Crime and Status: A Contribution To Strain Theory
Crime and Status: A Contribution To Strain Theory
Tim Friehe
Abstract
This paper studies the implications of strain theory for equilibrium crime rates in
a simple economic model by assuming that potential offenders care about their relative position. We establish that subjects status concerns allow for multiple crime
rate equilibria and that these positional preferences may modify the results regarding
predictions about how crime changes in response to a higher detection probability and
a higher sanction. In addition, we argue that the socially optimal level of the detection
probability and the sanction will often be higher when potential offenders care about
their relative position.
Keywords: Crime; Strain Theory; Status; Multiple Equilibria
JEL-Code: K42, H23
University of D
usseldorf, D
usseldorf Institute of Competition Economics, Universitatsstr. 1, 40225
D
usseldorf, Germany. E-mail: [email protected].
University of Bonn, Center for Advanced Studies in Law and Economics, Adenauerallee 24-42, 53113
Bonn, Germany. CESifo, Munich, Germany. E-mail: [email protected].
Introduction
1.1
trasted economic and sociological approach.1 In our setup, potential offenders are concerned
about their absolute level of consumption, but also about how their consumption compares
to that of similar others. Lagging behind the average level of consumption compared to a
reference group causes strain, which may make crime an efficient procedure in order to arrive at a more favorable relative comparison. However, the individual must take into account
the possibility that the offense will be detected and punished, a course of events that would
imply an even more unfavorable status comparison. In other words, when opting for criminal
activity, the individual accepts a gamble regarding both the absolute level of consumption
and status. Specifically, this paper will consider the possibility that social comparisons are
relevant due to the disutility from lagging behind in terms of status, and contrast it with
the scenario in which both being ahead and lagging behind is important for individual wellbeing and decision-making. Whereas the usual understanding of strain refers to somebody
suffering from being relatively worse off, Merton already suggested that personal success and
satisfaction derive not only from attaining goals but also from surpassing others (Lee and
Cohen 2008).
Potential offenders feeling strain from unfavorable comparisons and, possibly, satisfaction
from favorable ones implies in our operationalization via status considerations that multiple
crime rates may represent equilibrium crime participation. The intuition follows from the
fact that the concerns about relative consumption introduce an interdependency between
both the decision of a given potential offender regarding whether to participate in crime and
the decisions of all other individuals. This is of great interest because crime in the real-world
is indeed very unevenly distributed across space and time, despite similar characteristics of
respective locations. For example, there are numerous cases of twin cities in the US, where
cities with similar characteristics nevertheless witness very different crime rates (see, e.g.,
1
For example, Kelly (2000) summarizes his findings by stating that Property crime is well explained by
the economic theory of crime, while violent crime is better explained by strain and social disorganization
theories.
Marceau and Mongrain 2011). Interestingly, we find that crime may be either higher or
lower when compared to a scenario in which potential offenders do not attach value to their
relative position. This comes as a surprise since the presence of positional concerns is usually
considered an additional motive for crime in strain theory. Nevertheless, the explanation
for our finding is quite intuitive. When many other individuals engage in criminal activity
and thereby increase their expected income available for consumption, the individual who
complies with the law must tolerate an unfavorable status position. The strain implied may
push the individual into crime. At the same time, when only few other individuals engage
in crime, then average consumption is not too much different from what a norm-compliant
individual can afford. The strain caused by the small discrepancy is not necessarily conducive
to making crime privately optimal, whereas the anticipated status loss that would result upon
detection contributes to deterrence.
In addition, this paper shows that the presence of status considerations by potential
offenders may call into question that the standard comparative-statics predictions regarding
changes in the level of the sanction and the level of the detection probability hold. Indeed, it
may be that an increase in the detection probability, for example, brings about an increase in
the crime rate. This possibility follows from the influence that law enforcement parameters
have with regard to the average level of consumption, and the possibility that the varied
reference consumption level changes the desirability of crime.
When turning to welfare and policy implications, we argue that the presence of status
considerations will often imply higher marginal benefits of stricter law enforcement. The
explanation uses the fact that stricter enforcement lowers the average consumption level of
the comparison group, which benefits all individuals, while the adversely affected individuals
are concentrated among offenders.
Having outlined our main results, we summarize by saying that our article contributes
to the literature in several ways. First, we offer an economic analysis of the implications of
one of the principal sociological theories on criminal behavior, namely, strain theory. One of
our central findings in this regard is that the presence of positional concerns need not induce
more crime, as is often suggested in the literature. Second, the paper at hand contributes an
explanation for the variety of crime rates across time and space and thereby responds to a
recent call by Ferrer (2010), saying that differences in crime rates across locations ...remain
an open question in the law enforcement literature. Our explanation is complementary to
those established in the literature (see the discussion in the next section). Next, building on
our result regarding the possibility of multiple equilibria, we find that modest adjustments
in law enforcement policy may cause sizable changes in the crime rate, as society may move
from one stable equilibrium to another one. Finally, we establish status gains and losses
as additional aspects that ought to co-determine optimal law enforcement. We argue that
stricter law enforcement is likely to be the adequate response to potential offenders caring
about relative position in many circumstances.
1.2
Related literature
In addition to the literature on optimal law enforcement (see, e.g., Polinsky and Shavell
2009), the present paper is related to the literature that contributes to strain theory, to
articles exploring the implications of people having positional concerns, and to contributions
that establish the possibility of multiple crime equilibria.
Strain theory started with Merton (1938) and has since then been elaborated in several
ways. More recent contributions to strain and crime include the general strain theory by
Agnew (2006b) and the institutional anomie theory by Messner and Rosenfeld (2001). Strain
theory has accumulated a considerable amount of empirical support (Lee and Cohen 2008).
We contribute to the subject by offering a formal operationalization of important aspects of
the theory and studying the implications of these aspects. Our implementation relies on the
4
assumption that potential offenders compare their attainable level of consumption to the one
exhibited by peers. In this regard, our approach very closely mirrors the concept of relative
deprivation referred to before.
The idea that individuals care about their social rank with regard to their fellow human
beings has taken a stronghold in the economics literature.2 Both the fact that relative
concerns are important and that goods differ with regard to their positionality (i.e. that
certain goods have a higher relevance for relative standing in society) have been confirmed
in several empirical studies, among them Alpizar et al. (2005), Carlsson et al. (2007),
Carlsson and Qin (2010), Caporale et al. (2009), Clark et al. (2008), Clark and Senik
(2010), Johansson-Stenman et al. (2002), Solnick and Hemenway (1998, 2005), and Solnick
et al. (2007). When it comes to the identification of the reference point, there is evidence
that the respect and admiration one gets from interaction with face-to-face groups such as
colleagues and friends are a major determinant of status concerns (see Anderson et al. 2012,
Clark and Senik 2010, Senik 2009). Our study complements this literature by exploring the
repercussions of positional concerns for the choice regarding crime.
This paper establishes that the interdependence among potential offenders introduced
by status concerns may allow for multiple crime rates in equilibrium. It is an important
fact that crime is distributed quite unevenly across space and time, despite similar regional
characteristics (see, e.g., Glaeser et al. 1996). This has created an interest in the possibility
of multiple equilibria for a long time. The literature has succeeded in showing that there are
circumstances that permit multiple crime rates in equilibrium, and thereby can contribute to
an explanation of the empirical finding regarding the distribution of crime. One explanation
relies on the fact that a given enforcement budget creates a high detection probability when
2
For instance, Dohmen et al. (2011) provide evidence for the importance of relative income for subjective
well-being using functional magnetic resonance imaging (fMRI). Further empirical evidence for the importance of relative income positions for individual happiness and actions can be found in Stutzer (2004) and
Frey et al. (2008), for instance.
there is little criminal activity, but only a small detection probability when many individuals
engage in crime (see Bar-Gill and Harel 2001, Bond and Hagerty 2010, Conley and Wang
2006, Fender 1999, Ferrer 2010, Freeman et al. 1996, Helsley and Strange 1999, Sah 1991).
Another possible explanation uses asymmetric information about individuals characteristics
and self-fulfilling beliefs (Rasmusen 1996, Verdier and Zenou 2004). In contrast, the positive
covariance between the individual decisions to commit crime in our framework are due to
the fact that well-being depends partly on the comparison of the own consumption level to
that of others. There are other social interaction models such as Glaeser et al. (1996), Sah
(1991), Schrag and Scotchmer (1997), and Silverman (2004), relying on different reasons
responsible for the interdependence. For example, Glaser et al. (1996) consider the case in
which some individuals imitate their neighbors, while Schrag and Scotchmer (1997) consider
the consequences of several potential criminals having access to a crime opportunity where
an individual with access may be sanctioned erroneously. In another line of inquiry, Burdett
et al. (2003), Huang et al. (2004), and Marceau and Mongrain (2011) incorporate potential
labor market influences. The paper at hand provides a complementary explanation for the
empirically observed variance of crime across time and space building on the assumption
about the positional concerns of subjects that has convincing empirical support.
In addition to the connections to the literature described above, our paper is related
to studies exploring the interconnection of crime and inequality. In this realm, there are
theoretical contributions dealing with the fact that law enforcement may also be achieved by
granting social transfers, because this influences the opportunity costs of crime (Benoit and
Osborne 1995, Demougin and Schwager 2000, 2003). The empirical literature on the association of inequality and crime has in most contributions established a positive link between
the two (see Dahlberg and Gustavsson 2008, Demombynes and Ozler
2005, Fajnzylber et
al. 2002, Whitworth 2012, forthcoming), although other results have also been found (see
Kelly 2000 and Chintrakarn and Herzer 2012). In another line of inquiry, Chiu and Madden (1998) study the consequences of variations in the income distribution when private
protection against crime is taken into account.
The remainder of our paper is organized as follows. In Section 2, we describe the model
used for our analysis. Section 3 presents our equilibrium analysis, before Section 4 discusses
findings from a comparative-statics exercise. Potential welfare implications are addressed in
Section 5. The last section concludes.
The model
Consider a group of individuals who are identical regarding their preferences for consumption
and relative standing. We assume that individual well-being can be represented by the
following utility function:
U = v(c) + gw(S),
(1)
where c is the individuals consumption level and S is the individuals status. The utility
from consumption is increasing at a diminishing rate (i.e. v > 0 > v ). The marginal
utility from an improvement in relative standing is weakly positive (i.e. w 0) and may be
either constant, decreasing or increasing. For example, Robson (1992) assumes that utility
is strictly convex in status, whereas Corneo (2002) supposes a strictly concave relationship.3
The parameter g indicates the relative importance of status to absolute consumption. Our
specification of U implies that we consider utility separable in consumption utility and utility
from status considerations, a case between absolute consumption and relative standing being
complements or substitutes. Relative standing is determined by comparing the own level of
absolute consumption with the groups average level of absolute consumption c. We follow
Card et al. (2012), Falk and Knell (2004), and Konrad and Lommerud (1993), among others,
3
According to Robson (1992, p. 839) ... it seems plausible that the difference between a gold medal and
silver medal medal should be greater than that between a silver medal and a bronze medal.
when we specify
S c c
(2)
and w(0) = 0. The distinction of whether only lagging behind is behaviorally relevant or
both lagging behind and surpassing others affects utility can be incorporated by requiring
w (S) > 0 for S < 0 in both cases while w (S) = 0 for S > 0 in the first and w (S) > 0 for
S > 0 in the second case.
The individual level of consumption is constrained by available income. All individuals
have some income I > 0 from legal work. The assumption that all individuals (offenders
and non-offenders) have legal income has its empirical support since most criminals also
participate in lawful employment (see, e.g., Kleiman 2009). In addition, individuals may
take advantage of a criminal opportunity with gain b. In other words, the decision regarding
crime is binary (as is standard in the literature on optimal law enforcement, see, for example,
Polinsky and Shavell 2001, 2009). The level of the criminal gain differs between individuals,
where b [0, B] according to the cumulative distribution function F (b). Consideration of
individuals who are symmetric apart from their payoffs from crime is a standard procedure
in the economic analysis of crime (see, e.g., Bond and Hagerty 2010). Undertaking crime
entails the risk that both the criminal benefits are disgorged and a sanction s > 0 is imposed
with probability p, 0 < p < 1. As a result, we have to distinguish between three different
levels of available income for an individual with criminal benefits b:
In =I + b
(3)
Id =I s
(4)
IL =I
(5)
where the subscript n stands for no detection, d for detection, and L denotes the case in
which the individual does not engage in the illegal opportunity.
8
(6)
In contrast, individuals who opt for undertaking the offense have expected utility EUO
following from
EUO = (1 p)[v(In ) + gw(In c)] + p[v(Id ) + gw(Id c)].
(7)
When deciding on the commission of the crime, the individual takes the level of average
consumption c as exogenously given. The utility from leaving out the criminal opportunity
is the same for all individuals, whereas EUO varies across the population since b [0, B]
is drawn randomly for each subject. As a result, for a given average consumption level, a
crime rate between zero and one would mean that there is some individual with criminal
gain b for whom it holds that
(b, c) = EUO EUL = 0.
(8)
Denoting the critical level of criminal gain that solves (8) by b, we can conclude that individuals (do not) offend for a given level of comparison consumption when b (<) b. A
critical gain level b implies a crime rate given by 1 F (b).
Regarding the individual choice, it is interesting to note that offending implies a gamble
with respect to the utility from absolute consumption and the utility from relative standing.
While individuals are assumed to be risk-averse with respect to variations in the level of absolute consumption, it is possible that they are risk-seeking regarding status considerations.
Looking at a given potential offender with criminal gains b, we can calculate the implicit
risk premium R from
(1 p)[v(In ) + gw(In c)] + p[v(Id ) + g(Id c)] v(IL R) gw(IL R c) = 0, (9)
9
where R < (>)0 for individuals who do (not) offend for given parameter values. In other
words, individuals who would like to offend given the parameters must be paid R > 0 in
order to accept refraining from the criminal activity. Starting from this definition, we arrive
at
R
w (1 p)wn pwd
=g L
= g
c
vL + gwL
vL + gwL
R
(1 p)wn + pwd wL
=
.
g
vL + gwL
(10)
(11)
(12)
(13)
where it is assumed that the individual who is just indifferent between committing the offense
and refraining from crime has criminal gain b. The first term on the right-hand side of (13)
10
is the legal income that is common to all individuals. The second and the third term can
be explained as follows: All individuals with b b offend, either obtaining the benefits from
crime in the no detection state of the world (i.e., with probability 1 p) or experiencing a
reduction in available income due to the imposition of the sanction s in the detection state
of the world (i.e., with probability p). The definition of the reference consumption level as in
(13) makes clear that we suppose that potential offenders consider other potential offenders
as their relevant peers. This is consistent with the empirical literature on the identification
of the reference point discussed in Section 1.2. The fact that criminal acts cause social harm
will be taken into account when we turn to welfare considerations in Section 5.
The level of average consumption will be equal to the level of legal income when b = B
RB
(that is, when no crimes are committed), and maximal at I +(1p) b bdF (b)ps(1F (b ))
with b = ps/(1 p) (when we suppose that B > b ). The latter follows from
c
= f (b)[ps (1 p)b]
b
2 c
= f (b)[ps (1 p)b] (1 p)f (b).
2
(14)
(15)
It is intuitive that the level of average consumption increases when there is a positive level
of deterrence (ps > 0) and the critical benefit level increases from a very small level (i.e.,
when b 0). Similarly, average consumption decreases when b is increased at levels that
are high in view of what is required for maximizing expected payoffs (i.e., when b > b ).4
The level of average consumption is also a function of the law enforcement parameters, and
of legal income. Unsurprisingly, stricter deterrence lowers average consumption for a given
4
11
RB
barb
c
bdF (b) s(1 F (b)) < 0
=
p
b
c
= p(1 F (b)) < 0.
s
(16)
(17)
The direct effect of higher legal income on the level of average consumption is given by
c/I = 1 > 0.
This concludes the discussion of the two building blocks of our model, the critical gain
level that follows from individual utility maximization for a given amount of average consumption on the one hand and the level of average consumption as a function of the critical
gain level on the other.
Equilibrium
In this section, we first seek to establish that there is at least one equilibrium crime rate.
Next, we discuss the possibility of having multiple equilibria, and elaborate on their respective stability. The next section will present results from a comparative-statics analysis.
An individual with a crime opportunity that pays b in the no detection state of the
world decides whether or not to commit the offense given the average consumption level
c. The latter is a function of the critical gain level, the enforcement parameters, and the
level of legal income, c = c b, p, s, I . We assume that all individuals choose simultaneously
between offending and not offending. Their decision-making yields a critical gain level for a
given level of average consumption (i.e., for an assumed critical gain level). Accordingly, the
critical gain level in equilibrium must be consistent in the sense that b results from individual
decision-making that takes c = c b, p, s, I as given.
Proposition 1 There exists at least one equilibrium critical gain level b [0, B], implying
12
=0
(18)
(19)
It is interesting to inquire about the possibility of having more than one equilibrium in
our model, and whether a necessary condition for such an outcome can be readily identified
in our framework. The function b, c b reflects whether or not it is advantageous for an
individual with criminal gains given by b to engage in crime when the average consumption
amounts to c b , expressing the expectation that individuals with benefits b b will under
take the offense. For an interior equilibrium, we require that b, c b = 0, as in (18), and
(0, c(0)) > 0 or (B, c(B)) < 0 for corner solutions.
Considering b, c b as a function of b, we obtain
d b, c b
c
D = (1 p)[vn + gwn ] + g[wL (1 p)wn pwd ] .
{z
} | b
|
db
{z
}
X
(20)
The direct effect represented by Term X, reflecting higher income in the no-detection state
n, is always positive. The existence of more than one equilibrium would require that the
indirect effect expressed by Term Y is over some range of b of opposite sign and dominating the direct effect. Note that Term Y contains the term that determined whether or
not the status lottery will be more or less attractive as a result of a change in average consumption and was used already in determining how the risk premium changes when average
consumption increases, (10). With regard to the distinction between stable and unstable
equilibria, we suppose that D > 0 has to hold for an equilibrium with b, c b = 0 to be
stable. In that case, being to the left of the root means crime is not worthwhile at the given
benefit level as < 0 and accordingly this implies a movement toward the equilibrium level
b. A similar argument can be repeated regarding the case of being to the right of the root.
In the absence of status considerations (i.e., when g = 0), our model has only one
equilibrium since Y = 0 in this case. We obtain either (0, c(0)) > 0, b , c(b ) = 0 or
(B, c(B)) < 0 which determines the unique equilibrium crime rate. The intuition is clear.
The crime opportunity is only evaluated by trading off the material benefit with the expected
14
sanction and the associated risk implied, so that crime always becomes more attractive when
the material benefit is higher.
Similarly, there is only one equilibrium crime rate in case w (S) > 0 and w (S) = 0 for
all S. In that scenario, we obtain = 0 which rules out a negative Term Y. In other words,
a linear status utility that is equally relevant in all states of the world simply cancels out
in our setup. However, the same does not apply when only lagging behind is positionally
relevant (i.e., when w (S) = 0 for S > 0), as > 0 in that case.
A necessary condition for multiple equilibria is a negative Term D (which can only result
from a negative Term Y) over some range. From above, we know with respect to the first
part of Term Y that
c/b < (>) 0 for b > (<) b . From this, a negative Term Y would
follow for all levels of b when = wL (1 p)wn + pwd > (< 0) for b > (<) b . Referring
to the description of the evaluation of the status lottery used before, this implies that if
lower crime decreases average consumption, the status lottery implied by crime must get
less favorable with respect to crime and vice versa.
The difference in expected utility represented by b, c b may have a positive slope
initially (i.e., for low levels of b) and a negative slope ultimately (i.e., for high levels of b).
A Term Y that is initially positive (when b < b ) but negative for high levels of b follows if
> 0 for all b. When only lagging behind is relevant for the individual (i.e., when w (S) = 0
for S > 0), a positive will be possible for w being convex or for w being concave for
S < 0 when combined with a sufficiently small level of the detection probability. The case
in which only lagging behind regarding status is important is very suggestive of a negative
at b close to B for the following reason. Refraining from becoming a criminal implies
almost no status distutility, because average consumption is about the same as legal income.
However, taking advantage of the criminal opportunity would entail a rather hefty status
disutility in the detection state of the world. Regarding the possibility of having multiple
15
In the following, we will refer to an example that we will be useful also in our comparativestatics analysis. We suppose that v(Ij ) = 1 e2Ij , I = 7/4, g = 3/2, s = 1/2, B = 1,
and a uniform distribution for b on [0, 1]. In other words, we make use of a standard CARA
utility function regarding utility from absolute consumption with a coefficient of absolute
risk aversion equal to two. With respect to status utility, we rely on
+ (|S|)u S > 0
w(S) =
(|S|)u S 0
(21)
where the parameter + equals one in the event of above average consumption being relevant
for status concerns (+ = 1) and + = 0 otherwise.
Assuming that + = 0, we obtain Figure 1 when u = 1/2 and p = 25/100. Since
(0, c(0)) > 0, (B, c(B)) < 0, and D < 0 when b, c(b) = 0, this function implies an
equilibrium at b = 0 and one at b = B.5 In contrast, when there are no status concerns (i.e.,
when g = 0), the equilibrium critical benefit level is b = .425. In other terms, this scenario
allows for having a much higher crime rate only due to the presence of positional concerns. At
the same time, when we suppose that the no crime equilibrium materializes in the presence
of status effects, then the comparison points the other way. The use of u = 2 instead of
one half yields comparable results when the detection probability is adjusted accordingly. In
other words, the equilibria that arise for u = 1/2 (such as the case in which b = 0 and b = B
are equilibria) may also arise when status utility has the other curvature.
However, matters change when we assume that offenders are influenced by positional
concerns in all states of the world, i.e., when + = 1. Figure 2 illustrates for u = 1/2 and
5
The lower equilibrium will be interior for a higher detection probability, for example. When p = 55/100,
the two stable equilibria result at b = .07 and b = B.
16
Delta
0.4
0.2
0.2
0.4
0.6
0.8
1.0
-0.2
Figure 1: Multiple crime equilibria; horizontal axis: b, vertical axis: (b, c(b))
Figure 3 for u = 2. Simultaneously enjoying being ahead and suffering from lagging behind
causes (0, c(0)) < 0 and (B, c(B)) > 0, combined with D > 0 when b, c(b) = 0.
Accordingly, these constellations lead to one stable interior equilibrium crime rate that is
higher (when u = 1/2) or lower (when u = 2) than in the scenario without positional
concerns (where b = .425). In other words, potential offenders with status utility that is
convex on status losses and concave on status gains evaluate crime as more attractive than
peers without positional concerns, whereas potential offenders with status utility that is
concave on status losses and convex on status gains evaluate crime as less attractive than
peers without positional concerns.
For the case in which status utility is quadratic, we obtain two stable interior equilibria
by a slight modification of law enforcement parameters. Using p = 75/1000 and s = 1, we
obtain Figure 4. For our specification, we obtain three equilibria with an interior crime rate
(b = .278; b = .4; b = .463), whereas b = .365 results without status concerns. With
regard to the distinction between stable and unstable ones, we argue that stability requires
that D > 0 at the root in question (as described before). This leads us to conclude that we
obtain two stable equilibrium are at b = .278 and b = .463, whereas the equilibrium at
17
Delta
1.2
1.0
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1.0
18
Delta
1.0
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1.0
-0.2
Comparative-statics analysis
In the following, we will present results from a comparative-statics analysis. We will first
consider small variations in the exogenous parameters starting from a stable interior equilibrium. The focus on stable equilibria will be assured by maintaining that D > 0 must
hold. Next, we discuss briefly the possible repercussions that follow from having multiple
equilibria.
The equilibrium critical gain level is a function of the detection probability, the level of
the sanction, the level of legal income, the importance attached to relative standing, and
whether or not being ahead produces status utility (i.e., whether or not w (S) > 0 for S > 0).
Average consumption is directly affected by the first three of these exogenous variables and,
in addition, reacts to changes in the equilibrium critical gain level. In order to arrive at how
the critical benefit level responds to variations, we start from
b (p, s, I, g) , c p, s, I, b (p, s, I, g) = 0
(22)
In the absence of interaction effects, an increase in the level of the detection probability
and the level of the sanction unambiguously lead to an increase in b , which is synonymous
19
Delta
1.0
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1.0
-0.2
Figure 4: Multiple crime equilibria; horizontal axis: b, vertical axis: (b, c(b))
to a decrease in crime frequency. When we consider individuals who are partly motivated
by status considerations, we obtain
c
db
1
vn + gwn (vd + gwd) g
=D
dp
p
db
c
= D 1 p(vd + gwd ) g
ds
s
(23)
(24)
It follows that the analysis of our framework delivers standard conclusions whenever the
weight g attached to status utility or the response of average consumption to changes in law
enforcement is negligible, because both vn + gwn (vd + gwd ) > 0 and p(vd + gwd ) > 0.
However, the fact that both
c/s < 0 and
c/p < 0 implies that a counterintuitive
comparative-statics finding is possible whenever < 0, i.e., whenever the implied change in
the reference consumption makes the crime opportunity more appealing. The intuition for
this result goes as follows: The stricter law enforcement depresses the average consumption
benchmark. This fact influences an individuals status in at least two and possibly three
different states (depending on the level of w (S) for S > 0), and can make it more tempting
to opt for the status gamble implied by undertaking the offense. The condition states that
the marginal expected utility from an improvement in status is lower for the offender than the
20
non-offender. The comparative-statics finding will then arise locally whenever the influence
from the status gamble dominates the standard direct effect. Note that D > 0 always follows
when
c/bg > 0, where
c/b < 0 for b > b . In other words, the necessary condition for
a non-standard comparative-statics result is compatible with the sufficient condition for a
stable equilibrium.
Figure 5 highlights the possibility of the counterintuitive comparative-statics effect for
variations in the level of the detection probability, relying on our exemplary specification
used above and + = 1, s = 1, and u = 1/2. For some intermediate levels of p between 0.3
and 0.4, an increase in the detection probability actually entails a decrease in b, concomitant
with an increase in the crime rate.
1.0
0.8
0.6
0.4
0.2
0.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
1
vL (1 p)vn pvd g
=D
dI
I
(25)
This leads to the conclusion that the curvature of v and of w are critical for determining
the response of the critical gain level to a variation in the level of legal income, which is
ambiguous in general.
Finally, we seek to establish the consequence of increasing the weight attached to status
considerations for the level of crime in equilibrium.
db
= D 1 [wL (1 p)wn pwd ]
dg
(26)
In most circumstances, commission of crime for b will result in higher than average income
in the no-detection state. When w (S) > 0 for S > 0 and therefore wn > 0, we obtain a
finding that is in line with the central result of Konrad and Lommerud (1993). An increase
in the weight g will increase the level of risk-taking (that is, induce more crime) whenever
individuals are relatively less risk-averse regarding variations in status than they are with
respect to variations in the level of absolute consumption. The alternative case in which
wn = 0 rules out this outcome.
Implications of multiple equilibria:
In the comparative-statics analysis described above, we have considered only small variations
starting from a stable equilibrium. In Figure 6, we return to the example introduced at the
end of Section 3 where u = 2, s = 1, + = 1, and p = 75/1000 in order to assert that the
presence of multiple equilibria may allow for discontinuous changes in the level of deterrence.
When increasing the detection probability from .06 to .08, there is a discrete increase in the
level of deterrence. The figure additionally once more shows that status concerns may bring
about more or less crime, as becomes clear from comparing the equilibrium crime rate that is
represented by the strictly monotonous curve with g = 0 to the up to two stable equilibrium
22
crime rates we obtain when g = 3/2. A similar picture would emerge for variations in the
level of the sanction.
1.0
0.8
0.6
0.4
0.2
0.0
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
23
1.0
0.8
0.6
0.4
0.2
0.0
0.0
0.5
1.0
1.5
2.0
Figure 7: Equilibrium critical gains as a function of the weight g; horizontal axis g; vertical
axis b
Welfare
In the following, we briefly conjecture about the implications of positional concerns of potential offenders for welfare and the socially optimal levels of the sanction and the detection
probability. Our discussion will not consider discrete changes in the crime rate that may
be possible when there are multiple equilibria (as discussed at the end of Section 4) but
concentrate on small variations starting from a stable interior equilibrium. The level of
social welfare will be represented by a utilitarian welfare function that takes potential offenders utility into account and specifies that every criminal act implies social harm h, and
is thereby aligned with the standard representation used in the literature (see, e.g., Polinsky
24
b
(1 F (b))h K(p) p(1 F (b)s
(27)
by the use of the sanction s and the detection probability p, where c = c(p, s, I, b) and K(p),
K , K > 0, denotes enforcement costs from which the collected sanctions are deduced.
The change in the level of welfare in response to a variation in one of the law enforcement
parameters will be composed of a direct effect and the consequences of the variation for the
level of the critical benefit level and the level of average consumption. Introducing = s, p
and the indicator variable that is equal to one (zero) should represent p (s), we obtain
Z B
W
b
|
{z
}
x1
(1 ) p
|
Z
(vd
p(1 F (b))
{z
x2
Z B
c
b
b
b
|
{z
}
y
c
F (b)wL +
g
((1
{z
p)wn
(28)
using that = 0 at the critical equilibrium benefit level. The direct effect of the change in
law enforcement is represented either by the term x1 or the term x2 for p and s, respectively.
A higher detection probability makes the loss in utility consequent to a conviction more likely
for actual offenders and increases enforcement costs as well as sanctions collected. A higher
sanction lowers utility for actual offenders in the detection state of the world mirrored by an
increase in revenue for the state. The term y mirrors the consequences of the implied change
25
in the critical benefit level. In addition to the influence on the likelihood of incurring social
harm h and its effect on the total amount of sanctions, the change in the critical benefit level
causes the average consumption level to change. As soon as b > b , it holds that
c/b < 0. A
lower average consumption level implies marginal status gains for all individuals (including
compliant subjects). The same applies to the indirect effect represented by the term z,
which gives the repercussions of the decrease in the average consumption due to the change
in enforcement. In other words, the increase in the strictness of law enforcement has the
direct effect of lowering the level of the comparison consumption. This is beneficial for all
individuals. The negatively affected individuals are concentrated among offenders. These
individuals either suffer from a utility loss with a higher probability (when the level of the
detection probability is increased) or are worse off in the detection state of the world (when
the level of the sanction is increased). However, these adverse consequences are likely to be
welcomed from a social standpoint, given that they imply higher deterrence.
Remark: Assume that the marginal productivity of the detection probability and the sanction
with regard to the level of deterrence measured by b is relatively unaffected by the presence
of status concerns and that standard comparative-statics effects apply. Then, stronger status effects rationalize stricter law enforcement (i.e., relatively higher levels of the detection
probability and the sanction) when
c/b < 0 holds at the optimum and the additional costs
imposed on offenders are not dominant.
It is clear from (28) that stronger status effects (i.e., a higher level of g) magnify the
additional marginal benefits of an increase in the law enforcement parameter represented by
term z and the second part of term y. At the same time, a greater importance attached
to relative standing will imply direct costs on actual offenders (represented by either x1 or
x2 ). The fact that the marginal productivity of the detection probability and the sanction
will sometimes be strongly influenced by the presence of status concerns complicates the
26
Conclusion
Strain creates pressure that requires coping behaviors, one of which is criminal activity.
We consider the case in which a discrepancy between the individual consumption level and
the comparison consumption level causes strain that may induce criminality. The focus on
relative deprivation creates an interdependence between potential offenders that may cause
multiple stable crime rates. When many people resort to crime to improve their expected
income, this entails an unfavorable position for law-compliant individuals in expected terms
and may thereby push more individuals into criminal activity. Likewise, when most individuals refrain from crime, then average consumption is not too different from legal income,
implying little potential for strain. Interestingly, our exploration of the implications of strain
theory unearthes a surprising finding: Positional concerns need not cause more crime. This
is explained by inter alia the fact that incurring a sanction is more detrimental to individuals when it also connotes that they fall more behind with regard to status.
The individual concern for relative position influences how individuals respond to changes
in law enforcement parameters. Indeed, classic results regarding the additional deterrence
effects of higher detection probabilities or higher sanctions no longer follow unambiguously.
The change in law enforcement influences the comparison level of consumption and thereby
may provoke more offenses. When stricter law enforcement in fact increases deterrence, then
our setup presents reasons to implement higher levels of the detection probability and sanction when compared to a setup without status concerns. This finding can be attributed to
the fact that stricter law enforcement lowers the reference consumption level, which thereby
lowers disutility due to an otherwise high comparison standard.
27
References
Agnew, R., 2006a. Strain theory. In: McLaughlin, E., and J. Muncie (Eds.). The Sage
dictionary of Criminology. London: Sage
Agnew, R., 2006b. Pressured into crime: an overview of general strain theory. Cary: Roxbury Publishing.
Alpizar, F., Carlsson, F., and O. Johansson-Stenman, 2005. How much do we care about
absolute versus relative income and consumption? Journal of Economic Behavior and Organization 56, 405-421.
Anderson, C., Kraus, M.W., Galinsky, A.D., and D. Keltner, 2012. The local-ladder effect:
Social status and subjective well-being. Psychological Science 23, 764-771.
Bar-Gill, O., and A. Harel, 2001. Crime rates and expected sanctions: The economics of
deterrence revisited. Journal of Legal Studies 30, 485-501.
Baron, S.W., 2004. General strain, street youth and crime: A test of Agnews revised theory.
Criminology 42, 457-483.
Bond, P., and K. Hagerty, 2010. Preventing crime waves. American Economic Journal:
Microeconomics 2, 138-159.
Benoit, J.P., and M.J. Osborne, 1995. Crime, punishment, and social expenditure. Journal
of Institutional and Theoretical Economics 151, 326-347.
Burdett, K., Lagos, R., and R. Wright, 2003. Crime, inequality, and unemployment. American Economic Review 93, 1764-1777.
Caporale, G.M., Georgellis, Y., Tsitsianis, N., and Y.P. Yin, 2009. Income and happiness
across Europe: Do reference values matter? Journal of Economic Psychology 30, 42-51.
Card, D., Mas, A., Moretti, E., and E. Saez, 2012. Inequality at work: The effect of peer
salaries on job satisfaction. American Economic Review 102, 2981-3003.
Carlsson, F., Johansson-Stenman O., and P. Martisson, 2007. Do you enjoy having more
28
29
Dohmen, T., Falk, A., Fliessbach, K., Sunde, U., and B. Weber, 2011. Relative versus
absolute income, joy of winning, and gender: Brain imaging evidence. Journal of Public
Economics 95, 279-285.
Fajnzylber, P., Lederman, D., and N. Loayza, 2002. Inequality and violent crime. Journal
of Law and Economics 45, 1-39.
Falk, A., and M. Knell, 2004. Choosing the Joneses: Endogenous goals and reference standards. Scandinavian Journal of Economics 106, 417-435.
Fender, J., 1999. A general equilibrium model of crime and punishment. Journal of Economic Behavior and Organization 39, 437-453.
Ferrer, R., 2010. Breaking the law when others do: A model of law enforcement with neighborhood externalities. European Economic Review 54, 163-180.
Frey, B.S., Schmidt, S.L., and B. Torgler, 2008. Relative income position, inequality and
performance: An empirical panel analysis. In: Andersson, P., Ayton, P., and C. Schmidt
(Eds.). Myths and facts about football: The economics and psychology of the worlds greatest sport. Cambridge: Cambridge Scholars Publishing.
Glaeser, E.L., Sacerdote, B., and J.A. Scheinkman, 1996. Crime and social interactions.
Quarterly Journal of Economics 111, 507-548.
Helsley, R.W., and W.C. Strange, 1999. Gated communities and the economic geography of
crime. Journal of Urban Economics 46, 80-105.
Huang, C.C., Laing, D., and P. Wang, 2004. Crime and poverty: a search-theoretic approach. International Economic Review 45, 90-938.
Johansson-Stenman, O., Carlsson, F., and D. Daruvala, 2002. Measuring future grandparents preferences for equality and relative standing. Economic Journal 112, 362-383.
Kelly, M., 2000. Inequality and crime. Review of Economics and Statistics 82, 530-539.
Kleiman, M., 2009. When brute force fails. Princeton: Princeton University Press.
30
Konrad, K.A., and K.E. Lommerud, 1993. Relative standing comparisons, risk taking, and
safety regulations. Journal of Public Economics 51, 345-358.
Lee, D.R., and J.W. Cohen, 2008. Examining strain in a school context. Youth Violence
and Juvenile Justice 6, 115-135.
Marceau, N., and S. Mongrain, 2011. Competition in law enforcement and capital allocation.
Journal of Urban Economics 69, 136-147.
Merton, R.K., 1938. Social structure and anomie. American Sociological Review 3, 672-682.
Merton, R.K., 1968. Social theory and social structure. New York: Free Press.
Messner, S.F., and R. Rosenfeld, 2001. Crime and the American dream. Belmont: Wadsworth.
Polinsky, A.M., and S. Shavell, 2001. Corruption and optimal law enforcement. Journal of
Public Economics 81, 1-24.
Polinsky, A.M., and S. Shavell, 2009. Public enforcement of law. In: Garoupa, N. (Ed.).
Criminal Law and Economics. Northhampton, MA: Edward Elgar.
Rasmusen, E.B., 1996. Stigma and self-fulfilling expectations of criminality. Journal of Law
and Economics 29, 519-543.
Robson, A.J., 1992. Status, the distribution of wealth, private and social attitudes to risk.
Econometrica 60, 837-857.
Sah, R.K., 1991. Social osmosis and patterns of crime. Journal of Political Economy 94,
1272-1295.
Schrag, J., and S. Scotchmer, 1997. The self-reinforcing nature of crime. International Review of Law and Economics 17, 325-335.
Senik, C., 2009. Direct evidence on income comparisons and their welfare effects. Journal
of Economic Behavior and Organization 72, 408-424.
Silverman, D., 2004. Street crime and street culture. International Economic Review 45,
761-786.
31
Solnick, S., and D. Hemenway, 1998. Is more always better?: A survey on positional concerns. Journal of Economic Behavior and Organization 37, 373-383.
Solnick, S., and D. Hemenway, 2005. Are positional concerns stronger in some domains than
in others? American Economic Review 95, 147-151.
Solnick, S., Hong, L., and D. Hemenway, 2007. Positional goods in the United States and
China. Journal of Socio-Economics 36, 537-545.
Stiles, B.L., Liu, X., and H.B. Kaplan, 2000. Relative deprivation and deviant adaptations:
The mediating effects of negative self-feelings. Journal of Research in Crime and Delinquency
37, 64-90.
Stutzer, A., 2004. The role of income aspirations in individual happiness. Journal of Economic Behavior and Organization 54, 89-109.
Verdier, T., and Y. Zenou, 2004. Racial beliefs, location and the cause of crime. International Economic Review 45, 731-760.
Whitworth, A., 2012. Inequality and crime across England: a multilevel modeling approach.
Social Policy and Society 11, 27-40.
Whitworth, A., forthcoming. Local Inequality and crime: exploring how variation in the
scale of inequality measures affects relationships between inequality and crime. Urban Studies.
Young, J., 2006. Relative deprivation. In: McLaughlin, E., and J. Muncie (Eds.). The Sage
dictionary of Criminology. London: Sage.
32