sex and age. Available from https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1110002301

Stewart, L.A., Wardrop, K., Wilton, G., Thompson, J., Derkzen, D., & Motiuk, L.L. (2017). Reliability and validity of the Dynamic Factors Identification and Analysis – Revised. (Research Report R-395). Ottawa, ON: Correctional Service of Canada.

Varghese, F. P., Anderson, K. M., Cummings, D. L., & Fitzgerald, E. (2018). The offender job search selfefficacy scale: Development and initial validation. Psychological Services, 15, 477-485. doi:10.1037/ser0000170

Visher, C. A., Debus-Sherrill, S. A., & Yahner, J. (2011). Employment after prison: A longitudinal study of former prisoners. Justice Quarterly, 28(5), 698-718. doi: 10.1080/07418825.2010.535553

Visher, C., Debus, S., & Yahner, J. (2008). Employment after prison: A longitudinal study of releases in three states. Research Brief, 1, 1-9.

Von Bergen, C. W., & Bressler, M. S. (2016). "Ban the Box" gives ex-offenders a fresh start in securing employment. Labor Law Journal, 67(2), 383-395.

Webster, J. M., Staton-Tindall, M., Duvall, J. L., Garriety, T. F., & Leukefeld, C. G. (2007). Measuring employment among substance-using offenders. Substance Use & Misuse, 42, 1187-1205. doi: 10.1080/10826080701409800

Yang, M., Wong, S. C., & Coid, J. (2010). The efficacy of violence prediction: A meta-analytic comparison of nine risk assessment tools. Psychological Bulletin, 136(5), 740-767. doi: 10.1037/a0020473

Appendix A

Employment Discrimination Status Across Canada

Appendix B

Note. EXP(b):Standardized beta. Bolder values represent predictors that were statistically significant (p <.05). For dichotomous items, reference category is provided in parenthesis.

Note. N = 5,474 men, R² = .172; N = 345 women, R² = .126. EXP(b):Standardized beta. Bolder values represent predictors that were statistically significant (p <.05). For dichotomous items, reference category is provided in parenthesis.

Note. B =Standardized beta. Bolder values represent predictors that were statistically significant (p <.05). For dichotomous items, reference category is provided in parenthesis.

Note. EXP(b) =Standardized beta. Bolder values represent predictors that were statistically significant (p <.05). For dichotomous items, reference category is provided in parenthesis.

Date modified:

2022-07-25