EXAMINING THE GENERALITY OF THE UNEMPLOYMENT–CRIME ASSOCIATION

• Author: PEKKA MARTIKAINEN,MIKKO AALTONEN,JOHN M. MACDONALD,JANNE KIVIVUORI
• Date: 01 August 2013
• DOI: http://doi.org/10.1111/1745-9125.12012
• Published date: 01 August 2013

EXAMINING THE GENERALITY OF THE

UNEMPLOYMENT–CRIME ASSOCIATION∗

MIKKO AALTONEN

Criminological Unit

National Research Institute of Legal Policy

JOHN M. MACDONALD

Department of Criminology

University of Pennsylvania

PEKKA MARTIKAINEN

Population Research Unit

University of Helsinki

JANNE KIVIVUORI

Criminological Unit

National Research Institute of Legal Policy

KEYWORDS: unemployment, crime, within-individual analysis, adminis-

trative data

This article examines whether the relationship between unemploy-

ment and criminal offending depends on the type of crime analyzed.

We rely on fixed-effects regression models to assess the association be-

tween changes in unemployment status and changes in violent crime,

property crime, and driving under the influence (DUI) over a 6-year

period. We also examine whether the type of unemployment benefit re-

ceived moderates the link to criminal behavior. We find significantly pos-

itive effects of unemployment on property crime but not on other types

of crime. Our estimates also suggest that unemployed young males com-

mit less crime while participating in active labor market programs when

∗Additional supporting information can be found in the listing for this arti-

cle in the Wiley Online Library at http://onlinelibrary.wiley.com/doi/10.1111/

crim.2011.51.issue-3/issuetoc.

Direct correspondence to Mikko Aaltonen, National Research Institute of Legal

Policy Criminological Unit, P.O. Box 444 (Pitkansillanranta 3 a), 00531 Helsinki,

Finland (e-mail: mikko.e.aaltonen@om.fi).

C2013 American Society of Criminology doi: 10.1111/1745-9125.12012

CRIMINOLOGY Volume 51 Number 3 2013 561

562 AALTONEN ET AL.

compared with periods during which they receive standard unemploy-

ment benefits.

The association between unemployment and crime remains a matter of

debate, as both theory and empirical evidence suggest that selection mech-

anisms explain why individuals who are unemployed also commit a dispro-

portionate share of crime in the population (Caspi et al., 1998; Gottfredson

and Hirschi, 1990; Heckman, Stixrud, and Urzua, 2006). In other words, it

remains unclear whether unemployment causes crime or whether both are

a reflection of underlying traits in the people most likely to become unem-

ployed and criminally active. Selection into unemployment is then a fun-

damental issue when addressing the unemployment–crime link. A further

complication related to the unemployment–crime link is that theories that

view unemployment as a causal determinant of crime give varying hypothe-

ses on whether unemployment should increase all types of crime or whether

we should only expect levels of income-generating crime to increase as a

result of financial problems that joblessness creates. Supporting the latter

view, macrolevel evidence indicates that the unemployment rate is a better

predictor of property crime than of violent crime (Cantor and Land, 1985;

Chiricos, 1987; Raphael and Winter-Ebmer, 2001).

Macrolevel evidence aside, a substantial portion of individual-level re-

search on employment and crime focuses on the topic of criminal desistance

(Laub and Sampson, 2001; Uggen and Wakefield, 2008). Most experimen-

tal studies focus only on high-risk samples of people who are on probation

or parole, thereby addressing only the question of how employment pro-

vided to offenders influences crime (Berk, Lenihan, and Rossi, 1980; Uggen,

2000)—not whether periods of unemployment that happen through the nat-

ural life cycle influence criminal behavior. Evidence from experimental pro-

gram evaluations on the effects of employment and job placement on crime

is inconclusive (Bushway and Apel, 2012; Raphael, 2011; Visher, Winter-

field, and Goggeshall, 2005). Quasi-experimental evidence in the United

States that uses minimum age restrictions on work in different states as

an instrument for different age–work propensities in states shows that in-

creased work is associated with reduced youth crime (Apel et al., 2008).

However, few individual-level, quasi-experimental studies have addressed

the unemployment–crime link directly, and none of which we are aware of

examine this issue in a convincing fashion among adults in a general popu-

lation sample.

Given that most research on unemployment and crime has been con-

ducted in the United States, Nordic countries represent a different con-

text in which to examine the association between unemployment and crime.

In addition to universal access to free higher education, the tax policies in

these countries redistribute income and provide the entire population with

THE UNEMPLOYMENT–CRIME ASSOCIATION 563

a host of social welfare benefits not available in the United States. If the ef-

fect of unemployment on crime is related to financial hardship, then this ef-

fect should be smaller in countries that soften the financial shock that results

from job loss. Despite the comparatively generous social policies and more

even income distribution, recent research has shown that substantial so-

cioeconomic differences in crime persist in the Nordic countries (Aaltonen,

Kivivuori, and Martikainen, 2011; Galloway and Skardhamar, 2010;

Nilsson and Estrada, 2009). However, it remains unclear to what extent

selection processes contribute to socioeconomic differences in crime in

Nordic countries.

The current study examines whether periods of unemployment are asso-

ciated with changes in crime, holding constant stable individual differences

in criminal propensity. We focus specifically on three questions. First, we

examine whether periods of unemployment are associated with increased

levels of crime overall once stable individual differences are taken into

account. Second, we examine whether periods of unemployment and the

length of unemployment are associated with within-individual changes in

different types of crime. Third, we examine whether crime rates vary de-

pending on the type of unemployment benefits that individuals receive.

We use a quasi-experimental design to examine the effects of unemploy-

ment over relatively short time intervals as a method for identifying the

effects that shifts in unemployment have on crime (see Felson et al., 2011;

Horney, Osgood, and Marshall, 1995; Laub and Sampson, 2003; Osgood

et al., 1996). Longitudinal panel data with short time intervals can shed

more light on these questions with the help of fixed-effects models (Alli-

son, 2009) that control for time-stable individual differences and allow one

to identify the unemployment effect from only within-individual change.

As an addition to prior studies, we use Finnish register-based data on both

unemployment and criminal convictions to construct a longitudinal panel

of timing of unemployment, type of unemployment, and different cate-

gories of criminal convictions. We use a nationally representative sample

of Finnish men 20–30 years of age (N=15,658) whom we follow for 6 years

over 3-month intervals, thus allowing us to shed light on the effect of unem-

ployment on crime and types of crime in an unselected sample of the entire

population who commit a crime at least once during the study period.

THEORETICAL FRAMEWORK

Effects of life events on crime have been discussed under the concepts

of population heterogeneity and state dependence, ideas first introduced to

criminology as explanations of the relationship between past and future

offending (Nagin and Paternoster, 1991; Piquero, Farrington, and Blum-

stein, 2003). Gottfredson and Hirschi’s (1990) self-control theory is typically