Skip navigation

The Prison Journal, Crime Pays - the Connection Between Time in Prison and Future Criminal Earnings, 2012

Download original document:
Brief thumbnail
This text is machine-read, and may contain errors. Check the original document to verify accuracy.
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/258194428

Crime Pays The Connection Between Time in Prison
and Future Criminal Earnings
Article in The Prison Journal · September 2012
DOI: 10.1177/0032885512448607

CITATIONS

READS

23

396

1 author:
Donald Hutcherson
Ohio University - Lancaster Campus
5 PUBLICATIONS 68 CITATIONS
SEE PROFILE

All content following this page was uploaded by Donald Hutcherson on 28 January 2015.

The user has requested enhancement of the downloaded file.

448607
2448607HutchersonThe Prison Journal
© 2012 SAGE Publications

TPJ92310.1177/003288551

Reprints and permission:
sagepub.com/journalsPermissions.nav

Crime Pays: The
Connection Between
Time in Prison and
Future Criminal
Earnings

The Prison Journal
92(3) 315­–335
© 2012 SAGE Publications
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0032885512448607
http://tpj.sagepub.com

Donald T. Hutcherson II1

Abstract
This study draws on theories of stigma, social and human capital, and opportunity structure to assess the role of prior incarceration on illegal earnings.
Tobit regression models are estimated for young adult ex-offenders and
nonoffenders using the National Longitudinal Survey of Youth for 1997 to
2005. The findings reveal that individuals with an incarceration history earn
significantly higher annual illegal earnings than those who do not have such
a history. This is true net a variety of predictors of illegal income, including
race and ethnicity. The current research indicates that spending significant
time in jail or prison may force the ex-incarcerated into illegal opportunity
structures to obtain income.
Keywords
young adult offenders, prior incarceration, illegal earnings

Introduction
There are two paths that young adults involved in criminal activity can take
as they make the transition to adulthood. Most young adults are drawn into
1

Ohio University Department of Sociology and Criminal Justice, Lancaster, OH, USA

Corresponding Author:
Donald T. Hutcherson, Ohio University Department of Sociology and Criminal Justice,
Lancaster Campus, 524 Brasee Hall 1570 Granville Pike, Lancaster, OH 43130, USA
Email: hutchers@ohio.edu

316		

The Prison Journal 92(3)

conventional society by moving through a sequence of traditional life course
stages (e.g., completing high school, entering college or the military, gaining
conventional employment, getting married, having children, etc.). These life
course stages integrate young adults into mainstream society, and offer adult
offenders a way out of a life of crime. Some suggest that those involved in
crime that can find steady work and a stable marriage also become embedded
in a web of social supports and obligations (Sampson & Laub, 1993;
Western, 2006). It is these social bonds that help young adult criminals
refrain from furthermore offending.
The second path that crime-involved young adults can take does not end
so positively. For many Americans, incarceration has become a key life event
that can harmfully alter traditional life course stages. At the end of 2006, the
Nation’s jail and prison population stood at more than two million persons
(Sabol, Minton, & Harrison, 2007). This means that one out of every 150
U.S. residents is in jail or prison. The current United States rate of incarceration of 726 inmates per 100,000 population is the highest of any country in
the world (Garland, 2001).
At least 95% of all state prisoners will be released from prison at some
point (Hughes, Wilson, & Beck, 2001). Close to 70% of these offenders will
be rearrested in 3 years or less. It is evident that the ex-incarcerated have a
difficult time becoming a part of mainstream society. The story of what happens to these individuals after release from prison is not fully developed in
the research literature. However, we know that employment and related
income is a key factor in determining the direction of the life course of the
ex-incarcerated (see, for example, Sampson & Laub, 1993; Sampson & Laub,
2003; Western & Beckett, 1999; Western, 2002). This research highlights
that conventional employment and related income is a path out of crime for
young adults. These same studies reveal that conventional employment and
related income is difficult to obtain for the ex-incarcerated. Although the
affect of imprisonment on conventional employment prospects and related
earnings is clear, what is less clear is the extent to which imprisonment affects
opportunities in the illegal economy, specifically illegal earnings.

Theory Linking
Incarceration and Illegal Earnings
There are several causal mechanisms that explain how incarceration can lead
to increased illegal earnings. First, formerly incarcerated offenders are stigmatized by their incarceration past. The literature suggests that employers
are less likely to hire the ex-incarcerated compared with those without prison

Hutcherson

317

records (Boshier & Johnson, 1974; Buikhuisen & Dijkster-huis, 1971;
Holzer, 1996; Pager, 2003). A combination of criminal history and race can
be especially stigmatizing for many ex-incarcerated men of color (Pager,
2003). Second, due to spending significant time incarcerated, these individuals are prevented from acquiring human capital, or the job skills and experience necessary for conventional labor market success (Becker, 1968; Holzer,
Raphael, & Stoll, 2003; Kling, 1999). Third, spending significant time incarcerated can erode the social networks necessary for stable conventional
employment opportunities (Coleman, 1988; Hagan, 1993). Consequently,
due to the stigma of incarceration and race, and a lack of human and social
capital, the ex-incarcerated may be forced into illegal opportunity structures
that yield high illegal earnings. Cloward and Ohlin (1960) suggest that individuals are faced with two opportunity structures, one legitimate and the
other illegitimate. For those formerly incarcerated offenders that are denied
entry and success in the conventional labor market, illegitimate opportunity
structures and related criminal earnings may be an attractive and lucrative
option. This study will integrate these theoretical perspectives when analyzing the relationship between incarceration and illegal earnings.

The Present Study: Testing the Effect
of Past Incarceration on Illegal Earnings
In light of the paucity of research on the influence of past incarceration on
criminal earnings, this study will address the following research question:
How does incarceration influence criminal earnings for young adults? This
study estimates tobit regression models to examine criminal earnings for
young adult ex-offenders and nonoffenders using the National Longitudinal
Survey of Youth (NLSY97) for the years 1997 through 2005. This study
extends the research on the affect of incarceration on legal earnings. In sum,
these studies reveal that spending time in prison can lead to reduced employment and earnings in the conventional labor market (e.g., see Huebner, 2005;
Johnson, 2003; Sampson & Laub, 1993; Sampson & Laub, 2003; Western &
Beckett, 1999; Western, 2002). However, due to data and other limitations,
these studies fail to analyze the effect of incarceration on illegal earnings.
To date, only two studies have tested the relationship between incarceration and illegal earnings. In the first study, Levitt and Venkatesh (2001)
produce research on the illegal earnings of Chicago street gangs. These
researchers reveal that the formerly incarcerated are more likely to participate
in drug trafficking than individuals never incarcerated. In the second study,
Uggen and Thompson (2003) analyze a sample of ex-drug addicts and

318		

The Prison Journal 92(3)

offenders to predict month-to-month changes in illegal earnings. They find
that spending significant time in prison may reduce illegal earnings in the
short-term, because incarcerated offenders are blocked from earning income
outside of jail or prison. Unfortunately, both of these studies have significant
conceptual and methodological flaws that prevent them from being representative of populations beyond their limited samples. As a result, very little is
known about the influence of incarceration on illegal earnings once individuals are released from jail or prison back into the community. The current
research will begin to fill this literature gap.

The Sample
To examine the relationship between past incarceration and illegal earnings,
data from the NLSY97 were used, that is, the most recent survey in the
National Longitudinal Surveys program. The survey documents the transition from school to work for adolescents and young adults. The NLSY97
consists of two samples: (a) A cross-sectional sample of 6,748 respondents
designed to be representative of people living in the United States during the
initial survey round and born between January 1, 1980, and December 31,
1984; and (b) a supplemental sample of 2,236 respondents designed to oversample Hispanics and African Americans living in the United States during
the initial survey round and born during the same period as the crosssectional sample (Center for Human Resource Research, 2003). In sum, the
NLSY97 cohort includes 8,984 individuals.
This study uses nine rounds of the NLSY97 survey (1997 to 2005). It
contains detailed information on self-reported criminal behavior and subsequent criminal justice responses for young adults, including data on arrests,
convictions and incarceration experiences of the sample’s respondents. The
NLSY97 also includes data on the labor market experiences of its subjects,
both in the conventional labor market and from criminal activity.
Consequently, the NLSY97’s longitudinal design provides a unique opportunity to study the consequences of incarceration on both illegal and conventional labor market experiences of young adults. The first wave of the
NLSY97 includes adolescents aged 12 to 16. These same individuals are
aged between 20 and 24 by wave nine in 2005.
This study is restricted to a sample of young adults 18 years of age and
older. As stated in the theoretical section, the influence of human and social
capital is crucial during this stage of the life course. This study also uses a
person-period data structure. One of the advantages of using a person-period
data format is that individuals do not have to be excluded entirely if they are

Hutcherson

319

missing some observations on the dependent variable (see, for example,
Allison, 1994; Johnson, 2003). On average, each respondent in the sample
contributed 7.3 observations to the data set. The sample size for the personperiod dataset over the entire 9 year sampling period is 46,178 observations.
It should also be noted that nonrandom sample attrition can bias the analysis
of panel data using longtime periods (Western, 2002). However, furthermore
analysis of attrition from this sample finds that response rates are almost
identical for the ex-incarcerated versus never incarcerated individuals.

The Measures
Dependent Variable
Illegal income. Table 1 introduces the dependent, independent, and control
variables used in this analysis. For this study, the amount of raw illegal
income is taken from follow-up questions in each wave/round regarding
criminal behavior during the previous 12 months. If the respondent committed remunerative crimes (e.g., property crimes, drug trafficking, etc.) during
this period, they are asked about any monetary rewards (the total cash
received or the total cash he or she would have received) from selling these
items within the last year. There are three categories of illegal income captured in the NLSY97. First, respondents are asked about the frequency of
theft offenses over the last year and the amount of cash they received for the
items stolen or would have received if they had sold them. Second, respondents are queried about the frequency of activity in other property crimes
during the last year (e.g., fencing, receiving/possessing/selling stolen property, or cheating someone by selling them something that was worthless or
worth much less than what they stated). For these other property offenses,
respondents are also asked about the total cash income received from these
crimes. The final category of illegal earnings activity in the NLSY97 is the
frequency of drug selling activity by respondents in the last year and the
amount of cash income made from selling drugs. Annual raw illegal income
in each wave/round is calculated by adding all monetary rewards received
from these three categories of illegal earnings during the previous 12 months.
A note regarding zero earners. A debate that exists in the illegal earnings
literature is how to code zero earners, or those subjects that claim that they
have no income during a specified period. Some contend that whether to
restrict analyses to a minimum amount (e.g., US$1 or US$100), or include
zero earners is important conceptually to any study (Hauser, 1980; Uggen &
Thompson, 2003; Western, 2002). By counting zero earners, the earnings

320		

The Prison Journal 92(3)

Table 1. Descriptions of Dependent, Independent, and Control Variables, NLSY97
Variable
Dependent variable
 Annual illegal
income

Description

Year measured

Raw annual illegal income based on three
sources of criminal activity in a given year:
(1) theft offenses, (2) other property
crimes, and (3) drug trafficking.

All years

Independent variables
Prior incarceration Dummy variable for those spending at
least one month in jail or prison. Those
incarcerated in year t-1 or earlier are
coded as 1; those not incarcerated in
year t-1 or earlier are coded as 0.
Control variables
Prior illegal income Raw illegal income in year t-1 or earlier.
Illegal income is based on three sources
of criminal activity in a given year: (a)
theft offenses, (b) other property crimes,
and (c) drug trafficking.
 Current
Dummy variable for those spending at least
incarceration
one month in jail or prison in year t (the
last year). Those incarcerated in year t
are coded as 1; those not incarcerated in
year t are coded as 0.
 Current school
Dummy variable for full-time attendance in
attendance
junior high school, high school or college
in the last year. Those attending full time
are coded as 1; those not attending or
missing significant time are coded as 0.
Hardcore drug use Count of the frequency of use of powder
cocaine, crack, heroin and other drugs in
the last year.
Human capital
Legal income
Annual legal income from wages and salary
in the last year.
Employment status Dummy variable for employment in the
last year. Those employed are coded as 1;
those not employed are coded as 0.
ASVAB scores
Percentile score on the Armed Services
Vocational Aptitude Battery (ASVAB).
Scores range b/w 0 and 100.

All years

All years

All years

All years

All years

All years
All years
1999

(continued)

321

Hutcherson
Table 1. (continued)
Variable
Social capital
Criminal peers

Gang membership

Significant other

Demographic
variables
Age
Race/ethnicity
Gender

Description

Year measured

Dummy variable for friend or sibling gang
involvement in the last year. Those with
criminal peers coded as 1; those without
criminal peers coded as 0.
Dummy variable for respondent gang
involvement in the last year. Gang
members coded as 1; nongang members
coded as 0.
Dummy variable for being involved in a
relationship with a girlfriend or spouse in
the last year. Those with S/O coded as 1;
those without S/O coded as 0.

All years

Age at the time of the interview in years.
Dummy variables for non-Hispanic Blacks
and Hispanics. Black or Hispanic coded as
1; non-Black or non-Hispanic coded as 0.
Dummy variable for gender. Males coded as
1 and females coded as 0.

All years
1997

All years

All years

1997

distribution can be skewed and important questions can be raised about sample selectivity. The drawback of this approach to measuring earnings is that
one ignores unemployed individuals. This study includes zero earners in the
analysis of illegal earnings, as doing so highlights the distinction between
criminals and noncriminals.

Independent Variables
Prior incarceration. Prior incarceration is considered the primary independent variable for the models in this study. Information on crime and arrest in
the NLSY97 is collected in the self-administered section of the youth/young
adult instrument. For each wave/round, respondents are asked about criminal
behavior during the last year, including behavior that leads to official criminal justice processing. For each crime that results in an arrest, respondents are

322		

The Prison Journal 92(3)

asked about the sanction that is given, including arrests that lead to juvenile
and adult jail or prison time. Therefore, NLSY97 data are good for comparing the incarceration experiences of young adults with both never incarcerated offenders and nonoffenders. The prior incarceration measure is a dummy
variable, with those spending at least one month or more in jail or prison as a
juvenile and/or adult in the year t – 1 or earlier coded as 1. Respondents in the
study who did not spend significant time in jail and/or prison as a youth or
young adult during this period are coded as 0.

Control Variables
A number of control variables are included in the analyses because prior
research has found them to be associated with criminal earnings.
Prior illegal income. The primary control variable in this analysis is prior
illegal income, as there could be a spurious relationship between incarceration and illegal income. It is very likely that illegal income earned prior to
incarceration could explain both incarceration and present illegal income, so
controlling for prior illegal income will highlight the independent effect of
incarceration on illegal income for formerly incarcerated offenders. Prior
illegal income is calculated by adding all monetary rewards received from
illegal earnings during the years t – 1. Current incarceration. The current incarceration measure accounts for the contemporaneous effect of incarceration
on the respondent’s ability to earn illegal income. This is a dummy variable
scored 1 for respondents who spent at least 1 month or more in jail or prison
in year t, and 0 otherwise. Current school attendance. It is suggested earlier
that being confined in a secure environment such as jail or prison during the
same year that respondents earn illegal income reduces their ability to earn
illegal income. The same is argued for spending significant time attending
school. Full-time students have much less time to earn illegal income compared with individuals not in school full-time. The current school attendance
variable is a dummy variable that captures full-time attendance in high school
or college. Individuals attending school full-time in these educational settings with close to perfect attendance records are coded as 1, while those not
attending school or missing a significant number of months of school are
coded as 0.
Hardcore drug use. All respondents in the NLSY97 are surveyed on their
experience with marijuana, powder cocaine, crack, heroin, and other substances not prescribed by a doctor and used in order to get high or achieve an
altered state. The substance abuse measure in this study is a count of how
often subjects used hardcore drugs (powder cocaine, crack, heroin, etc.) during the survey year.

Hutcherson

323

Human capital. Conventional human capital captures ability and work
experience at the individual-level. Conventional human capital measures
used in this study are described below.
Legal income. The amount of raw legal income used in this study is collected from a NLSY97 question asking respondents to provide all income
from wages and salary in the last year.
Employment status. Employment status is measured in this study based on
a question inquiring whether the respondent received salary from conventional employment in the months prior to the interview. Employment status is
dummied, with those employed coded as 1 and the nonemployed coded as 0.
ASVAB scores. As a measure of conventional human capital, scores from
the Armed Services Vocational Aptitude Battery (ASVAB), a national
achievement test, will be controlled for in this study. In round one of the
NLSY97, most respondents participated in the administration of the ASVAB.
The NLS Program staff computed a percentile score to represent the average
performance on both the math and verbal sections of the ASVAB. ASVAB
scores range between 0 and 100, with higher scores suggesting greater
achievement. These scores are included in this analysis.
Social capital. The measures for social capital are described below.
Criminal peers. To measure the type of social capital or networks that
would be more likely to influence criminal earnings, this analysis includes
direct measures of criminal peer associations. The criminal peer measure in
this analysis is a dummy variable taken from a question that asks if the
respondent’s siblings or friends belonged to a criminal gang in the previous
year. Those with siblings or friends who participated in gang activity are
coded as 1, and respondents without gang-involved siblings and friends are
coded as 0.
Gang membership. As a measure of criminal social capital, respondent
gang membership represents a good proxy variable for the influence of criminal peers. The gang membership measure used in this study is a dummy variable taken from a question asking if the respondent belonged to a criminal
gang in the previous year. Respondents involved in a gang are coded as 1,
with those not involved in gang activity coded as 0.
Significant other. As a measure of social capital, the significant other measure used in this study is taken from a NLSY97 question asking how attached
or close respondents felt toward their girlfriend or spouse in the previous
year. This study measures significant other as a dummy variable. Thus,
respondents with a significant other are coded as 1, and those without a significant other are coded as 0 in this research.
Demographics/age. Age is measured here as the value of age of the respondent in year t (at the time of the interview).

324		

The Prison Journal 92(3)

Race/ethnicity. The race and ethnicity of each respondent in the NLSY97 is
identified separately from the first wave/round of the study. The ethnicity
question identifies individuals of Hispanic origin. For the purposes of this
analysis, each category of race and ethnicity is measured as a dummy variable. African American is coded as 1, and is distinguished from Whites
(coded as 0). Hispanic is coded as 1, and is distinguished from non-Hispanic
Whites and African Americans (coded as 0).
Gender. For the variable gender, males are coded as 1, while females are
coded as 0.

Analytic Strategy
This study estimates and compares both random effects and tobit regression
models to examine illegal income for young adult ex-offenders and nonoffenders using the NLSY97 for 1997 to 2005. The two most common analytic
strategies considered for a longitudinal, person-period data format are
random-effects models and fixed-effects models. Random effects models are
selected over fixed-effects models in this study because variables with
unchanging values cannot be used in a fixed-effects model. Race and ethnicity (as measures of stigma) are two such variables with unchanging values
over time. As these variables are crucial to this study’s theoretical model,
random effects models will be used as the analytic strategy of this research
(Johnston & DiNardo, 1997; Long & Freese, 2003; Wooldridge, 2002).
Although random effects models are useful for analyzing longitudinal,
person-period data with unchanging values over time (see, for example,
Johnston & DiNardo, 1997; Long & Freese, 2003; Wooldridge, 2002), tobit
regression techniques are useful when the dependent variable consists of a
large proportion of zero values. Close to 10% of the study’s 46,178 observations earn illegal income over the 9-year sampling period. Tobit regression
addresses the limited floor value of the dependent variable in this analysis,
illegal income, by censoring all cases with zero values (Roncek, 1992).
Therefore, cases with real dollar values can be analyzed. Beta estimates in
tobit regression represent the marginal effect of x on y*, the latent variable
(observed illegal income amounts in this study), and not y.

Results
Table 2 presents means and standard deviations of the dependent, independent, and control variables used in the analyses for the total sample and for
the ex-incarcerated compared with the never incarcerated. The ex-incarcerated,

325

Hutcherson

Table 2. Means and Standard Deviations (in Parentheses) of Dependent,
Independent, and Control Variables, NLSY 1997-2005 (N = 46,178 observations)

Variable

Ex-incarcerated
(U.S. dollar)

Never
incarcerated(U.S.
dollar)

Total sample
(U.S. dollar)

Dependent variable
Annual illegal income $1,070 ($8,986)
$120 ($2,623)
$162 ($3,195)
Independent variables
Prior incarceration
—
—
—
Control variables
Prior illegal income
$20,801 ($100,145) $1,362 ($26,799) $2,230 ($33,902)
Current
0.07 (0.25)
0.01 (0.08)
0.01 (0.10)
incarceration
School attendance
0.97 (0.18)
0.91 (0.29)
0.91 (0.29)
Hardcore drug use
6.61 (43.80)
1.91 (22.83)
2.12 (24.17)
Human capital
Annual legal income
$4,604 ($8,988)
$5,311 ($9,237) $5,278 ($9,226)
Employment status
0.38 (0.49)
0.50 (0.50)
0.49 (0.50)
ASVAB scores
25.54 (22.86)
45.98 (29.10)
45.16 (29.15)
Social capital
Criminal peers
0.12 (0.33)
0.05 (0.22)
0.05 (0.23)
Gang membership
0.05 (0.22)
0.01 (0.09)
0.01 (0.10)
Significant other
0.27 (0.44)
0.17 (0.37)
0.17 (0.38)
Demographic variables
Age
21.23 (1.98)
20.56 (1.94)
20.59 (1.96)
Race
  White
0.39 (0.49)
0.52 (0.50)
0.52 (0.50)
   African American
0.35 (0.48)
0.26 (0.44)
0.26 (0.44)
  Hispanic
0.24 (0.43)
0.21 (0.41)
0.21 (0.41)
Gender
0.79 (0.41)
0.50 (0.50)
0.51 (0.50)

on average, have higher annual illegal income than the never incarcerated
(US$1,070 vs. US$120, respectively). The ex-incarcerated have also accumulated US$20,801 of past illegal income, compared with US$1,362 for
those never incarcerated. Of those with an incarceration history, 7% are
incarcerated during the year of the interview (year t). In contrast, only 1% of
those never incarcerated prior to year t are incarcerated during the year of the
interview. The ex-incarcerated are more likely to use hardcore drugs (6.61 vs.
1.91 on the use frequency scale) than those never incarcerated.

326		

The Prison Journal 92(3)

Based on the summary statistics, the ex-incarcerated earn less legal income
annually from wages than their counterparts who were never incarcerated
(US$4,604 vs. US$5,311, respectively). Also, the ex-incarcerated are less
likely to be employed than those who were never incarcerated (38% vs. 50%,
respectively). Regarding social capital measures, the ex-incarcerated are
much more likely to have criminal peers that are involved in gangs (12% vs.
5%, respectively). The ex-incarcerated are also involved in gangs more frequently than those never incarcerated (5% vs. 1%, respectively). The formerly incarcerated are much more likely to have a significant other (27%)
than those never incarcerated (17%).
This table also shows that the ex-incarcerated in this sample are slightly
older compared with those never incarcerated (21.23 years vs. 20.56 years,
respectively). Finally, while White male respondents consist of 52% of the
total sample, they comprise a much smaller percentage of those ever incarcerated (39%). Conversely, while African Americans make up 26% of the
overall sample, they consist of a much higher percentage of the ex-incarcerated
(35%). Compared with the overall sample, the percentage of Hispanics that
are ex-incarcerated is slightly higher (21% vs. 24%, respectively). Finally,
while males make up roughly one half of the entire sample, they consist of a
much higher percentage of those incarcerated in the past (79%) compared
with females.

Main Model Comparison: Random
Effects Versus Tobit Regression
For comparison with the tobit model, this study conducted a random effects
regression analysis of the data. In both the random effects and tobit regression main model, annual illegal income is predicted to be a product of incarceration, net of other predictors of illegal income. To predict the amount of
annual illegal income in raw dollars from respondents in the sample, the
following predictors are considered: past illegal income, past incarceration,
current incarceration, school attendance, substance abuse, measures of
human and social capital, age, race/ethnicity, and gender. Table 3 shows the
unstandardized coefficients and the standard errors (in parentheses) for both
the random effects and tobit regression of annual income on incarceration.
The statistically significant predictors of illegal income in the random
effects model are past illegal income, past incarceration, current incarceration, hardcore drug use, legal income, ASVAB scores, criminal peers, gang
membership, significant other, age, Hispanic origin, and gender. The main
effects results show that the past incarceration and illegal earnings relationship

327

Hutcherson

Table 3. Unstandardized Coefficients From the Regression of Annual Illegal
Income on Incarceration, Random Effects Versus Tobit Models, NLSY Men, 1997 to
2005
Random effects
model
Variable
Intercept
Past illegal income
Past incarceration
Current incarceration
School attendance
Hardcore drug use
Human capital
Legal income
Employment status
ASVAB scores
Social capital
Criminal peers
Gang membership
Significant other
Demographic variables
Age
Race
  White
   African American
  Hispanic
Gender
R2
Number of observations

Tobit model

b

SE

B

SE

760***
0.01***
471***
1,795***
32
11***

198
0.01
107
168
52
0.64

9,613***
0.03***
6,294***
11,729***
–0.558
63***

2,176
0.01
667
992
494
3

1***
33
–2**

1
51
0.88

0.01*
973**
12*

0.02
409
7

343***
3,294***
98*

76
174
47

7,862***
10,964***
–269*

544
965
497

–0.37***

10

–1,904***

114

—
–25
–100*
185***

60
63
46

.05
37,338

—
–2,619***
462
–1,764***
463
5,171***
360
.04 (pseudo R2)
37,338

Note: Standard errors are in parentheses.
*p < .05. **p < .01. ***p < .001 (one-tailed).

is statistically significant at the .001 level. Those with a past incarceration
earn statistically significantly higher illegal income than those who were
never incarcerated. In fact, the ex-incarcerated earn just US$471 more illegal
income than those never incarcerated, on average.
The random effects regression model also reveals that individuals that use
hardcore drugs earn significantly higher illegal income than those who do not

328		

The Prison Journal 92(3)

use such drugs. In terms of human capital variables, there is a significant and
positive relationship between legal and illegal earnings. Those with lower
ASVAB scores also earn significantly higher illegal income than individuals
with higher scores. Regarding the social capital measures, those with criminal peers earn significantly higher illegal income compared with those without criminal peers. Gang members, too, earn significantly higher illegal
income than nongang members. Individuals with a significant other earn
higher illegal income than those without a significant other. For the demographic variables, younger respondents earn significantly higher illegal
income than older individuals. Non-Latinos earn significantly higher illegal
income than Latinos. Finally, males earn significantly higher illegal income
than females.
This study then conducted a tobit regression analysis of the data (Table 3).
The tobit regression analysis, which modeled the underlying amount of illegal income earned, proved to be more robust than the random effects regression analysis. Focusing exclusively on the main effects results from the tobit
regression model, it is found that the past incarceration and illegal earnings
relationship is statistically significant at the .001 level. Those with a past
incarceration earn significantly higher illegal income than those who were
never incarcerated. In fact, the ex-incarcerated earn US$6,294 more illegal
income than those never incarcerated, on average. The coefficients for the
tobit regression model are consistently much larger than the coefficients from
the random effects regression model.
The tobit regression analysis above reports more predictors of annualized
illegal income compared with the random effects model. The statistically significant predictors of illegal income in the tobit model are past illegal income,
past incarceration, current incarceration, hardcore drug use, legal income,
employment status, ASVAB scores, criminal peers, gang membership, significant other, age, race/African American, Hispanic origin and gender.
Individuals that use hardcore drugs earn significantly higher illegal income
than those who do not use such drugs. In terms of human capital variables,
there is a significant and positive relationship between legal and illegal earnings. Employed individuals earn significantly higher illegal income than
unemployed respondents, although this same relationship is nonsignificant in
the random effects regression model. Finally, those with lower ASVAB
scores earn significantly higher illegal income than individuals with higher
ASVAB scores.
Regarding the social capital measures in the tobit regression model, those
with criminal peers earn significantly higher illegal income compared with
those without criminal peers. In addition, gang members earn significantly

Hutcherson

329

higher illegal income than nongang members. Individuals without a significant other earn higher illegal income than those without a significant other.
For the demographic variables, younger respondents earn significantly higher
illegal income than older individuals. Non-African Americans earn significantly higher illegal income than African Americans in the tobit model,
although this same relationship is nonsignificant in the random effects model.
Non-Latinos earn significantly higher illegal income than Latinos. Finally,
males earn significantly higher illegal income than females.
Interaction effects are appropriate when there is reason to believe that the
affect of a given independent variable may depend or be conditional on
another independent variable (Aiken & West, 1991). The interaction effects
tobit regression analysis (Table 4, Model 2) reveals that the only statistically
significant interactions are between Prior incarceration × African American,
Incarceration × Legal income, and Incarceration × Gang membership.
Specifically, the model shows that the Prior incarceration × African American
interaction is statistically significant and positive at the .05 level. Also, the
Prior incarceration × Legal income interaction is statistically significant and
positive at the .05 level. The strongest interaction is between Prior incarceration × Gang membership, which is significant and negative at the .001 level.

Discussion
This study compares random effects models with tobit regression models to
examine illegal earnings for young adult ex-offenders and nonoffenders
using the NLSY97. Specifically, this study is interested in whether individuals with an incarceration history earn higher illegal income than those without an incarceration history.
This study hypothesized that the formerly incarcerated will earn significantly higher illegal income than individuals never incarcerated, controlling
for other predictors of illegal income. The analysis clearly shows that individuals with an incarceration history earn significantly higher annual illegal
income from criminal activity compared with respondents without an incarceration history. This is true when controlling for several predictors of illegal
income. Also, respondents that use hardcore drugs earn significantly more
illegal income than those that do not use hardcore drugs. This finding concurs
with prior research offering evidence for a strong, positive relationship
between serious drug use and illegal earnings (Fagan, 1992; Uggen &
Thompson, 2003).
This study presents evidence that human and social capital measures are
linked to annual illegal income. In terms of human capital, the relationship

330		

The Prison Journal 92(3)

Table 4. Unstandardized Coefficients From the Regression of Annual Illegal
Income on Incarceration, Main and Interaction Effects of Tobit Models, NLSY 1997
to 2005
Model 1
Variable
Intercept
Past illegal income
Past incarceration
Current incarceration
School attendance
Hardcore drug use
Human capital
Legal income
Employment status
ASVAB scores
Social capital
Criminal peers
Gang membership
Significant other
Demographic variables
Age
Race
  White
   African American
  Hispanic
Gender
Interactions
Past incarceration × Drug use
Past incarceration × African American
Past incarceration × Hispanic
Past incarceration × Gender
Past incarceration × Legal income
Past incarceration × Employment status
Past incarceration × ASVAB
Past incarceration × Criminal peers
Past incarceration × Gang membership
Past incarceration × Significant other
Sigma (ancillary parameter)
Pseudo R2
Number of observations

Model 2

b

SE

B

SE

9,613***
0.03***
6,294***
11,729***
–558
63***

2,176
0.01
667
992
494
3

9,552***
0.03***
8,566***
11,852***
–567
64***

2,183
0.01
1,826
996
494
4

0.01*
973**
12*

0.02
409
7

0.01*
971**
10*

0.02
424
7

7,862***
10,964***
–269*

544
965
497

7,778***
12,514***
–298*

578
1,079
532

–1,904***

114

–1,893***

114

—
–2,619***
–1,764***
5,171***

—
462
463
361

—
–2,899***
–1,567***
5,209***

—
485
1,717
370

—
—
—
—
—
—
—
—
—
—
230

–10
9
3,398* 1,656
–1,517 1,717
–2,117 1,591
0.15*
0.08
139 1,571
35
29
–277 1,656
–6,742*** 2,438
–130 1,464
14,439
230
.05
37,338

—
—
—
—
—
—
—
—
—
—
14,463
.04
37,338

Note: All regressions are estimated using tobit regression.
*p < .05. **p < .01. ***p < .001 (one-tailed).

Hutcherson

331

between logged legal and logged illegal earnings is positive and statistically
significant. Also, those that are employed earn significantly higher illegal
income than those that are unemployed. Finally, this study finds that respondents with high ASVAB scores earn significantly more illegal income compared with those with low ASVAB scores. One can conclude based on the
above findings that, although studies have found that the ex-incarcerated are
prevented from acquiring human capital due to time spent in jail or prison
(Holzer, Raphael, & Stoll, 2003; Kling, 1999), the same human capital variables that lead to success in the conventional labor market also lead to success in the illegal economy.
Regarding the social capital measures, there is strong evidence from this
study that having criminal peers and gang membership increases criminal
earnings. This is consistent with research that shows that the development of
criminal social capital, or associations with skilled offenders, is important for
offenders involved in crime as a source of income (McCarthy & Hagan,
2001). Also, this study finds that respondents with a significant other earn
less illegal income than those without a significant other. This result supports
research that show that strong social bonds to spouses help to facilitate the
criminal desistance process (Horney, Osgood, & Marshall, 1995; Laub,
Nagin, & Sampson, 1998; Sampson & Laub, 1993).
The demographic variables highlight that males earn significantly higher
illegal income than females. This is not surprising, as prior research suggests
that young men are more involved in the underground economy than women
(Freeman, 1996; Short, 1997). Also, younger respondents earn significantly
higher illegal income than older individuals. Keep in mind that the sample
consists of young adults that are 18 years of age or older. Life course criminology has produced some revealing facts about crime. First, there exists an
age–crime curve (Farrington, 1986). This curve shows that the peak age of
criminal activity is 17 years old, whereas the peak age of desistance is
between 20 and 28 years (Farrington, 2003). It is expected that younger
adults would earn more income from crime than their older adult counterparts. Finally, non–African Americans and non-Latinos earn more illegal
income than their African American and Latino counterparts. It is revealed in
this study that those with more human and social capital earn both more legal
and illegal income, regardless of race and ethnicity. If White respondents
possess more human and social capital than their ethnic and racial minority
counterparts, this may explain their relatively higher earnings from crime.
Although the experiences of African Americans and Latinos in the illegal
economy dominate most of the research in this area, some suggest that the
recent expansion of the drug economy have created new opportunities for

332		

The Prison Journal 92(3)

economically disadvantaged Whites in the illicit labor market (Freeman &
Fagan, 1999). This study supports this hypothesis.
Western (2006) asserts that incarceration is a pathway to the secondary
sector labor market because the ex-incarcerated earn lower hourly wages and
annual income and are at greater risk of unemployment than their never
incarcerated counterparts (for furthermore evidence, also see Freeman,
1992; Freeman, 1996; Kling, 1999; Nagin & Waldfogel, 1998; Pager, 2003;
Sampson & Laub, 2003; Waldfogel, 1994; Western & Beckett, 1999;
Western, Kling, & Weiman, 2001; Western, Lopoo, & McLanahan, 2004).
Some offer that crime as a source of income provides an attractive alternative
to closed opportunities in the legitimate labor market (Cloward & Ohlin,
1960). The current research adds to the growing body of literature on the collateral consequences of incarceration by showing that spending significant
time in jail or prison may force the ex-incarcerated into illegal opportunity
structures to obtain income.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research,
authorship, /or publication of this article.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.

References
Aiken, L., & West. S. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: SAGE.
Allison, P. (1994). Using panel data to estimate the effects of events. Sociological
Methods and Research, 23, 174-199.
Becker, G. S. (1968). Crime and punishment: An economic approach. Journal of
Political Economy, 76, 169-217.
Boshier, R., & Johnson, D. R. (1974). Does conviction affect employment opportunities? British Journal of Criminology, 14, 264-268.
Buikhuisen, W., & Dijksterhuis, F. P. H. (1971). Delinquency and stigmatization.
British Journal of Criminology, 11, 185-187.
Cloward, R., & Ohlin, L. E. (1960). Delinquency and opportunity. Glencoe, IL: Free
Press.
Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, 95-120.

Hutcherson

333

Fagan, J. (1992). Drug Selling and licit income in distressed neighborhoods:
The economic lives of street-level drug users and dealers. In G. E. Peterson
& A. V. Harrell (Eds.), Drugs, crime and social isolation: Barriers to urban
opportunity (pp. 99-141). Washington, DC: Urban Institute.
Fagan, J. (1992). The dynamics of crime and neighborhood change: The ecology
of crime and drug use in inner cities. New York, NY: Social Science Research
Council.
Farrington, D. P. (1986). “Age and Crime.” In M. Tonry & N. Morris, (Eds.), Crime
and justice: An annual review of research, (Vol. 7, pp. 189-250). Chicago, IL:
University of Chicago.
Farrington, D. P. (2003). Developmental and life-course criminology: Key theoretical and empirical issues—The 2002 Sutherland Award address. Criminology, 41,
221-255.
Freeman, R. B. (1992). Crime and the economic status of disadvantaged young men.
In G. E. Peterson & W. Vroman (Eds.), Urban labor markets and job opportunities (pp. 201-237). Washington, DC: Urban Institute.
Freeman, R. B. (1996). The supply of youths to crime. In S. Pozo (Ed.), Exploring the
underground economy (pp. 25-41). Kalamazoo, MI: W. E. Upjohn Institute for
Employment Research.
Freeman, R. B. (1996). Why do so many young American men commit crimes and
what might we do about it? Journal of Economic Perspectives, 10(1), 25-42.
Freeman, R. B., & Fagan, J. (1999). Crime and work. In M. Tonry (Ed.), Crime and
justice: A review of research (Vol. 25, pp. 225-290). Chicago, IL: University of
Chicago Press.
Garland, D. (2001). Mass imprisonment: Social causes and consequences. Thousand
Oaks, CA: SAGE.
Hagan, J. (1993). The social embeddedness of crime and unemployment. Criminology, 31, 465-491.
Hauser, R. M. (1980). On stratification in a dual economy. American Sociological
Review, 45, 702-712.
Holzer, H. J. (1996). What employers want: Job prospects for less-educated workers.
New York, NY: Russell Sage.
Holzer, H. J, Raphael, S. & M. A. Stoll. (2003). Employment barriers facing ex-offenders.
New York, NY: Urban Institute Reentry Roundtable.
Horney, J., Osgood, D. W., & Marshall, I. H. (1995). Criminal careers in the shortterm: Intra-individual variability in crime and its relation to local life circumstances. American Sociological Review, 60, 655-673.
Huebner, B. M. (2005). The effect of incarceration on marriage and work over the life
course. Justice Quarterly, 22, 281-303.

334		

The Prison Journal 92(3)

Hughes, T. A., Wilson, D. J., & Beck, A. J. (2001). Trends in state parole, 1990-2000.
Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics.
Johnson, K. (2003). Prison, race and space: The impact of incarceration on career
trajectories and labor market outcomes (Doctoral dissertation). North Carolina
State University, Raleigh
Johnston, J., & DiNardo, J. (1997). Econometric methods (4th ed.). New York, NY:
McGraw-Hill.
Kling, J. (1999). The effect of prison sentence length on subsequent employment and
earnings of federal criminal defendants (Woodrow Wilson School Economics
Discussion Paper 20). Philadelphia, PA: Woodrow Wilson School.
Krivo, L. J., & Kaufman, R. L. (2004). Housing and wealth inequality: Racial-ethnic
differences in home equity in the United States. Demography, 41, 585-605.
Laub, J., Nagin, D., & Sampson, R. J. (1998). Trajectories of change in criminal
offending: Good marriages and the desistance process. American Sociological
Review, 63, 225-238.
Levitt, S. D., & Venkatesh, S. (2001). The financial activities of urban street gang.
Quarterly Journal of Economics, 115, 755-789.
Levitt, S. D., & Venkatesh, S. (2001). Growing up in the projects: The economic
lives of a cohort of men who came of age in Chicago public housing. American
Economic Review, 91, 79-84.
Long, S., & Freese, J. (2003). Regression models for categorical variables using
Stata. College Station, TX: Stata Press.
McCarthy, B., & Hagan, J. (2001). When crime pays: Capital, competence, and criminal success. Social Forces, 79, 1035-1059.
Nagin, D., & Waldfogel, J. (1998). The effect of conviction on income through the
life cycle. International Review of Law and Economics, 18, 25-40.
Pager, D. (2003). The mark of a criminal record. American Journal of Sociology, 108,
937-975.
Roncek, D. W. (1992). Learning more from tobit coefficients: Extending a comparative analysis of political protest. American Sociological Review, 57, 503-507.
Sabol, W. J., Minton, T. D., & Harrison, P. M. (2007). Prison and jail inmates at
midyear 2006. Washington, DC: U.S. Department of Justice, Bureau of Justice
Statistics.
Sampson, R. J., & Laub, J. H. (1993). Crime in the making: Pathways and turning
points through life. Cambridge, MA: Harvard University Press.
Sampson, R. J., & Laub, J. H. (2003). Life-course desisters? Trajectories of crime
among delinquent boys followed to age 70. Criminology, 41, 555-592.
Short, J. F. Jr. (1997). Poverty, ethnicity and violent crime. Boulder, CO: Westview.

Hutcherson

335

Uggen, C., & Thompson, M. (2003). The socioeconomic determinants of ill-gotten
gains: Within-person changes in drug use and illegal earnings. American Journal
of Sociology, 109(1), 146-185.
Waldfogel, J. (1994). The effect of criminal conviction on income and the trust
reposed in the workmen. Journal of Human Resources, 29, 62-81.
Western, B. (2002). The impact of incarceration on wage mobility and inequality.
American Sociological Review, 67, 526-546.
Western, B. (2006). Punishment and inequality in America. New York, NY: Russell
Sage.
Western, B., & Beckett, K. (1999). How unregulated is the U.S. labor market: The
penal system as a labor market institution. American Journal of Sociology, 104,
1030-1060.
Western, B., Kling, J. R., & Weiman, D. F. (2001). The labor market consequences of
incarceration. Crime and Delinquency, 47, 410-427.
Western, B., Lopoo, L., & McLanahan, S. (2004). Incarceration and the bonds between
parents in fragile families. In M. Patillo, D. Weiman, & B. Western (Eds.), Imprisoning America: The social effects of mass incarceration (pp. 21-45). New York,
NY: Russell Sage.
Wooldridge, J. (2002). Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.

Bio
Donald T. Hutcherson II is an assistant professor of sociology and criminal justice
at Ohio University, Lancaster Campus. His research interests include the effect of
extralegal factors on the criminal justice system and the collateral consequences of
incarceration on individuals.

View publication stats