It might seem obvious that unemployment and crime are related and that one causes the other. Yet the literature on this topic shows that the relationship is much more complex and nuanced than it might seem at first. It is essential that policymakers understand these nuances as they develop both economic and criminal justice policy.

We examine three major theoretical frameworks:

• Economic choice theory argues that crime and unemployment are causally linked because employment, particularly high-paid employment, acts as a potential opportunity cost that an individual must weigh in the decision to carry out a crime.
• Social control theory agrees that unemployment and crime are causally linked, but for a different reason. It argues that what deters someone from committing a crime is the social bonds that they have with members of their family and, importantly for this discussion, their workplace.
• Self-control theory, on the other hand, rejects the notion of a causal link between crime and unemployment, arguing that an individual’s life course is determined early in childhood by whether or not they develop self-control. According to this view, both crime and unemployment are manifestations of this character trait.

We also briefly touch on other theories, including strain theory, learning theory, and labelling theory.

The truth is likely to be found in a synthesis of these ideas. Indeed, the literature has produced data in support of all of these theories. Despite an early “consensus of doubt” among researchers who investigated the relationship between crime and unemployment, recent work has suggested that, on the whole, there is a statistically significant relationship between the two. That said, there are important distinctions to be aware of:

• The strength of the relationship depends greatly on whether one is examining data at the national, community, or individual levels.
• Research shows that the relationship is much stronger when property crime is tested separately from violent crime.
• There are also important nuances relating to the age of offenders: while crime tends to be positively correlated with unemployment for adults, there is evidence that it is negatively correlated with unemployment for adolescents.

All this comes to a head in recidivism and the capacity of employment to lead ex-offenders away from the commission of future crimes. There has been a push in recent years to ban employers from asking about criminal history early in the job application process. The idea is to shield applicants from the stigma associated with their criminal records. However, recent research shows evidence that, in the absence of information about individual criminal histories, some employers discriminate against populations that are statistically more likely to have a criminal record, such as some racial minorities. So-called ban-the-box initiatives therefore remain controversial and potentially counterproductive. Employment nevertheless has the potential to offer hope to those who have been caught up in the justice system. Therefore, we believe that there is ample room for more research and debate on ban-the-box initiatives and alternative policies to increase the employability of those with criminal records. Cardus intends to develop such alternative policy options that will be informed by the research in this paper.

Work is often seen as giving people a purpose in life and described as having dignity beyond the benefit of a paycheque. Previous Cardus research has demonstrated that those who have a job are more likely to be in good physical and mental health, less likely to commit suicide or self-harm, and generally have increased life satisfaction compared to those people who do not. Unemployment can have negative consequences, such as an increase in family conflict and divorce, domestic abuse and neglect, and alcohol and substance abuse (Dijkema and Gunderson 2019). It is not surprising, therefore, that our intuition would make the connection between crime and unemployment.

The Canadian government estimated in 2019 that over one in ten Canadians had a criminal record, totalling nearly four million people across Canada (Public Safety Canada 2019). 1 1 The statistics are difficult to determine. The Canadian Police Information Centre may overestimate the number in some cases because its database includes tourists and deceased Canadians. It may also underestimate the number because there may be records in local police services that are not (yet) reported. This is a large constituency of people, nearly equivalent in size to the population of Alberta. If anything, this number understates those who are affected by crime, as it does not include the victims of crime nor the families of both victims and perpetrators. The sheer size of this population ought to move society to investigate the causes of crime and seek to reduce its effects.

This paper examines whether there is a link between employment and crime, and if so, what the nature of this link may be. This is a crucial question for public policy. On the one hand, government financial resources are scarce, so their expenditure on pro-employment policies for those at risk of committing crimes must be tested and justified on the basis of sound analysis. On the other hand, if such a basis can be established, then a pro-employment policy for this group of people may be a vital component of both an economic and a criminal justice agenda.

The paper identifies a link and goes on to provide policymakers with an overview of how and why unemployment and crime are linked. While it might seem obvious at first blush that unemployment causes crime, and vice versa, the academic literature on the topic is much more complex and nuanced than one might think. It is important for policymakers to have a proper understanding of the theoretical literature, because their policies must be grounded in a clear-eyed view of human nature. The data examined in this paper show that the relationship manifests differently based on a number of different factors, including the type of crime, the type of unemployment, and the age of the offender. All of these factors are important for policymakers to understand so that they can enact the right kinds of policies and target them accurately toward those who would be helped the most by them.

This is also a crucial question for employers, who, as the hiring decision-makers in our economy, are on the front lines of this issue. As we shall see in this paper, many employers’ perceptions of those with previous involvement in the criminal justice system do not match the reality. Correcting this misperception would open these businesses up to an under-tapped pool of labour. Cardus intends to publish further research, including policy recommendations, on promoting employment to reduce recidivism. This research paper, therefore, provides the theoretical justification for this future work.

But even more fundamentally, promoting employment among ex-offenders is a profoundly moral responsibility. The previously incarcerated are, by definition, those who have paid their debt to society for the crimes they have committed. They ought to have the opportunity, following their release from prison, to rebuild their lives, to contribute to society in constructive ways, and to obtain the dignity and satisfaction that come from meaningful work.

This paper does not purport to address other factors that affect crime, such as family strife, housing issues, poverty, and drug addiction, except tangentially when the occasion requires it. While this paper is a contribution along one track of inquiry—the economic track—it is clear that the factors that lead an individual into the commission of a crime are varied and complex.

The economic track is nonetheless important. The primary purpose of this report is to arm policymakers and employers with facts about the relationship between crime and unemployment. As such, this paper is a survey of some of the research that has already been done on the intersection of employment and crime. It is not intended to be an exhaustive review, as the literature is vast and deep. It does, however, seek to be representative of major strains of thought that have influenced contemporary thinking on this topic. The paper also relies heavily on studies conducted in the United States because researchers there have done more work on this topic. While there are differences between the United States and Canada, the societies are not so different that certain basic conclusions about the nature of work and its effect on crime cannot be drawn.

The first section is an exposition of the key theoretical frameworks and how they have attempted to explain how unemployment may (or may not) relate to criminal behaviour. The second section investigates some of the empirical research that has been conducted to determine the nature of the relationship between crime and employment. The third and final section takes stock of these results to consider the particular relationship between employment and recidivism through a case study on ban-the-box initiatives, a pro-employment policy targeted at ex-offenders that has garnered significant popularity among policymakers in the United States.

We begin with a brief overview of the major theories that have influenced the literature on the relationship between crime and employment. Each theory tells a different story about whether and, if so, how, crime and employment are related to each other. Each therefore recommends different paths for policy.

We have converged on three theories: economic choice theory, social control theory, and self-control theory. Each represents a significant contribution to how scholars have conceived of the relationship between employment and crime. While all three expect to find a positive correlation between unemployment and crime in one form or another, they differ considerably in their explanations of this relationship, with self-control theory even rejecting the idea that one causes the other at all.

While these three theories represent important frameworks for our discussion, some may argue that the discussion below is incomplete. Taxonomies of the theories of crime vary from one source to another, with authors placing a greater emphasis on some than others (cf. Apel 2009; Bushway and Reuter 1997; and Pratt and Cullen 2005). There being no consensus list of theories that captures all the different theoretical frameworks, the overview below seeks to emphasize those theories that have garnered the most recent attention of scholars, while also acknowledging others that are worthy of mention. At the end of this section, we highlight a handful of other theories—namely, strain theory, learning theory, and labelling theory—of which readers should be aware.

The seminal paper on modern economic choice theory is Becker (1968), although the application of economic theory to criminology goes back, as Becker himself points out, to Jeremy Bentham in the early nineteenth century and even to Cesare Beccaria in the eighteenth. Becker’s theory has profoundly influenced the thinking about the crime-employment relationship, with Bushway and Reuter (2002, 191) noting that economic choice theory is the “theoretical justification . . . that receives the most attention.”

Based on classical liberal economic theory, economic choice theory posits that the existence and prevalence of crime can best be understood as the result of marginal decision-making of rational actors seeking to maximize pleasure and minimize pain. According to this view, crime can be a rational response to a weighing of cost, benefit, and risk. An individual’s decision to commit a crime would take into consideration a number of factors, such as:

• What the individual hopes to gain from the commission of the crime (the benefit);
• The time, effort, and other resources associated with committing the crime (an immediate cost);
• The likelihood of getting caught (a risk calculation); and
• The legal penalty for committing the crime, whether a fine or incarceration, including any forgone income during a period of incarceration (a potential cost).

According to this theory, if the expected benefit of the crime is higher than the combination of its risk and cost, then the crime is a rational act. If not, then the rational individual would not commit the crime. Economic choice theory therefore sees crime primarily as the result of the interaction of incentives acting on the individual. As Chiricos (1987, 203) puts it, “Work and crime are the principal alternatives for most people to generate an income.”

This theory may provide a useful framework for understanding how employment and crime may interact. In this analysis, employment acts as an opportunity cost of crime. For example, going to prison removes from the individual the opportunity to generate income in a job, effectively raising the stakes in the fourth factor listed above. The better the employment, the higher the opportunity cost of going to prison. Employment, especially well-compensated employment, therefore, is a disincentive to crime. The greater the compensation, the higher the potential cost, and the more the job acts as a disincentive to crime.

A weakness of this theory is that it helps to explain some crimes better than others. In their summary of the various theories about crime and unemployment, Bushway and Reuter (1997, 6–4) argue that while economic choice theory may provide an account for property crime that provides an income that could otherwise be obtained through legal means, “the theory offers no account of non-income generating crime,” such as domestic abuse or sexual violence.

Pushing the point further, Bushway and Reuter distinguish between “instrumental” and “expressive” crime, using physical assault as an example. If the assault is a means to an end, perhaps the collateral damage of a robbery, then it is “instrumental” in the commission of another offence. Economic choice theory would attempt to explain the crime by showing that the benefit of the robbery to the criminal outweighs the penalty associated with assault and robbery balanced off with the risk of getting caught. On the other hand, the assault could be a crime of passion, the manifestation of a deep rage brought on by an unexpected situation, the “expression” of an emotional outburst. Economic choice theory would have a much more difficult time explaining this, as there is no income to be generated by the crime. Even the nature of a crime of passion would seem to discount the possibility of a rational calculation of cost and benefit.

Bushway and Reuter acknowledge that economic choice theory may nevertheless go some way in explaining “expressive” crimes. While there may not be any income to be gained from crimes of this nature, there is income to be lost through the penalties associated with the crime, whether directly through fines or indirectly through forgone income during periods of incarceration or through lost social capital. Thus the negative consequences associated with a crime may stay an individual’s hand and deter them from the commission of an offence, even from the commission of a crime of passion.

This explanation comes back to the opportunity cost of crime. Economic choice theorists posit that, even in expressive crime, those with gainful employment are less likely to engage in it because they have more income or potential income to lose. Proponents of this theory would therefore expect to find that crime has a strong correlation with unemployment and with wages, as the direct result of the economic incentives acting on the utility-maximizing individual.

Where economic choice theory emphasizes the rational decision-making of the individual, social control theory emphasizes the social bonds among persons. Rather than starting with the individual as such, social control theorists see the human being as essentially placed within a social or institutional setting. The relationships with others that emerge from these settings condition the individual to conform to social norms to greater or lesser degrees. The nature and the strength of those ties determine the nature and strength of the conditioning. According to Sampson and Laub (1993, 18), social control theory argues that “crime and deviance result when an individual’s bond to society is weak or broken.”

Arguing against the idea that crime can be understood in economic terms, Toby (1957, 13) points to the “trifling” material benefits accrued in a life of crime. The “Al Capones” of the world “who steal because enormous profits outweigh the risks of apprehension” (a clear reference to the thought behind the economic choice theory of crime), are exceedingly rare. The reality of the crime world, full of petty crime and “seemingly irrational behavior,” cannot be explained by criminals weighing the costs and benefits.

Instead, Toby seeks to explain it by the social environments in which individuals find themselves. Those from “good” families and “respectable” neighbourhoods have a “stake in conformity” to the social norms of those communities (16). Losing that stake by violating the norms of the community is a powerful deterrent to acting out criminally. By contrast, those who hail from families and neighbourhoods lower down the socioeconomic ladder do not feel the same stake in conformity and thus have less to lose. In fact, an adolescent who falls into the wrong crowd may find that their stake in conformity rests in criminal activity. “For him, stealing is not primarily a way to make money,” says Toby. “It is primarily a means of gaining approval within a clique of outcasts” (15).

Sampson and Laub (1993) developed a modern and influential interpretation of social control theory, employing the concept of social capital to refine its earlier interpretations. 2 2 For now, we consider Sampson and Laub as proponents of social control theory. That said, they see their theory as a synthesis of several theories, as we discuss below. According to this theory, when people forge bonds with others, they become more invested in each other. They also become more invested in the institutions in which these social bonds are created. These social bonds then hold people back from actions, such as crime, that would break them. Stronger bonds mean less crime and weaker bonds mean more.

Sampson and Laub identified two institutions that they believed were powerful in creating interdependent social bonds: the family and the workplace. They suggest that it is not the mere participation in these institutions that is important for keeping individuals from falling into crime. Rather, it is the quality and strength of those bonds that are the important determining factors. Thus, simply being married does not prevent anti-social behaviour; having a loving, committed, and stable marriage does. The same is true, they argued, for the workplace:

Employment alone does not increase social control. It is employment coupled with job stability, job commitment, and mutual ties to work (that is, employee-employer interdependence) that should increase social control and, all else being equal, lead to a reduction in criminal and deviant behavior (140).

According to this theory, the workplace is “a potent source of informal social control” (Apel 2009), because the prospect of losing a job and the social standing that comes from it is a deterrent to the commission of a crime. The better the job and the higher the social standing, the stronger the disincentive. Conversely, the loss of employment causes “a breakdown of positive social bonds” and can lead one to crime (Bushway and Reuter 1997, 6–4). Toby (1957) points out that a similar phenomenon happens with adolescents who drop out of school: they tend to get unskilled jobs, leading them to switch jobs frequently and “lose interest in steady work” (15). 3 3 Note that the individuals described here would fall in the working class as defined by Cardus’s recent report “Canada’s New Working Class.” This report provides a deeper analysis of the precarity of work among those in jobs that do not require post-secondary credentials and offers some policy solutions to address this issue. See Speer, Bezu, and Nauta (2022). This then leads the person further to lose a stake in conformity.

A benefit of this theory is that its explanatory power is broader, in terms of the different kinds of crime, than economic choice theory. Yet the two theories nonetheless bear a strong similarity, namely, that both attribute the avoidance of crime to the fear (or cost) of the loss of something else. Both present the benefit of employment as an opportunity cost of crime, although one sees the benefit of employment primarily as income, while the other sees it primarily as a source of social capital. Importantly, both theories understand unemployment as having a causal link with crime.

In contrast to the two theories examined above, self-control theory contests the notion that there is a causal link between crime and unemployment at all, arguing that they are both manifestations of a third, more fundamental character trait.

The main proponents of this theory, Gottfredson and Hirschi (1990, 85), place it in diametric opposition to the theories discussed above, contrasting their idea with “theories emphasizing the prevention of crime through consequences painful to the individual.” They deny the contention shared by the above theories that the contemporaneous circumstances of an adult’s life have much of anything to do with their propensity to commit crime.

Gottfredson and Hirschi argue instead that the cause of criminality is to be found in an individual’s capacity (or lack thereof) for self-control. They note that criminal acts usually offer “immediate gratification” but “few or meager long-term benefits.” They are also usually “easy or simple” and require “little skill or planning” (89). These are the hallmarks of an act committed by an individual who lacks the capacity to delay gratification. Moreover, they state that criminal behaviour is merely one possible manifestation of a lack of self-control. Some with low self-control may not manifest criminal behaviour but instead “tend to smoke, drink, use drugs, gamble, have children out of wedlock, and engage in illicit sex” (90).

Crucially, in Gottfredson and Hirschi’s view, persons with low self-control also tend to be unemployed, but there is no direct relationship between crime and unemployment. Rather, the two are related to each other through the determining variable of self-control. They place a great deal of importance on a famous empirical study by Sheldon and Eleanor Glueck, which is discussed further below, and conclude that “employment does not explain, or help to explain, the reduction in crime with age, and that it is not relevant to theories that differentiate between offenders and non-offenders” (139).

Furthermore, they believe that self-control is a character trait that is developed quite early in life. This leads them to argue essentially that there is little hope for those with low self-control to succeed in the labour market, going so far as to say that these individuals will find the rigours of “good” work too restrictive for their natures. Any attempts at rehabilitation would have to target the offender’s self-control and would have to occur “very early in development.” Unfortunately, by the time an offender is in a position to receive treatment, it is, in their view, almost certainly too late to effect any meaningful change. Interventions should focus instead on parents and caregivers to instill the virtue of self-control in their children. As for the criminal justice system, they believe “it is unlikely that . . . rehabilitation programs will themselves reduce criminal behavior sufficiently to justify their cost” (269).

It is important to note again that this theory would explain a correlation between crime and unemployment not as the result of a causal relationship between the two but rather as the result of a shared correlation with a third variable: low self-control. Therefore, this theory would expect that the correlation would disappear if this third variable is controlled for.

This exposition of theories of crime as they relate to employment has necessarily been non-exhaustive. It nevertheless provides a sufficient basis to initiate a discussion of the empirical evidence. Before proceeding to the data, though, it is worth noting a few observations.

At first glance, many of these theories may seem reductive or to fail in capturing the complexity of the human condition. For example, economic choice theory seems to reduce human beings to pleasure-maximizing economic agents, while self-control theory seems to reduce adult decision-making to pathways pre-determined in childhood. For this reason, it is better to understand each of these theories as explaining a different aspect of human nature. In this way, we can begin to see how these theories can work together in explaining what makes a given individual more likely to go down a path of crime.

There are some similarities in these ostensibly opposing theoretical frameworks. For example, notice that both the economic choice theory and the social control theory place a value on the quality of a job in determining an individual’s likelihood to commit crime, though in different ways. The economic choice theory would place emphasis on the quality of the job in terms of financial compensation: the higher the compensation, the more a potential criminal has to lose from being caught in the commission of a crime. For social control theory, the quality of the job has more to do with the quantity and quality of the relationships and human interactions it provides. In both theories, however, there is an element of a “stake” that a potential criminal could lose if they commit a crime, and the fear of losing that stake is what keeps them from doing so.

Perhaps these two theories are not so different. It is possible, for example, to broaden the understanding of economic choice theory to include the non-financial benefits of employment. The stake in conformity of social control theory can be understood as one of the non-financial benefits of work. In this view, anything can have subjective value to an individual, including intangible assets that do not have a price per se. Thus, the stake in conformity can be seen as an economic good of some value. From this broader economic perspective, a worker who is tempted to commit a crime would weigh the benefits of the crime against all manner of costs, including the non-financial cost of losing their standing among their peers. This is one important aspect of social control theory. The two theories may therefore not be as far apart as they might seem, if we understand the social bonds created in the workplace as having some economic value that can be weighed against other economic goods.

This stands to reason. Human beings are complex, with intricate, sometimes competing, sets of motivations. It is difficult if not impossible to boil down human motivations to a single cause. Our own experiences surely suggest that it is difficult to discern even for our own selves the extent to which each of our personal impulses motivates any given action. The same may very well be true in the pathway of an individual toward the commission of a crime. Elements of the three main theories discussed here could co-exist. For example, it is possible that some people, having grown up in an environment that did not instill self-control, find themselves in an isolated social existence with a low-paying job and with no prospect for advancement. Such a situation would represent the confluence of the main factors identified by these three theories and may represent the “perfect storm” that may lead someone toward crime.

This is, in fact, the view of Sampson and Laub (1993). Their theory is in direct response to the self-control theory of Gottfredson and Hirschi (1990), whose contributions they take seriously and incorporate into their own thinking. Sampson and Laub accept the contention that behaviours learned in early childhood have a strong predictive power for behaviours in adulthood, adding that these behaviours are explained by the family and school contexts in which children grow up. Where they differ from Gottfredson and Hirschi is that they reject the notion that adult social bonds have no relationship with crime. They argue that the data (which is considered further below) cannot be explained entirely by “early delinquency,” suggesting that there are other factors—that is, adult social bonds—at play in crime committed as an adult.

Sampson and Laub therefore present their theory as a synthesis of the theoretical literature. Of course, the abstracted versions of these theories are in contradiction with each other on the question of whether unemployment causes crime, so they cannot co-exist in their pure forms. Nevertheless, it is probable that no one theory can account for the totality of the human propensity for crime, but that elements of each might help to explain it.

There are other theories that are worth mentioning in brief (cf. Apel 2009; Bushway and Reuter 1997).

• Strain theory suggests that social pressures force some individuals to commit crimes. The unemployed and those in low-paying jobs resort to crime to achieve the expectation of a quality of life that has been imposed by society.
• Learning theory argues that individuals learn criminal behaviour by associating with other criminals. According to this view, a stable job in a good workplace provides an environment in which an individual is unlikely to be associating with criminals. Therefore, work is favourable to non-criminal behaviour and could keep someone out of a life of crime.
• Labelling theory suggests that there is a negative feedback loop between crime and unemployment. A criminal acquires a stigmatic label that makes it difficult to participate normally in the labour force (Bushway and Reuter 1997). This, in turn, pushes them further into a life of crime, continuing a vicious circle of unemployed criminality.

Note that each of these theories would expect a causal relationship between unemployment and crime, similar to economic choice theory and social control theory. It is also important to note that they are not necessarily at odds with the other theories. For instance, in our hypothetical example of someone in the “perfect storm” described above, it is conceivable that their situation could be made even worse by having heavy social expectations placed on them, being associated with criminals, and carrying the stigma of a prior conviction. It is possible, therefore, to see the elements of these theories of crime as factors that compound to push someone further and further along the road to crime.

All theories outlined above would expect to find a link between employment and crime, though they would each take a different perspective on whether, and if so, how, one causes the other. In this section, we examine some of the empirical studies that have tested this relationship. As one might expect, the real world is more complicated than these theories might suggest, but we can still identify some trends in the data and begin to draw certain conclusions about the relationship between crime and employment.

If there is one thing on which all the theories outlined above agree, it is that we would expect to find a correlation between unemployment and crime, although self-control theory would expect it to disappear if certain variables were controlled for. But even this has been difficult to establish. While the most current research strongly suggests that there is a significant, positive relationship between the two, it took some time for scholars to arrive there. Early research in this area led to what Chiricos (1987, 188) calls “a consensus of doubt” regarding the relationship between unemployment and crime, with the scholarship calling “into question the strength, the significance, and even the direction of the [unemployment-crime] relationship.” He argues, however, that this consensus of doubt was premature and based on weak methodology.

Chiricos was onto something, because since then new studies using more refined methodologies tell a different story. Mustard (2010, 1), for example, notes that since the late 1990s scholars have made “substantial progress in resolving this disconnect between the theory and empirics.” Mustard examines a number of studies from that period and shows that, by controlling for a larger number of variables and using more modern statistical methodologies, this “new generation of studies” uncovered the relationship one would expect from theory (27). We turn to some of these developments now.

One important development was to separate the research into distinct levels of analysis. Bushway and Reuter (1997) identify three:

• Research at the national level examines how macroeconomic data on labour markets (e.g., the unemployment rate) track aggregate national data on crime rates.
• Research at the community level looks at labour-market data for municipalities or neighbourhoods and how they track that community’s crime rates.
• Research at the individual level examines data on individual persons to see whether their employment status tracks their propensity to commit crime.

Bushway and Reuter’s survey reveals that the strength of the relationship depends on what level is being examined. There is at best a weak connection at the national level, but stronger results appear when studies examine the other two levels. A majority of studies that they reviewed observe a relationship between crime and employment at the individual level. Studies show the strongest link at the community level, with between half and three-quarters finding a statistically significant correlation between crime and employment.

It may be surprising that Bushway and Reuter do not find a stronger relationship at the individual level. After all, the criminological theories outlined in this paper, especially the economic choice theory, suggest a relationship between the two variables at the individual level. If the relationship exists primarily at the community level, this would suggest that the employment level of one’s neighbours is more determinative of one’s propensity to commit crime than one’s own employment status.

Yet it would be a mistake to downplay the relationship at the individual level. In another review a few years later, these researchers acknowledged that “fairly strong evidence indicates that an individual’s criminal behavior is responsive to changes in his or her employment status, independent of what is occurring with the demand for labor at a macro level” (Bushway and Reuter 2002, 206). Similarly, Apel (2009) notes that while “the strength of the employment-crime correlation is not nearly as impressive as a number of theoretical accounts would suggest” (123), the empirical studies still “confirm the expectation from a variety of theories that having a job is associated with less crime than not having a job” (120).

Chiricos (1987) finds similar results in his review of the empirical literature. He points out that, contrary to the “consensus of doubt” against which he was arguing, the strength of the relationship between crime and unemployment depended on which level was being examined, with stronger results being found in studies that zeroed in on micro levels of analysis. Furthermore, Mustard’s (2010) review of the literature concludes that “research that utilizes smaller units of analysis generally shows a tighter nexus between labor markets and crime” (8). He reports that “nearly every study that uses panel data,” that is, data on the same individuals over time, “finds economically and statistically significant relationships between labor markets and crime” (25).

We are left with the question of why the national-level statistics may not show a strong correlation between crime and unemployment. After all, if the relationship between crime and unemployment holds at the individual level, then we would expect to find that having more people unemployed would mean more people with a greater likelihood of committing a crime. Yet the data don’t provide as strong support for this. Cantor and Land (1985) propose one explanation. They note that there are two effects of an increasing unemployment rate: a “criminal motivation effect” and a “criminal opportunity effect” (317). The former is the effect that is predicted by the theoretical frameworks outlined above, that is, that unemployment would bring about stronger incentives to commit crimes. The latter suggests that because the unemployment rate rises during economic downturns, there are fewer worthwhile targets (or “opportunities”) for property crime. The criminal opportunity effect, therefore, attenuates the criminal motivation effect, meaning that the expected correlation would be weaker than if only the latter were considered. According to Cantor and Land, this distinction helps to explain the empirical studies which to that point had failed to achieve a consensus on unemployment and crime. Importantly, it maintains the fundamental insight from the theories above that expect criminal motivations to increase as unemployment rises.

A recurring theme in the literature is that the effect of employment on crime depends heavily on the kind of crime being investigated. When property crimes are analyzed separately from violent crimes, the results can be starkly different.

In their review of the literature, Bushway and Reuter (1997) note that while only 50 percent of studies in general showed a statistically significant relationship between crime and unemployment, that ratio shot up to 75 percent when property crime was isolated in the study. Similarly, 90 percent of the studies surveyed by Chiricos (1987) find a positive relationship between unemployment and crimes such as burglary, larceny, and property crime, with over 60 percent of them also showing a statistically significant positive relationship. By contrast, studies considering assault and violent crimes tend to show a negative relationship with unemployment, that is, that an increase in unemployment is associated with a decrease in violent crime. That said, very few (at most, 3 percent of the studies) found that this negative relationship was statistically significant. Raphael and Winter-Ebmer (2001, 280) find that studies “consistently indicate that unemployment is an important determinant of property crime rates” but that with the exception of violent robbery, they “did not find such consistency for violent crimes.”

Taken together, the literature suggests that property crime is more strongly correlated with employment than other types of crime, such as violent crime. We saw in our earlier discussion that the economic choice theory predicts a stronger connection between unemployment and crimes that seek to replace income, particularly property crime, than between employment and other kinds of crime. That there is a differential between property crime and violent crime in the explanatory power of employment offers some support for the economic choice theory as described above.

Other theories of crime would have more difficulty explaining this differential in correlations. Self-control theory, for instance, would predict a correlation between unemployment and any crime, because both are manifestations of the same problem of lack of self-control. One would expect a similar correlation from social control theory, because the weak social embeddedness caused by unemployment would presumably drag a person into any sort of crime. Of course, as discussed above, there is room for multiple theories of crime to work together. It would seem, however, that the differential between property crime and other types of crime with respect to the correlation with employment provides at least some evidence for the influence of economic choices made by those involved in crime.

Scholars have long paid close attention to whether an individual’s age is a factor in their propensity to commit crime. Bushway and Reuter (2002, 221) go so far as to say that getting older is “the only completely convincing mechanism” for offenders to desist from crime (emphasis added).

Scholars have even noticed that age can alter how employment affects crime. Uggen (2000) finds that the correlation between crime and employment is stronger among older individuals than it is among the young. He compares the effects of employment on recidivism between those of different ages: those aged 27 and older and those under the age of 27. He finds that, among the older cohort of ex-offenders, those who were employed re-offended less often than those who were unemployed. Employment did not have the same effect for the younger cohort, however. Among those in their teens and early twenties, the effect of employment was small. “Work,” he concludes, “thus appears to be a turning point for older, but not younger, offenders” (529).

In fact, a body of research has found a positive correlation between crime and employment among adolescents—the opposite of what one would expect based on the theoretical frameworks outlined above and the opposite finding of the research on the general population. According to Apel (2009), the hypothesis of this research is that working a significant number of hours during one’s high school years interferes with other activities that help adolescents grow into successful adults. He cites research by Greenberger and Steinberg that associates adolescent employment, particularly employment of more than twenty hours per week, with other forms of misbehaviour, including non-criminal misbehaviour, and showed that these were exacerbated as the number of hours worked per week increased. While the emphasis here is on adolescent employment as a cause of these high rates of delinquency, the actual relationship may be more complex. For example, an adolescent engaged in these behaviours may also be growing up in poverty, requiring them to get a job working longer hours than their peers. But this does not change the underlying point of the theory, which suggests that adolescent employment may, as Uggen and Staff (2001, 6) put it, represent “a premature or precocious transition to adult roles.”

Gottfredson and Hirschi (1990, 138) use these findings to support their theory that there is no direct causal relationship between crime and employment. While recognizing that there is a reduction in crime rates as people age, they note that, at least among teenagers, “people with jobs attached to them are more rather than less likely to be delinquent.” They argue that, when other factors are controlled for and the effects of employment are tested in isolation, “differences in rates of crime are small, nonexistent, or even in the wrong direction,” concluding that employment is irrelevant in explaining why people tend to commit less crime as they age.

However, Apel (2009) points to other explanations for this apparent anomaly in adolescent employment. Economic choice theory, for example, would argue that the jobs that tend to be held by adolescents of high school age are low-paying jobs that would offer a fairly low disincentive to committing crimes. Therefore, it is not that employment leads to crime for adolescents; it is that low-wage, low-quality jobs, in which adolescents disproportionately happen to be employed, don’t provide as strong a disincentive to a life of crime as other, higher-quality jobs. This is not unlike Toby’s (1957) description, outlined above, of an adolescent dropping out of school and taking an unskilled job, leading to frequent job changes and the loss of a stake in conformity.

Social control theory provides another possible explanation for the apparent anomaly of adolescent employment (Apel 2009). Recall that according to social control theory, people are kept from committing crimes by a set of social bonds to other individuals and institutions of society. The stronger one’s bonds to prosocial influences, the less likely they are to slip into crime. What, then, are the strongest bonds that deter an adolescent from engaging in illegal behaviour? According to the research highlighted by Apel, it is not the workplace that provides teenagers with the strongest bulwark against crime; it is rather their family ties that are the strongest preventative medicine. In this view, employment competes with family for time and intensity of social relationships. Moreover, it may be that those teenagers with weaker ties to their family (or to their school, for that matter) are the ones who are most likely to work more than twenty hours per week at a job outside of school. Therefore, according to this theory, it is not employment itself that is leading these adolescents into crime. Rather, high levels of employment at this young age are a symptom of weak social bonds in the family, which social control theory proposes as the primary cause of crime.

This finding has some important consequences for policy, particularly youth employment strategies. It suggests that policies to promote work among youth should not overlook more important factors that prevent crime. As important as work is to avoiding crime, family bonds are more so. Thus, pro-employment policies must not lose sight of other efforts to improve family life, which may be more fundamental to setting youth up for success.

The central question of this paper is not whether crime and employment are related. That is a relatively easy question to answer, and we have seen that despite doubts arising from early empirical research, most of the recent literature is converging on the expected significant relationship between the two. The more interesting problem is how they are related and whether we can say that one causes the other.

While Bushway and Reuter (1997, 6–6) cite studies that find a correlation between crime and unemployment at the individual level, they are quick to note that “the causality is uncertain.” We have already seen that Gottfredson and Hirschi (1990) deny that there is any causality directly between crime and unemployment, arguing that both are caused by an individual’s degree of self-control. In support of their thesis, they cite the study conducted by Sheldon and Eleanor Glueck, two pioneering researchers in criminology, that followed 1,000 individuals from the Boston area, half of whom were “delinquent” and half of whom were “nondelinquent.” The key for this study was that it occurred through the Second World War, when there was effectively full employment. According to Gottfredson and Hirschi, this study showed that those with a history of delinquency never changed, despite what was effectively a “full-employment treatment program” presented by the onset of the war (164). The “delinquents” were unable to secure employment with the armed forces or couldn’t maintain those positions for long.

Sampson and Laub (1993) look at the Glueck researchers’ data and find those subjects with low job stability when they are between the ages of seventeen and twenty-five tend to have high levels of excessive use of alcohol, arrest, and general deviance when they are between the ages of twenty-five and thirty-two. Thus, Sampson and Laub conclude that “young-adult job stability has substantial predictive power” on these variables later in life. This, they suggest, is evidence that lack of “job stability in the transition to young adulthood significantly modifies trajectories of crime and deviance” (147). Therefore, contrary to Gottfredson and Hirschi, they argue that crime cannot be explained solely by factors developed in childhood and that a lack of stable employment can be a significant factor in adult crime.

Even if there is a causal link between crime and unemployment, however, the direction that the causality flows can be unclear. Does unemployment cause crime? Or does crime cause unemployment? Or does each cause the other to some degree? If the causality is in fact bidirectional, then studies may overestimate the impact of unemployment on crime if they fail to consider the possible causation going the other direction. Apel (2009) notes that labelling theory is particularly sensitive to this issue: it highlights the stigma of being labelled a criminal and the ensuing difficulty of, among other things, obtaining gainful employment.

Studies have attempted to control for this problem. Good, Pirog-Good, and Sickles (1986), for example, use modelling to distinguish the two effects. They find that, while there is a reciprocal relationship between  employment and crime, the effect of unemployment on crime was stronger than the effect of crime on  unemployment. Thornberry and Christenson (1984) study this question while controlling for various demographic characteristics and find a statistically significant relationship between crime and unemployment in both directions. They also find that unemployment can lead to crime more immediately, whereas crime can lead to unemployment on a more long-term basis over the course of an individual’s life.

Freeman (1991, 15) conducted a longitudinal analysis of disadvantaged youth and found that “jail/probation has a striking adverse effect on an individual’s employment status.” Importantly, Freeman argues that, by controlling for a number of other characteristics that one would expect to be correlated with crime and employment, his research demonstrates a causal relationship between the two. Moreover, his research shows that the effects of prior incarceration on employment persist for a long time, such that offenders and ex-offenders constitute a “sizeable relatively permanent population . . . outside the mainstream of society” (1).

Overall, while causality is difficult to determine with any degree of certainty, these studies nonetheless provide evidence that there is a causal, bidirectional relationship between crime and unemployment. The differences in both the strength and immediacy of the effects in either direction suggest that elements of multiple theories of crime may be at work in the link between these two variables. Economic choice theory, social control theory, or both may go some way in explaining the link from employment to crime, whereas labelling theory may do so in the other direction.

The foregoing review has shown that the recent academic literature has increasingly found evidence of a relationship between crime and unemployment, with a number of studies even suggesting the existence of a causal relationship. It is highly likely, therefore, that tackling the issue of unemployment would be a policy agenda that would yield fruit in deterring people from going down a path that leads to crime and in re-establishing themselves on a more prosocial foundation.

Everything that has been said about the link between crime and employment comes to a head in the topic of recidivism and the ability of employment to reduce it. As stated in the introduction to this paper, the successful reintegration of ex-offenders into the labour market is not only a pro-growth and anti-crime policy but also a profoundly moral responsibility of our society.

Yet ex-offenders struggle to enter or re-enter the labour market. Babchishin, Keown, and Mularcyzk (2021) conducted a study of 11,158 ex-offenders who had been admitted to Correctional Service of Canada facilities and who had been released from prison for an average of fourteen years. They find that ex-offenders are substantially less likely to report employment income. Of those who did report employment income, average yearly earnings were $14,000, substantially lower than those of the general Canadian population. They are also more likely to receive welfare payments, with 41 percent reporting social assistance income on their tax forms.

Similar results can be found in the United States. Carson et al. (2021) find that 33 percent of offenders released from custody in the US in 2010 were unable to find any employment in the four years that followed their release. The results were worse for those convicted of non-drug-related offences and for members of racial minority communities.

Our hope should be that ex-offenders, having paid their debt to society, can re-enter the community, including the labour market, and contribute meaningfully to it. Unfortunately, these statistics suggest that this is not happening for many of those exiting the criminal justice system.

In this context, it is important to examine what barriers may exist to those who are trying to join the workforce. Here this paper transitions into a discussion of one particular policy proposal. There are potentially a great number of policy responses that can be taken here, including skills development and education programs in prisons. However, we have chosen to focus on the proposal to “ban the box,” which has garnered a high degree of attention, especially in the United States.

The reason for this focus is two-fold. First, it is topical and hotly contested in the public policy sphere, so it serves as a useful policy case study. Second, this section of the paper lays important groundwork for future Cardus research on the problems that ban-the-box initiatives are trying to solve.

In recent years, several states and municipalities in the United States have pursued so-called ban-the-box initiatives. The “box” in this case represents the checkbox on application forms where applicants attest that they do not have a criminal record. The idea of “ban the box” is to ban prospective employers from asking about a job applicant’s criminal history—or at least postponing a criminal record check until as late in the application process as a conditional offer of employment—so as to shield ex-offenders from the stigma of a criminal record.

Recall that the labelling theory of crime posits that the stigma of being labelled a criminal pushes an individual further and further into crime. In this light, a criminal record is a label imposed by none other than a governmental authority. 4 4 The stigma of a criminal record in the context of a job application acts not only on the presuppositions of a prospective employer but even on the applicant themselves. Consider a job applicant with a criminal record who comes across a question on an application about criminal records. This ex-offender may simply decide not to pursue the application so as not to face the question, even if the employer may have been willing to hire them. Thus, ban-the-box initiatives can also be intended to mitigate this issue of how stigma affects ex-offenders’ own willingness to pursue employment opportunities.

Studies have shown that employers are less willing to hire persons with criminal records than those without. Pager (2003), for example, finds that employers were 50 percent less likely to call a job applicant for an interview if it was known that the applicant had a criminal record. This is unsurprising, of course. Hiring any new employee is a risky decision. When that person carries a perceived risk of bringing crime into the workplace, employers understandably seek to reduce the level of risk by seeking as much information as possible about candidates. Moreover, even if an applicant is unlikely to bring crime into the workplace, employers may nevertheless view a criminal record as a signal that a job applicant is more likely to be a problematic employee in general. This could be based on a view that ex-offenders are less likely to be agreeable, hard-working, or ethical. It is not surprising, therefore, that from an employer’s perspective the existence of a criminal record would be a relevant piece of information and a reasonable query in the hiring process.

The question, crucially, is whether employers are using this information fairly and in a way that is rationally tied to reality. Research suggests that employers tend to overestimate the impact of a criminal record on the ability of an ex-offender to perform well in a workplace. Lundquist, Pager, and Strader (2018) study the experiences of US military recruits with and without criminal records. They find that while ex-offenders were slightly more likely than their counterparts to commit an offence in the military, there was no difference between ex-offenders and non-offenders in poor performance leading to dismissal. In fact, ex-offenders were promoted to higher ranks more quickly than non-offenders, suggesting that they demonstrated behaviour pleasing to their superiors.

Of course, the military may be a unique case that is not transferrable to other workplaces. After all, perhaps the military’s strict regimen of obedience to orders is a perfect environment for a certain kind of ex-offender. Yet other studies show similar trends. In a study of recidivism rates of 317 ex-offenders, Soothill, Humphreys, and Francis (2013, 417–418) find that, while 40 percent were convicted of subsequent offences, only 8 percent committed an offence that “directly and adversely affected an employer,” suggesting that employers had “exaggerated fears” of hiring previously incarcerated persons. Griffith and Harris (2020) examine data from the customer-service industry to find that there was no difference in objective standards of job performance between ex-offenders and others employed in the sector. 5 5 It should be noted that Minor et al. (2018), whose study we discuss in more detail below, identify customer service as one of the sectors in which ex-offenders score the best in comparison to non-offenders. This is important context that may call into question the applicability of Griffith and Harris’s (2020) conclusions to other industries. Nonetheless, their main finding—that there was no relationship between criminal history and job performance in this particular case—shows that exaggerated fears exist in at least one industry. In fact, Minor, Persico, and Weiss (2018) find evidence that ex-offenders in some jobs scored better on some workplace metrics than other employees.

Moreover, criminal record checks may unfairly elevate one particular characteristic of a prospective employee over any other criterion. Some employers refuse to hire a job applicant if they have any criminal record at all (John Howard Society of Ontario 2018). Official Canadian statistics show that recidivism among federal ex-offenders is on the decline and that, of those who do recidivate, subsequent offences are less severe than the original offences (Department of Justice 2020). Prior behaviour is thus not necessarily indicative of future behaviour.

There is also evidence that employers err in their perceptions of how likely different age categories of ex-offenders are to re-offend. For example, Griffith, Rade, and Anazodo (2019) point, on the one hand, to studies that show that prospective employers are more likely to overlook past offences among younger job applicants and, on the other hand, to studies that show, as described in an earlier section, that youthful ex-offenders are more likely to re-offend than older ex-offenders. This is not to suggest that employers should be less willing to hire youthful offenders (though perhaps they should be more willing to hire older ex-offenders); it is simply to show that employers’ expectations of recidivism rates do not necessarily match reality.

In short, allowing the existence of a criminal record effectively to veto an applicant would appear to be at odds with what a number of studies say about the risks to employers of recidivism. Of course, this may not allay employers’ concerns that a criminal record is a general signal of the quality of a worker, but it does highlight a disconnect between employers’ beliefs and reality.

It has become popular, particularly in some American jurisdictions, to prevent employers from asking about criminal records, at least early in the hiring process. Yet serious, negative, and unintended consequences with such ban-the-box initiatives, particularly relating to racial discrimination, have been unearthed by some researchers.

There has been an increasing interest in recent years on the effects of racial characteristics on the relationship between crime and employment. In an analysis of fifty-eight academic articles, Griffith, Rade, and Anazodo et al. (2019) note that a large number of these studies find that employers were less likely to hire African Americans with criminal records than they were to hire whites with criminal records.

Agan and Starr (2018) find a troubling result suggesting that ban-the-box initiatives had the potential to discriminate against minority job applicants who did not even have criminal records. They studied how the treatment of white and Black job applicants by employers differed before and after the implementation of ban-the-box policies in New Jersey and New York City. They find that, before ban-the-box was implemented, white applicants were called back 7 percent more often than Black applicants. After ban-the-box was implemented, the call-back gap between white and Black applicants jumped to 43 percent. They argue that this discrepancy may be explained by the possibility that “employers are relying on exaggerated impressions of real-world racial differences in felony conviction rates” (195). In other words, in the absence of information about the criminal histories of individual applicants, employers may rely on impressions of racial groups that are based on macro-level data on conviction rates for minority populations, particularly Black people. In other words, employers may be using racial identification as a proxy for determining whether an applicant has a criminal history.

In response to these findings, some advocacy groups for ex-offenders have expanded the scope of their campaigns. While they have not given up the goal of banning the box, they have paired these efforts with education campaigns targeted at employers on the reality of hiring ex-offenders, in part so as to avoid the unintended consequences of ban-the-box initiatives. The John Howard Society of Ontario, for example, recently established the Fair Chances Coalition, based on a similar coalition in the United States, to educate private-sector employers on this matter and to increase employer buy-in on policies to reduce the stigma of criminal records (John Howard Society of Ontario, n.d.a). In addition, as of the date of this publication, the John Howard Society of Ontario provides numerous resources on its website for employers to understand the criminal records system and to encourage them to institute human resources policies that do not introduce undue discrimination against ex-offenders into their hiring processes (John Howard Society of Ontario, n.d.b).

Another criticism of ban-the-box initiatives is that they can be overly uniform. Minor, Persico, and Weiss (2018) show that ex-offenders perform differently depending on the kind of job in which they are employed. Among the ex-offenders studied in their paper, those in customer-service jobs tended to do reasonably well, while those in other jobs were much more likely to leave or to be dismissed for reasons of misconduct. They conclude that “not all workforces are the same” and that laws and regulations applied uniformly will not work (33). They point instead to other methods of combatting unemployment among the previously incarcerated, such as employer-education campaigns and incentives for hiring ex-offenders.

Of course, ban-the-box initiatives could be scaled back and tailored for certain industries or certain kinds of offenders. It would be unreasonable, for example, to do away with some form of criminal record check for work that brings the employee into contact with vulnerable populations. Yet these studies demonstrate that ban-the-box initiatives are not a silver bullet to the problem of ex-offender unemployment. They remain controversial and a matter of significant debate in the United States.

Reducing the stigma of a criminal record is a major undertaking that will involve multiple spheres of society, including government, the private sector, and ex-offenders themselves. In Ontario, Monte McNaughton, current Minister of Labour, Immigration, Training and Skills Development, has recently placed a considerable emphasis on supporting the re-entry of ex-offenders into the labour force through a top-up of the Skills Development Fund. Supporting the labour-market re-entry of ex-offenders is only one of the goals of this fund (Employment Ontario 2021), but the Minister’s communications have regularly and explicitly highlighted how this funding will help ex-offenders specifically (McNaughton 2021; 2022a; 2022b; 2022c). Efforts such as these by high-profile leaders, alongside education campaigns such as those promoted by the John Howard Society, are important in changing minds and eliminating unjust prejudices toward those previously involved in the criminal justice system.

This survey of existing literature on the intersection of crime and employment has yielded some interesting observations:

• Despite an early “consensus of doubt,” modern research studies more consistently find a positive link between unemployment and crime;
• The strength of the link depends substantially on the type of crime, with unemployment being more strongly associated with property crime than with violent crime;
• The relationship varies with age, with employment among adolescents shown to be positively linked with crime;
• A number of studies have sought and have claimed to find evidence of a causal link between unemployment and crime; and
• These studies offer evidence that the causality is bidirectional, that is, unemployment can cause crime, but crime can also cause unemployment.

This evidence suggests that well-designed pro-employment policies will go some way in addressing the root causes of crime.

Initiatives that encourage employers to hire ex-offenders may well lead to a reduction in criminal behaviour among this population. Some have proposed ban-the-box initiatives as a way to protect those previously involved in the justice system from the stigma of their criminal record when they apply for jobs. As we saw in this paper, however, these policies lead to serious, negative, and unintended consequences, as there is evidence that some employers engage in racial profiling in the absence of individual information from criminal records. While some scholars and entities have proposed strategies to mitigate these effects, such as employer-education programs, we believe that there is substantial room for further research and debate on this subject. As such, Cardus intends to contribute to this discussion in the near future with policy recommendations.

Of course, the relationship between crime and employment is complex, with many other social and economic factors playing compounding roles. Nevertheless, employment appears to have the potential to lead people away from a life of crime and offer them hope for a better future. It therefore behooves us to exploit the ability of work and employment to assist those caught in the criminal justice system in leading new lives of greater meaning, direction, and dignity.