Unemployment, GDP, and crime: the importance of multiple measurements of the economy.

AuthorAndresen, Martin A.
PositionGross domestic product


Research that investigates the relationship between the state of the economy and criminal activity dates back over 80 years to the work of Shaw and McKay (1931,1942). Because of the length of time this phenomenon has been under investigation, there are many approaches that have been used to explain the theoretical connection: strain theory, utilitarian or rational-choice theories, conflict theories, and opportunity theories (Cantor and Land 1985). As outlined by Cantor and Land (1985), one of the limitations of this older research is that, while many of these theories are related to one another, they were not considered in a common framework.

In an effort to rectify this situation, Cantor and Land (1985) put forth a model of unemployment and crime that integrated previously disparate theoretical approaches. The Cantor and Land (1985) model of unemployment and crime identified two mechanisms that comprised the total effect of unemployment on crime: the motivation effect and the opportunity/ guardianship effect. As discussed further below, Cantor and Land (1985) hypothesized that an increase in unemployment had a lagged positive effect on crime through increased motivation, and a contemporaneous negative effect on crime because of increased guardianship and reduced opportunity as more people stayed home. This model of the relationship between unemployment and crime has proven to be most instructive, generating a significantly large body of literature. Though there is much disagreement over the various empirical validations of the Cantor and Land (1985) model, the theories behind the model are not often questioned.

Testing their proposed mechanisms, Cantor and Land (1985) found that the opportunity effect dominated the motivation effect, particularly for property crime. Much of the subsequent literature regarding the unemployment and crime relationship has involved this empirical validation of the model; disagreement with the finding that opportunity dominated motivation has led to analyses in many different contexts, different spatial scales, and different empirical methods. Some of the more recent research on the relationship between unemployment and crime has argued that the unemployment rate is not a good/appropriate measurement of the state of the economy for testing the Cantor and Land (1985) model; additionally, there are complications that are inherent in the statistical models used to identify the various predictions of the Cantor and Land (1985) model, leading to unreliable empirical results (Arvanites and DeFina 2006).

The present article considers both of these issues, using the Canadian provinces (1981-2009), variables representing both unemployment and gross domestic product, and a methodological approach that allows for the separation and identification of motivation from opportunity effects. By statistically identifying the short- and long-run effects of the relationship between economic performance and crime, an appropriate test of the Cantor and Land (1985) model can be performed. The use of both unemployment and gross domestic product allows for a comparison of these two variables, not previously undertaken, and the methodological approach addresses the complications in previous empirical research. We find that both motivation and guardianship matter for crime, but they operate differently, based not only on the time frame of analysis but on the crime type under analysis.

The unemployment and crime relationship

The unemployment and crime relationship put forth by Cantor and Land (1985) is not direct in the sense that unemployment directly affects crime. Rather, as shown in Figure 1, changes in the unemployment rate (a conventional indicator of macroeconomic activity) affect criminal motivation (system activity effect) and criminal opportunity (guardianship effect). Subsequently, increases in either motivation or opportunity lead to increases in crime. Unemployment is related to criminal motivation through changes in illegitimate versus legitimate means of securing income. With increased and sustained unemployment, it becomes increasingly difficult to pay bills. This leads to economic hardship, which leads to increased motivation to commit property crime types. Consequently, unemployment and criminal motivation have a positive relationship, particularly with property crime types.

Criminal opportunity, however, has a negative relationship with unemployment for reasons Kennedy and Forde (1990) make clear. Kennedy and Forde (1990) showed definitively that those who spend more of their time away from the relatively protective environment of the home are more likely to suffer criminal victimization, both property and violent crime types. But when unemployment increases, those who are unemployed (as well as those who may feel threatened with unemployment) are likely to spend more time at home and less time in public, due to decreased disposable income. In consequence, their exposure to others and, hence, their likelihood of violent crime victimization is reduced. Also, staying home increases guardianship over personal property, reducing their exposure to property crime victimization.


These two opposing effects of unemployment on crime are shown in Figure 1. Which effect dominates the other is, of course, an empirical question. Cantor and Land (1985) assumed that the system activity and guardianship effects operated in different time frames: the system activity effect (motivation) had a lagged effect because people do not turn to illegitimate activity as soon as economic hardship occurs, and the guardianship effect occurred immediately (a contemporaneous effect) because people were at home guarding person and property immediately. The opposing effects in the Cantor and Land (1985) model are most easily understood within the context of routine activity theory. In their seminal publication, Cohen and Felson (1979) stated that, in order for a crime event to occur, there must be three factors present: a motivated offender, a suitable target, and the lack of a capable guardian.

The system activity effect (motivation) is clearly related to the motivated offender. In the original formulation of routine activity theory, motivation was held constant to explain the trend of increasing crime rates from the late 1940s through to the early 1970s. However, it should be noted that motivation for crime is a variable--Cohen and Felson (1979) did not need to invoke changes in motivation to explain the dramatic rise in crime rates during their study period. As stated above, the effect of motivation on criminal choices made by motivated offenders is lagged. This lagged effect exists because it may take some time for economic hardship to lead individuals to commit criminal events; these criminal events may also either directly or indirectly increase the demand for crime through the purchase of illegitimate goods. However, this stated relationship between criminal motivation and the changes in crime trends is more complex than it appears. The hypothesized relationship implies that there is a significant portion of the population (large enough to have an impact on official crime statistics) who move in and out of criminal activities as the strength of the economy waxes and wanes. Such a phenomenon cannot be true in the long run because of the sociological paradox discussed by Cohen and Felson (1979): economic conditions improved significantly, on average, during the 1950s and 1960s; yet crime rates rose drastically.

In an effort to have theoretical expectations that are consistent with known facts, a more conservative interpretation is used in the present analysis: individuals who face higher levels of structural unemployment and/ or poorer economic conditions have greater motivation for criminal activity; particularly, for property crime types. In the aggregate, members of this population are further along the continuum of motivation for criminal behaviour. This is consistent with the Cantor and Land (1985) model because they state that a complete model of the relationship between unemployment and crime must consider how "unemployment [relates] to the prevalence of potential offenders" (Cantor and Land 1985: 319; emphasis added). Consequently, areas, such as provinces, that have poorer economic conditions or develop poorer economic conditions are expected to have relatively more motivated offenders than other areas simply because there are fewer legitimate economic opportunities. This is a long-run effect.

The guardianship effect (opportunity) refers to the concepts of guardianship and suitable targets in routine activity theory. This effect is expected to be contemporaneous because if more individuals are not working, then more individuals will be spending more time at home protecting themselves and their homes from criminal events as economic conditions worsen. Consequently, if there are increasing numbers of people spending more time in the relatively protective environment of the home, there will be an increase in the level of guardianship--this reduces the number of criminal opportunities. In addition, because there will also be fewer people converging in time and space outside of the relatively protective environment of the home, there will be a reduction in the number of suitable targets for violent crime. Therefore, in the short run, economic conditions are expected to negatively affect both property crime and violent crime.

Recent research

In their original work on unemployment and crime, Cantor and Land (1985) stated that, once the model of unemployment and crime is properly specified in a temporal context, the impact of contemporaneous unemployment (opportunity) on crime is negative. This statement was in contrast to years of research that most often stated the opposite. Needless to say, researchers have argued over the unit of analysis, over statistical methods...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT