Local Labour Markets and Theft: New Evidence from Canada
| Author | Fraser Summerfield |
| Published date | 01 February 2019 |
| DOI | http://doi.org/10.1111/obes.12256 |
| Date | 01 February 2019 |
146
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 81, 1 (2019) 0305–9049
doi: 10.1111/obes.12256
Local Labour Markets and Theft: New Evidence from
Canada*
Fraser Summerfield†
†Economics Department, St. Francis Xavier University, 2329 Notre Dame Avenue,
Antigonish NS, Canada (email: fsummerf@stfx.ca)
Abstract
This paper provides causal evidence that labour market opportunities affect theft-related
crime rates in Canada. Synthetic panel data from 2007–2011 combine the Labour Force
Survey and Uniform Crime Reports microdata. Low-skill unemployment rates and corre-
sponding crime rates are measured for age-city-specific groups of young males. IV esti-
mates exploit the exposure of low-skill employment to exogenous demand for exports to
the US. Causal estimates of the elasticity of theft-related crimes with respect to low-skill
unemployment range from 0.357 to 0.654. The use of aggregated unemployment rates
appears to bias OLS estimates downward. IV estimates are found to mitigate this aggrega-
tion bias.
I. Introduction
Identifying factors that cause crime is a longstanding preoccupation among academic
researchers (Block and Heineke, 1975;Wolpin, 1978; Ehrlich, 1977; Freeman, 1983; Levitt,
1997).1One reason is the significant cost of crime. Shoplifting, for example, cost US
businesses an estimated $1.6 billion in 2007 (Bressler, 2009). Thaler (1978) finds that
each property crime has a negative externality on nearby housing prices of about $2,500,
in 2017 terms. Crime is also costly in terms of public expenditure. Justice system spending
in 2002 was $613 per-capita in Canada and $630 per-capita in the US (Story and Yalkin,
2013; Chalfin, 2015). Theft crimes contribute significantly to this spending in Canada
where they account for approximately 40% of all crimes (Perreault, 2013). Recent upward
JEL Classification numbers: J23; K42.
*The author thanks DianaAlessandrini, Michael Baker, James Fenske, Louise Grogan,Tina Hotton, TengWah Leo,
Miana Plesca, Chris Robinson, Anindya Sen, MikeShannon, Stan Veuger, the editor and two anonymousreferees for
helpful comments. The research has also benefitted from suggestions through a departmental seminar at St. Francis
Xavier University and presentations at the 2011 CEA, 2012 CLEA and 2016 SEA meetings. Financialsuppor t from
a Social Sciences and Humanities Research Council Doctoral Award is acknowledged. Data access provided by the
Canadian Research Data Centre Network (CRDCN). The services and activities providedby the CRDCN are made
possible by the financial or in-kind support of the SSHRC, the CIHR, the CFI, Statistics Canada and contributing
Universities.The views expressed in this paper do not necessarily represent the CRDCN’s or that of its partners’.All
errors and opinions are the author’s.
1Ehrlich (1996) notes that interest in crime can be traced back to Adam Smith’s “Wealthof Nations”.
Local labour markets and theft 147
trends in policing and corrections costs combined with stable incarceration rates imply that
the traditional “punitive” approach to crime reduction may not be sustainable in Canada
(Story and Yalkin, 2013; Easton, Furness and Brantingham, 2014). The situation may be
even less favourable in other countries such as the US where incarceration rates are rising.
These conditions motivate the exploration of alternative policy levers. The findings in this
paper suggest that one alternative is to improve labour market opportunities.
The economic analysis of crime is rooted in the seminal work of Becker (1968), with
later additions from Ehrlich (1973) and others. In Becker’s model, offenses, o, are supplied
by individual iaccording to the function:
oi=oi(pi,fi,ui).(1)
This function depends on the probability of conviction, pi, the penalty if convicted, fi, and
a composite variable, ui, that captures individual characteristics and other aspects of the
economy. Individuals at the margin of committing crime supply more offenses when any
of these arguments decreases the opportunity cost of offending.
Labour market conditions are one element of uithat affect opportunity costs. Indi-
viduals in unfavourable labour markets can expect to forgo less income if they are in-
carcerated. Since conviction may decrease future employability, even short incarceration
periods could significantly affect future income. Theft, itself a potential income source,
might also increase if unfavourable labour markets render legitimateear ning opportunities
relatively less lucrative. Grogger (1998) illustrates this tradeoff by extending the standard
labour-leisure framework and Raphael and Winter-Ebmer (2001) shows that unemploy-
ment increases time allocated to crime among some workers. A correlation between theft
crime and labour market conditions is visible in Canadian data for young males during
2007–2011. Figure 1 shows that both series decreased in 2007 as the economy boomed
and subsequently peaked between 2009 and 2010 in the aftermath of the Great Recession.
Offense rates may be higher among certain demographic groups if these groups con-
tain proportionally more individuals at the margin of offending. Stylized facts indicate that
groups most likely to commit theft-related crimes are low-skill or less-educated individuals
(Gould, Weinberg and Mustard, 2002; Lochner and Moretti, 2004; Machin and Meghir,
2004), and in particular, young males (Phillips,Votey and Maxwell, 1972; Foug`ere, Pouget
and Kramarz, 2009; Trussler, 2012).These groups often have less-favourable labour mar-
ket opportunities and thus face lower opportunity costs for theft crimes.2If the perceived
opportunity cost of offending is heavilyinfluenced by local labour market outcomes among
similar individuals, the individual decision to offend may not depend on aggregate unem-
ployment trends. The literature has raised similar concerns about the use of aggregate data
in crime regressions, including potential social multiplier effects (Glaeser, Scheinkman
and Sacerdote, 2003) and the possibility that aggregate unemployment rates will not ade-
quately capture labour marketprospects for individuals at the margin of offending (Mustard,
2010). The latter may be a particularly salient concern for the US and Canada where many
2Males aged 15–25 represent about 8% of the population from 2007–2011 (Statistics Canada, 2017a) but commit
49% of all theft crimes in the data.
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
148 Bulletin
550 600 650 700 750
Theft Crimes
0 4 8 121620
Unemployment Rate (%)
2007 2008 2009 2010 2011 2012
Monthly unemployment rate for males 15-24
Monthly count of theft crimes per 100,000
Figure 1. Aggregateunemployment and thefts in Canada 2007–2012
Source: Statistics Canada CANSIM tables282-0001 and UCR microdata. Crimes include theft related violation
codes for which offenders are observed in 80% or more of all incidents. These crimes include shoplifting,
possession of stolen goods and trafficking in stolen goods. Details are provided in the text Section II.
major layoffs since the 1990s have affected high-skilled workers.3The scarcity of crime
microdata has limited evidence from the US to the self-reports of particular cohorts in the
NLSY79 (Grogger, 1998; Gould et al., 2002) and a cohort of adolescents in grades 7–12
(Mocan and Rees, 2005). Trumbull (1989) and Yang (2016) provide related evidence on
recidivism using microdata limited to past offenders. However, these studies do not use
labour market conditions specific to particular demographic groups.
This paper estimates the effect of unemployment on a subset of property crimes using
recently available Canadian microdata for the period 2007–2011. A synthetic panel of
males aged 15–25 is created by merging confidential versions of Canada’s Uniform Crime
Reports (UCR) and the Labour Force Survey (LFS) at the minimum level of aggregation
possible. Observations in the resulting cell-leveldata represent workers grouped by age and
Census Metropolitan Area (CMA).This level of detail increases the likelihood of capturing
labour market conditions affecting those at the margin of offending. For example, the low-
skill unemployment rate among 18-year-oldmales in Toronto is related directlyto the crime
rate among 18-year-old males inToronto. The crime rates used in this paper are constructed
from a subset of all property crimes and include shoplifting, possession of stolen goods and
trafficking in stolen goods. These categories provide the most reliable estimates because
3Since 1993, mass layoffs have been announced by companies that employ a significant number of higher skill
workers. IBM, Citigroup, AT&T, Boeing, Bank of America, HP, Alameda School District, Merrill Lynch, Lucent
Technologies,Pfizer, & RIM announced layoffs of 19,000 to 60,000 workers (McGregor, 2015).
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial