Examining the link between crime and unemployment: a time-series analysis for Canada

Abstract

We use national and regional Canadian data to analyse the relationship between economic activity (as reflected by the unemployment rate) and crime rates. Given potential aggregation bias, we disaggregate the crime data and look at the relationship between six different types of crimes rates and unemployment rate; we also disaggregate the data by region. We employ an error correction model in our analysis to test for short-run and long-run dynamics. We find no evidence of long-run relationship between crime and unemployment, when we look at both disaggregation by type of crime and disaggregation by region. Lack of evidence of a long-run relationship indicates we have no evidence of the motivation hypothesis. For selected types of property crimes, we find some evidence of a significant negative short-run relationship between crime and unemployment, lending support to the opportunity hypothesis. Inclusion of control variables in the panel analysis does not alter the findings, qualitatively or quantitatively.

Keywords:

• crime rates in Canada
• time-series analysis
• unemployment

JEL Classification:

• C22
• K1

Acknowledgements

We are grateful to Mustafa Caglayan, Juan Carlos Cuestas and the participants at the Royal Economic Conference (April 2014) for providing helpful comments. We would also like to thank an anonymous referee for very valuable comments, which have much improved the article.

Notes

¹ Wu and Wu (2012) find mixed results, with unemployment and fraud having a negative relationship and unemployment and drug offences having a positive relationship.

² Most of the empirical studies which find a positive link between unemployment and crime (consistent with the motivation hypothesis) use annual date and are mainly looking at the relationship between first difference of crime and contemporaneous first difference of unemployment, that is, ${\beta }_2$.

³ Incarceration works in two ways: deterrence (threat of sanctions deters people from engaging in crime) and incapacitation (while incarcerated criminals are unable to commit crime). INCAR is included with a lag to avoid the problem of reverse causality; see Corman and Mocan (2000, 2005) and Levitt (1996).

⁴ The recession dates mentioned are those obtained from the work done by the Economic Cycle Research Institute and empirical research by Louis and Simons (2007).

⁵ Criminologists have observed a falling trend in crime rates in other countries as well. Levitt (2004) analyses the causes of decreases in crime rates in the US from 1991 to 2001. Ward and Carmichael (2001) consider the case of England and Wales. Similar trends are observed for Canada; see Boyce et al. (2014).

⁶ The panel co-integration test was done using the xtwest command of STATA (Persyn and Westerlund, 2008). Results are available from authors on request.

⁷ Glaeser et al. (1996) and Machin and Meghir (2004) also find significant persistence in crime over time across areas.